English Syntax: An Introduction

Jong-Bok Kim and Peter Sells
November 20, 2007
CENTER FOR THE STUDY
OF LANGUAGE
AND INFORMATION
Contents
1 Some Basic Properties of English Syntax 1
1.1 Some Remarks on the Essence of Human Language 1
1.2 How We Discover Rules 4
1.3 Why Do We Study Syntax and What Is It Good for? 7
1.4 Exercises 9
2 From Words to Major Phrase Types 11
2.1 Introduction 11
2.2 Lexical Categories 12
2.2.1 Determining the Lexical Categories 12
2.3 Grammar with Lexical Categories 17
2.4 Phrasal Categories 19
2.5 Phrase Structure Rules 22
2.5.1 NP: Noun Phrase 22
2.5.2 VP: Verb Phrase 23
2.5.3 AP: Adjective Phrase 25
2.5.4 AdvP: Adverb Phrase 25
2.5.5 PP: Preposition Phrase 26
2.6 Grammar with Phrases 26
2.7 Exercises 32
3 Syntactic Forms, Grammatical Functions, and Semantic Roles 35
3.1 Introduction 35
3.2 Grammatical Functions 36
3.2.1 Subjects 36
3.2.2 Direct and Indirect Objects 38
3
3.2.3 Predicative Complements 39
3.2.4 Oblique Complements 40
3.2.5 Modifiers 41
3.3 Form and Function Together 41
3.4 Semantic Roles 43
3.5 Exercises 46
4 Head, Complements, and Modifiers 49
4.1 Projections from Lexical Heads to Phrases 49
4.1.1 Internal vs. External Syntax 49
4.1.2 Notion of Head, Complements, and Modifiers 50
4.2 Differences between Complements and Modifiers 52
4.3 PS Rules, X

-Rules, and Features 55
4.4 Lexicon and Feature Structures 62
4.4.1 Feature Structures and Basic Operations 62
4.4.2 Feature Structures for Linguistic Entities 65
4.4.3 Argument Realization 66
4.4.4 Verb Types and Argument Structure 67
4.5 Exercises 71
5 More on Subjects and Complements 73
5.1 Grammar Rules and Principles 73
5.2 Feature Specifications on the Complement Values 76
5.2.1 Complements of Verbs 76
5.2.2 Complements of Adjectives 80
5.2.3 Complements of Common Nouns 82
5.3 Feature Specifications for the Subject 83
5.4 Clausal Complement or Subject 84
5.4.1 Verbs Selecting a Clausal Complement 84
5.4.2 Verbs Selecting a Clausal Subject 90
5.4.3 Adjectives Selecting a Clausal Complement 91
5.4.4 Nouns Selecting a Clausal Complement 93
5.4.5 Prepositions Selecting a Clausal Complement 94
5.5 Exercises 95
6 Noun Phrases and Agreement 99
6.1 Classification of Nouns 99
6.2 Syntactic Structures 100
4
6.2.1 Common Nouns 100
6.2.2 Pronouns 102
6.2.3 Proper Nouns 102
6.3 Agreement Types and Morpho-syntactic Features 103
6.3.1 Noun-Determiner Agreement 103
6.3.2 Pronoun-Antecedent Agreement 105
6.3.3 Subject-Verb Agreement 105
6.4 Semantic Agreement Features 107
6.5 Partitive NPs and Agreement 111
6.5.1 Basic Properties 111
6.5.2 Two Types of Partitive NPs 113
6.5.3 Measure Noun Phrases 118
6.6 Modifying an NP 120
6.6.1 Adjectives as Prenominal Modifiers 120
6.6.2 Postnominal Modifiers 121
6.7 Exercises 123
7 Raising and Control Constructions 127
7.1 Raising and Control Predicates 127
7.2 Differences between Raising and Control Verbs 128
7.2.1 Subject Raising and Control 128
7.2.2 Object Raising and Control 131
7.3 A Simple Transformational Approach 132
7.4 A Nontransformational Approach 134
7.4.1 Identical Syntactic Structures 134
7.4.2 Differences in Subcategorization Information 136
7.4.3 Mismatch between Meaning and Structure 140
7.5 Explaining the Differences 143
7.5.1 Expletive Subject and Object 143
7.5.2 Meaning Preservation 144
7.5.3 Subject vs. Object Control Verbs 145
7.6 Exercises 147
8 Auxiliary Constructions 151
8.1 Basic Issues 151
8.2 Transformational Analyses 153
8.3 A Lexicalist Analysis 154
8.3.1 Modals 154
5
8.3.2 Be and Have 157
8.3.3 Periphrastic do 159
8.3.4 Infinitival Clause Marker to 162
8.4 Explaining the NICE Properties 162
8.4.1 Auxiliaries with Negation 162
8.4.2 Auxiliaries with Inversion 166
8.4.3 Contracted Auxiliaries 169
8.4.4 Auxiliaries with Ellipsis 171
8.5 Exercises 174
9 Passive Constructions 177
9.1 Introduction 177
9.2 Relationships between Active and Passive 178
9.3 Approaches to Passive 180
9.3.1 From Structural Description to Structural Change 180
9.3.2 A Transformational Approach 181
9.3.3 A Lexicalist Approach 182
9.4 Prepositional Passives 188
9.5 Exercises 192
10 Wh-Questions 195
10.1 Clausal Types and Interrogatives 195
10.2 Movement vs. Feature Percolation 197
10.3 Feature Percolation with No Abstract Elements 199
10.3.1 Basic Systems 199
10.3.2 Non-subject Wh-questions 201
10.3.3 Subject Wh-Questions 206
10.4 Indirect Questions 210
10.4.1 Basic Structure 210
10.4.2 Non-Wh Indirect Questions 215
10.4.3 Infinitival Indirect Questions 216
10.4.4 Adjunct wh-questions 219
10.5 Exercises 222
11 Relative Clause Constructions 225
11.1 Introduction 225
11.2 Non-subject Wh-Relative Clauses 226
11.3 Subject Relative Clauses 231
6
11.4 That-relative clauses 233
11.5 Infinitival and Bare Relative Clauses 235
11.6 Restrictive vs. Nonrestrictive Relative Clauses 238
11.7 Constraints on the GAP 241
11.8 Exercises 245
12 Special Constructions 247
12.1 Introduction 247
12.2 ‘Easy’ Constructions 248
12.2.1 Basic Properties 248
12.2.2 Transformational Analyses 249
12.2.3 A Lexicalist Analysis 250
12.3 Extraposition 254
12.3.1 Basic Properties 254
12.3.2 Transformational Analysis 255
12.3.3 A Lexicalist Analysis 256
12.4 Cleft constructions 260
12.4.1 Basic Properties 260
12.4.2 Distributional Properties of the Three clefts 261
12.4.3 Syntactic Structures of the Three Types of Cleft: Movement Analyses 262
12.4.4 Lexically-Based Analyses 264
12.5 Exercises 272
Index 287
7
Preface
One important aspect of teaching English syntax (to native and nonnative undergraduate stu-
dents alike) involves the balance in the overall approach between facts and theory. We under-
stand that one important goal of teaching English syntax to undergraduate students is to help
students enhance their understanding of the structure of English in a systematic and scientific
way. Basic knowledge of this kind is essential for students to move on the next stages, in which
they will be able to perform linguistic analyses for simple as well as complex English phe-
nomena. This new introductory textbook has been developed with this goal in mind. The book
focuses primarily on the descriptive facts of English syntax, presented in a way that encourages
students to develop keen insights into the English data. It then proceeds with the basic, theoret-
ical concepts of generative grammar from which students can develop abilities to think, reason,
and analyze English sentences from linguistic points of view.
We owe a great deal of intellectual debt to the previous textbooks and literature on English
syntax. In particular, much of the content, as well as our exercises, has been inspired by and
adopted from renowned textbooks such as Aarts (1997), Baker (1997), Borsley (1991, 1996),
Radford (1988, 1997, 2004), Sag et al. (2003), to list just a few. We acknowledge our debt to
these works, which have set the course for teaching syntax over the years.
Within this book, Chapters 1 to 5 cover the fundamental notions of English grammar. We
start with the basic properties of English words, and then rules for combining these words to
form well-formed phrases and, ultimately, clauses. These chapters guide students through the
basic concepts of syntactic analysis such as lexical categories, phrasal types, heads, comple-
ments, and modifiers. In Chapter 4, as a way of formalizing the observed generalizations, the
textbook introduces the feature structure system of Head-Driven Phrase Structure Grammar
(HPSG, Pollard and Sag (1994), Sag et al. (2003)) which places strong emphasis on the role of
lexical properties and the interactions among grammatical components.
From Chapter 6 on, the book discusses major constructions of English within a holistic view
of grammar allowing interactions of various grammatical properties including syntactic forms,
their grammatical functions, their semantic roles, and overall aspects of clausal meaning. In
Chapter 6, we introduce English subject verb agreement, and concentrate on interrelationships
i
among different grammatical components which play crucial interacting roles in English agree-
ment phenomena. In particular, this chapter shows that once we allow morphological informa-
tion to interface with the system of syntax, semantics, or even pragmatics, we can provide good
solutions for some puzzling English agreement phenomena, within a principled theory. Chapter
7 covers raising and control phenomena, and provides insights into the properties of the two
different constructions, which are famously rather similar in terms of syntactic structures, but
different in terms of semantics. Chapter 8 deals with the English auxiliary system, itself remark-
able in that a relatively small number of elements interact with each other in complicated and
intriguing ways. This chapter assigns the precise lexical information to auxiliary verbs and con-
structional constraints sensitive to the presence of an auxiliary verb. This allows us to express
generalizations among auxiliary-sensitive phenomena such as negation, inversion, contraction,
and ellipsis, which we would otherwise be missed.
From Chapter 9 through Chapter 12, the textbook discusses how to capture systematic re-
lations between related constructions. Chapter 9 deals with the relationships between active
and passive voice clauses. Studying this chapter, students will be able to fully understand why,
how, and when to choose between canonical and passive constructions. Chapters 10 and 11 deal
with wh-questions and relative clause constructions, often called non-local or long-distance de-
pendency constructions, in the sense that a gap and its filler are in a potentially long-distance
relationship. These two chapters present the basic properties of these constructions and show
how the mechanism of feature percolation is a crucial part of a systematic account for them.
The final chapter of the book covers the so-called ‘tough’ constructions, extraposition, and cleft
constructions. These constructions are also based on long-distance dependencies, but different
from the constructions in chapters 10 and 11. The goal of all these chapters is the present a
groundwork of facts, which students will then have in hand, in order to consider theoretical
accounts which apply in precise ways.
We have tried to make each chapter maximally accessible. We provide clear, simple tree
diagrams which will help students understand the structures of English and develop analytic
skills to English syntax. The theoretical notions are kept as simple yet precise as possible so
that students can apply and use them in analyzing English sentences. Each chapter also contains
exercises ranging from simple to challenging, aiming to promote deeper understanding of the
factual and theoretical contents of each chapter.
Numerous people have helped us in writing this textbook, in various ways. We thank for
their comments in various places, help and interest in our textbook: [...................................] We
also thank teachers and colleagues in Kyung Hee University and Stanford University for their
constant encouragement over the years. Our gratitude also goes to undergraduate and graduate
students at Kyung Hee University who used the draft of this as the textbook and raised so many
questions that help us reshape its structure as well as contents. We also thank Jinyoung Kim,
Dongjun Lee, and Juwon Lee for their administrative help. We also owe out thanks to Dikran
Karagueuzian, Director of CSLI Publications, for his patience and support, as well as Lauri
Kanerva for his help in matters of production. We also thank Kaunghi Un for helping us with
ii
LaTex problems.
Lastly, but not the least, we also truly thank our close friends and family members who gave
us unconditional love and support in every possible regard. We dedicate this book to our beloved
ones who with true love and refreshing and comforting words have lead us to think ‘wise and
syntactic’ when we are spiritually and physically down.
iii
1
Some Basic Properties of English Syntax
1.1 Some Remarks on the Essence of Human Language
One of the crucial functions of any human language, such as English or Korean, is to convey
various kinds of information from the everyday to the highly academic. Language provides a
means for us to describe how to cook, how to remove cherry stains, how to understand English
grammar, or how to provide a convincing argument. We commonly consider certain properties
of language to be key essential features from which the basic study of linguistics starts.
The first well-known property (as emphasized by Saussure 1916) is that there is no moti-
vated relationship between sounds and meanings. This is simply observed in the fact that
the same meaning is usually expressed by a different sounding-word in a different language
(think of house, maison, casa). For words such as hotdog, desk, dog, bike, hamburger, cran-
berry, sweetbread, their meanings have nothing to do with their shapes. For example, the word
hotdog has no relationship with a dog which is or feels hot. There is just an arbitrary relation-
ship between the word’s sound and its meaning: this relationship is decided by the convention
of the community the speakers belong to.
The second important feature of language, and one more central to syntax, is that language
makes infinite use of finite set of rules or principles, the observation of which led the de-
velopment of generative linguistics in the 20th century (cf. Chomsky 1965). A language is a
system for combining its parts in infinitely many ways. One piece of evidence of the system can
be observed in word-order restrictions. If a sentence is an arrangement of words and we have
5 words such as man, ball, a, the, and kicked, how many possible combinations can we have
from these five words? More importantly, are all of these combinations grammatical sentences?
Mathematically, the number of possible combinations of 5 words is 5! (factorial), equalling
120 instances. But among these 120 possible combinations, only 6 form grammatical English
sentences:
1
(1) a. The man kicked a ball.
1
Examples like (1e) and (1f) are called ‘topicalization’ sentences in which the topic expression (the ball and the
man), already mentioned and understood in the given context, is placed in the sentence initial position. See Lambrecht
(1994) and references therein.
1
b. A man kicked the ball.
c. The ball kicked a man.
d. A ball kicked the man.
e. The ball, a man kicked.
f. The man, a ball kicked.
All the other 114 combinations, a few of which are given in (2), are unacceptable to native
speakers of English. We use the notation * to indicate that a hypothesized example is ungram-
matical.
(2) a. *Kicked the man the ball.
b. *Man the ball kicked the.
c. *The man a ball kicked.
It is clear that there are certain rules in English for combining words. These rules constrain
which words can be combined together or how they may be ordered, sometimes in groups, with
respect to each other.
Such combinatory rules also play important roles in our understanding of the syntax of an
example like (3a).
2
Whatever these rules are, they should give a different status to (3b), an
example which is judged ungrammatical by native speakers even though the intended meaning
of the speaker is relatively clear and understandable.
(3) a. Kim lives in the house Lee sold to her.
b. *Kim lives in the house Lee sold it to her.
The requirement of such combinatory knowledge also provides an argument for the assumption
that we use just a finite set of resources in producing grammatical sentences, and that we do not
just rely on the meaning of words involved. Consider the examples in (4):
(4) a. *Kim fond of Lee.
b. Kim is fond of Lee.
Even though it is not difficult to understand the meaning of (4a), English has a structural re-
quirement for the verb is as in (4b).
More natural evidence of the ‘finite set of rules and principles’ idea can be found in cognitive,
creative abilities. Speakers are unconscious of the rules which they use all the time, and have no
difficulties in producing or understanding sentences which they have never heard, seen, or talked
about before. For example, even though we may well not have seen the following sentence
before, we can understand its meaning if we have a linguistic competence in English:
(5) In January 2002, a dull star in an obscure constellation suddenly became 600,000
times more luminous than our Sun, temporarily making it the brightest star in our
galaxy.
2
Starting in Chapter 2, we will see these combinatory rules.
2
A related part of this competence is that a language speaker can produce an infinite number
of grammatical sentences. For example, given the simple sentence (6a), we can make a more
complex one like (6b) by adding the adjective tall. To this sentence, we can again add another
adjective handsome as in (6c). We could continue adding adjectives, theoretically enabling us
to generate an infinitive number of sentences:
(6) a. The man kicked the ball.
b. The tall man kicked the ball.
c. The handsome, tall man kicked the ball.
d. The handsome, tall, nice man kicked the ball.
e. . . .
One might argue that since the number of English adjectives could be limited, there would be a
dead-end to this process. However, no one would find themselves lost for another way to keep
the process going (cf. Sag et al. 2003):
(7) a. Some sentences can go on.
b. Some sentences can go on and on.
c. Some sentences can go on and on and on.
d. Some sentences can go on and on and on and on.
e. . . .
To (7a), we add the string and on, producing a longer one (7b). To this resulting sentence (7c),
we once again add and on. We could in principle go on adding without stopping: this is enough
to prove that we could make an infinite number of well-formed English sentences.
3
Given these observations, how then can we explain the fact that we can produce or under-
stand an infinite number of grammatical sentences that we have never heard or seen before? It
seems implausible to consider that we somehow memorize every example, and in fact we do not
(Pullum and Scholz 2002). We know that this could not be true, in particular when we consider
that native speakers can generate an infinite number of infinitely long sentences, in principle. In
addition, there is limit to the amount of information our brain can keep track of, and it would
be implausible to think that we store an infinite number of sentences and retrieve whenever we
need to do so.
These considerations imply that a more appropriate hypothesis would be something like (8):
4
(8) All native speakers have a grammatical competence which can generate an infinite
set of grammatical sentences from a finite set of resources.
3
Think of a simple analogy: what is the longest number? Yet, how many numbers do you know? The second question
only makes sense if the answer is 0–9 (ten digits).
4
The notion of ‘competence’ is often compared with that of ‘performance’ (Chomsky 1965). Competence refers
to speakers’ internalized knowledge of their language, whereas performance refers to actual usage of this abstract
knowledge of language.
3
This hypothesis has been generally accepted by most linguists, and has been taken as the subject
matter of syntactic theory. In terms of grammar, this grammatical competence is hypothesized
to characterize a generative grammar, which we then can define as follows (for English, in
this instance):
(9) Generative Grammar:
An English generative grammar is the one that can generate an infinite set of well-
formed English sentences from a finite set of rules or principles.
The job of syntax is thus to discover and formulate these rules or principles.
5
These rules tell us
how words are put together to form grammatical phrases and sentences. Generative grammar,
or generative syntax, thus aims to define these rules which will characterize all of the sentences
which native speakers will accept as well-formed and grammatical.
1.2 How We Discover Rules
How can we then find out what the generative rules of English syntax are? These rules are
present in the speakers’ minds, but are not consciously accessible; speakers cannot articulate
their content, if asked to do so. Hence we discover the rules indirectly, and of the several meth-
ods for inferring these hidden rules, hypotheses based on the observed data of the given lan-
guage are perhaps the most reliable. These data can come from speakers’ judgments – known
as intuitions – or from collected data sets – often called corpora. Linguistics is in one sense
an empirical science as it places a strong emphasis on investigating the data underlying a phe-
nomenon of study.
The canonical steps for doing empirical research can be summarized as follows:
.
Step I: Data collection and observation.
.
Step II: Make a hypothesis to cover the first set of data.
.
Step III: Check the hypothesis with more data.
.
Step IV: Revise the hypothesis, if necessary.
Let us see how these steps work for discovering one of the grammar rules in English, in partic-
ular, the rule for distinguishing count and non-count nouns:
6
[Step I: Observing Data] To discover a grammar rule, the first thing we need to do is to
check out grammatical and ungrammatical variants of the expression in question. For example,
5
In generative syntax, ‘rules’ refers not to ‘prescriptive rules’ but to ‘descriptive rules’. Prescriptive rules are those
which disfavor or even discredit certain usages; these prescribe forms which are generally in use, as in (i). Meanwhile,
descriptive rules are meant to characterize whatever forms speakers actually use, with any social, moral, or intellectual
judgement.
(i) a. Do not end a sentence with a preposition.
b. Avoid double negatives.
c. Avoid split infinitives.
The spoken performance of most English speakers will often contain examples which violate such prescriptive rules.
6
Much of the discussion and data in this section are adopted from Baker, C.L. (1995).
4
let us look at the usage of the word evidence:
(10) Data Set 1: evidence
a. *The professor found some strong evidences of water on Mars.
b. *The professor was hoping for a strong evidence.
c. *The evidence that John found was more helpful than the one that Smith found.
What can you tell from these examples? We can make the following observations:
(11) Observation 1:
a. evidence cannot be used in the plural.
b. evidence cannot be used with the indefinite article a(n).
c. evidence cannot be referred to by the pronoun one.
In any scientific research one example is not enough to draw any conclusion. However, we
can easily find more words that behave like evidence:
(12) Data Set 2: equipment
a. *We had hoped to get three new equipments every month, but we only had enough
money to get an equipment every two weeks.
b. *The equipment we bought last year was more expensive than the one we bought
this year.
We thus extend Observation 1 a little bit further:
(13) Observation 2:
a. evidence/equipment cannot be used in the plural.
b. evidence/equipment cannot be used with the indefinite article a(n).
c. evidence/equipment cannot be referred to by the pronoun one.
It is usually necessary to find contrastive examples to understand the range of a given observa-
tion. For instance, words like clue and tool act differently:
(14) Data Set 3: clue
a. The professor gave John some good clues for the question.
b. The student was hoping for a good clue.
c. The clue that John got was more helpful than the one that Smith got.
(15) Data Set 4: tool
a. The teacher gave John some good tools for the purpose.
b. The student was hoping for a tool.
c. The tool that Jones got was more helpful than the one that Smith got.
Unlike equipment and evidence, the nouns clue and tool can be used in the test linguistic con-
texts we set up. We thus can add Observation 3, different from Observation 2:
(16) Observation 3:
5
a. clue/tool can be used in the plural.
b. clue/tool can be used with the indefinite article a(n).
c. clue/tool can be referred to by the pronoun one.
[Step II: Forming a Hypothesis] From the data and observations we have made so far, can
we make any hypothesis about the English grammar rule in question? One hypothesis that we
can make is something like the following:
(17) First Hypothesis:
English has at least two groups of nouns, Group I (count nouns) and Group II
(non-count nouns), diagnosed by tests of plurality, the indefinite article, and the
pronoun one.
[Step III: Checking the Hypothesis] Once we have formed such a hypothesis, we need to
check out if it is true of other data, and also see if it can bring other analytical consequences.
A little further thought allows us to find support for the two-way distinction for nouns. For
example, consider the usage of much and many:
(18) a. much evidence, much equipment, information, much furniture, much advice
b. *much clue, *much tool, *much armchair, *much bags
(19) a. *many evidence, *many equipment, *many information, *many furniture, *many
advice
b. many clues, many tools, many suggestions, many armchairs
As observed here, count nouns can occur only with many, whereas non-count nouns can com-
bine with much. Similar support can be found from the usage of little and few:
(20) a. little evidence, little equipment, little advice, little information
b. *little clue, *little tool, *little suggestion, *little armchair
(21) a. *few evidence, *few equipment, *few furniture, *few advice, *few information
b. few clues, few tools, suggestions, few armchairs
The word little can occur with non-count nouns like evidence, yet few cannot. Meanwhile, few
occurs only with count nouns.
Given these data, it appears that the two-way distinction is quite plausible and persuasive.
We can now ask if this distinction into just two groups is really enough for the classification of
nouns. Consider the following examples with cake:
(22) a. The mayor gave John some good cakes.
b. The president was hoping for a good cake.
c. The cake that Jones got was more delicious than the one that Smith got.
Similar behavior can be observed with a noun like beer, too:
6
(23) a. The bartender gave John some good beers.
b. No one knows how to tell from a good beer to a bad one.
These data show us that cake and beer may be classified as count nouns. However, observe
the following:
(24) a. My pastor says I ate too much cake.
b. The students drank too much beer last night.
(25) a. We recommend to eat less cake and pastry.
b. People now drink less beer.
The data mean that cake and beer can also be used as non-count nouns since that can be used
with less or much.
[Step IV: Revising the Hypothesis] The examples in (24) and (25) imply that there is an-
other group of nouns that can be used as both count and non-count nouns. This leads us to revise
the hypothesis in (17) as following:
(26) Revised Hypothesis:
There are at least three groups of nouns: Group 1 (count nouns), Group 2 (non-count
nouns), and Group 3 (count and non-count).
We can expect that context will determine whether a Group 3 noun is used as count or as non-
count.
As we have observed so far, the process of finding finite grammar rules crucially hinges on
finding data, drawing generalizations, making a hypothesis, and revising this hypothesis with
more data.
1.3 Why Do We Study Syntax and What Is It Good for?
There are many reasons for studying syntax, from general humanistic or behavioral motivations
to much more specific goals such as those in the following:
.
To help us to illustrate the patterns of English more effectively and clearly.
.
To enable us to analyze the structure of English sentences in a systematic and explicit way.
For example, let us consider how we could use the syntactic notion of head, which refers
to the essential element within a phrase. The following is a short and informal rule for English
subject-verb agreement.
7
(27) In English, the main verb agrees with the head element of the subject.
This informal rule can pinpoint what is wrong with the following two examples:
(28) a. *The recent strike by pilots have cost the country a great deal of money from
tourism and so on.
7
The notion of ‘subject’ is further discussed in Chapter 3 and that of ‘head’ in Chapter 4.
7
b. *The average age at which people begin to need eyeglasses vary considerably.
Once we have structural knowledge of such sentences, it is easy to see that the essential element
of the subject in (28a) is not pilots but strike. This is why the main verb should be has but not
have to observe the basic agreement rule in (27). Meanwhile, in (28b), the head is the noun age,
and thus the main verb vary needs to agree with this singular noun. It would not do to simply
talk about ‘the noun’ in the subject in the examples in (28), as there is more than one. We need
to be able to talk about the one which gives its character to the phrase, and this is the head. If
the head is singular, so is the whole phrase, and similarly for plural. The head of the subject and
the verb (in the incorrect form) are indicated in (29):
(29) a. *[The recent strike by pilots] have cost the country a great deal of money from
tourism and so on.
b. *[The average age at which people begin to need eyeglasses] vary considerably.
Either example can be made into a grammatical version by pluralizing the head noun of the
subject.
Nowlet us look at some slightly different cases. Can you explain why the following examples
are unacceptable?
(30) a. *Despite of his limited educational opportunities, Abraham Lincoln became one of
the greatest intellectuals in the world.
b. *A pastor was executed, notwithstanding on many applications in favor of him.
To understand these examples, we first need to recognize that the words despite and notwith-
standing are prepositions, and further that canonical English prepositions combine only with
noun phrases. In (30), these prepositions combine with prepositional phrases again (headed by
of and on respectively), violating this rule.
A more subtle instance can be found in the following:
(31) a. Visiting relatives can be boring.
b. I saw that gas can explode.
These examples each have more than one interpretation. The first one can mean either that the
event of seeing our relatives is a boring activity, or that the relatives visiting us are themselves
boring. The second example can either mean that a specific can containing gas exploded, which
I saw, or it can mean that I observed that gas has a possibility of exploding. If one knows English
syntax, that is, if one understands the syntactic structure of these English sentences, it is easy to
identify these different meanings.
Here is another example which requires certain syntactic knowledge:
(32) He said that that ‘that’ that that man used was wrong.
This is the kind of sentence one can play with when starting to learn English grammar. Can you
analyze it? What are the differences among these five thats? Structural (or syntactic) knowledge
8
can be used to diagnose the differences. Part of our study of syntax involves making clear exactly
how each word is categorized, and how it contributes to a whole sentence.
When it comes to understanding a rather complex sentence, knowledge of English syntax
can be a great help. Syntactic or structural knowledge helps us to understand simple as well as
complex English sentences in a systematic way. There is no difference in principle between the
kinds of examples we have presented above and (33):
(33) The government’s plan, which was elaborated in a document released by the Trea-
sury yesterday, is the formal outcome of the Government commitment at the Madrid
summit last year to put forward its ideas about integration.
Apart from having more words than the examples we have introduced above, nothing in this
example is particularly complex.
1.4 Exercises
1. For each of the following nouns, decide if it can be used as a count or as a non-count
(mass) noun. In doing so, construct acceptable and unacceptable examples using the tests
(plurality, indefinite article, pronoun one, few/little, many/much tests) we have discussed
in this chapter.
(i) activity, art, cheese, discussion, baggage, luggage, suitcase, religion, sculpture,
paper, difficulty, cheese, water, experience, progress, research, life
2. Check or find out whether each of the following examples is grammatical or ungrammat-
ical. For each ungrammatical one, provide at least one (informal) reason for its ungram-
maticality, according to your intuitions or ideas.
(i) a. Kim and Sandy is looking for a new bicycle.
b. I have never put the book.
c. The boat floated down the river sank.
d. Chris must liking syntax.
e. There is eager to be fifty students in this class.
f. What is John eager to do?
g. What is John easy to do?
h. Is the boy who holding the plate can see the girl?
i. Which chemical did you mix the hydrogen peroxide and?
j. There seem to be a good feeling developing among the students.
k. Strings have been pulled many times to get students into that university.
3. Consider the following set of data, focusing on the usage of ‘self’ reflexive pronouns and
personal pronouns:
(i) a. He washed himself.
9
b. *He washed herself.
c. *He washed myself.
d. *He washed ourselves.
(ii) a. *He washed him. (‘he’ and ‘him’ referring to the same person)
b. He washed me.
c. He washed her.
d. He washed us.
Can you make a generalization about the usage of ‘self’ pronouns and personal pronouns
like he here? In answering this question, pay attention to what the pronouns can refer to.
Also consider the following imperative examples:
(iii) a. Wash yourself.
b. Wash yourselves.
c. *Wash myself.
d. *Wash himself.
(iv) a. *Wash you!
b. Wash me!
c. Wash him!
Can you explain why we can use yourself and yourselves but not you as the object of
the imperatives here? In answering this, try to put pronouns in the unrealized subject
position.
4. Read the following passage and identify all the grammatical errors. If you can, discuss
the relevant grammar rules that you can think of.
(i) Grammar is important because it is the language that make it possible for
us to talk about language. Grammar naming the types of words and word
groups that make up sentences not only in English but in any language. As
human beings, we can putting sentences together even as children–we can
all do grammar. People associate grammar for errors and correctness. But
knowing about grammar also helps us understood what makes sentences and
paragraphs clearly and interesting and precise. Grammar can be part of lit-
erature discussions, when we and our students closely reading the sentences
in poetry and stories. And knowing about grammar means finding out that
all language and all dialect follow grammatical patterns.
8
8
Adapted from “Why is Grammar Important?” by The Assembly for the Teaching of English Grammar
10
2
From Words to Major Phrase Types
2.1 Introduction
In Chapter 1, we observed that the study of English syntax is the study of rules which generate
an infinite number of grammatical sentences. These rules can be inferred from observations
about the English data. One simple mechanism we recognize is that in forming grammatical
sentences, we start from words, or ‘lexical’ categories. These lexical categories then form a
larger constituent ‘phrase’; and phrases go together to form a ‘clause’. A clause either is, or is
part of, a well-formed sentence:
(1)
sentence

¸
¸
¸
¸
¸
¸
¸
. . . clause

¸
¸
¸
¸
¸
¸
¸
. . .
. . . phrase

¸
¸
¸
¸
¸
¸
¸
. . .
. . . word . . .
Typically we use the term ‘clause’ to refer to a complete sentence-like unit, but which may be
part of another clause, as a subordinate or adverbial clause. Each of the sentences in (2b)–(2d)
contains more than one clause, in particular, with one clause embedded inside another:
(2) a. The weather is lovely today.
b. I am hoping that [the weather is lovely today].
c. If [the weather is lovely today] then we will go out.
d. The birds are singing because [the weather is lovely today].
This chapter deals with what kind of combinatorial rules English employs in forming these
phrases, clauses, and sentences.
11
2.2 Lexical Categories
2.2.1 Determining the Lexical Categories
The basic units of syntax are words. The first question is then what kinds of words (also known
as parts of speech, or lexical categories, or grammatical categories) does English have? Are
they simply noun, verb, adjective, adverb, preposition, and maybe a few others? Most of us
would not be able to come up with simple definitions to explain the categorization of words.
For instance, why do we categorize book as a noun, but kick as a verb? To make it more difficult,
how do we know that virtue is a noun, that without is a preposition, and that well is an adverb
(in one meaning)?
Words can be classified into different lexical categories according to three criteria: meaning,
morphological form, and syntactic function. Let us check what each of these criteria means,
and how reliable each one is.
At first glance, it seems that words can be classified depending on their meaning. For exam-
ple, we could have the following rough semantic criteria for N (noun), V (verb), A (adjective),
and Adv (adverb):
(3) a. N: referring to an individual or entity
b. V: referring to an action
c. A: referring to a property
d. Adv: referring to the manner, location, time or frequency of an action
Though such semantic bases can be used for many words, these notional definitions leave a
great number of words unaccounted for. For example, words like sincerity, happiness, and pain
do not simply denote any individual or entity. Absence and loss are even harder cases.
There are also many words whose semantic properties do not match the lexical category that
they belong to. For example, words like assassination and construction may refer to an action
rather than an individual, but they are always nouns. Words like remain, bother, appear, and
exist are verbs, but do not involve any action.
A more reliable approach is to characterize words in terms of their forms and functions. The
‘form-based’ criteria look at the morphological form of the word in question:
(4) a. N: + plural morpheme -(e)s
b. N: + possessive ’s
c. V: + past tense -ed or 3rd singular -(e)s
d. V: + 3rd singular -(e)s
e. A: + -er/est (or more/most)
f. A: + -ly (to create an adverb)
According to these frames, where the word in question goes in the place indicated by , nouns
allow the plural marking suffix -(e)s to be attached, or the possessive ’s, whereas verbs can have
the past tense -ed or the 3rd singular form -(e)s. Adjectives can take comparative and superlative
12
endings -er or -est, or combine with the suffix -ly. (5) shows some examples derived from these
frames:
(5) a. N: trains, actors, rooms, man’s, sister’s, etc.
b. V: devoured, laughed, devours, laughs, etc.
c. A: fuller, fullest, more careful, most careful, etc.
d. Adv: fully, carefully, diligently, clearly, etc.
The morphological properties of each lexical category cannot be overridden; verbs cannot have
plural marking, nor can adjectives have tense marking. It turns out, however, that these morpho-
logical criteria are also only of limited value. In addition to nouns like information and furniture
that we presented in Chapter 1, there are many nouns such as love and pain that do not have
a plural form. There are adjectives (such as absent and circular) that do not have comparative
-er or superlative -est forms, due to their meanings. The morphological (form-based) criterion,
though reliable in many cases, is not a necessary and sufficient condition for determining the
type of lexical categories.
The most reliable criterion in judging the lexical category of a word is based on its syntactic
function or distributional possibilities. Let us try to determine what kind of lexical categories
can occur in the following environments:
(6) a. They have no .
b. They can .
c. They read the book.
d. He treats John very .
e. He walked right the wall.
The categories that can go in the blanks are N, V, A, Adv, and P (preposition). As can be seen
in the data in (7), roughly only one lexical category can appear in each position:
(7) a. They have no TV/car/information/friend.
b. They have no *went/*in/*old/*very/*and.
(8) a. They can sing/run/smile/stay/cry.
b. They can *happy/*down/*door/*very.
(9) a. They read the big/new/interesting/scientific book.
b. They read the *sing/*under/*very book.
(10) a. He treats John very nicely/badly/kindly.
b. He treats John very *kind/*shame/*under.
(11) a. He walked right into/on the wall.
b. He walked right *very/*happy/*the wall.
As shown here, only a restricted set of lexical categories can occur in each position; we can then
assign a specific lexical category to these elements:
13
(12) a. N: TV, car, information, friend, . . .
b. V: sing, run, smile, stay, cry, . . .
c. A: big, new, interesting, scientific, . . .
d. Adv: nicely, badly, kindly, . . .
e. P: in, into, on, under, over, . . .
In addition to these basic lexical categories, does English have other lexical categories? There
are a few more. Consider the following syntactic environments:
(13) a. student hits the ball.
b. John sang a song, Mary played the piano.
c. John thinks Bill is honest.
The only words that can occur in the open slot in (13a) are words like the, a, this, that, and so
forth, which are determiner (Det). (13b) provides a frame for conjunctions (Conj) such as and,
but, so, for, or, yet.
1
In (13c), we can have the category we call ‘complementizer’, here the word
that – we return to these in (17) below.
Can we find any supporting evidence for such lexical categorizations? It is not so difficult to
construct environments in which only these lexical elements appear. Consider the following:
(14) We found out that very lucrative jobs were in jeopardy.
Here we see that only words like the, my, his, some, these, those, and so forth can occur here.
These articles, possessives, quantifiers, and demonstratives all ‘determine’ the referential prop-
erties of jobs here, and for this reason, they are called determiners. One clear piece of evidence
for grouping these elements as the same category comes from the fact that they cannot occupy
the same position at the same time:
(15) a. *[My these jobs] are in jeopardy.
b. *[Some my jobs] are in jeopardy.
c. *[The his jobs] are in jeopardy.
Words like my and these or some and my cannot occur together, indicating that they compete
with each other for just one structural position.
Now consider the following examples:
(16) a. I think learning English is not easy at all.
b. I doubt you can help me in understanding this.
c. I am anxious you to study English grammar hard.
Once again, the possible words that can occur in the specific slot in (17) are strictly limited.
(17) a. I think that [learning English is not all that easy].
1
These conjunctions are ‘coordinating conjunctions’ different from ‘subordinating conjunctions’ like when, if, since,
though, and so forth. The former conjoins two identical phrasal elements whereas the latter introduces a subordinating
clause as in [Though students wanted to study English syntax], the department decided not to open that course this year.
14
b. I doubt if [you can help me in understanding this].
c. I am anxious for [you to study English grammar hard].
The italicized words here are different from the other lexical categories that we have seen so
far. They introduce a complement clause, marked above by the square brackets, and may be
sensitive to the tense of that clause. A tensed clause is known as a ‘finite’ clause, as opposed to
an infinitive. For example, that and if introduce or combine with a tensed sentence (present or
past tense), whereas for requires an infinitival clause marked with to. We cannot disturb these
relationships:
(18) a. *I think that [learning English to be not all that easy].
b. *I doubt if [you to help me in understanding this].
c. *I am anxious for [you should study English grammar hard].
The term ‘complement’ refers to an obligatory dependent clause or phrase relative to a head.
2
The italicized elements in (18) introduce a clausal complement and are consequently known as
‘complementizers’ (abbreviated as ‘C’). There are only a few complementizers in English (that,
for, if , and whether), but nevertheless they have their own lexical category.
Now consider the following environments:
(19) a. John not leave.
b. John drink beer last night.
c. John leave for Seoul tomorrow?
d. John will study syntax, and Mary , too.
The words that can appear in the blanks are neither main verbs nor adjectives, but rather words
like will, can, shall and must. In English, there is clear evidence that these verbs are different
from main verbs, and we call them auxiliary verbs (Aux). The auxiliary verb appears in front
of the main verb, which is typically in its citation form, which we call the ‘base’ form. Note the
change in the main verb form in (20b) when the negation is added:
(20) a. He left.
b. He did not leave.
There is also one type of to which is auxiliary-like. Consider the examples in (21) and (22):
(21) a. Students wanted to write a letter.
b. Students intended to surprise the teacher.
(22) a. Students objected to the teacher.
b. Students sent letters to the teacher.
It is easy to see that in (22), to is a preposition. But how about the infinitival marker to in (21),
followed by a base verb form? What lexical category does it belong to? Though the detailed
properties of auxiliary verbs will not be discussed until Chapter 8, we treat the infinitival marker
2
See Chapter 4 for a fuller discussion of ‘head’ and ‘complement’.
15
to as an auxiliary verb. For example, we can observe that to behaves like an auxiliary verb
should:
(23) a. It is crucial for John to show an interest.
b. It is crucial that John should show an interest.
(24) a. I know I should [go to the dentist’s], but I just don’t want to.
b. I don’t really want to [go to the dentist’s], but I know I should.
In (23), to and should introduce the clause and determines the tenseness of the clause. In (24),
they both can license the ellipsis of its VP complement.
3
Another property to shares with other auxiliary verbs like will is that it requires a base verb
to follow. Most auxiliary verbs are actually finite (tensed) forms which therefore pattern with
that in a finite clause, while the infinitival clause introduced by for is only compatible with to:
(25) a. She thought it was likely [that everyone *to/might/would fit into the car].
b. She thought it was easy [for everyone to/*might/*would fit into the car].
Finally, there is one remaining category we need to consider, the ‘particles’ (Part), illustrated
in (26):
(26) a. The umpire called off the game.
b. The two boys looked up the word.
Words like off and up here behave differently from prepositions, in that they can occur after the
object.
(27) a. The umpire called the game off .
b. The two boys looked the word up.
Such distributional possibilities cannot be observed with true prepositions:
(28) a. The umpire fell off the deck.
b. The two boys looked up the high stairs (from the floor).
(29) a. *The umpire fell the deck off .
b. *The students looked the high stairs up (from the floor).
We can also find differences between particles and prepositions in combination with an object
pronoun:
(30) a. The umpire called it off . (particle)
b. *The umpire called off it.
(31) a. *The umpire fell it off .
b. The umpire fell off it. (preposition)
3
See Chapter 8 for detailed discussion on the ellipsis.
16
The pronoun it can naturally follow the preposition as in (31b), but not the particle in (30b).
Such contrasts between prepositions and particles give us ample reason to introduce another
lexical category Part (particle) which is differentiated from P (preposition). In the next section,
we will see more tests to differentiate these two types of word.
2.3 Grammar with Lexical Categories
As noted in Chapter 1, the main goal of syntax is building a grammar that can generate an
infinite set of well-formed, grammatical English sentences. Let us see what kind of grammar
we can develop now that we have lexical categories. To start off, we will use the examples in
(32):
(32) a. A man kicked the ball.
b. A tall boy threw the ball.
c. The cat chased the long string.
d. The happy student played the piano.
Given only the lexical categories that we have identified so far, we can set up a grammar rule
for sentence (S) like the following:
(33) S →Det (A) N V Det (A) N
The rule tells us what S can consist of: it must contain the items mentioned, except that those
which are in parentheses are optional. So this rule characterizes any sentence which consists of
a Det, N, V, Det, and N, in that order, possibly with an A in front of either N. We can represent
the core items in a tree structure as in (34):
(34)
S
.
.
.
.
.
.
.
.
.
.
.
.
.

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
Det N V Det N
. . . . . . . . . . . . . . .
We assume a lexicon, a list of categorized words, to be part of the grammar along with the rule
in (33):
(35) a. Det: a, that, the, this, . . .
b. N: ball, man, piano, string, student, . . .
c. V: kicked, hit, played, sang, threw, . . .
d. A: handsome, happy, kind, long, tall, . . .
By inserting lexical items into the appropriate pre-terminal nodes in the structure, where the
labels above . . . are, we can generate grammatical examples like those (32) as well as those like
the following, not all of which describe a possible real-world situation:
(36) a. That ball hit a student.
17
b. The piano played a song.
c. The piano kicked a student.
d. That ball sang a student.
Such examples are all syntactically well-formed, even if semantically in some cases, implying
that syntax is rather ‘autonomous’ from semantics. Note that any anomalous example can be
preceded by the statement “Now, here’s something hard to imagine: . . . ”.
4
Notice that even this simple grammar rule can easily extend to generate an infinite number
of English sentences by allowing iteration of the A:
5
(37) S →Det A

N V Det A

N
The operator allows us to repeat any number of As, thereby generating sentences like (38). Note
that the parentheses around ‘A’ in (34) are no longer necessary in this instance, for the Kleene
Star operator means any number including zero.
(38) a. The tall man kicked the ball.
b. The tall, handsome man kicked the ball.
c. The tall, kind, handsome man kicked the ball.
One could even generate a sentence like (39):
(39) The happy, happy, happy, happy, happy, happy man sang a song.
A grammar using only lexical categories can be specified to generate an infinite number of
well-formed English sentences, but it nevertheless misses a great deal of basic properties that
we can observe. For example, this simple grammar cannot capture the agreement facts seen in
examples like the following:
(40) a. The mother of the boy and the girl is arriving soon.
b. The mother of the boy and the girl are arriving soon.
Why do the verbs in these two sentences have different agreement patterns? Our intuitions tell
us that the answer lies in two different possibilities for grouping the words:
(41) a. [The mother of [the boy and the girl]] is arriving soon.
b. [The mother of the boy] and [the girl] are arriving soon.
The different groupings shown by the brackets indicate who is arriving: in (41a), the mother,
while in (41b) it is both the mother and the girl. The grouping of words into larger phrasal units
which we call constituents provides the first step in understanding the agreement facts in (41).
Now, consider the following examples:
(42) a. John saw the man with a telescope.
b. I like chocolate cakes and pies.
4
See Exercise 9 of this chapter and the discussion of ‘selectional restrictions’ in Chapter 4.
5
This iteration operator

is called the ‘Kleene Star Operator’, and is a notation meaning ‘zero to infinitely many’
occurrences. It should not be confused with the * prefixed to a linguistic example, indicating ungrammaticality.
18
c. We need more intelligent leaders.
These sentences have different meanings depending on how we group the words. For example,
(42a) will have the following two different constituent structures:
(43) a. John saw [the man with a telescope].
(the man had the telescope)
b. John [[saw the man] with a telescope].
(John used the telescope)
Even these very cursory observations indicate that a grammar with only lexical categories is
not adequate for describing syntax. In addition, we need a notion of ‘constituent’, and need to
consider how phrases may be formed, grouping certain words together.
2.4 Phrasal Categories
In addition to the agreement and ambiguity facts, our intuitions may also lead us to hypothesize
constituency. If you were asked to group the words in (44) into phrases, what constituents would
you come up with?
(44) The student enjoyed his English syntax class last semester.
Perhaps most of us would intuitively assign the structure given in (45a), but not those in (45b)
or (45c):
(45) a. [The student] [enjoyed [his English syntax class last semester]].
b. [The] [student enjoyed] [his English syntax class] [last semester].
c. [The student] [[enjoyed his English] [syntax class last semester]].
What kind of knowledge, in addition to semantic coherence, forms the basis for our intuitions of
constituency? Are there clear syntactic or distributional tests which demonstrate the appropriate
grouping of words or specific constituencies? There are certain salient syntactic phenomena
which refer directly to constituents or phrases.
Cleft: The cleft construction, which places an emphasized or focused element in the X posi-
tion in the pattern ‘It is/was X that . . . ’, can provide us with simple evidence for the existence
of phrasal units. For instance, think about how many different cleft sentences we can form from
(46).
(46) The policeman met several young students in the park last night.
With no difficulty, we can cleft almost all the constituents we can get from the above sentence:
(47) a. It was [the policeman] that met several young students in the park last night.
b. It was [several young students] that the policeman met in the park last night.
c. It was [in the park] that the policeman met several young students last night.
d. It was [last night] that the policeman met several young students in the park.
19
However, we cannot cleft sequences that not form constituents:
6
(48) a. *It was [the policeman met] that several young students in the park last night.
b. *It was [several young students in] that the policeman met the park last night.
c. *It was [in the park last night] that the policeman met several young students.
Constituent Questions and Stand-Alone Test: Further support for the existence of phrasal
categories can be found in the answers to ‘constituent questions’, which involve a wh-word such
as who, where, when, how. For any given wh-question, the answer can either be a full sentence
or a fragment. This stand-alone fragment is a constituent:
(49) A: Where did the policeman meet several young students?
B: In the park.
(50) A: Who(m) did the policeman meet in the park?
B: Several young students.
This kind of test can be of use in determining constituents; we will illustrate with example (51):
(51) John put old books in the box.
Are either old books in the box or put old books in the box a constituent? Are there smaller
constituents? The wh-question tests can provide some answers:
(52) A: What did you put in your box?
B: Old books.
B: *Old books in the box.
(53) A: Where did you put the book?
B: In the box.
B: *Old books in the box.
(54) A: What did you do?
B: *Put old books.
B: *Put in the box.
B: Put old books in the box.
Overall, the tests here will show that old books and in the box are constituents, and that put old
books in the box is also a (larger) constituent.
The test is also sensitive to the difference between particles and prepositions. Consider the
similar-looking examples in (55), including looked and up:
(55) a. John looked up the inside of the chimney.
b. John looked up the meaning of ‘chanson’.
6
The verb phrase constituent met . . . night here cannot be clefted for independent reasons (see Chapter 12).
20
The examples differ, however, as to whether up forms a constituent with the following material
or not. We can again apply the wh-question test:
(56) A: What did he look up?
B: The inside of the chimney.
B: The meaning of ‘chanson’.
(57) A: Where did he look?
B: Up the inside of the chimney.
B: *Up the meaning of ‘chanson’.
(58) A: Up what did he look?
B: The inside of the chimney.
B: *The meaning of ‘chanson’.
What the contrasts here show is that up forms a constituent with the inside of the chimney in
(55a) whereas it does not with the meaning of ‘chanson’ in (55b).
Substitution by a Pronoun: English, like most languages, has a system for referring back to
individuals or entities mentioned by the use of pronouns. For instance, the man who is standing
by the door in (59a) can be ‘substituted’ by the pronoun he in (59b).
(59) a. What do you think the man who is standing by the door is doing now?
b. What do you think he is doing now?
There are other pronouns such as there, so, as, and which, which also refer back to other con-
stituents.
(60) a. Have you been [to Seoul]? I have never been there.
b. John might [go home], so might Bill.
c. John might [pass the exam], and as might Bill.
d. If John can [speak French fluently] – which we all know he can – we will have no
problems.
A pronoun cannot be used to refer back to something that is not a constituent:
(61) a. John asked me to put the clothes in the cupboard, and to annoy him I really stuffed
them there [there=in the cupboard].
b. John asked me to put the clothes in the cupboard, and to annoy him I stuffed them
there [them=the clothes].
c. *John asked me to put the clothes in the cupboard, but I did so [=put the clothes] in
the suitcase.
Both the pronoun there and them refer to a constituent. However, so in (61c), referring to a VP,
refers only part of a constituent put the clothes, making it unacceptable.
21
Coordination: Another commonly-used test is coordination. Words and phrases can be co-
ordinated by conjunctions, and each conjunct is typically the same kind of constituent as the
other conjuncts:
(62) a. The girls [played in the water] and [swam under the bridge].
b. The children were neither [in their rooms] nor [on the porch].
c. She was [poor] but [quite happy].
d. Many people drink [beer] or [wine].
If we try to coordinate unlike constituents, the results are typically ungrammatical.
(63) a. *Mary waited [for the bus] and [to go home].
b. *Lee went [to the store] and [crazy].
Even though such syntactic constituent tests are limited in certain cases, they are often
adopted in determining the constituent of given expressions.
2.5 Phrase Structure Rules
We have seen evidence for the existence of phrasal categories. We say that phrases are projected
from lexical categories, and hence we have phrases such as NP, VP, PP, and so on. As before,
we use distributional evidence to classify each type, and then specify rules to account for the
distributions we have observed.
2.5.1 NP: Noun Phrase
Consider (64):
(64) [liked ice cream].
The expressions that can occur in the blank position here are once again limited. The kinds of
expression that do appear here include:
(65) Mary, I, you, students, the students, the tall students, the students from Seoul, the
students who came from Seoul, etc.
If we look into the sub-constituents of these expressions, we can see that each includes at least
an N and forms an NP (noun phrase). This leads us to posit the following rule:
7
(66) NP →(Det) A* N (PP/S)
This rule characterizes a phrase, and is one instance of a phrase structure rule (PS rule). The rule
indicates that an NP can consist of an optional Det, any number of optional A, an obligatory N,
and then an optional PP or a modifying S.
8
The slash indicates different options for the same
place in the linear order. These options in the NP rule can be represented in a tree structure:
7
The relative clause who came from Seoul is kind of modifying sentence (S). See Chapter 11.
8
To license an example like the very tall man, we need to make A* as AP*. For simplicity, we just use the former in
the rule.
22
(67)
NP
.
.
.
.
.
.
.
.
.
.
.
.
.

¸
¸
¸
¸
¸
¸
¸
¸
(Det) A* N (PP/S)
. . . . . . . . . . . .
Once we insert appropriate expressions into the pre-terminal nodes, we will have well-formed
NPs; and the rule will not generate the following NPs:
(68) *the whistle tune, *the easily student, *the my dog, . . .
One important point is that as only N is obligatory in NP, a single noun such as Mary, you, or
students can constitute an NP by itself. Hence the subject of the sentence She sings will be an
NP, even though that NP consists only of a pronoun.
2.5.2 VP: Verb Phrase
Just as N projects an NP, V projects a VP. A simple test environment for VP is given in (69).
(69) The student .
(70) lists just a few of the possible phrases that can occur in the underlined position.
(70) snored, ran, sang, loved music, walked the dog through the park, lifted 50 pounds,
thought Tom is honest, warned us that storms were coming, etc.
These phrases all have a V as their head – as projections of V, they form VP. VP can be charac-
terized by the rule in (71), to a first level of analysis:
(71) VP →V (NP) (PP/S)
This simple VP rule says that a VP can consist of an obligatory V followed by an optional NP
and then any number of PPs or an S. The rule thus does not generate ill-formed VPs such as
these:
(72) *leave the meeting sing, *the leave meeting, *leave on time the meeting, . . .
We can also observe that the presence of a VP is essential in forming a grammatical S, and the
VP must be finite (present or past tense). Consider the following examples:
(73) a. The monkey wants to leave the meeting.
b. *The monkey eager to leave the meeting.
(74) a. The monkeys approved of their leader.
b. *The monkeys proud of their leader.
(75) a. The men practice medicine.
b. *The men doctors of medicine.
These examples show us that an English well-formed sentence consists of an NP and a (finite)
VP, which can be represented as a PS rule:
23
(76) S →NP VP
We thus have the rule that English sentences are composed of an NP and a VP, the precise
structural counterpart of the traditional ideas of a sentence being ‘a subject and predicate’ or ‘a
noun and a verb’.
One more aspect to the structure of VP involves the presence of auxiliary verbs. Think of
continuations for the fragments in (77):
(77) a. The students .
b. The students want .
For example, the phrases in (78a) and (78b) can occur in (77a) whereas those in (78c) can appear
in (77b).
(78) a. run, feel happy, study English syntax, . . .
b. can run, will feel happy, must study English syntax, . . .
c. to run, to feel happy, to study English syntax, . . .
We have seen that the expressions in (78a) all form VPs, but how about those in (78b) and
(78c)? These are also VPs, which happen to contain more than one V. In fact, the parts after
the auxiliary verbs in (78b) and (78c) are themselves regular VPs. In the full grammar we
will consider to and can and so on as auxiliary verbs, with a feature specification [AUX +] to
distinguish them from regular verbs. Then all auxiliary verbs are simply introduced by a second
VP rule:
9
(79) VP →V[AUX +] VP
One more important VP structure involves the VP modified by an adverb or a PP:
(80) a. John [[read the book] loudly].
b. The teacher [[met his students] in the class].
In such examples, the adverb loudly and the PP in the class are modifying the preceding VP. To
form such VPs, we need the PS rule in (81):
(81) VP →VP Adv/PP
This rule, together with (76) will allow the following structure for (80b):
10
(82)
S
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
NP

¸
¸
¸
¸
¸
¸
VP
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
The teacher VP

¸
¸
¸
¸
¸
¸
¸
¸
PP

¸
¸
¸
¸
¸
¸
met his students in the class
9
The detailed discussion of English auxiliary verbs is found in Chapter 8.
10
We use a triangle when we need not represent the internal structure of a phrase.
24
2.5.3 AP: Adjective Phrase
The most common environment where an adjective phrase (AP) occurs is in ‘linking verb’
constructions as in (83):
(83) John feels .
Expressions like those in (84) can occur in the blank space here:
(84) happy, uncomfortable, terrified, sad, proud of her, proud to be his student, proud that
he passed the exam, etc.
Since these all include an adjective (A), we can safely conclude that they all form an AP. Look-
ing into the constituents of these, we can formulate the following simple PS rule for the AP:
(85) AP →A (PP/VP/S)
This simple AP rule can easily explain the following:
(86) a. John sounded happy/uncomfortable/terrified/proud of her.
b. John felt proud that his son won the game.
c. John sounded *happily/*very/*the student/*in the park.
The verb sounded requires an AP to be followed, but in (86c) we have no AP. In addition,
observe the contrasts in the following examples:
(87) a. *The monkeys seem [want to leave the meeting].
b. The monkeys seem [eager to leave the meeting].
(88) a. *John seems [know about the bananas].
b. John seems [certain about the bananas].
These examples tell us that the verb seem combines with an AP, but not with a VP.
2.5.4 AdvP: Adverb Phrase
Another phrasal syntactic category is adverb phrase (AdvP), as exemplified in (89).
(89) soundly, well, clearly, extremely, carefully, very soundly, almost certainly, very
slowly, etc.
These phrases are often used to modify verbs, adjectives, and adverbs themselves, and they can
all occur in principle in the following environments:
(90) a. He behaved very .
b. They worded the sentence very .
c. He treated her very .
Phrases other than an AdvP cannot appear here. For example, an NP the student or AP happy
cannot occur in these syntactic positions. Based on what we have seen so far, the AdvP rule can
be given as follows:
(91) AdvP →(AdvP) Adv
25
2.5.5 PP: Preposition Phrase
Another major phrasal category is preposition phrase (PP). PPs like those in (92), generally
consist of a preposition plus an NP.
(92) from Seoul, in the box, in the hotel, into the soup, with John and his dog, under the
table, etc.
These PPs can appear in a wide range of environments:
(93) a. John came from Seoul.
b. They put the book in the box.
c. They stayed in the hotel.
d. The fly fell into the soup.
One clear case in which only a PP can appear is the following:
(94) The squirrel ran straight/right .
The intensifiers straight and right can occur neither with an AP nor with an AdvP:
(95) a. The squirrel ran straight/right up the tree.
b. *The squirrel is straight/right angry.
c. *The squirrel ran straight/right quickly.
From the examples in (92), we can deduce the following general rule for forming a PP:
11
(96) PP →P NP
The rule states that a PP consists of a P followed by an NP. We cannot construct unacceptable
PPs like the following:
(97) *in angry, *into sing a song, *with happily, . . .
2.6 Grammar with Phrases
We have seen earlier that the grammar with just lexical categories is not adequate for captur-
ing the basic properties of the language. How much further do we get with a grammar which
includes phrases? A set of PS rules, some of which we have already seen, is given in (98).
12
(98) a. S →NP VP
b. NP →(Det) A* N (PP/S)
c. VP →V (NP) (PP/S/VP)
d. AP →A (PP/S)
11
Depending on how we treat the qualifier straight and right, we may need to extend this PP rule as “PP → (Qual) P
NP” so that the P may be preceded by an optional qualifier like right or straight. However, this means that we need to
introduce another lexical category ‘Qual’. Another direction is to take the qualifier categorically as an adverb carrying
the feature QUAL while allowing only such adverbs to modify a PP.
12
The grammar consisting of such form of rules is often called a ‘Context Free Grammar’, as each rule may apply
any time its environment is satisfied, regardless of any other contextual restrictions.
26
e. AdvP →(AdvP) Adv
f. PP →P NP
The rules say that a sentence is the combination of NP and VP, and an NP can be made up of
a Det, any number of As, an obligatory N, and any number of PPs, and so on.. Of the possible
tree structures that these rules can generate, the following is one example:
(99)
S
.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
_
NP

¸
¸
¸
¸
¸
¸
¸
VP
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
Det A N V NP

¸
¸
¸
¸
¸
¸
¸
PP

¸
¸
¸
¸
¸
¸
¸
. . . . . . . . . . . . Det N P NP

¸
¸
¸
¸
¸
¸
¸
. . . . . . . . . Det N
. . . . . .
With the structural possibilities shown here, let us assume that we have the following lexical
entries:
(100) a. Det: a, an, this, that, any, some, which, his, her, no, etc.
b. A: handsome, tall, fat, large, dirty, big, yellow, etc.
c. N: book, ball, hat, friend, dog, cat, man, woman, John, etc.
d. V: kicked, chased, sang, met, believed, thinks, imagines, assumes etc.
Inserting these elements in the appropriate pre-terminal nodes (the places with dots) in (99), we
are able to generate various sentences like those in (101):
13
(101) a. This handsome man chased a dog.
b. A man kicked that ball.
c. That tall woman chased a cat.
d. His friend kicked a ball.
There are several ways to generate an infinite number of sentences with this kind of grammar.
As we have seen before, one simple way is to repeat a category (e.g., adjective) infinitely. There
are also other ways of generating an infinite number of grammatical sentences. Look at the
following two PS rules from (98) again:
13
The grammar still generates semantically anomalous examples like#The desk believed a man or#A man sang her
hat. For such semantically distorted examples, we need to refer to the notion of ‘selectional restrictions’ (see Chapter
7).
27
(102) a. S →NP VP
b. VP →V S
As we show in the following tree structure, we can ‘recursively’ apply the two rules, in the sense
that one can feed the other, and then vice versa:
(103)
S

¸
¸
¸
¸
¸
¸
¸
7654 0123
NP VP

¸
¸
¸
¸
¸
¸
¸
?>=< 89:;
N V S

¸
¸
¸
¸
¸
¸
¸
7654 0123
John believes NP VP

¸
¸
¸
¸
¸
¸
¸
?>=< 89:;
N V S

¸
¸
¸
¸
¸
¸
¸
Mary thinks Tom is honest
It is not difficult to expand this sentence by applying the two rules again and again:
(104) a. Bill claims John believes Mary thinks Tom is honest.
b. Jane imagines Bill claims John believes Mary thinks Tom is honest.
There is no limit to this kind of recursive application of PS rules: it proves that this kind of
grammar can generate an infinite number of grammatical sentences.
One structure which can be also recursive involves sentences involving auxiliary verbs. As
noted before in (79), an auxiliary verb forms a larger VP after combining with a VP:
(105)
S

¸
¸
¸
¸
¸
¸
¸
NP VP

¸
¸
¸
¸
¸
¸
¸
7654 0123
N V[AUX +] VP

¸
¸
¸
¸
¸
¸
¸
7654 0123
They will V NP
study English syntax
This means that we will also have a recursive structure like the following:
14
14
Due to the limited number of auxiliary verbs, and restrictions on their cooccurrence, the maximum number of
auxiliaries in a single English clause is 3. See Chapter 8.
28
(106)
S
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
NP VP
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
?>=< 89:;
N V [AUX +] VP
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
7654 0123
They will V[AUX +] VP
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
?>=< 89:;
have V[AUX +] VP
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
been studying English syntax
Another important property that PS rules bring us is the ability to make reference to hierar-
chical structures within given sentences, where parts are assembled into sub-structures of the
whole. One merit of such hierarchical structural properties is that they enable us to represent the
structural ambiguities of sentences we have seen earlier in (42). Let us look at more examples:
(107) a. The little boy hit the child with a toy.
b. Chocolate cakes and pies are my favorite desserts.
Depending on which PS rules we apply, for the sentences here, we will have different hierar-
chical tree structures. Consider the possible partial structures of (107a) which the grammar can
generate:
15
(108) a.
VP
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
_
_
VP
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
PP

¸
¸
¸
¸
¸
¸
V NP

¸
¸
¸
¸
¸
with the toy
hit the child
b.
VP
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
V NP
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
hit Det N PP

¸
¸
¸
¸
¸
¸
the child with the toy
15
One can draw a slight different structure for (108b) with the introduction of the rule ‘NP → NP PP’.
29
The structures clearly indicate what with the toy modifies: in (108a), it modifies the whole VP
phrase whereas (108b) modifies just the noun child. The structural differences induced by the
PS rules directly represent these meaning differences.
In addition, we can easily show why examples like the following are not grammatical:
(109) a. *The children were in their rooms or happily.
b. *Lee went to the store and crazy.
We have noted that English allows two alike categories to be coordinated. This can be written
as a PS rule, for phrasal conjunction, where XP is any phrase in the grammar.
16
(110) XP →XP
+
Conj XP
The ‘coordination’ rule says two identical XP categories can be coordinated and form the same
category XP. Applying this PS rule, we will then allow (111a) but not (111b):
(111) a.
PP
.
.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
_
_
PP

¸
¸
¸
¸
¸
¸
¸
Conj PP

¸
¸
¸
¸
¸
¸
in their rooms or on the porch
b.
*PP
.
.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
_
_
PP

¸
¸
¸
¸
¸
¸
Conj AP

¸
¸
¸
¸
to the store and crazy
Unlike categories such as PP and AP may not be coordinated.
The PS rules further allow us to represent the difference between phrasal verb (e.g, call off )
constructions and prepositional verb constructions (e.g. rely on), some of whose properties we
have seen earlier. Consider a representative pair of contrasting examples:
(112) a. John suddenly got off the bus.
b. John suddenly put off the customers.
By altering the position of off , we can determine that off in (112a) is a preposition whereas off
in (112b) is a particle:
(113) a. *John suddenly got the bus off.
b. John suddenly put the customers off.
This in turn means that off in (112a) is a preposition, forming a PP with the following NP,
whereas off in (112b) is a particle that forms no constituent with the following NP the customers.
16
Different from the Kleene star operator

, the plus operator
+
here means the XP here occurs at least once.
30
This in turn means that in addition to the PP formation rule, the grammar needs to introduce the
following VP rules:
(114) a. VP →V Part NP
b. VP →V NP Part
c. VP →V PP
Equipped with these rules, we then can easily represent the differences of these grammatical
sentences (112a), (112b) and (113b) in tree structures:
(115) a.
VP
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
V PP
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
get P NP

¸
¸
¸
¸
¸
off the bus
b.
VP
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
V Part NP

¸
¸
¸
¸
¸
¸
¸
put off the customers
c.
VP
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
V NP

¸
¸
¸
¸
¸
¸
¸
Part
put the customers off
As represented here, the particle does not form a constituent with the following or preceding
NP whereas the preposition does form a constituent with it.
In summary, we have seen that a grammar with lexical categories can not only generate
an infinite number of grammatical English sentences, but also account for some fundamental
properties, such as agreement and constituency.
17
This motivates the introduction of phrases
into the grammar.
17
In this chapter, we have not discussed the treatment of agreement with PS rules. Chapter 6 discusses the subject-
verb agreement in detail.
31
2.7 Exercises
1. Determine the lexical category of the italicized words in the following. In doing so, use
the three criteria (morphological, semantic, and syntactic) to provide the evidence for
your answer and state which one is the most reliable one.
(i) a. His second book came out earlier this year and became an instant best-
seller.
b. When you book something such as a hotel room, you arrange to have it.
c. Price quotes on selected categories will be sent out upon request.
d. No doubt that he was forced to leave his family against his will.
e. He intended to will the large amount of money to Frank.
f. Jane stood aside to let her pass.
g. He has a rail pass that’s right for you.
h. It is important for us to spend time with children.
i. He was arrested for being drunk.
j. I think that person we met last week is insane.
k. We believe that he is quite reasonable.
l. I forgot to return the book that I borrowed from the teacher.
2. Consider the following data carefully and describe the similarities and differences among
that, for, if and whether. In so doing, first compare that and for and then see how these
two are different from if and whether.
(i) a. I am anxious that you should arrive on time.
b. *I am anxious that you to arrive on time.
(ii) a. I am anxious for you to arrive on time.
b. *I am anxious for you should arrive on time.
(iii) a. I don’t know whether/if I should agree.
b. I wonder whether/if you’d be kind enough to give us information.
(vi) a. If students study hard, teachers will be happy.
b. Whether they say it or not, most teachers expect their students to study
hard.
3. Check if the italic parts form a constituent or not, using at least two constituenthood tests
(e.g., cleft, pronoun substitution, stand-alone, etc.). Also provide tree structures for each
of the following examples.
(i) a. John bought a book on the table.
b. John put a book on the table.
(ii) a. She turned down the side street
b. She turned down his offer.
(iii) a. He looked at a book about swimming.
b. He talked to a girl about swimming.
32
c. He talked with a girl about swimming.
(iv) a. I don’t know the people present.
b. John called the president a fool.
4. Explain why the examples in (i) are ungrammatical. As part of the exercise, first draw
structure for each example and try to determine the applicability of the the PS rules such
as the coordination rule in (110), presented earlier in this chapter.
(i) a. *Could you turn off the fire and on the light?
b. *A nuclear explosion would wipe out plant life and out animal life.
c. *He ran down the road and down the President.
d. *I know the truth and that you are innocent.
e. *Lee went to the store and crazy.
5. Provide a tree structure for each of the following sentences and suggest what kind of VP
rules are necessary. In doing so, pay attention to the position of modifiers like proudly,
by the park, and so forth.
(i) a. John refused the offer proudly.
b. I consider Telma the best candidate.
c. I saw him leaving the main building.
d. He took Masako to the school by the park.
e. John sang a song and danced to the music.
f. John wants to study linguistics in near future.
g. They told Angelica to arrive early for the award.
h. That Louise had abandoned the project surprised everyone.
6. Each of the following sentences is structurally ambiguous – it has at least two structures.
Represent the structural ambiguities by providing different tree structures for each string
of words.
18
(i) a. I know you like the back of my hand.
b. I forgot how good beer tastes.
c. I saw that gas can explode.
d. Time flies like an arrow.
e. I need to have that report on our webpage by tomorrow.
7. Provide tree structures for each of the following sentences and see if there are any new
PS rules that we need to add, to supplement those we covered in this chapter. If there are
any places you cannot assign structures, please use triangles.
(i) Different languages may have different lexical categories, or they might
associate different properties to the same one. For example, Spanish uses
adjectives almost interchangeably as nouns while English cannot. Japanese
has two classes of adjectives whereas English has one; Korean, Japanese,
18
For i-e, to help tease out the ambiguity, consider related potential interpretations like Please put that book on my
desk. and That report on our webpage alleges that it does not function well..
33
and Chinese have measure words while European languages have nothing
resembling them; many languages don’t have a distinction between adjec-
tives and adverbs, or adjectives and nouns, etc. Many linguists argue that
the formal distinctions between parts of speech must be made within the
framework of a specific language or language family, and should not be
carried over to other languages or language families.
19
19
Adapted from http://en.wikipedia.org/wiki/Part_of_speech
34
3
Syntactic Forms, Grammatical Functions, and
Semantic Roles
3.1 Introduction
In the previous chapter, we analyzed English sentences with PS rules. For example, the PS
rule ‘S →NP VP’ represents the basic rule for forming well-formed English sentences. As we
have seen, such PS rules allow us to represent the constituent structure of a given sentence in
terms of lexical and phrasal syntactic categories. There are other dimensions of the analysis
of sentences; one such way is using the notion of grammatical functions such as subject and
object:
(1) a. Syntactic categories: N, A, V, P, NP, VP, AP, . . .
b. Grammatical functions: SUBJ (Subject), OBJ (Object), MOD (Modifier), PRED
(Predicate), . . .
The notions such as SUBJ, OBJ and PRED represent the grammatical function each constituent
plays in the given sentence. For example, consider one simple sentence:
(2) The monkey scratched a boy on Monday.
This sentence can be structurally represented in terms of either syntactic categories or gram-
matical functions as in the following:
(3) a. [
S
[
NP
The monkey] [
VP
scratched [
NP
a boy] [
PP
on Monday]]].
b. [
S
[
SUBJ
The monkey] [
PRED
scratched [
OBJ
a boy] [
MOD
on Monday]]].
As shown here, the monkey is an NP in terms of its syntactic form, but is the SUBJ (subject) in
terms of its grammatical function. The NP a boy is the OBJ (object) while the verb scratched
functions as a predicator. More importantly, we consider the entire VP to be a PRED (predicate)
which describes a property of the subject. On Monday is a PP in terms of its syntactic category,
but serves as a MOD (modifier) here.
We also can represent sentence structure in terms of semantic roles. Constituents can be
considered in terms of conceptual notions of semantic roles such as agent, patient, location,
35
instrument, and the like. A semantic role denotes the underlying relationship that a participant
has with the relation of the clause, expressed by the main verb. Consider the semantic roles of
the NPs in the following two sentences:
1
(4) a. John tagged the monkey in the forest.
b. The monkey was tagged in the forest by John.
Both of these sentences describe a situation in which someone named John tagged a particular
monkey. In this situation, John is the agent and the monkey is the patient of the tagging event.
This in turn means that in both cases, John has the semantic role of agent (agt), whereas the
monkey has the semantic role of patient (pat), even though their grammatical functions are
different. We thus can assign the following semantic roles to each constituent of the examples:
(5) a. [[
agt
John] [
pred
tagged [
pat
the monkey] [
loc
in the forest]]].
b. [[
pat
The monkey] [
pred
was tagged [
loc
in the wood] [
agt
by John]]].
As noted here, in addition to agent and patient, we have predicate (pred) and locative (loc)
which also express the semantic role that each phrase performs in the described situation.
Throughout this book we will see that English grammar refers to these three different levels
of information (syntactic category, grammatical function, and semantic role), and they interact
with each other. For now, it may appear that they are equivalent classifications: for example, an
agent is a subject and an NP, and a patient is an object and an NP. However, as we get further
into the details of the grammar, we will see many ways in which the three levels are not simply
co-extensive.
3.2 Grammatical Functions
How can we identify the grammatical function of a given constituent? Several tests can be used
to determine grammatical function, as we show here.
3.2.1 Subjects
Consider the following pair of examples:
(6) a. [The cat] [devoured [the rat]].
b. [The rat] [devoured [the cat]].
These two sentences have exactly the same words and have the same predicator devoured. Yet
they are significantly different in meaning, and the main difference comes from what serves as
subject or object with respect to the predicator. In (6a), the subject is the cat, whereas in (6b)
it is the rat, and the object is the rat in (6a) but the cat in (6b).
The most common structure for a sentence seems to be one in which the NP subject is the one
who performs the action denoted by the verb (thus having the semantic role of agent). However,
this is not always so:
1
Semantic roles are also often called ‘thematic roles’ or ‘θ-roles (“theta roles”) in generative grammar (Chomsky
1982, 1986).
36
(7) a. My brother wears a green overcoat.
b. This car stinks.
c. It rains.
d. The committee disliked her proposal.
Wearing a green overcoat, stinking, raining, or disliking one’s proposal are not agentive activi-
ties; they indicate stative descriptions or situations. Such facts show that we cannot rely on the
semantic roles of agent for determining subjecthood.
More reliable tests for subjecthood come from syntactic tests such as agreement, tag ques-
tions, and subject-auxiliary inversion.
Agreement: The main verb of a sentence agrees with the subject in English:
(8) a. She never writes/*write home.
b. These books *saddens/sadden me.
c. Our neighbor takes/*take his children to school in his car.
As we noted in Chapter 1, simply being closer to the main verb does not entail subjecthood:
(9) a. The book, including all the chapters in the first section, is/*are very interesting.
b. The effectiveness of teaching and learning *depend/depends on several factors.
c. The tornadoes that tear through this county every spring *is/are more than just a
nuisance.
The subject in each example is book, effectiveness, and tornadoes respectively, even though
there are nouns closer to the main verb. This indicates that it is not simply the linear position of
the NP that determines agreement; rather, agreement shows us what the subject of the sentence
is.
Tag questions: A tag question, a short question tagged onto the end of an utterance, is also a
reliable subjecthood test:
(10) a. The lady singing with a boy is a genius, isn’t she/*isn’t he?
b. With their teacher, the kids have arrived safely, haven’t they/ *hasn’t he?
The pronoun in the tag question agrees with the subject in person, number, and gender – it
refers back to the subject, but not necessarily to the closest NP, nor to the most topical one. The
pronoun she in (10a) shows us that lady is the head of the subject NP in that example, and they
in (10b) leads us to assign the same property to kids. The generalization is that a tag question
must contain a pronoun which identifies the subject of the clause to which the tag is attached.
Subject-auxiliary inversion: In forming questions and other sentence-types, English has
subject-auxiliary inversion, which applies only to the subject.
(11) a. This teacher is a genius.
b. The kids have arrived safely.
37
c. It could be more detrimental.
(12) a. Is this teacher a genius?
b. Have the kids arrived safely?
c. Could it be more detrimental?
As seen here, the formation of ‘Yes/No questions’ such as these involves the first tensed auxil-
iary verb moving across the subject: more formally, the auxiliary verb is inverted with respect to
the subject, hence the term ‘subject-auxiliary inversion’. This is not possible with a non-subject:
(13) a. The kids in our class have arrived safely.
b. *Have in our class the kids arrived safely?
Subject-auxiliary inversion provides another reliable subjecthood test.
3.2.2 Direct and Indirect Objects
A direct object (DO) is canonically an NP, undergoing the process denoted by the verb:
(14) a. His girlfriend bought this computer.
b. That silly fool broke the teapot.
However, this is not a solid generalization. The objects (OBJ) in (15a) and (15b) are not really
affected by the action. In (15a) the dog is experiencing something, and in (15b) the thunder is
somehow causing some feeling in the dog:
(15) a. Thunder frightens [the dog].
b. The dog fears [thunder].
Once again, the data show us that we cannot identify the object based on semantic roles. A
much more firm criterion is the syntactic construction of passivization, in which a notional
direct object appears as subject. The sentences in (16) can be turned into passive sentences in
(17):
(16) a. His girlfriend bought this computer for him.
b. The child broke the teapot by accident.
(17) a. This computer was bought for him by his girlfriend.
b. The teapot was broken by the child by accident.
What we can notice here is that the objects in (16) are ‘promoted’ to subject in the passive
sentences. The test comes from the fact that non-object NPs cannot be promoted to the subject:
(18) a. This item belongs to the student.
b. *The student is belonged to by this item.
(19) a. He remained a good friend to me.
b. *A good friend is remained to me (by him).
38
The objects that undergo passivization are direct objects, distinct from indirect objects.
An indirect object (IO) is one which precedes a direct object (DO), as in (20); IOs are NPs
and have the semantic roles of goal, recipient, or benefactive:
(20) a. I threw [the puppy] [the ball]. (IO = goal)
b. John gave [the boys] [the CDs]. (IO = recipient)
c. My mother baked [me] [a birthday cake]. (IO = benefactive)
In examples like (20), passive has the property of making the IO into the subject.
(21) a. The boys were given the CDs (by John).
b. She was sent a review copy of the book (by the publisher).
Note that sentences with the IO-DO order are different from those where the semantic role
of the IO is expressed as an oblique PP, following the DO (such a process is sometimes called
dative alteration):
(22) a. John gave the CDs to the boys.
b. The publisher sent a review copy of the book to her.
c. My mother baked a cake for me.
In this kind of example, it is once again the DO which can be passivized, giving examples like
the following:
(23) a. The CDs were given to the boys by John.
b. A review copy of the book was sent to her by the publisher.
c. This nice cake was baked for me by my mother.
3.2.3 Predicative Complements
There also are NPs which follow a verb but which do not behave as DOs or IOs. Consider the
following sentences:
(24) a. This is my ultimate goal.
b. Michelle became an architect.
(25) a. They elected Graham chairman.
b. I consider Andrew the best writer
The italicized elements here are traditionally called ‘predicative complements’ in the sense that
they function as the predicate of the subject or the object. However, even though they are NPs,
they do not passivize:
(26) a. *Chairman was elected Graham.
b. *The best writer was considered Andrew.
The difference between objects and predicative complements can also be seen in the following
contrast:
(27) a. John made Kim a great doll.
39
b. John made Kim a great doctor.
Even though the italicized expressions here are both NPs, they function differently. The NP a
great doll in (27a) is the direct object, as in John made a great doll for Kim, whereas the NP
a great doctor in (27b) cannot be an object: it serves as the predicate of the object Kim. If we
think of part of the meaning informally, only in the second example would we say that the final
NP describes the NP Kim.
(28) a. (27)a: Kim = a great doll
b. (27)b: Kim = a great doctor
In addition, phrases other than NPs can serve as predicative complements:
(29) a. The situation became terrible.
b. This map is what he wants.
c. The message was that you should come on time.
(30) a. I made Kim angry.
b. I consider him immoral.
c. I regard Andrew as the best writer.
d. They spoil their kids rotten.
The italicized complements function to predicate a property of the subject in (29) and of the
object in (30).
3.2.4 Oblique Complements
Consider now the italicized expressions in (31):
(31) a. John put books in the box.
b. John talked to Bill about the exam.
c. She reminded him of the last time they met.
d. They would inform Mary of any success they have made.
These italicized expressions are neither objects nor predicative complements. Since their pres-
ence is obligatory, for syntactic well-formedness, they are called oblique complements. Roughly
speaking, ‘oblique’ contrasts with the ‘direct’ functions of subject and object, and oblique
phrases are typically expressed as PPs in English.
As we have seen before, most ditransitive verbs can also take oblique complements:
(32) a. John gave a book to the student.
b. John bought a book for the student.
The PPs here, which cannot be objects since they are not NPs, also do not serve as predicate of
the subject or object – they relate directly to the verb, as oblique complements.
40
3.2.5 Modifiers
The functions of DO, IO, predicative complement, and oblique complement all have one com-
mon property: they are all selected by the verb, and we view them as being present to ‘comple-
ment’ the verb to form a legitimate VP. Hence, these are called complements (COMPS), and
typically they cannot be omitted.
Unlike these COMPS, there are expressions which do not complement the predicate in the
same way, and which are truly optional:
(33) a. The bus stopped suddenly.
b. Shakespeare wrote his plays a long time ago.
c. They went to the theater in London.
d. He failed chemistry because he can’t understand it.
The italicized expressions here are all optional and function as modifiers (also called ‘adjuncts’
or ‘adverbial’ expressions). These modifiers specify the manner, location, time, or reason,
among many other properties, of the situations expressed by the given sentences – informally,
they are the (how, when, where, and why) phrases.
One additional characteristic of modifiers is that they can be stacked up, whereas comple-
ments cannot.
(34) a. *John gave Tom [a book] [a record].
b. I saw this film [several times] [last year] [during the summer].
As shown here, temporal adjuncts like several times and last year can be repeated, whereas the
two complements a book and a record in (34a) cannot. Of course, temporal adjuncts do not
become the subject of a passive sentence, suggesting that they cannot serve as objects.
(35) a. My uncle visited today.
b. *Today was visited by my uncle.
3.3 Form and Function Together
We now can analyse each sentence in terms of grammatical functions as well as the structural
constituents. Let us see how we can analyze a simple sentence along these two dimensions:
(36)
S

_
_
_
_
_
_
_
_
_
_
_
_
_
_
NP: SUBJ

¸
¸
¸
¸
¸
¸
¸
¸
VP: PRED

¸
¸
¸
¸
¸
¸
¸
¸
Det A N V NP: OBJ

¸
¸
¸
¸
¸
¸
¸
¸
The little cat devoured Det N
a mouse
41
As represented here, the expressions the little cat and a mouse are both NPs, but they have
different grammatical functions, SUBJ and OBJ. The VP as a whole functions as the predicate
of the sentence, describing the property of the subject.
2
Assigning grammatical functions within complex sentences is no different:
(37)
S
.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
_
NP: SUBJ VP: PRED
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
N V CP: OBJ
.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
_
John believes C S
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
that NP: SUBJ

¸
¸
¸
¸
¸
VP: PRED

¸
¸
¸
¸
¸
¸
¸
the cat V NP: OBJ
devoured a mouse
Each clause has its own SUBJ and PRED: John is the subject of the higher clause, whereas the
cat is the subject of the lower clause. We also can notice that there are two OBJs: the CP is the
object of the higher clause whereas the NP is that of the lower clause.
3
Every category in a given sentence has a grammatical function, but there is no one-to-one
mapping between category such as NP or CP and its possible grammatical function(s). The
following data set shows us how different phrase types can function as SUBJ or OBJ:
4
(38) a. [
NP
The termites] destroyed the sand castle.
b. [
VP
Being honest] is not an easy task.
c. [
CP
That John passed] surprised her.
d. [
VP
To finish this work on time] is almost unexpected.
e. [
PP
Under the bed] is a safe place to hide.
(39) a. I sent [
NP
a surprise present] to John.
b. They wondered [
S
what she did yesterday].
c. They believed [
CP
that everybody would pass the test].
2
A word of caution is in order here. We should not confuse the functional term ‘adverbial’ with the category term
‘adverb’. The term ‘adverbial’ is used interchangeably with ‘adjunct’ or ‘modifier’, whereas ‘adverb’ only designates a
part of speech. In English almost any kind of phrasal category can function as an adverbial element, but only a limited
set of words are adverbs.
3
The phrase CP is headed by the complementizer that.
4
In due course, we will discuss in detail the properties of each phrase type here.
42
d. Are you going on holiday before or after Easter? I prefer [
PP
after Easter].
As the examples in (38) and (39) show, not only NPs but also infinitival VPs and CPs can also
function as SUBJ and OBJ. The following tag-question, subject-verb agreement, and subject-
hood tests show us that an infinitival VP and CP can function as the subject.
(40) a. [That John passed] surprised her, didn’t it?
b. [[That the march should go ahead] and [that it should be cancelled]] have/*has
been argued by different people at different times.
(41) a. [To finish it on time] made quite a statement, didn’t it?
b. [[To delay the march] and [to go ahead with it]] have/*has been argued by different
people at different times.
The same goes for MOD, as noted before. Not only AdvP, but also phrases such as NP, S,
VP, or PP can function as a modifier:
(42) a. The little cat devoured a mouse [
NP
last night].
b. John left [
AdvP
very early].
c. John has been at Stanford [
PP
for four years].
d. John studied hard [
VP
to pass the exam].
e. She disappeared [
S
when the main party arrived].
The sentence (42a) will have the following structure:
(43)
S
.
.
.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
_
_
_
NP: SUBJ

¸
¸
¸
¸
¸
¸
¸
¸
VP: PRED

_
_
_
_
_
_
_
_
_
_
_
_
_
_
Det A N VP: PRED

¸
¸
¸
¸
¸
¸
¸
¸
NP: MOD

¸
¸
¸
¸
¸
¸
The little cat V NP: OBJ

¸
¸
¸
¸
¸
last night
devoured a mouse
Here the expression last night is an adverbial NP in the sense that it is categorically an NP but
functions as a modifier (adjunct) to the VP.
As we go through this book, we will see that the distinction between grammatical functions
and categorical types is crucial in the understanding of English syntax.
3.4 Semantic Roles
As noted before, semantic roles were introduced as a way of classifying the arguments of pred-
icators (mostly verbs and adjectives) into a closed set of participant types. Even though we
43
cannot make any absolute generalizations about the relationship between grammatical func-
tions and semantic roles, the properties of semantic roles do interact in regular ways with cer-
tain grammatical constructions. A list of the most relevant thematic roles and their associated
properties is given below.
5
• Agent: A participant which the meaning of the verb specifies as doing or causing some-
thing, possibly intentionally. Examples: subject of eat, kick, hit, hammer, etc.
(44) a. John ate his noodle quietly.
b. A boy hit the ball.
c. A smith hammered the metal.
• Patient: A participant which the verb characterizes as having something happen to it, and
as being affected by what happens to it. Examples: object of kick, hit, hammer, etc.
6
(45) a. A boy hit the ball.
b. A smith hammered the metal.
• Experiencer: A participant who is characterized as aware of something. Examples: sub-
ject of perception verbs like feel, smell, hear, see, etc.
(46) a. The students felt comfortable in the class.
b. The student heard a strange sound.
• Theme: A participant which is characterized as changing its position or condition, or
as being in a state or position. Examples: direct object of give, hand, subject of come,
happen, die, etc.
(47) a. John gave a book to the students.
b. John died last night.
• Benefactive: The entity that benefits from the action or event denoted by the predicator.
Examples: oblique complement of make, buy, etc.
(48) a. John made a doll for his son.
b. John bought a lot of books for his sons.
• Source: The one from which motion proceeds. Examples: subject of promise, object of
deprive, free, cure, etc.
(49) a. John promised Bill to leave tomorrow morning.
b. John deprived his sons of game cards.
• Goal: The one to which motion proceeds. Examples: subject of receive, buy, indirect
object of tell, give, etc.
(50) a. Mary received an award from the department.
b. John told the rumor to his friend.
5
The definition of semantic roles given here is adopted from Dowty (1989).
6
Patient and theme are often unified into ‘undergoer’ in the sense that both the patient and theme individual undergo
the action in question.
44
• Location: The thematic role associated with the NP expressing the location in a sentence
with a verb of location. Examples: subject of keep, own, retain, locative PPs, etc.
(51) a. John put his books in the attic.
b. The government kept all the money.
• Instrument: The medium by which the action or event denoted by the predicator is carried
out. Examples: oblique complement of hit, wipe, hammer, etc.
(52) a. John hit the ball with a bat.
b. John wiped the window with a towel.
An important advantage of having such semantic roles available to us is that it allows us to
capture the relationship between two related sentences, as we have already seen. As another
example, consider the following pair:
(53) a. [
agt
The cat] chased [
pat
the mouse].
b. [
pat
The mouse] was chased by [
agt
the cat].
Even though the above two sentences have different syntactic structures, they have essentially
identical interpretations. The reason is that the same semantic roles assigned to the NPs: in both
examples, the cat is the agent, and the mouse is the patient. Different grammatical uses of verbs
may express the same semantic roles in different arrays.
The semantic roles also allow us to classify verbs into more fine-grained groups. For exam-
ple, consider the following examples:
(54) a. There still remains an issue to be solved.
b. There lived a man with his grandson.
c. There arrived a tall, red-haired and incredibly well dressed man.
(55) a. *There sang a man with a pipe.
b. *There dances a man with an umbrella.
All the verbs are intransitive, but not all are acceptable in the there-construction. The dif-
ference can come from the semantic role of the postverbal NP, as assigned by the main verb.
Verbs like arrive, remain, live are taken to assign the semantic role of ‘theme’ (see the list of
roles above), whereas verbs like sing, dance assign an ‘agent’ role. We thus can conjecture that
there-constructions do not accept the verb whose subject carries an agent semantic role.
While semantic roles provide very useful ways of describing properties across different con-
structions, we should point out that the theoretical status of semantic roles is still unresolved.
7
For example, there is no agreement about exactly which and how many semantic roles are
needed. The problem is illustrated by the following simple examples:
(56) a. John resembles his mother.
b. A is similar to B.
7
See Levin and Rappaport Hovav (2005) for further discussion of this issue.
45
What kind of semantic roles do the arguments here have? Both participants seem to be playing
the same role in these examples – they both cannot be either agent or patient or theme. They are
also cases where we might not be able to pin down the exact semantic role:
(57) a. John runs into the house.
b. Mary looked at the sky.
The subject John in (57a) is both agent and theme: it is agent since it initiates and sustains the
movement but also theme since it is the object that moves.
8
Also, the subject Mary in (57b) can
either be an experiencer or an agent depending on her intention – one can just look at the sky
with no purpose at all.
9
Even though there are theoretical issues involved in adopting semantic roles in the grammar,
there are also many advantages of using them. We can make generalizations about the grammar
of the language: typically the ‘agent’ takes the subject position, while an NP following the
word from is serving as the ‘source’. As we will see in the next chapter, semantic roles are
also recognized as the standard concepts used for organizing predicate-argument structures for
predicates within the lexicon. In the subsequent chapters, we will refer to semantic roles in
various places.
3.5 Exercises
1. Construct sentences containing the following grammatical functions:
(i) a. subject, predicator, direct object
b. subject, predicator, indirect object, direct object
c. subject, predicator, adjunct
d. adjunct, subject, predicator
e. adjunct, subject, predicator, direct object
f. subject, predicator, direct object, oblique complement
g. subject, predicator, predicative complement
h. subject, predicator, direct object, predicative complement
i. subject, predicator, predicative complement, adjunct
j. subject, predicator, direct object, predicative complement, adjunct
2. Give the grammatical function of the italicized phrases in the following examples:
(i) a. All of his conversation was reported to me.
8
Jackendoff (1987) develops an account of thematic roles in which agency and motion are two separate dimensions,
so, in fact, a single NP can be agent and theme.
9
To overcome the problemof assigning the correct semantic role to an argument, one can assume that each predicator
has its own (individual) semantic roles. For example, the verb kick, instead of having an agent and a patient, has two
individualized semantic roles ‘kicker’ and ‘kicked’. See Pollard and Sag (1987).
46
b. Sandy removed her ballet shoes.
c. The school awarded a few of the girls in Miss Kim’s class scholarships.
d. She was the nicest teacher in the Senior School.
e. They elected him America’s 31st President.
f. The next morning we set out for Seoul.
g. Doing syntax is not easy.
h. This is the place to go to.
i. He saw the man with the stick.
j. This week will be a difficult one for us.
k. We need to finish the project this week.
3. Draw tree structures for the following sentences and then assign an appropriate gram-
matical function to each phrase.
(i) a. They parted the best of friends.
b. In the summer we always go to France.
c. Benny worked in a shoe factory when he was a student.
d. Last year I saw this film several times.
e. He baked Tom the bread last night.
f. That they have completed the course is amazing.
g. Everyone hoped that she would sing.
h. The gang robbed her of her necklace.
i. They helped us edit the script.
j. The teacher made students happy.
k. We reminded him of the agreement.
4. Consider the following examples:
(i) a. There is/*are only one chemical substance involved in nerve transmission.
b. There *is/are more chemical substances involved in nerve transmission.
With respect to the grammatical function of there, what can we infer from these data? Try
out more subjecthood tests such as the tag-question test to determine the grammatical
function of there in these examples. In addition, try to decide what is the subject in
the following so-called ‘locative inversion’ examples and provide at least three different
locative inversion examples that you can find from naturally-occurring material.
(ii) a. In the garden stands/*stand a statue.
b. Among the guests was/*were sitting my friend Louise.
5. Determine the grammatical function of the italiczed phrase, providing at least one syn-
tactic test we have discussed in the chapter.
10
(i) a. This proved a decisive factor.
b. This proved my hypothesis.
10
This exercise is adopted from Huddleston and Pullum (2002).
47
c. The students all enjoyed that summer.
d. The students all worked that summer.
e. The scientist made her a robot.
f. The students called me a teacher.
6. Assign a semantic role to each argument in the following sentences.
(i) a. A big green insect flew into the soup.
b. John’s mother sent a letter to Mary.
c. John smelled the freshly baked bread.
d. We placed the cheese in the refrigerator.
e. Frank threw himself into the sofa.
f. The crocodile devoured the doughnut.
g. John came from Seoul.
h. John is afraid of Bill.
i. The ice melted.
j. The vacuum cleaner terrifies the child.
7. Determine the grammatical functions for the underlined expressions in the following
text.
(i) Scientists found that the birds sang well in the evenings, but performed
badly in the mornings. After being awake several hours, however, the
young males regained their mastery of the material and then improved on
the previous day’s accomplishments. To see whether this dip in learning
was caused by the same kind of pre-coffee fog that many people feel in
the morning, the researchers prevented the birds from practicing first thing
in the morning. They also tried keeping the birds from singing during the
day, and they used a chemical called melatonin to make the birds nap at
odd times. The researchers concluded that their study supports the idea
that sleep helps birds learn. Studies of other animals have also suggested
that sleep improves learning.
11
11
From Science News Online, Feb 2, 2007
48
4
Head, Complements, and Modifiers
4.1 Projections from Lexical Heads to Phrases
4.1.1 Internal vs. External Syntax
As we have seen in the previous chapters, both syntactic categories (NP, AP, VP, PP, etc.) and
grammatical functions (subject, complement, and modifier) play important roles in the analysis
of English sentences. We have also observed that the grammatical function and form of each
constituent depend on where it occurs or what it combines with.
The combinatory properties of words and phrases involve two aspects of syntax: internal
and external syntax.
1
Internal syntax deals with how a given phrase itself is constructed in a
well-formed manner whereas external syntax is concerned with how a phrase can be used in a
larger construction. Observe the following examples:
(1) a. *John [put his gold].
b. *John [put under the bathtub].
c. *John [put his gold safe].
d. *John [put his gold to be under the bathtub].
e. John [put his gold under the bathtub].
Why is only (1e) acceptable? Simply, because only it satisfies the condition that the verb put
selects an NP and a PP as its complements, and it combines with them in the VP. In the other
examples, this condition is not fulfilled. This combinatory requirement starts from the internal
(or lexical) properties of the verb put, and is not related to any external properties of the VP.
In contrast, the external syntax is concerned with the external environment in which a phrase
occurs. Some of the unacceptable examples in (1) can be legitimate expressions if they occur in
the proper (syntactic) context.
(2) a. This is the box in which John [put his gold]. (cf. (1a))
b. This is the gold that John [put under the bathtub]. (cf. (1b))
1
The terms ‘internal’ and ‘external’ syntax are from Baker (1995).
49
Meanwhile, the well-formed VP in (1e) can be unacceptable, depending on external contexts.
For example, consider frame induced by the governing verb kept in (3):
(3) a. *The king kept [put his gold under the bathtub].
b. The king kept [putting his gold under the bathtub].
The VP put his gold under the bathtub is a well-formed phrase, but cannot occur in (3a) since
this is not the environment where such a finite VP occurs. That is, the verb kept requires the
presence of a gerundive VP like putting his gold under the bathtub, and therefore imposes an
external constraint on VPs.
4.1.2 Notion of Head, Complements, and Modifiers
One important property we observe in English phrase-internal syntax is that in building up any
phrase, there is one obligatory element in each phrase. That is, each phrase has one essential
element as represented in the diagrams in (4):
(4) a.
NP

¸
¸
¸
¸
¸
¸
¸
. . . N
7654 0123
b.
VP

¸
¸
¸
¸
¸
¸
¸
V
7654 0123
. . .
c.
AP

¸
¸
¸
¸
¸
¸
¸
A
7654 0123
. . .
The circled element here is the essential, obligatory element within the given phrase. We call
this essential element the head of the phrase.
2
The head of each phrase thus determines its
‘projection’ into a larger phrasal constituent. The head of an NP is thus N, the head of a VP is
V, and the head of an AP is A.
The notion of headedness plays an important role in the grammar. For example, the verb
put, functioning as the head of a VP, dictates what it must combine with – two complements,
NP and PP. Consider other examples:
(5) a. The defendant denied the accusation.
b. *The defendant denied.
(6) a. The teacher handed the student a book.
b. *The teacher handed the student.
The verb denied here requires an NP object whereas handed requires two NP complements, in
this use. The properties of the head verb itself determine what kind of elements it will combine
with. As noted in the previous chapter, the elements which a head verb should combine with
are called complements. The complements include direct object, indirect object, predicative
complement, and oblique complement since these are all potentially required by some verb or
other.
The properties of the head become properties of the whole phrase. Why are the examples in
(7b) and (8b) ungrammatical?
2
See section 1.3 in Chapter 1 also.
50
(7) a. They [want to leave the meeting].
b. *They [eager to leave the meeting].
(8) a. The senators [know that the president is telling a lie].
b. *The senators [certain that the president is telling a lie].
The examples in (7b) and (8b) are unacceptable because of the absence of the required head.
The unacceptable examples lack a finite (tensed) VP as the bracketed part, but we know that
English sentences require a finite VP as their immediate constituent, as informally represented
as in (9):
(9) English Declarative Sentence Rule:
Each declarative sentence must contain a finite VP as its head.
Each finite VP is headed by a finite verb. If we amend the ungrammatical examples above to
include a verb but not a finite one, they are still ungrammatical:
(10) a. *They [(to) be eager to leave the meeting].
b. *The senators [(to) be certain that the president is telling a lie].
The VP is considered to be the (immediate) head of the sentence, with the verb itself as the head
of the VP. In this way, we can talk about a finite or non-finite sentence, one which is ultimately
headed by a finite or nonfinite verb, respectively.
3
In addition to the complements of a head, a phrase may also contain modifiers (or also called
adjuncts):
(11) a. Tom [
VP
[
VP
offered advice to his students] in his office].
b. Tom [
VP
[
VP
offered advice to his students] with love].
The PPs in his office or with love here provide further information about the action described
by the verb, but are not required as such by the verb. These phrases are optional and func-
tion as modifiers, and they function to augment the minimal phrase projected from the head
verb offered. The VP which includes this kind of modifier forms a maximal phrase. We might
say that the inner VP here forms a ‘minimal’ VP which includes all the ‘minimally’ required
complements, and the outer VP is the ‘maximal’ VP which includes optional modifiers.
What we have seen can be summarized as follows:
(12) a. Head: A lexical or phrasal element that is essential in forming a phrase.
b. Complement: A phrasal element that a head must combine with or a head select.
These include direct object, indirect object, predicative complement, and oblique
complement.
c. Modifier: A phrasal element not selected by the verb functions as a modifier to
the head phrase.
d. Minimal Phrase: A minimal phrase is the phrase including this head and all of
its complements.
3
See Chapter 5.2 for the detailed discussion of English verb form (VFORM) values including finite and nonfinite.
51
e. Maximal Phrase: A maximal phrase is the phrase that includes complements as
well as modifiers.
4.2 Differences between Complements and Modifiers
Given these notions of complements and modifiers, the question that follows is then how we can
distinguish between complements and modifiers. There are several tests to determine whether a
phrase is a complement or a modifier.
4
Obligatoriness: As hinted at already, complements are strictly-required phrases whereas mod-
ifiers are not. The examples in (13)–(15) show that the verb placed requires an NP and a PP as
its complements, kept an NP and a PP or an AP, and stayed a PP.
(13) a. John placed Kim behind the garage.
b. John kept him behind the garage.
c. *John stayed Kim behind the garage.
(14) a. *John placed him busy.
b. John kept him busy.
c. *John stayed him busy.
(15) a. *John placed behind the counter.
b. *John kept behind the counter.
c. John stayed behind the counter.
In contrast, modifiers are optional. Their presence is not required by the grammar:
(16) a. John deposited some money in the bank.
b. John deposited some money in the bank on Friday.
In (16b), the PP on Friday is optional here, serving as a modifier.
Iterability: The possibility of iterating identical types of phrase can also distinguish between
complements and modifiers. In general two or more instances of the same modifier type can
occur with the same head, but this is impossible for complements.
(17) a. *The UN blamed global warming [on humans] [on natural causes].
b. Kim and Sandy met [in Seoul] [in the lobby of the Lotte Hotel] in March.
In (17a) on humans is a complement and thus the same type of PP on natural causes cannot
co-occur. Yet in Seoul is a modifier and we can repeatedly have the same type of PP.
4
Most of the criteria we discuss here are adopted from Pollard and Sag (1987).
52
Do-so Test: Another reliable test often used to distinguish complements frommodifiers is the do
so or do the same thing test. As shown in (18), we can use do the same thing to avoid repetition
of an identical VP expression:
(18) a. John deposited some money in the checking account and Mary did the same thing
(too).
b. John deposited some money in the checking account on Friday and Mary did the
same thing (too).
What we can observe in (18b) is that the VP did the same thing can replace either the minimal
phrase deposited some money in the checking account or the maximal phrase including the
modifier on Friday. Notice that this VP can replace only the minimal phrase, leaving out the
modifier.
(19) John deposited some money in the checking account on Friday and Mary did the
same thing on Monday.
From these observations, we can draw the conclusion that if something can be replaced by do
the same thing, then it is either a minimal or a maximal phrase. This in turn means that this
‘replacement’ VP cannot be understood to leave out any complement(s). This can be verified
with more data:
(20) a. *John [deposited some money in the checking account] and Mary did the same
thing in the savings account.
b. *John [gave a present to the student] and Mary did the same thing to the teacher.
Here the PPs in the checking account and to the student are both complements, and thus they
should be included in the do the same thing phrase. This gives us the following informal gener-
alization:
(21) Do-so Replacement Condition:
The phrase do so or do the same thing can replace a verb phrase which includes at
least any complements of the verb.
This condition explains why we cannot have another locative complement phrase in the
savings account or to the teacher in (20). The unacceptability of the examples in (22) also
supports this generalization about English grammar:
(22) a. *John locked Fido in the garage and Mary did so in the room.
b. *John ate a carrot and Mary did so a radish.
Constancy of semantic contribution: An adjunct can cooccur with a relatively broad range of
heads whereas a complement is typically limited in its distribution. Note the following contrast:
(23) a. Kim camps/jogs/mediates on the hill.
53
b. Kim jogs on the hill/under the hill/over the hill.
(24) a. Kim depends/relies on Sandy.
b. Kim depends on Sandy/*at Sandy/*for Sandy.
The semantic contribution of the adjunct on the hill in (23a) is independent of the head whereas
that of the complement on Sandy is idiosyncratically dependent upon the head.
Structural Difference: We could distinguish complements and modifiers by tree structures,
too: complements combine with a lexical head (not a phrase) to form a minimal phrase whereas
modifiers combine with a phrase to form a maximal phrase. This means that we have structures
of the following forms:
(25)
XP

¸
¸
¸
¸
¸
¸
¸
XP

¸
¸
¸
¸
¸
¸
¸
Modifier
X Complement(s)
As represented in the tree structures, complements are sisters of the lexical head X, whereas
modifiers are sisters of a phrasal head. This structural difference between complements and
modifiers provides a clean explanation for the contrast in do-so test. Given that the verb ate
takes only an NP complement whereas put takes an NP and a PP complement, we will have the
difference in the two structures shown in (26):
(26) a.
VP
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
_
_
VP
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
PP

¸
¸
¸
¸
¸
¸
V NP

¸
¸
¸
¸
¸
¸
in the room
ate some food
b.
VP
.
.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
_
_
V NP

¸
¸
¸
¸
¸
¸
PP

¸
¸
¸
¸
¸
¸
put the money in the room
In this way, we represent the difference between complements and modifiers.
54
Ordering Difference: Another difference that follows from the structural distinction between
complements and modifiers is an ordering difference. As a complement needs to combine with
a lexical head first, modifiers follow complements:
(27) a. John met [a student] [in the park].
b. *John met [in the park] [a student].
A similar contrast can be observed in the following contrast:
(28) a. the student [of linguistics] [with long hair]
b. *the student [with long hair] [of linguistics]
The PP with long hair is a modifier whereas the of linguistics is the complement of student. This
is why with long hair cannot occur between the head student and its complement of linguistics.
5
As such, observed ordering restrictions can provide more evidence for the distinction be-
tween complements and modifiers.
4.3 PS Rules, X

-Rules, and Features
We have seen in Chapter 2 that PS rules can describe how English sentences are formed. How-
ever, two main issues arise with the content of PS rules.
6
The first is related to the headedness
of each phrase, often called the ‘endocentricity’ property of each phrase.
Let us consider the PS rules that we saw in the previous chapters. We have seen that PS
rules such as those in (29) can characterize well-formed phrases in English, together with an
appropriate lexicon:
(29) a. S →NP VP
b. NP →Det AdjP

N
c. VP →V (NP) (VP)
d. VP →V NP AP
e. VP →V NP NP
f. VP →V S
g. AP →A VP
h. PP →P NP
i. VP →Adv VP
One common property of all these rules is, as we have discussed, that every phrase has its own
head. In this sense, each phrase is the projection of a head, and thereby has the endocentricity.
However, we can ask the theoretical question of whether or not we can have rules like the
following, in which the phrase has no head at all:
(30) a. VP →P NP
5
See (49) for the structural differences between with long hair and of linguistics.
6
The discussion of this section is based on Sag et al. (2003).
55
b. NP →PP S
Nothing in the grammar makes such PS rules unusual, or different in any way from the set in
(29). Yet, if we allow such ‘non-endocentric’ PS rules in which a phrase does not have a lexical
head, the grammar would then be too powerful to generate only the grammatical sentences of
the language.
Another limit that we can find from the simple PS rules concerns an issue of redundancy.
Observe the following:
(31) a. *The problem disappeared the accusation.
b. The problem disappeared.
(32) a. *The defendant denied.
b. The defendant denied the accusation.
(33) a. *The boy gave the book.
b. The boy gave the baby the book.
What these examples show is that each verb has its own requirement for its complement(s).
For example, deny requires an NP, whereas disappear does not, and gave requires two NPs as
its complements. The different patterns of complementation are said to define different subcate-
gories of the type verb. The specific pattern of complements is known as the ‘subcategorization’
requirement of each verb, which can be represented as following (IV: intransitive, TV: transitive,
DTV: ditransitive):
(34) a. disappear: IV,
b. deny: TV, NP
c. give: DTV, NP NP
In addition, in order to license the grammatical sentences in (31)–(33), we need to have the
following three VP rules:
(35) a. VP →IV
b. VP →TV NP
c. VP →DTV NP NP
We can see here that in each VP rule, only the appropriate verb can occur. That is, a DTV cannot
form a VP with the rules in (35a) or (35b): It forms a VP only according to the last PS rule.
Each VP rule thus also needs to specify the kind of verb that can serve as its head.
Taking these all together, we see that a grammar of the type just suggested must redundantly
encode the subcategorization information both in the lexical type of each verb (e.g., DTV) and
in the PS rule for that type of verb.
A similar issue of redundancy arises in accounting for subject-verb agreement:
(36) a. The bird devours the worm.
b. The birds devour the worm.
56
To capture the fact that the subject NP agrees with the predicate VP, we need to differentiate the
S rule into the following two:
(37) a. S →NP
sing
VP
sing
(for (36)a)
b. S →NP
pl
VP
pl
(for (36)b)
The two PS rules ensures that the singular (sing) subject combines with a singular VP whereas
the plural (pl) subject NP with a plural VP.
Descriptively, there is no problem with a grammar with many specific parts. From a theoret-
ical perspective, though, we have a concern about the the endocentricity and redundancy issues.
A more particular related question is that of how many PS rules English has. For example, how
many PS rules do we need to characterize English VPs?—Presumably there are as many rules
as there are subcategories of verb.
We need to investigate the abstract content of PS rules, in order to develop a theoretical view
of them. For example, it seems to be the case that each PS rule must have a ‘head’. This will
disallow many possible PS rules which we can write using the rule format, from being actual
rules of any language.
In order to understand more about the structures that rules describe, we need two more no-
tions, ‘intermediate category/phrase’ and ‘specifier (SPR)’. We motivate the idea of the inter-
mediate category, and then specifier is a counterpart of it. Consider the examples in (38):
(38) a. Every photo of Max and sketch by his students appeared in the magazine.
b. No photo of Max and sketch by his students appeared in the magazine.
What are the structures of these two sentences? Do the phrases every photo of Max and sketch
by his students form NPs? It is not difficult to see sketch by his students is not a full NP by itself,
for if it was, it should be able to appear as subject by itself:
(39) *Sketch by his students appeared in the magazine.
In terms of the semantic units, we can assign the following structures to the above sentences, in
which every and no operate over the meaning of the rest of the phrase:
(40) a. [Every [[photo of Max] and [sketch by his students]]] appeared in the magazine.
b. [No [[photo of Max] and [sketch by his students]]] appeared in the magazine.
The expression photo of Max and sketch by his students are phrasal elements but not full NPs
— so what are they? We call these ‘intermediate phrases’, notationally represented as N-bar or
N

. The phrase N

is thus intuitively bigger than a noun, but smaller than a full NP, in the sense
that it still requires a determiner from the class the, every, no, some, and the like.
The complementary notion that we introduce at this point is ‘specifier’ (SPR), which can
include the words just mentioned as well as phrases, as we illustrate in (41):
(41) a. [the enemy’s] [
N
destruction of the city]
b. [The enemy] [
VP
destroyed the city].
57
The phrase the enemy’s in (41a) and the subject the enemy in (41b) are semantically similar
in the sense that they complete the specification of the event denoted by the predicate. These
phrases are treated as the specifiers of N

and of VP, respectively.
As for the possible specifiers of N

, observe the following:
(42) a. a little dog, the little dogs (indefinite or definite article)
b. this little dog, those little dogs (demonstrative)
c. my little dogs, their little dog (possessive adjective)
d. every little dog, each little dog, some little dog, either dog, no dog (quantifying)
e. my friend’s little dog, the Queen of England’s little dog (possessive phrase)
The italicized expressions here all function as the specifier of N

. However, notice that though
most of these specifiers are determiners, some consist of several words as in (42e) (my friend’s,
the Queen of England’s) . This motivates us to introduce the new phrase type DP (determiner
phrase) that includes the possessive phrase (NP + ’s) as well as determiners. This new phrase
then will give us the generalization that the specifier of N

is a DP.
7
Now let us compare the syntactic structures of (41a) and (41b):
(43)
NP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
DP

¸
¸
¸
¸
¸
¸
¸
N

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
the enemy’s N PP

¸
¸
¸
¸
¸
¸
¸
destruction of the city
(44)
S

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
NP

¸
¸
¸
¸
¸
¸
¸
VP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
The enemy V
·
·
·
·
·
·
·
¸
¸
¸
¸
¸
¸
¸
NP

¸
¸
¸
¸
¸
¸
destroyed the city
Even though the NP and S are different phrases, we can notice several similarities. In the NP
structure, the head N destruction combines with its complement and forms an intermediate
7
Some analyses take the whole expression in (43) to be a DP (e.g., a little dog, my little dogs) in which expressions
like little dog is not an N

but an NP.
58
phrase N

which in turn combines with the specifier DP the enemy’s. In the S structure, the head
V combines with its complement the city and forms a VP. This resulting VP then combines with
the subject the enemy, which is also a specifier. In a sense, the VP is an intermediate phrase that
requires a subject in order to be a full and complete S.
Given these similarities between NP and S structures, we can generalize over them as in (45),
where X is a variable over categories such as N, V, P, and other grammatical categories:
8
(45)
XP

¸
¸
¸
¸
¸
¸
¸
Specifier X

¸
¸
¸
¸
¸
¸
¸
X Complement(s)
This structure in turn means the grammar now includes the following two rules:
9
(46) a. XP →Specifier, X

(Head-Specifier Rule)
b. XP →X, YP

(Head-Complement Rule)
These Head-Specifier and Head-Complement Rules, which form the central part of ‘X

-theory’,
account for the core structure of NP as well as that of S. In fact, these two general rules can also
represent most of the PS rules we have seen so far. In addition to these two, we just need one
more rule:
10
(47) XP →Modifier, X

(Head-Modifier Rule)
This Head-Modifier Rule allows a modifier to combine with its head as in the PS rule VP →
VP Adv/PP, as presented in the following:
(48)
XP

¸
¸
¸
¸
¸
¸
¸
Specifier X

¸
¸
¸
¸
¸
¸
¸
X

¸
¸
¸
¸
¸
¸
¸
Modifier
X Complement(s)
8
We can assume that the head of S is VP and that VP is an intermediate phrase in the sense that it still requires a
subject as its specifier.
9
Unlike the PS rules we have seen so far, the rules here are further abstracted, indicated by the comma notation
between daughters on the right-hand side. We assume that the relative linear order of a head and complements etc. is
determined by a combination of general and language-specific ordering principles, while the hierarchical X

-structures
themselves are universal.
10
The comma indicates that the modifier can appear either before the head or after the head as in always read books
or read books always.
59
One thing to notice in the Head-Complement Rule is that the head must be a lexical el-
ement. This in turn means that we cannot apply the Head-Modifier Rule first and then the
Head-Complement Rule. This explains the following contrast:
(49) a. the king [of Rock and Roll] [with a hat]
b. *the king [with a hat] [of Rock and Roll]
The badness of (49b) is due to the fact that the modifier with a hat is combined with the head
king first.
(50) a.
NP
.
.
.
.
.
.
.
.
.
.
.
.
¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸
DP N

.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
_
the
N

.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
_
PP

¸
¸
¸
¸
¸
¸
N
PP

¸
¸
¸
¸
¸
¸
¸
¸
with a hat
king of Rock and Roll
b.
NP
.
.
.
.
.
.
.
.
.
.
.
.
¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸
DP *N

.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
_
the
N

.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
_
PP

¸
¸
¸
¸
¸
¸
¸
¸
N
PP

¸
¸
¸
¸
¸
¸
of Rock and Roll
king with a hat
We can observe in (50b) that the combination of king with with a hat forms an N

, but the com-
bination of the complement of Rock and Roll with this N

will not satisfy the Head-Complement
Rule.
The existence and role of the intermediate phrase N

, which is a larger than a lexical cate-
gory but still not a fully-fledged phrase, can be further supported from the pronoun substitution
examples in (51):
(51) a. The present king of country music is more popular than the last one.
b. *The king of Rock and Roll is more popular than the one of country music.
Why do we have the contrast here? One simple answer is that the pronoun one here replaces an
N

but not an N or an NP. This will also account for the following contrast, too:
60
(52) A: Which student were you talking about?
B: The one with long hair.
B: *The one of linguistics with long hair.
The phrase of linguistics is the complement of student. This means the N-bar pronoun one
should include this. However, the modifier with long hair cannot be within the N

.
There are several more welcome consequences that the three X

rules that bring to us. The
grammar rules can account for the same structures as all the PS rules we have seen so far: with
those rules we can identify phrases whose daughters are a head and its complement(s), or a
head and its specifier, or a head and its modifier. The three X

rules thereby greatly minimize
the number of PS rules that need to characterize well-formed English sentences.
In addition, these X

rules directly solve the endocentricity issue, for they refer to ‘Head’.
Assume that X is N, then we will have N, N

, and NP structures. We can formalize this more
precisely by introducing the feature POS (part of speech), which has values such as noun, verb,
adjective. The structure (53) shows how the values of the features in different parts of a structure
are related:
(53)
XP[POS 1 ]

¸
¸
¸
¸
¸
¸
¸
Specifier X

[POS 1 ]

¸
¸
¸
¸
¸
¸
¸
X

[POS 1 ]

¸
¸
¸
¸
¸
¸
¸
Modifier
X[POS 1 ] Complement(s)
The notation 1 shows that whatever value the feature has in one place in the structure, it has
the same value somewhere else. This is a representational tag, in which the number 1 has no
significance: it could as easily be 7 or 437 . We provide more details of the formal feature system
in the following section.
So (53) indicates that the phrase’s POS value is identical to its head daughter, capturing
the headedness of each phrase: the grammar just does not allow any phrase without a head.
The redundancy issue mentioned above for agreement is now a matter of introducing another
feature, NUMBER. That is, with the new feature NUMBER, with values singular and plural,
we can add a detail to the Head-Specifier Rule as following:
(54) XP →Specifier[NUMBER 1 ], X

[NUMBER 1 ]
The rule states that the subject’s NUMBER value is identical with that of the predicate VP’s
NUMBER value. The two rules in (37) are both represented in (54).
61
4.4 Lexicon and Feature Structures
In the previous section, we have seen that the properties of a lexical head determine the compo-
nents of the minimal phrase, in terms of complements, and that other properties of the head are
directly properties of the phrase. This information is encoded in a lexical entry, for each word
in the lexicon.
Every lexical entry at least includes phonological (but in practice, orthographic), morpho-
logical, syntactic, and semantic information. For example, the word puts will have at least the
following information:
(55) Minimal Lexical Information for puts:
a. phonological information: <puts>
b. syntactic information: verb, finite, 3rd singular
c. argument information: <agent
i
, theme
j
, location
k
>
d. semantic information: put

(i,j,k)
The phonological information is the information about how the word is pronounced; the syn-
tactic information indicates that this particular word is a verb and is in the 3rd singular present
(finite) form. The argument structure represents the number of arguments which the verb se-
lects, to indicate the participants that are minimally involved in the event expressed by the verb.
This argument information is linked to its more precise meaning as indicated by the indexes i,
j and k. These indexes refer to the participants denoted by the arguments. Finally, the semantic
structure represents that the verb’s meaning relates three participants – someone i who is doing
the action of putting, something j being put in a place, and someplace k it is put in. These lexi-
cal entries can be represented in a more systematic and precise way with the system of feature
structures, which we now introduce.
4.4.1 Feature Structures and Basic Operations
Most modern grammars rely on a representation of lexical information in terms of features
and their values.
11
We present here a formal and explicit way of representing it with feature
structures. Each feature structure is an attribute-value matrix (AVM):
(56)
_
¸
¸
¸
_
Attribute1 value1
Attribute2 value2
Attribute3 value3
. . . . . .
_
¸
¸
¸
_
The value of each attribute can be an atomic element, a list, a set, or a feature structure:
11
In particular, grammars such as Head-driven Phrase Structure Grammar (HPSG) and Lexical Functional Grammar
(LFG) are couched upon mathematically well-defined feature structure systems. The theory developed in this textbook
heavily relies upon the feature structure system of HPSG. See Sag et al. (2003).
62
(57)
_
¸
¸
¸
¸
¸
¸
¸
_
Attribute1 atomic
Attribute2
Attribute3
_ _
Attribute4
_
. . .
_
_
¸
¸
¸
¸
¸
¸
¸
_
One important property of every feature structure is that it is typed.
12
That is, each feature
structure is relevant only for a given type. A simple illustration can show why feature structure
needs to be typed. The upper left declaration in italics is the type of the feature structure:
(58) a.
_
¸
_
university
NAME kyunghee univ.
LOCATION seoul
_
¸
_
b.
*
_
¸
_
university
NAME kyunghee univ.
MAJOR linguistics
_
¸
_
The type university may have many properties, including its name and location, but having a
MAJOR, as a subject of study, is inappropriate. In the linguistic realm, we might declare that
TENSE is appropriate only for verb, for example.
Now consider the following example of a typed feature structure, information about one
author of this book:
(59)
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
author
NAME kim
SONS Edward, Richard
HOBBIES
_
swimming, jogging, reading, . . .
_
ADVANCED-DEGREE
_
¸
_
FIELD linguistics
AREA syntax-semantics
YEAR 1996
_
¸
_
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
This illustrates the different types of values that attributes (feature names) may have. Here, the
value of the attribute NAME is atomic, whereas the value of SONS is a list which represents
something relative about the two values, in this case that one is older than the other. So, for
example ‘youngest son’ would be the right-most element in the list value of SONS. Meanwhile,
the value of HOBBIES is a set, showing that there is no significance in the relative ordering.
Finally, the value of the feature ADVANCED-DEGREE is a feature structure which in turn has
three attributes.
One useful notion in the feature structure is structure-sharing, which we have already seen
above in terms of the 1 notation (see (53)). This is to represent cases where two features (or
12
Even though every feature structure is typed in the present grammar, we will not specify the type of each feature
structure unless it is necessary for the discussion.
63
attributes) have an identical value:
(60)
_
¸
¸
¸
¸
¸
¸
¸
¸
_
individual
NAME kim
TEL 1
SONS
_
_
¸
_
individual
NAME richard
TEL 1
_
¸
_
,
_
¸
_
individual
NAME edward
TEL 1
_
¸
_
_
_
¸
¸
¸
¸
¸
¸
¸
¸
_
For the type individual, attributes such as NAME and TEL and SONS are appropriate. (60)
represents a situation in which the particular individual kimhas two sons, and their TEL attribute
has a value which is the same as the value of his TEL attribute, whatever the value actually is.
In addition to this, the notion of subsumption is also important in the theoretical use of
feature structures; the symbol represents subsumption. The subsumption relation concerns
the relationship between a feature structure with general information and one with more specific
information. In such a case, the general one subsumes the specific one. Put differently, a feature
structure A subsumes another feature structure B if A is not more informative than B.
(61)
A:
_
individual
NAME kim
_
B:
_
¸
_
individual
NAME kim
TEL 961-0892
_
¸
_
In (62), A represents more general information than B. This kind of subsumption relation is
used to represent ‘partial’ information, for in fact we cannot represent the total information
describing all possible worlds or states of affairs. In describing a given phenomenon, it will be
more than enough just to represent the particular or general aspects of the facts concerned. Each
small component of feature structure will provide partial information, and as the structure is
built up, the different pieces of information are put together.
The most crucial operation in feature structures is unification, represented by

. Feature
unification means that two compatible feature structures are unified, conveying more coherent
and rich information. Consider the feature structures in (62); the first two may unify to give the
third:
(62)
_
individual
NAME kim
_

_
individual
TEL 961-0892
_

_
¸
_
individual
NAME kim
TEL 961-0892
_
¸
_
The two feature structures are unified, resulting in a feature structure with both NAME and TEL
information. However, if two feature structures have incompatible feature values, they cannot
be unified:
64
(63)
_
individual
NAME edward
_

_
individual
NAME richard
_

*
_
¸
_
individual
NAME edward
NAME richard
_
¸
_
Since the two smaller feature structures have different NAME values, they cannot be unified.
Unification will make sure that information is consistent as it is built up in the analysis of a
phrase or sentence.
4.4.2 Feature Structures for Linguistic Entities
Any individual or entity including a linguistic expression can be represented by a feature struc-
ture. For example, the word puts, whose general type is verb, can have a feature structure like
the following:
13
(64)
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
verb
PHON puts
SYN
_
_
POS verb
VFORM fin
_
_
ARG-ST
_
[agt]
i
, [th]
j
, [loc]
k
_
SEM
_
¸
¸
¸
¸
¸
_
PRED put-relation
AGENT i
THEME j
LOCATION k
_
¸
¸
¸
¸
¸
_
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
This feature structure has roughly the same information as the informal representation in (55).
The verb puts, like any verb, has its own phonological (PHON) value, syntactic (SYN), ar-
gument structure (ARG-ST), and semantic (SEM) information. The SYN attribute indicates
that the POS (part of speech) value is verb, that it has a finite verbal inflectional form value
(VFORM).The ARG-ST attribute indicates that the verb selects for three arguments (with the-
matic roles agent, theme, location), which will be realized as the subject and two complements
in the full analysis. The SEM feature represents the information this verb denotes the predicate
relation put-relation, whose three participants are linked to the elements in the ARG-ST via the
indexing values i, j, and k.
One thing to note here is that since there are some cases where we have difficulties in assign-
ing a specific named semantic role to a selected argument discussed in Chapter 3, we typically
just indicate the number of arguments each predicate is selecting in ARG-ST: we underspecify
13
Later on, we will not represent the PHON and SEM values unless relevant to the discussion at hand.
65
the information unless it is necessary to show more details. So, for example, verbs like smile,
devour and give will have the following ARG-ST representations, respectively:
(65) a.
_
ARG-ST [ ]
_
b.
_
ARG-ST [ ], [ ]
_
c.
_
ARG-ST [ ], [ ], [ ]
_
One-place predicates like smile select for just one argument, two-place predicates like devour
take two arguments, and three-place predicates take three arguments. Eventually, the arguments
selected by each predicate are linked to grammatical functions, to the core semantic properties,
and to other parts of the representation of the grammatical properties.
4.4.3 Argument Realization
Each element on the ARG-ST list is realized as SPR (specifier) or COMPS (complements),
through one of the rules in (46).
14
In general, the basic pattern is that the first element on the list
is realized as subject and the rest as complements:
(66) Argument Realization Constraint (ARC, first approximation):
The first element on the ARG-ST list is realized as SPR, the rest as COMPS in
syntax.
This realization is obligatory in English; for example, the three arguments of put are realized as
subject and complements, with the putter (agent) as subject:
15
(67) a. John put the book in the box.
b. *John put in the box.
c. *In the box put John the book.
d. #The book put John in the box.
We see that the arguments selected by a lexical head should be all realized as SPR and COMPS,
which are combined in the notion of valence (VAL) features.
16
More formally, we can represent
this constraint as applied to put as the following feature structure:
(68)
_
¸
¸
¸
_
VAL
_
_
SPR 1 NP
COMPS 2 NP, 3 PP
_
_
ARG-ST 1 , 2 , 3
_
¸
¸
¸
_
The boxed tags show the different identities in the overall structure. For example, the first el-
ement of ARG-ST and of SPR have the boxed tag 1 , ensuring that the two are identical. The
14
Once again, remember that the term SPR includes subject as well as the noun’s specifier.
15
The notation # indicates that the structure is technically well-formed from a syntactic perspective, but semantically
anomalous.
16
The term ‘valence’ refers to the number of arguments that a lexical item can combine with, to make a syntactically
well-formed sentence, often along with a description of the categories of those constituents. It is inspired by the notion
of valence as used in atomic theory in chemistry.
66
general ARC constraint blocks examples like (67c) in which the location argument is realized
as the subject, as shown in (69):
(69) *
_
¸
¸
¸
_
VAL
_
_
SPR 3 PP
COMPS 1 NP, 2 NP
_
_
ARG-ST 1 , 2 , 3
_
¸
¸
¸
_
This violates the ARC, which requires the first element of ARG-ST be realized as the SPR (the
subject of a verb or the specifier of a noun).
Notice that the arguments can be realized into different categories, depending on the proper-
ties of the given verb:
(70) a. The election results surprised everybody.
b. That he won the election surprised everybody.
The data indicate that verbs like surprise will take two arguments, but the first argument can be
realized either as an NP subject as in (70a) or a CP subject as in (70b). This difference in the
argument realization can be represented as the following, respectively:
(71) a.
_
¸
¸
¸
_
VAL
_
_
SPR 1 NP
COMPS 2 NP
_
_
ARG-ST 1 , 2
_
¸
¸
¸
_
b.
_
¸
¸
¸
_
VAL
_
_
SPR 1 CP
COMPS 2 NP
_
_
ARG-ST 1 , 2
_
¸
¸
¸
_
Though there is no difference in terms of the number of arguments that surprise select, the
arguments can be realized in a different phrase. As the book goes on, we will see how the
argument realization is further constrained by the lexical properties of the verb in question or
by other grammatical components.
4.4.4 Verb Types and Argument Structure
As mentioned earlier, lexical elements in the classes V, A, N, and P, select one or more comple-
ment(s) to form a minimal phrase. With the construct of ARG-ST, we know that every lexical
element has ARG-ST information which will be realized in surface form through the SPR and
COMPS values. Verb types can be differentiated by looking only at the COMPS value since
every verb will have one SPR (subject) element. This is exactly the way that verbs are differen-
tiated using the traditional notion of subcategorization.
Intransitive: This is a type of verb that does not have any COMPS:
(72) a. John disappeared.
b. *John disappeared Bill.
67
(73) a. John sneezed.
b. *John sneezed the money.
These verbs have no COMPS element—the list is necessarily empty. Such a verb will have just
one argument that is realized as subject:
(74)
_
¸
¸
¸
¸
¸
¸
_
disappear
VAL
_
_
SPR 1 NP
COMPS
_
_
ARG-ST 1
_
¸
¸
¸
¸
¸
¸
_
Linking verbs: Verbs such as look, seem, remain, and feel require different complements that
are typically of category AP:
(75) a. The president looked [weary].
b. The teacher became [tired of the students].
c. The lasagna tasted [scrumptious].
d. John remained [somewhat calm].
e. The jury seemed [ready to leave].
These verbs also can select other phrases (here, NP):
(76) a. John became a success.
b. John seemed a fool.
c. John remained a student.
Though each verb may select different types of phrases, they all at least select a predicative
complement, where a property is ascribed to the subject. (Compare John remained a student
and John revived a student.) This subcategorization requirement can be represented as follows:
(77)
_
¸
¸
¸
¸
¸
¸
_
become
VAL
_
_
SPR 1 NP
COMPS 2 AP/NP[PRD +]
_
_
ARG-ST 1 , 2
_
¸
¸
¸
¸
¸
¸
_
This kind of verb selects two arguments: one is canonically an NP to be realized as the subject
and the other is any phrase (XP) that can function as a predicate (PRD +) (see also the examples
in (84)). Of course, this presupposes an accurate characterization of which phrases can be [PRD
+], which we simply assume here.
Transitive verbs: Unlike linking verbs, pure transitive verbs select a referential, non-predicative
NP as their complement, functioning as direct object:
(78) a. John saw Fred.
b. Alice typed the letter.
68
c. Clinton supported the health care bill.
d. Raccoons destroyed the garden.
Such verbs will have the following lexical information:
(79)
_
¸
¸
¸
¸
¸
¸
_
destroy
VAL
_
_
SPR 1 NP
COMPS 2 NP
_
_
ARG-ST 1 , 2
_
¸
¸
¸
¸
¸
¸
_
Ditransitive: There are also ‘ditransitive’ verbs that require IO and DO:
(80) a. The school board leader asked the students a question.
b. The parents bought the children non-fiction novels.
c. John taught new students English Syntax.
Such verbs have three arguments: one subject and two complement NPs functioning as indi-
rect and direct object, respectively:
(81)
_
¸
¸
¸
¸
¸
¸
_
teach
VAL
_
_
SPR 1 NP
COMPS 2 NP, 3 NP
_
_
ARG-ST 1 , 2 [goal], 3 [theme]
_
¸
¸
¸
¸
¸
¸
_
These verbs typically have a related realization with the goal argument realized as a dative
PP complement:
(82) a. The school board leader asked a question of the students.
b. The parents bought non-fiction novels for the children.
c. John taught English Syntax to new students.
In this realization, the second argument has the theme role while the third one has the goal role.
(83)
_
¸
¸
¸
¸
¸
¸
_
teach
VAL
_
_
SPR 1 NP
COMPS 2 NP, 3 PP
_
_
ARG-ST 1 , 2 [theme], 3 [goal]
_
¸
¸
¸
¸
¸
¸
_
Complex Transitive: There is another type of transitive verb which selects two complements,
one functioning as a direct object and the other as a predicative phrase (NP, AP, or VP), describ-
ing the object:
(84) a. John regards Bill as a good friend.
b. The sexual revolution makes some people uncomfortable.
c. Ad agencies call young people Generation X-ers.
69
d. Historians believe FDR to be our most effective president.
In (84a), the predicative PP as a good friend follows the object Bill; in (84b), the AP uncom-
fortable serves as a predicate phrase of the preceding object some people. In (84c), the NP
Generation X-ers is the predicative phrase. In (84d), the predicative phrase is an infinitive VP.
Just like linking verbs, these verbs require a predicative ([PRD +]) XP as complement:
(85)
_
¸
¸
¸
¸
¸
¸
_
call
VAL
_
_
SPR 1 NP
COMPS 2 NP, 3 XP
_
_
ARG-ST 1 , 2 , 3 [PRD +]
_
¸
¸
¸
¸
¸
¸
_
This means that the verbs in (84) all select an object NP and an XP phrase that function as a
predicate.
Even though these five types of verb that we have seen so far represent many English verb
types, there are other verbs that do not fit into these classes; for instance, the use of the verb
carry in (86).
(86) a. *John carried to the door.
b. John carried her on his back.
The examples in (86) illustrate that carried requires an NP and a PP, as represented in the feature
structure:
(87)
_
¸
¸
¸
¸
¸
¸
_
carry
VAL
_
_
SPR 1 NP
COMPS 2 NP, 3 PP
_
_
ARG-ST 1 [agt], 2 [th], 3 [loc]
_
¸
¸
¸
¸
¸
¸
_
The PP here cannot be said to be predicate of the object her; it denotes the location to which
John carries her.
Of course, there are various other verb types that we have not described here, in terms of
complementation patterns. As the book goes on, we will see yet more different types of verbs.
70
4.5 Exercises
1. For each of the verbs below, use it in a grammatical example; write down the example
and provide its lexical entry including the COMPS value.
(i) appear, consider, award, borrow, owe, explain, introduce, discuss, tell, say
2. Provide tree structures for the following pairs of sentences while checking the gram-
matical function of each phrase with valid distributional tests. In doing so, state what
differences we need to represent for the underlined expressions.
(i) a. Tom locked Fido in the garage.
b. Tom bathed Fido in the garage.
(ii) a. Tom placed it under the table.
b. Tom played it under the table.
(iii) a. I wonder if you will come back tomorrow.
b. You would have a reply if you come back tomorrow.
3. For each example below, draw its structure and then give the lexical entry (SPR and
COMPS value) of the main verb.
(i) a. Tom hid the manuscript in the cupboard.
b. Fred hired Sharon to change the oil.
c. They pushed the prisoners into the truck.
d. Frank hopes to persuade Harry to make the cook wash the dishes.
e. George mailed the attorney his photograph of the accident.
f. Tom keeps asking Karen’s sister to buy the car.
g. Jane left the book on the table.
h. We have not confirmed whether the flight had been booked.
i. We saw him beaten by the champion.
j. They confined his remarks to the matter under discussion.
4. The verbs in the following examples are used incorrectly. Correct the errors or replace
the verb with another one, and write out each new example. In addition, provide the
COMPS value for each verb (in its use in your grammatical examples).
(i) a. *Oliver ascribed his longevity there.
b. *Oliver mentioned Charles the problem.
c. *Oliver fined ten pounds to the prisoner.
d. *Oliver drove me a lunatic.
e. *Oliver addressed the king the letter.
f. *Oliver absented his brother from the meeting.
5. Draw tree structures for the following two sentences. In particular, provide detailed NP
structures using the intermediate phrase N

:
(i) a. The students of English from Seoul faced many issues in the process of
interpreting, transcribing, and editing the poems.
71
b. The love of my life and father of my children would never do such a thing.
c. The museum displayed no painting by Miro or drawing by Klee.
d. By law, every dog and cat in the area has to be spayed or neutered.
6. Read the following texts and provide the ARG-ST, SPR and COMPS values of the un-
derlined expressions.
(i) Learning to use a language freely and fully is a lengthy and effortful pro-
cess. Teachers cannot learn the language for their students. They can set
their students on the road, helping them to develop confidence in their own
learning powers. Then they must wait on the sidelines, ready to encourage
and assist.
(ii) Deep ecologists put a reign on human exploitation of natural “resources”
except to satisfy vital needs. Thus, the use of a field by an African tribe to
growgrain for survival is an example of a vital need whereas the conversion
of a swamp to an exclusive golf course would not. Rest assured that much
of the mining, harvesting, and development of our technological age would
not meet the requirement of this principle. Rather than being concerned
about how to raise automobile production, this ethic would be interested
in solving the problem of human mobility in a way that would not require
the disruption of highways, roads, and parking lots. It rebels against an
industrialist world view: “Before it is possessed and used, every plant is a
weed and every mineral is just another rock.”
17
17
From http://www.unitedearth.com.au/deepecology.html
72
5
More on Subjects and Complements
5.1 Grammar Rules and Principles
As we have seen in the previous chapter, the arguments in ARG-ST are realized as the syntactic
elements SPR (subject of a verb and determiner of a noun) and COMPS. The X

rules control
their combination with a relevant head:
(1) a. XP →Specifier, Head
b. XP →Head, Complement(s)
c. XP →Modifier, Head
The rule (1a) represents the case where a head combines with its specifier (e.g., a VP with
its subject and an N

with its determiner), whereas (1b) says that a head combines with its
complement(s) and forms a phrase. (1c) allows a combination of a head with its modifier. As
noted earlier, in order to guarantee that the head’s POS (part of speech) value is identical with
its mother phrase, we need to introduce the category variable X and the feature POS:
(2) a. Head-Specifier Rule:
XP
_
POS 1
_
→Specifier, X

[POS 1 ]
b. Head-Complement Rule:
XP
_
POS 1
_
→X[POS 1 ], Complement(s)
c. Head-Modifier Rule:
XP
_
POS 1
_
→Modifier, XP
_
POS 1
_
The POS feature is thus a head feature which passes up to a ‘mother’ phrase from its head
‘daughter’, as shown in (3):
(3)
VP[POS verb]

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
V[POS verb]
¸¸
PP
73
This percolation from a head to its mother is ensured by the following Head Feature Principle:
(4) The Head Feature Principle (HFP):
A phrase’s head feature (e.g., POS, VFORM, etc.) is identical with that of its head.
The HFP thus ensures that every phrase has its own lexical head with the identical POS value.
The HFP will apply to any features that we declare to be ‘head features’, VFORMbeing another.
The grammar does not allow hypothetical phrases like the following:
(5)
*VP[POS verb]
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
A[POS adj] PP
We have not yet spelled out clearly what ensures a lexical head to combine not just with one
of its complements but with all of its COMPS elements. Consider the following examples:
(6) a. Kim put the book in the box.
b. *Kim put the book.
c. *Kim put in the box.
As seen from the contrast here and as noted in the lexical entry in (7), the verb put selects two
complements and must combine with all of its complements.
(7)
_
¸
¸
¸
_
HEAD| POS verb
VAL
_
_
SPR NP
COMPS NP, PP
_
_
_
¸
¸
¸
_
We can also see that a finite verb must combine with its subject:
(8) a. *Is putting the book in the box.
b. *Talked with Bill about the exam.
Such combinatorial requirements can be formally stated in the revised grammar rules as given
in (9):
(9) a. Head-Specifier Rule:
XP
_
SPR
_
→ 1 , H[SPR 1 ]
b. Head-Complement Rule:
XP
_
COMPS
_
→H[COMPS 1 ,. . . , n ], 1 ,. . . , n
c. Head-Modifier Rule:
XP →[MOD 1 ], 1 H
The grammar rules here are well-formedness conditions on possible phrases in English, indi-
cating what each head combines with and then what happens as the result of the combination.
74
For example, in (9a) when a head, requiring a SPR, combines with it, we have a well-formed
head-specifier phrase with the SPR value discharged; in (9b), a head combines with all of its
COMPS value, it forms a Head-Complement phrase with no further COMPS value; in (9c),
when a modifier combines (carrying the MOD feature) with the head it modifies, the resulting
phrase forms a well-formed head-modifier phrase.
1
These three grammar rules, interacting with the general principles such as the HFP, license
grammatical sentences in English. Let us consider one example in a little more detail:
2
(10)
S
_
¸
¸
¸
_
HEAD 4 | POS verb
SPR
COMPS
_
¸
¸
¸
_

_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
Subj
H
1 NP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
VP
_
¸
¸
¸
_
HEAD 4 | POS verb
SPR 1
COMPS
_
¸
¸
¸
_
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
H
C
C
Kim
V
_
¸
¸
¸
_
HEAD 4 | POS verb
SPR 1
COMPS 2 , 3
_
¸
¸
¸
_
2 NP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
3 PP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
deposits some money in the bank
The finite verb deposits selects a subject (a specifier) and two complements. The HFP ensures
that the head feature POS values of the verb, its mother VP and S are all identical. When the
lexical head combines with its two complements, the COMPS value becomes empty, forming
a VP in accordance with the Head-Complement Rule. This VP will still need to combine with
its SPR in order to form a complete sentence. This kind of ‘discharging’ mechanism is further
1
In addition to these three grammar rules, English employs the Head-Filler Rule that licenses the combination of
a head missing one phrasal element with a filler that matches this missing element, as in What did John eat ?. See
Chapter 10 for discussion of this rule.
2
For convenience reason, we adopt a shorthand system in representing feature structures, suppressing unrelated
features. For example, the fully specified feature structure in (10) will include VAL as well as PHON, SYN, SEM, etc.
75
ensured by the following general principle:
(11) The Valence Principle (VALP):
The mother’s SPR and COMPS value is identical with its head daughters minus the
discharged value(s).
This principle thus ensures that when the VP in (10) combines with the subject SPR, it forms a
complete S in accordance with the Head-Specifier Rule. More generally, the VALP ensures that
each verb combines in the syntactic structure with exactly all and only the syntactic dependents
that its SPR and COMPS values indicate.
5.2 Feature Specifications on the Complement Values
5.2.1 Complements of Verbs
Intuitively, English verbs have 6 grammatical forms. For example, the verb drive can take these
forms: drives, drove, drive, driving, driven, to drive, in addition to the citation form. The present
and past tense forms are usually classified together as fin(ite), with all the rest being nonfin(ite)
in some way. Using this division, we might lay out the forms as in (12):
(12) Types of English Verb Forms:
Finiteness Verb forms Example
pres He drives a car.
fin pst He drove a car.
pln They drive a car.
bse He wants to drive a car.
Driving a car, he sang a song.
ing He was driving.
nonfin He is proud of driving a car.
Driven by the mentor, he worked.
en The car was driven by him.
He has driven the car.
inf He has to drive.
The fin forms have three subtypes pres (present), pst (past), and pln (plain) which are typically
realized as -(e)s, -ed form, and null, respectively. The plain form, though identical to the citation
form, is used for present tense when the subject is non-3rd person. The nonfin forms have the
basic forms of bse (base), ing (present participle), and en (past participle), and inf (infinitive).
This classification means that the plain and base forms are identical with the lexical base (or
citation) of the lexeme.
3
Even though the two forms are identical in most cases, substitution
tests by the past form or by the verb be can show us a clear difference:
3
A lexeme is thus an abstract unit of morphological analysis. For example, drive, drives, driving, drove, drive are
forms of the same lexeme DRIVE. In this sense, we can take a lexicon to consist of lexemes as headwords.
76
(13) a. They write/wrote to her.
b. They are/*be kind to her.
(14) a. They want to write/*wrote to her.
b. They want to be/*are kind to her.
The verbs write and are in (13) are plain forms whereas write and be in (14) are base forms with
no tense information.
We also follow the fairly standard generative grammatical analysis of English ‘infinitives’
in which the infinitive part is a head (to) which takes as its complement a verb in the bseform.
This has the consequence that there is only one verb in English with an infinitive form: to itself.
With these classifications, we propose the following hierarchy for the values of the attribute
VFORM:
(15)
vform
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
fin

¸
¸
¸
¸
¸
¸
¸
¸
nonfin

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
pres pst pln bse ing en inf
The classification of VFORM values here means that the values of VFORM can be typed, and
those types have different subtypes. Sometimes we want to be able to refer to the type of a value,
as in (16a), and sometimes to a particular form, as in (16b).
(16) a. [VFORM fin]
b. [VFORM ing]
The need to distinguish between fin and nonfin is easily determined. Every declarative sen-
tence in English needs to have a finite verb with tense information:
(17) a. The student [knows the answers].
b. The student [knew the answers].
c. The students [know the answers].
(18) a. *The student [knowing the answers].
b. *The student [known the answers].
The unacceptability of the examples in (18) is due to the fact knowing and known have no
expression of tense – they are not finite. This in turn shows us that only finite verb forms can
be used as the head of the highest VP in a declarative sentence, satisfying a basic requirement
placed on English declarative sentences:
(19) English Declarative Sentence Rule:
For an English declarative sentence to be well-formed, its verb form value
(VFORM) must be finite.
77
The finiteness of a sentence or a VP comes from the head verb, showing that the finiteness of
the VFORM value is a head feature:
(20)
S[VFORM fin]

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
NP
VP[VFORM fin]
¸
One thing we need to remember is that the two participle forms (present and past) have many
different uses, in different constructions, as partially exemplified in (21) and (22).
(21) Usages of the present participle (ing) form:
a. He is writing another long book about beavers.
(part of the the progressive aspect construction)
b. Broadly speaking, the project was successful.
(used as a sentence modifier)
c. He is proud of his son’s passing the bar exam.
(used in a gerundive construction)
(22) Usages of the past participle (en) form:
a. The chicken has eaten.
(part of the perfect aspect construction)
b. The chicken was eaten.
(part of the passive voice construction)
c. Seen from this perspective, there is no easy solution.
(used as a sentence modifier)
Some of these usages have been introduced as VFORM values for the ease of exposition (cf.
Gazdar et al. (1985) or Ginzburg and Sag (2000).) though strictly speaking, there are only two
participle forms in English, which each have several functions or constructional usages.
Every verb will be specified for a value of the head feature VFORM. For example, let us
consider a simple example like The student knows the answer. Here the verb knows will have
the following lexical information:
(23)
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
knows
HEAD
_
_
POS verb
VFORM pres
_
_
VAL
_
¸
_
SPR
_
1 NP
_
COMPS
_
2 NP
_
_
¸
_
ARG-ST
_
1 , 2
_
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
78
This [VFORM pres] value will be projected to the S in accordance with the HFP, as represented
in the following:
(24)
S
_
VFORM pres
_
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
NP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
VP
_
VFORM pres
_
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
The student
V
_
VFORM pres
_
¸
NP

¸
¸
¸
¸
¸
¸
¸
¸
knows the answer
It is easy to verify that if we have knowing instead of knows here, the S would have the [VFORM
ing] and the result could not be a well-formed declarative sentence. This is simply because the
value ing is a subtype of nonfin.
There are various constructions in English where we need to refer to VFORM values, such
as:
(25) a. The monkeys kept [forgetting/*forgot/*forgotten their lines]. (ing)
b. We caught them [eating/*ate/*eat/*eaten the bananas]. (ing)
c. John made Mary [cook/*to cook/*cooking Korean food]. (bse)
Even though each main verb here requires a VP as its complement (the part in brackets), the
required VFORM value could be different, as illustrated by the following lexical entries:
(26) a.
_
¸
¸
¸
_
keep
HEAD| POS verb
COMPS
_
VP[ing]
_
_
¸
¸
¸
_
b.
_
¸
¸
¸
_
make
HEAD| POS verb
COMPS
_
NP, VP[bse]
_
_
¸
¸
¸
_
Such lexical specifications on the VP’s VFORM value will make sure that these verbs only
combine with a VP complement with the appropriate VFORM value. The following structure
represents one example:
79
(27)
S
_
¸
¸
¸
¸
¸
¸
_
HEAD 4
_
_
POS verb
VFORM pst
_
_
SPR
COMPS
_
¸
¸
¸
¸
¸
¸
_

_
_
_
_
_
_
_
_
_
_
_
_
_
_
1 NP
VP
_
¸
¸
¸
_
HEAD 4
SPR 1 NP
COMPS
_
¸
¸
¸
_
.
.
.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
_
_
_
N
V
_
¸
¸
¸
_
HEAD 4
SPR 1 NP
COMPS 2 VP
_
¸
¸
¸
_
2 VP
_
¸
¸
¸
_
HEAD 5 | VFORM ing
SPR 2 NP
COMPS
_
¸
¸
¸
_
.
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
John kept
V
_
¸
¸
¸
_
HEAD 5
SPR 2 NP
COMPS 5
_
¸
¸
¸
_
5 NP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
forgetting his lines for the play
The finite verb kept selects as its complement a VP whose VFORM value is ing. The verb
forgetting has this VFORM value which is passed up to its mother VP in accordance with the
HFP. The Head-Complement Rule allows the combination of the head verb kept with this VP.
In the upper part of the structure, the VFORM value of the verb kept is also passed up to its
mother node VP, ensuring that the VFORM value of the S is a subtype of fin, satisfying the
basic English rule for declarative sentences.
5.2.2 Complements of Adjectives
There are at least two types of adjectives in English in terms of complement selection: those
selecting no complements at all, and those taking complements. As shown in the following
examples, an adjective like despondent optionally takes a complement, while intelligent does
80
not take any complements:
(28) a. The monkey seems despondent (that it is in a cage).
b. He seems intelligent (*to study medicine).
Adjectives such as eager, fond and compatible each select a complement, possibly of different
categories (for example, VP or PP).
(29) a. Monkeys are eager [to leave/*leaving the compound].
b. The chickens seem fond [of/*with the farmer].
c. The foxes seem compatible [with/*for the chickens].
d. These are similar [to/* with the bottles].
e. The teacher is proud [of/*with his students].
f. The contract is subject [to/*for approval by my committee].
One thing we can note again is that the complements also need to be in a specific VFORM
and PFORM value, where PFORM indicates the form of a specific preposition, as illustrated in
examples (29b–f). Just like verbs, adjectives also place restrictions on the VFORM or PFORM
value of their complement. Such restrictions are also specified in the lexical information, for
example:
(30) a.
_
¸
¸
¸
¸
¸
_
eager
HEAD| POS adj
SPR NP
COMPS VP[VFORM inf ]
_
¸
¸
¸
¸
¸
_
b.
_
¸
¸
¸
¸
¸
_
fond
HEAD| POS adj
SPR NP
COMPS PP[PFORM of ]
_
¸
¸
¸
¸
¸
_
Such lexical entries will project sentences like the following:
81
(31)
S
_
VFORM pres
_

_
_
_
_
_
_
_
_
_
_
_
_
3 NP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
VP
_
VFORM pres
_
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
Monkeys
V
_
_
VFORM pres
COMPS 2
_
_
2 AP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
are
A
_
COMPS 1
_
1 VP
_
VFORM inf
_
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
eager
V
_
VFORM inf
_
VP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
to leave the meeting
As represented in this simplified tree structure, the adjective eager combines with its VP[inf ]
complement in accordance with the Head-Complement Rule. In addition, this rule also licenses
the combination of the infinitival marker to with its VP[bse] complement (see also section 8.3.4)
and the combination of the copula are with its AP complement. The HFP ensures the HEAD
features, POS and VFORM, are passed up to the final S. Each structure will satisfy all the
relevant constraints and principles.
5.2.3 Complements of Common Nouns
Nouns do not usually select complements, though they often may have specifiers. For example,
common nouns like idea, book, beer etc. require only a specifier, but no complement. Yet there
also are nouns which do require a specific type of complement, such as proximity, faith, king,
desire, and bottom:
(32) a. their proximity to their neighbors/*for their neighbors
b. Bill’s faith in/*for Fred’s sister
c. the king of/*in English
d. the desire to become famous/*for success
e. the bottom of/*in the barrel
82
Although these complements can be optional in the right context, they are grammatically classi-
fied as complements of the nouns, and should be represented in the following simplified lexical
entries:
4
(33) a.
_
¸
¸
¸
¸
¸
¸
_
proximity
HEAD| POS noun
SPR
_
1 DP
_
COMPS
_
2 PP[PFORM to]
_
_
¸
¸
¸
¸
¸
¸
_
b.
_
¸
¸
¸
¸
¸
¸
_
faith
HEAD| POS noun
SPR
_
1 DP
_
COMPS
_
2 PP[PFORM in]
_
_
¸
¸
¸
¸
¸
¸
_
Though many more details remain to be covered for the various complement types of lexical cat-
egories, the discussion so far has given an idea of what kinds of complement lexical categories
select for.
5.3 Feature Specifications for the Subject
In general, most verbs select a regular NP as subject:
(34) a. John/Some books/The spy disappeared.
b. The teacher/The monkey/He fooled the students.
However, as noted in the previous chapter, certain English verbs select only it or there as subject:
(35) a. It/*John/*There rains.
b. There/*The spy lies a man in the park.
The pronouns it and there are often called ‘expletives’, indicating that they do not have or con-
tribute any meaning. The use of these expletives is restricted to particular contexts or verbs,
though both forms have regular pronoun uses as well. One way to specify such lexical speci-
fications for subjects is to make use of a form value specification for nouns: all regular nouns
[NFORM norm(al)] as default specification; overall we classify nouns as having three different
NFORM values: normal, it, and there. Given the NFORM feature, we can have the following
lexical entries for the verbs above:
(36) a.
_
¸
¸
_
rained
SPR NP[NFORM it]
COMPS
_
¸
¸
_
4
DP covers not only simple determiners like a, the, and that, but also includes a possessive phrase like John’s. In
Chapter 6 we cover NP structures in detail.
83
b.
_
¸
¸
_
fooled
SPR NP[NFORM norm
COMPS NP
_
¸
¸
_
We can also observe that only a limited set of verbs require their subject to be [NFORM
there]:
5
(37) a. There exists only one truly amphibian mammal.
b. There arose a great storm.
(38) a. There exist few solutions which are cost-effective.
b. There is a riot in the park.
c. There remained just a few problems to be solved.
The majority of verbs do not allow there as subject:
(39) a. *There runs a man in the park.
b. *There sings a man loudly.
For the sentences with there subjects, we first consider the verb forms which have regular sub-
jects. A verb like exist in (37a) takes one argument when used in an example like Only one truly
amphibian mammal exists, and the argument will be realized as the SPR, as dictated by the
entry in (40a). In addition, such verbs can introduce there as the subject, through the Argument
Realization option given in (40b), which is the form that occurs in the structure of (37a).
(40) a.
_
¸
¸
¸
¸
¸
_
exists
SPR 1 NP
COMPS
ARG-ST 1 NP
_
¸
¸
¸
¸
¸
_
b.
_
¸
¸
¸
¸
¸
_
exists
SPR 1 NP[NFORM there]
COMPS 2 NP
ARG-ST 1 NP, 2 NP
_
¸
¸
¸
¸
¸
_
5.4 Clausal Complement or Subject
5.4.1 Verbs Selecting a Clausal Complement
We have seen that the COMPS list includes predominantly phrasal elements. However, there are
verbs selecting not just a phrase but a whole clause as a complement, either finite or nonfinite.
For example, consider the complements of think or believe:
(41) a. I think (that) the press has a check-and-balance function.
b. They believe (that) Charles Darwin’s theory of evolution is just a scientific theory.
5
Some verbs such as arise or remain sound a little archaic in these constructions.
84
The C (complementizer) that here is optional, implying that this kind of verb selects for a finite
complement clause of some type, which we will notate as a [VFORM fin] clause. That is, these
verbs will have one of the following two COMPS values:
(42) a.
_
COMPS
_
S[VFORM fin]
_
_
b.
_
COMPS
_
CP[VFORM fin]
_
_
If the COMPS value only specifies a VFORM value, the complement can be either S or CP. This
means that we can subsume these two uses into the following single lexical entry, suppressing
the category information of the sentential complement:
6
(43)
_
¸
¸
¸
_
believe
HEAD| POS verb
COMPS
_
[VFORM fin]
_
_
¸
¸
¸
_
We can also find somewhat similar verbs like demand and require:
(44) a. John demanded [that she stop phoning him].
b. The rules require [that the executives be polite].
Unlike think or believe, these verbs which introduce a subjunctive clause typically only take a
CP[VFORM bse] as their complement: the finite verb itself is actually in the bse form. Observe
the structure of (44b):
6
Although the categories V or VP are also potentially specified as [VFORM fin], such words or phrases cannot
be complements of verbs like think or believe. This is because complements are typically saturated phrases, with no
unsatisfied requirements for complements or specifiers. While S and CP are saturated categories projected from V, VP
and V are not saturated.
85
(45)
S

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
NP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
VP
_
COMPS
_
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
The rules
V
_
COMPS 2
_
2 CP
_
VFORM bse
_
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
require
C
_
VFORM bse
_
S
_
VFORM bse
_
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
that the executives be polite
The verb require selects a bse CP complement, and this COMPS requirement is discharged at
its mother VP: this satisfies the Head-Complement Rule. There is one issue here with respect to
the percolation of the VFORM value: the CP must be bse, but this information comes from the
head C, not from its complement S. One way to make sure this is to assume that the VFORM
value of C is identical with that of its complement S, as represented in this lexical entry:
(46)
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
that
HEAD
_
_
POS comp
VFORM 1
_
_
SPR
COMPS S[VFORM 1 ]
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
This lexical information will then allow us to pass on the VFORM value of S to the head C
and then be percolated up to the CP according to the HFP. This encodes the intuition that a
complementizer ‘agrees’ with regard to VFORM value with its complement sentence.
There are also verbs which select a sequence of an NP followed by a CP as complements. NP
and CP are abbreviations for feature structure descriptions that include the information [POS
noun] and [POS comp], respectively:
(47) a. Joe warned the class that the exam would be difficult.
b. We told Tom that he should consult an accountant.
c. Mary convinced me that the argument was sound.
The COMPS value of such verbs will be as in (48):
86
(48)
_
COMPS NP, CP[VFORM fin]
_
In addition to the that-type of CP, there is an infinitive type of CP, headed by the comple-
mentizer for. Some verbs select this nonfinite CP as the complement:
(49) a. Tom intends for Sam to review that book.
b. John would prefer for the children to finish the oatmeal.
The data show that verbs like intend and prefer select an infinitival CP clause. The structure of
(49a) is familiar, but now has a nonfinite VFORM value within it:
(50)
S

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
NP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
VP
_
COMPS
_
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
Tom
V
_
COMPS 2
_
2 CP
_
VFORM inf
_
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
intends
C
_
VFORM inf
_
S
_
VFORM inf
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
for NP
VP
_
VFORM inf
_
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
Sam
V
_
VFORM inf
_
VP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
to review that book
The structure given here means that the verb intends will have the following lexical information:
(51)
_
¸
¸
¸
¸
¸
¸
_
intends
HEAD
_
_
POS verb
VFORM pres
_
_
COMPS CP[VFORM inf ]
_
¸
¸
¸
¸
¸
¸
_
87
To fill out the analysis, we need explicit lexical entries for the complementizer for and for to,
which we treat as an (infinitive) auxiliary verb. In fact, to has a distribution very similar to the
finite modal auxiliaries such as will or must, differing only in the VFORM value (see section
8.3.4).
(52) a.
_
¸
¸
¸
¸
¸
¸
_
for
HEAD
_
_
POS comp
VFORM inf
_
_
COMPS S[VFORM inf ]
_
¸
¸
¸
¸
¸
¸
_
b.
_
¸
¸
¸
¸
¸
¸
_
to
HEAD
_
_
POS verb
VFORM inf
_
_
COMPS VP[VFORM bse]
_
¸
¸
¸
¸
¸
¸
_
Just like the complementizer that, the complementizer for selects an infinitival S as its comple-
ment, inheriting its VFORM value too. The evidence that the complementizer for requires an
infinitival S can be found from coordination data:
(53) a. For John to either [make up such a story] or [repeat it] is outrageous.
(coordination of bse VPs)
b. For John either [to make up such a story] or [to repeat it] is outrageous.
(coordination of inf VPs)
c. For [John to tell Bill such a lie] and [Bill to believe it] is outrageous.
(coordination of inf Ss)
Given that only like categories (constituents with the same label) can be coordinated, we can
see that base VPs, infinitival VPs, and infinitival Ss are all constituents.
7
One thing to note here is that the verbs which select a CP[VFORM inf ] complement can also
take a VP[VFORM inf ] complement:
(54) a. John intends to review the book.
b. John would prefer to finish the oatmeal.
By underspecifying the category information of complements, we can generalize this subcate-
gorization information:
(55)
_
¸
¸
¸
_
intend
HEAD| POS verb
COMPS
_
[VFORM inf ]
_
_
¸
¸
¸
_
7
Tensed VPs can be coordinated regardless of their different tense values as in Kim [alienated cats] and [beats his
dog].
88
Since the specification [VFORM inf ] is quite general, it can be realized either as CP[VFORM
inf ] or VP[VFORM inf ].
However, this does not mean that all verbs behave alike: not all verbs can take variable
complement types such as an infinitival VP or S. For examples, try, tend, hope, and others
select only a VP[inf ] as attested by the data:
(56) a. Tom tried to ask a question.
b. *Tom tried for Bill to ask a question.
(57) a. Tom tends to avoid confrontations.
b. *Tom tends for Mary to avoid confrontations.
(58) a. Joe hoped to find a solution.
b. *Joe hoped for Beth to find a solution.
Such subcategorization differences are hard to predict just from the meaning of verbs: they are
simple lexical specifications which language users need to learn.
There is another generalization that we need to consider with respect to the property of verbs
that select a CP: most verbs that select a CP can at first glance select an NP, too:
(59) a. John believed it/that he is honest.
b. John mentioned the issue to me/mentioned to me that the question is an issue.
Should we have two lexical entries for such verbs or can we have a simple way of representing
such a pattern? To reflect such lexical patterns, we can assume that English parts of speech come
in families and can profitably be analyzed in terms of a type hierarchy as following:
8
(60)
part-of-speech
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

¸
¸
¸
¸
¸
¸
¸
¸
¸
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
nominal

¸
¸
¸
¸
¸
¸
¸
¸
¸
verbal

¸
¸
¸
¸
¸
¸
¸
¸
¸
adj prep . . .
noun comp verb
According to the hierarchy, the type nominal is a supertype of both noun and comp. In accor-
dance with the basic properties of systems of typed feature structures, an element specified as
[POS nominal] can be realized either as [POS noun] or [POS comp]. These will correspond to
the phrasal types NP and CP, respectively.
9
The hierarchy implies that the subcategorization pattern of English verbs will refer to (at
least) each of these types. Consider the following patterns:
(61) a. She pinched [his arm] as hard as she could.
8
This type hierarchy is adopted from Kim and Sag (2006).
9
For the function of verbal, see Chapter 12.
89
b. *She pinched [that he feels pain].
(62) a. We hope [that such a vaccine could be available in ten years].
b. *We hope [the availability of such a vaccine in ten years].
(63) a. Cohen proved [the independence of the continuum hypothesis].
b. Cohen proved [that the continuum hypothesis was independent].
The part-of-speech type hierarchy in (60) allows us to formulate simple lexical constraints that
reflect these subcategorization patterns, making reference to noun, verbal, and nominal:
(64) a.
_
ARG-ST NP, NP[POS noun], . . .
_
b.
_
ARG-ST NP, CP[POS comp], . . .
_
c.
_
ARG-ST NP, [POS nominal], . . .
_
In each class, the ARG-ST list specifies the argument elements that the verbs select (in the order
Subject, Direct Object, . . . ). The POS value of a given element is the part-of-speech type that
a word passes on to the phrases it projects. These three patterns illustrate that English transitive
verbs come in at least three varieties.
5.4.2 Verbs Selecting a Clausal Subject
In addition to CP as a complement, we also find some cases where a CP is the subject of a verb:
(65) a. [John] bothers me.
b. [That John snores] bothers me.
(66) a. [John] loves Bill
b. *[That John snores] loves Bill.
The contrast here means that verbs like bother can have two realizations of the ARG-ST,
whereas those like love allow only one:
(67) a.
_
¸
¸
¸
¸
¸
_
bothers
SPR 1 [POS nominal]
COMPS 2 NP
ARG-ST 1 , 2
_
¸
¸
¸
¸
¸
_
b.
_
¸
¸
¸
¸
¸
_
loves
SPR 1 NP
COMPS 2 NP
ARG-ST 1 , 2
_
¸
¸
¸
¸
¸
_
These different realizations all hinge on the lexical properties of the given verb, and only some
verbs allow the dual realization described by (67a).
A clausal subject is not limited to a finite that-headed CP, but there are other clausal types:
90
(68) a. [That John sold the ostrich] surprised Bill.
(that-clause CP subject)
b. [(For John) to train his horse] would be desirable.
(infinitival CP or VP subject)
c. [That the king or queen be present] is a requirement on all Royal weddings.
(subjunctive that-clause CP subject)
d. [Which otter you should adopt first] is unclear.
(wh-question CP subject)
Naturally, each particular predicate dictates which kinds of subjects are possible, as in (68),
and which are not, as in (69):
(69) a. *That Fred was unpopular nominated Bill.
b. *That Tom missed the lecture was enjoyable.
c. *For John to remove the mother is undeniable.
d. *How much money Gordon spent is true.
For example, the difference between the two verbs nominate and surprise can be seen in these
partial lexical entries:
(70) a.
_
¸
¸
¸
¸
¸
¸
_
nominate
VAL
_
_
SPR 1 NP
COMPS 2 NP
_
_
ARG-ST 1 , 2
_
¸
¸
¸
¸
¸
¸
_
b.
_
¸
¸
¸
¸
¸
¸
_
surprise
VAL
_
_
SPR 1 [POS nominal]
COMPS 2 NP
_
_
ARG-ST 1 , 2
_
¸
¸
¸
¸
¸
¸
_
Unlike nominate, the first argument of surprised can be a nominal. This means that its subject
can be either an NP or a CP.
5.4.3 Adjectives Selecting a Clausal Complement
Like verbs, certain adjectives can also select CPs as their complements. For example, confident
and insistent select a finite CP, whereas eager selects an infinitival CP:
(71) a. Tom is confident [that the elephants respect him].
b. Tom is insistent [that the witnesses be truthful].
(72) a. Tom seems eager [for her brother to catch a cold].
b. Tom seems eager [to catch a cold].
We can easily find more adjectives which select a CP complement:
91
(73) a. I am ashamed that I neglected you.
b. I am delighted that Mary finished his thesis.
c. We are content for the cleaners to return the drapes next week.
d. We were thankful that no one had been hurt.
e. We were glad it was over.
The lexical entries for some adjectives are given in (74):
(74) a.
_
¸
¸
_
confident
HEAD| POS adj
COMPS CP[VFORM fin]
_
¸
¸
_
b.
_
¸
¸
_
insistent
HEAD| POS adj
COMPS CP[VFORM bse]
_
¸
¸
_
c.
_
¸
¸
_
eager
HEAD| POS adj
COMPS [VFORM inf ]
_
¸
¸
_
Such lexical entries, interacting with the Head-Complement Rule, the Head-Specifier Rule, and
the HFP, can license analyses such as (75), for (72b):
(75)
S

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
NP

¸
¸
¸
¸
¸
¸
¸
VP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
Tom
V
_
COMPS 2
_ 2 AP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
seems
A
_
COMPS 3
_
3 VP
_
VFORM inf
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
eager V
VP
_
VFORM bse
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
to V NP

¸
¸
¸
¸
¸
catch a cold
92
When the head adjective eager combines with its complement, VP[inf ], it satisfies the Head-
Complement Rule. The same rule allows the combination of the verb seem with its AP comple-
ment.
5.4.4 Nouns Selecting a Clausal Complement
Nouns can also select a CP complement, for example, eagerness:
(76) a. (John’s) eagerness [for Harry to win the election]
b. (John’s) eagerness [to win the election]
These examples imply that eagerness will have the following lexical information:
(77)
_
¸
¸
¸
_
eagerness
HEAD| POS noun
COMPS
_
[VFORM inf ]
_
_
¸
¸
¸
_
This lexical entry will allow a structure like the following:
(78)
NP
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
DP

¸
¸
¸
¸
¸
¸
¸
N

.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
John’s
N
_
COMPS 2
_
2 VP
_
VFORM inf
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
eagerness to win the election
One pattern that we can observe is that when a verb selects a CP complement, if there is a
corresponding noun, it also selects a CP:
(79) a. Bill alleged that Fred signed the check.
b. We believe that the directors were present.
c. We convinced him that the operation is safe.
(80) a. the allegation that Fred signed the check
b. the belief that the directors were present
c. his conviction that the operation is safe
This shows us that the derivational process which derives a noun from a verb preserves the
COMPS value of that verb.
A caution here is that not all nouns can of course select a CP complement:
(81) a. *his attention that the earth is round
b. *his article that the earth is flat
93
c. *the ignorance that James can play the flute
d. *the expertise that she knows how to bake croissants
These nouns cannot combine with a CP complement, indicating that they do not have CP in the
value of COMPS.
5.4.5 Prepositions Selecting a Clausal Complement
In general, prepositions in English cannot select a CP complement.
(82) a. *Alan is thinking about [that his students are eager to learn English].
b. *Fred is counting on [for Tom to make an announcement].
However, wh-CPs, sometimes known as indirect questions, may serve as prepositional comple-
ments.
(83) a. The outcome depends on [how many candidates participate in the election].
b. Fred is thinking about [whether he should stay in Seoul].
These facts show us that indirect questions have some feature which distinguishes them from
canonical that- or for-CPs, and makes them somehow closer to true nouns (for NP is the typical
complement for a preposition).
94
5.5 Exercises
1. Check if each of the following expressions selects for a clausal complement:
(i) ignore, doubt, deny, prefer, intend, glad, unsure, confident, allegation, ig-
norance
Write out the examples which justify your classifications.
2. Check if each of the following expressions selects a clausal subject or not.
(i) annoy, vanish, remain, admire, select, mandatory, enjoyable, apparent
Write out the examples which justify your classifications.
3. For each sentence, draw a tree structure and then give the COMPS value (including
VFORM and PFORM value) for the italicized word.
(i) a. The offer made Smith admire the administrators.
b. John tried to make Sam let George ask Bill to keep delivering the mail.
c. The soldiers must enforce Bill to make the baby be quiet.
d. John enjoyed drawing trees for his syntax homework.
e. The picture on the wall reminded him of his country.
f. Free enterprise is compatible with American values and traditions.
g. They were taking a hard look at possible FTA.
h. We need to be in frequent contact with the clients.
i. The contract is subject to approval by my committee.
j. Acknowledge that everyone has limits.
k. We are aware of the existing problems.
4. Identify errors in the following sentences, focusing on the form values of verbs, adjec-
tives, and nouns, and/or their COMPs values.
(i) a. *Why don’t you leaving me concentrate on my work?
b. *The general commended that all troops was in dress uniform.
c. *My morning routine features swim free styles slowly for one hour.
d. *You should avoid to travel in the rush hour.
e. *You should attempt answering every question.
f. *The authorities blamed Greenpeace with the bombing.
g. *The authorities charged the students of the cheating.
h. *Sharon has been eager finishing the book.
i. *We respect Mary’s desire for becoming famous.
j. *John referred from the building.
k. *John died to heart disease.
l. *John paid me against the book.
m. *We were glad what to do.
n. *She was busy to make lunch.
5. Draw trees for the following examples with detailed NP structures.
95
(i) a. The constant rain forced the abandonment of the next day’s competitions.
b. The committee will study the feasibility of setting up a national computer
network.
c. Aloe may have an analgesic effect on inflammation and minor skin irrita-
tions.
d. The public never had faith in his ability to handle the job.
e. He repeated his claim that the people backed his action.
6. In general the present and base forms of a main verb are identical, but can be distin-
guished in the case of be, with its present form are:
(i) a. We made them take the money.
b. *We made them are rude.
c. We made them be rude.
Do the same substitution test for the following sentences and determine whether the
italicized verb is the base or present form.
(ii) a. Do not use these words in the beginning of a sentence.
b. We know the witnesses seem eager to testify against the criminal.
c. Jane isn’t sure whether the students keep the books.
d. Why not try to catch the minnows?
7. Read the following text and provide the lexical entries for the underlined words. In doing
so, try to specify the VFORM or PFORM value of the complement(s).
(i) The study of grammar helps us to communicate more effectively. Quite
simply, if we know how English works, then we can make better use of it.
For most purposes, we need to be able to construct sentences which are far
more complicated than David plays the piano. A knowledge of grammar
enables us to evaluate the choices which are available to us during compo-
sition. In practice, these choices are never as simple as the choice between
David plays the piano and *plays David piano the. If we understand the
relationship between the parts of a sentence, we can eliminate many of
the ambiguities and misunderstandings which result from poor construc-
tion. In the interpretation of writing, too, grammatical knowledge is often
crucially important. The understanding of literary texts, for example, often
depends on careful grammatical analysis. Other forms of writing can be
equally difficult to interpret. Scientific and academic writing, for instance,
may be complex not just in the ideas they convey, but also in their syntax.
These types of writing can be difficult to understand easily without some
familiarity with how the parts relate to each other. The study of grammar
enables us to go beyond our instinctive, native-speaker knowledge, and to
96
use English in an intelligent, informed way.
10
10
From ‘Introducing the Internet Grammar of English’ at http://www.ucl.ac.uk/internet-grammar/
intro/intro.htm.
97
6
Noun Phrases and Agreement
6.1 Classification of Nouns
Nouns not only represent entities like people, places, or things, but also denote abstract and
intangible concepts such as happiness, information, hope, and so forth. Such diversity of ref-
erence renders it difficult to classify nouns solely according to their meanings. The following
chart shows the canonical classification of nouns taking into account semantic differences, but
also considering their formal and grammatical properties:
(1) Types of Nouns in English:
common countable desk, book, difficulty, remark, etc.
noun uncountable butter, gold, music, furniture, laziness, etc.
proper noun
Seoul, Kyung Hee, Stanford, Palo Alto,
January, etc.
personal he, she, they, his, him, etc.
relative that, which, what, who, whom, etc.
pronoun
interrogative who, where, how, why, when, etc.
indefinite
anybody, everybody, somebody, nobody,
anywhere, etc.
As shown here, nouns fall into three major categories: common nouns, proper nouns, and pro-
nouns. One important aspect of common nouns is that they are either count or non-count.
Whether a noun is countable or not does not fully depend on its reference; examples like diffi-
culty which is mass (non-count) but difficulties which is count suggest how subtle the distinction
can be, and we have nouns like furniture/*furnitures which are only mass and a chair/chairs
which are only count.
Proper nouns denote specific people or places and are typically uncountable. Common nouns
and proper nouns display clear contrasts in terms of the combinatorial possibilities with deter-
miners as shown in the following chart:
99
(2) Combinatory Possibilities with Determiners:
Common N
Proper N
countable uncountable neutral
Only N Einstein *book music cake
the + N *the Einstein the book the music the cake
a + N *an Einstein a book *a music a cake
some + N *some Einstein *some book some music some cake
N + s *Einsteins books *musics cakes
Proper nouns do not combine with any determiner, as can be seen from the chart. Meanwhile,
count nouns have singular and plural forms (e.g., a book and books), whereas uncountable
nouns combine only with some or the. As noted in Chapter 1, some common nouns may be
either count or non-count, depending on the kind of reference they have. For example, cake can
be countable when it refers to a specific one as in I made a cake, but can be noncountable when
it refers to ‘cake in general’ as in I like cake.
Together with verbs, nouns are of pivotal importance in English, forming the semantic and
structural components of sentences. This chapter deals with the structural, semantic, and func-
tional dimensions of NPs, with focus on the agreement relationships of nouns with determiners
and of subjects with verbs.
6.2 Syntactic Structures
6.2.1 Common Nouns
As noted before, common nouns can have a determiner as a specifier, unlike proper and pro-
nouns. In particular, count nouns cannot be used without a determiner when they are singular:
(3) a. *Book is available in most countries.
b. *Student studies English for 4 hours a day.
(4) a. Rice is available in most countries.
b. Students study English for 4 hours a day.
We can see here that mass nouns, or plural count nouns, are fully grammatical as bare nouns
phrases.
1
This has the consequence for our grammatical analysis that singular countable nouns like
student must select a determiner as specifier. As we have seen in Chapters 2 and 4, there are
various kinds of expressions which can serve as determiners including a, an, this, that, any,
some, his, how, which, some, no, much, few, . . . as well as a possessive phrase:
(5) a. His friend learned dancing.
1
The style of English used in headlines does not have this restriction, e.g., Student discovers planet, Army receives
high-tech helicopter.
100
b. My bother’s friend learned dancing.
c. The president’s bodyguard learned surveillance.
d. The King of Rock and Roll’s records led to dancing.
These possessive NPs my brother’s or the president’s are not determiners, because they are
phrases. We take such phrases as DPs headed by the Det ’s (cf. Abney (1987)). Let’s consider
the lexical entries for the relevant words:
(6)
a.
_
¸
¸
¸
¸
¸
_
my
HEAD| POS det
SPR
COMPS
_
¸
¸
¸
¸
¸
_
b.
_
¸
¸
¸
¸
¸
_
’s
HEAD| POS det
SPR NP
COMPS
_
¸
¸
¸
¸
¸
_
c.
_
¸
¸
¸
¸
¸
_
brother
HEAD| POS noun
SPR DP
COMPS
_
¸
¸
¸
¸
¸
_
d.
_
¸
¸
¸
¸
¸
_
friend
HEAD| POS noun
SPR DP
COMPS
_
¸
¸
¸
¸
¸
_
As given here, determiners like my and uncountable nouns like rice select neither specifier
nor complement whereas the possessive marker ’s and the countable noun friend requires a
NP and DP specifier, respectively. Meanwhile, countable noun brother requires a DP as its
specifier. This means that not only a simple determiner but also a possessive NP can function
as a specifier. Reflecting this generalization, we can assume that a countable noun selects not a
simple determiner but a DP (determiner phrase) as its specifier: These lexical entries will project
NP structures like the following for (5b):
2
(7)
NP

¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸
1 DP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
N

_
_
SPR 1 DP
COMPS
_
_
2 NP
·
·
·
·
·
·
·
·
¸
¸
¸
¸
¸
¸
¸
¸
Det
_
_
SPR 2 NP
COMPS
_
_
N
_
_
SPR 1 DP
COMPS
_
_
my brother ’s friend
2
Keen readers may have noticed that we allow a vacuous projection from N to N

and to NP when no specifier or
complement is combined. We can also allow an N to directly project into an NP.
101
As represented here, the noun friend does not select a complement, and thus projects into an
intermediate phrase which in turn combines with the specifier DP my brother’s whose head is
the possessive determiner.
6.2.2 Pronouns
The core class of pronouns in English includes at least three main subgroups:
(8) a. Personal pronouns: I, you, he, she, it, they, we
b. Reflexive pronouns: myself, yourself, himself, herself, itself
c. Reciprocal pronoun: each other
Personal pronouns refer to specific persons or things and take different forms to indicate person,
number, gender, and case. They participate in agreement relations with their antecedent, the
phrase which they are understood to be referring to (indicated by the underlined parts of the
examples in (9)).
(9) a. President Lincoln delivered his/*her Gettysburg Address in 1863.
b. After reading the pamphlet, Judy threw it/*them into the garbage can.
c. I got worried when the neighbors let their/*his dogs out.
Reflexive pronouns are special forms which typically are used to indicate a reflexive activity
or action, which can include mental activities.
(10) a. After the party, I asked myself why I had faxed invitations to everyone in my office
building.
b. Edward usually remembered to send a copy of his e-mail to himself.
As noted earlier, these personal or reflexive pronouns neither take a determiner nor combine
with an adjective except in very restricted constructions.
3
.
6.2.3 Proper Nouns
Since proper nouns usually refer to something or someone unique, they do not normally take a
plural form and cannot occur with a determiner:
(11) a. John, Bill, Seoul, January, . . .
b. *a John, *a Bill, *a Seoul, *a January, . . .
However, proper nouns can be converted into countable nouns when they refer to a particular
individual or type of individual:
(12) a. No John Smiths attended the meeting.
b. This John Smith lives in Seoul.
c. There are three Davids in my class.
d. It’s nothing like the America I remember.
e. My brother is an Einstein at maths.
3
These restricted constructions involve some indefinite pronouns (e.g., a little something, a certain someone)
102
In such cases, proper nouns are converted into common nouns, may select a specifier, and take
other nominal modifiers.
6.3 Agreement Types and Morpho-syntactic Features
6.3.1 Noun-Determiner Agreement
Common nouns in English participate in three types of agreement. First, they are involved in
determiner-noun agreement. All countable nouns are used either as singular or plural. When
they combine with a determiner, there must be an agreement relationship between the two:
(13) a. this book/that book
b. *this books/*that books/these books/those books
c. *few dog/few dogs
The data in turn means that the head noun’s number value should be identical to that of its
specifier, leading us to revise the Head-Specifier Rule as following:
(14) Head-Specifier Rule:
XP →SPR
_
AGR 1
_
, H
_
AGR 1
_
This revised rule guarantees that English head-specifier phrases require their head and spec-
ifier to share agreement features, implying that determiners and nouns have NUM (number)
information as their syntactic AGR (agreement) value:
(15)
a.
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
a
HEAD
_
_
POS det
AGR| NUM sing
_
_
SPR
COMPS
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
b.
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
book
HEAD
_
_
POS noun
AGR | NUM sing
_
_
SPR DP[NUM sing]
COMPS
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
Common nouns thus impose a specific NUM value on the specifier:
103
(16)
NP
_
AGR| NUM sing
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
1 DP[NUM sing]
N

_
¸
¸
¸
_
AGR| NUM sing
SPR 1 DP
COMPS
_
¸
¸
¸
_
¸¸
a
N
_
¸
¸
¸
_
AGR| NUM sing
SPR 1 DP
COMPS
_
¸
¸
¸
_
¸¸
book
The singular noun book selects a singular determiner like a. Notice that the AGR value on the
head noun book is passed up to the whole NP, marking the whole NP as singular, so that it can
combine with a singular VP, if it is the subject.
In addition, there is nothing preventing a singular noun from combining with a determiner
which is not specified at all for a NUM value:
(17) a. *those book, *these book, . . .
b. no book, the book, my book, . . .
Determiners like the, no and my are not specified for a NUM value. Formally, their NUM value
is underspecified as num(ber). That is, the grammar of English has the underspecified value num
for the feature NUM, with two subtypes sing(ular) and pl(ural):
(18)
num

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
sing pl
Given this hierarchy, nouns like book requiring a singular Det can combine with determiners
like the whose AGR value is num. This is in accord with the grammar since the value num is a
supertype of sing.
104
6.3.2 Pronoun-Antecedent Agreement
As noted earlier, a second type of agreement is pronoun-antecedent agreement, as indicated in
(19).
(19) a. In the book, he talks about his ups and downs at McLaren. Throughout it all he
seeks to answer the questions about himself.
b. If John wants to succeed in corporate life, he/*she has to know the rules of the
game.
c. The critique of Plato’s Republic was written from a contemporary point of view. It
was an in-depth analysis of Plato’s opinions about possible governmental forms.
The pronoun he or it here needs to agree with its antecedent not only with respect to the number
value but also with respect to person (1st, 2nd, 3rd) and gender (masculine, feminine, or neuter)
values too. This shows us that nouns have also information about person and gender as well as
number in the AGR values:
(20)
a.
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
book
HEAD
_
¸
¸
¸
¸
¸
_
POS noun
AGR
_
¸
¸
_
NUM sing
GEN neut
PER 3rd
_
¸
¸
_
_
¸
¸
¸
¸
¸
_
VAL
_
_
SPR DP[NUM sing]
COMPS
_
_
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
b.
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
he
HEAD
_
¸
¸
¸
¸
¸
_
POS noun
AGR
_
¸
¸
_
NUM sing
PER 3rd
GEN masc
_
¸
¸
_
_
¸
¸
¸
¸
¸
_
VAL
_
_
SPR
COMPS
_
_
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
As we have briefly shown, nouns have NUM, PER(SON), and GEN(DER) for their AGRvalues.
The PER value can be 1st, 2nd or 3rd; the GEN value can be masc(uline), fem(inine) or neut(er).
The NUM values are shown in (18) above.
6.3.3 Subject-Verb Agreement
The third type of agreement is subject-verb agreement, which is one of the most important
phenomena in English syntax. Let us look at some slightly complex examples:
(21) a. The characters in Shakespeare’s Twelfth Night *lives/live in a world that has been
turned upside-down.
b. Students studying English read/*reads Conrad’s Heart of Darkness while at uni-
versity.
As we can see here, the subject and the verb need to have an identical number value; and the
person value is also involved in agreement relations, in particular when the subject is a personal
pronoun:
(22) a. You are/*is the only person that I can rely on.
105
b. He is/*are the only person that I can rely on.
These facts show us that a verb lexically specifies the information about the number as well as
person values of the subject that it selects for.
To show how the agreement system works, we will use some simpler examples:
(23) a. The boy swims/*swim.
b. The boys swim/*swims.
English verbs will have at least the following selectional information:
(24)
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
swims
HEAD
_
_
POS verb
VFORM pres
_
_
VAL| SPR
_
NP
_
_
PER 3rd
NUM sing
_
_
_
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
The present-tense verb swims thus specifies that its subject (SPR’s value) carries a 3rd singular
AGR information. This lexical information will license a structure like the following:
(25)
S
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
2 NP
_
_
AGR
_
_
PER 3rd
NUM sing
_
_
_
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
VP
_
¸
_
SPR
_
2 NP
_
_
AGR
_
_
PER 3rd
NUM sing
_
_
_
_
_
_
¸
_
The boy
V
_
SPR 2 NP
_
swims
Only when the verb combines with a subject satisfying its AGR requirement will we have a
well-formed head-subject phrase. In other words, if this verb were to combine with a subject
with an incompatible agreement value, we would generate an ungrammatical example like *The
boys swims in (23b). In this system, what subject-verb agreement is is structure-sharing between
the AGR value of the subject (SPR value of the verb) and that of the NP that the VP combines
with.
106
The acute reader may have noticed that there are similarities between noun-determiner agree-
ment and subject-verb agreement, that is, in the way that agreement works inside NP and inside
S. Both NP and S require agreement between the head and the specifier, as reflected in the
revised Head-Specifier Rule in (14).
6.4 Semantic Agreement Features
What we have seen so far is that the morphosyntactic AGR values of noun or verb can be spec-
ified, and may be inherited by phrases built out of them. However, consider now the following
examples adopted from Nunberg (1995):
(26) a. [The hash browns at table nine] are/*is getting cold.
b. [The hash browns at table nine] is/*are getting angry.
When (26b) is spoken by a waiter to another waiter, the subject refers to a person who ordered
hash browns. A somewhat similar case is found in (27):
(27) King prawns cooked in chili salt and pepper was very much better, a simple dish
succulently executed.
Here the verb form was is singular to agree with the dish being referred to, rather than with a
plurality of prawns. If we simply assume that the subject phrase inherits the morphosyntactic
agreement features of the head noun (hash) browns in (26b) and (King) prawns in (27), and
requires that these features match those of the verb, we would not expect the singular verb form
to be possible at all in these examples. In the interpretation of a nominal expression, it must
be anchored to an individual in the situation described. We call this anchoring value the noun
phrase’s ‘index’ value. The index of hash browns in (26a) must be anchored to the plural entities
on the plate, whereas that of hash browns in (26b) is anchored to a customer who ordered the
food.
English agreement is not purely morpho-syntactic as described in the sections above, but
context-dependent in various ways, via the notion of ‘index’ that we have just introduced. Often
what a given nominal refers to in the real world is important for agreement – index agreement.
Index agreement involves sharing of referential indexes, closely related to the semantics of a
nominal, and somewhat separate from the syntactic agreement feature AGR. This then requires
us to distinguish the morphological AGR value and semantic (SEM(ANTIC)) IND(EX) value.
So, in addition to the morphological AGR value introduced above, each noun will also have a
semantic IND value representing what the noun refers to in the actual world.
(28) a.
_
¸
¸
¸
¸
¸
¸
_
boy
SYN| HEAD
_
_
POS noun
AGR| NUM sing
_
_
SEM| IND| NUM sing
_
¸
¸
¸
¸
¸
¸
_
107
b.
_
¸
¸
¸
¸
¸
¸
_
boys
SYN| HEAD
_
_
POS noun
AGR| NUM pl
_
_
SEM| IND| NUM pl
_
¸
¸
¸
¸
¸
¸
_
The lexical entry for boy indicates that it is syntactically a singular noun (through the feature
AGR) and semantically also denotes a singular entity (through the feature IND). And the verb
will place a restriction on its subject’s IND value rather than its morphological AGR value:
4
(29)
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
swims
SYN
_
¸
¸
¸
¸
_
HEAD
_
_
POS verb
AGR| NUM sing
_
_
VAL| SPR
_
NP[IND| NUM sing]
_
_
¸
¸
¸
¸
_
SEM| IND s0
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
The lexical entry for swims here indicates that it is morphologically marked as singular (the
AGR feature) and selects a subject to be linked to a singular entity in the context (by the feature
IND). Different from nouns, the verb’s own IND value is a situation index (s0) in which the in-
dividual referred to through the SPR value is performing the action of swimming. If the referent
of this subject (its IND value) does not match, we would generate an ungrammatical example
like *The boys swims:
(30)
*S
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
2 NP
_
_
IND
j
_
_
PER 3rd
NUM pl
_
_
_
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
VP
_
¸
_
SPR
_
2 NP
i
_
_
PER 3rd
NUM sing
_
_
_
_
¸
_
The boys
V
_
SPR 2 NP
i

_
swims
4
The IND value of a noun will be an individual index (i, j, k, etc) whereas that of a verb or predicative adjective will
be a situation index such as s0, s1, s2, etc.
108
As we can see here, the required subject has the IND value i, but the subject has a different IND
value j.
In the most usual cases, the AGR and IND value are identical, but they can be different, as in
examples like (26b). This means that depending on the context, hash browns can have different
IND values:
5
(31)
a.
_
¸
¸
¸
¸
¸
¸
_
hash browns
SYN| HEAD
_
_
POS noun
AGR| NUM pl
_
_
SEM| IND 1 | NUM pl
_
¸
¸
¸
¸
¸
¸
_
(when referring to the food itself)
b.
_
¸
¸
¸
¸
¸
¸
_
hash browns
SYN| HEAD
_
_
POS noun
AGR| NUM pl
_
_
SEM| IND 1 | NUM sing
_
¸
¸
¸
¸
¸
¸
_
(when referring to a customer, or to a dish)
In the lexical entry (31b), the AGR’s NUM value is plural but its IND’s NUM value is singu-
lar. As shown by (26), the reference hash browns can be transferred from cooked potatoes to
the customer who ordered it. This means that given an appropriate context, there could be a
mismatch between the morphological form of a noun and the index value that the noun refers
to.
What this indicates is that subject-verb agreement and noun-specifier agreement are dif-
ferent. In fact, English determiner-noun agreement is only a reflection of morpho-syntactic
agreement features between determiner and noun, whereas both subject-verb agreement and
pronoun-antecedent agreement are index-based agreement. This is represented in (32).
(32) Morpho-syntactic agreement (AGR)
Det

head-noun
¸¸ ¸¸
verb . . .
Index agreement (IND)
Such agreement patterns can be clearly found in examples like the following where the un-
derlined parts have singular agreement with four pounds, which is internally plural.
(33) [Four pounds] was quite a bit of money in 1950 and it was not easy to come by.
Given the separation of the morphological AGR value and the semantic IND value, nothing
blocks mismatches between the two (AGR and IND) as long as all the other constraints are
satisfied. Observe further examples in the following:
(34) a. [Five pounds] is/*are a lot of money.
5
As indicated here, the lexical expression now has two features SYN (SYNTAX) and SEM (SEMANTICS). The
feature SYN includes HEAD and SPR and COMPS. The feature SEM is for semantic information. As our discussion
goes on, we will add more to this part.
109
b. [Two drops] deodorizes/*deodorize anything in your house.
c. [Fifteen dollars] in a week is/*are not much.
d. [Fifteen years] represents/*represent a long period of his life.
e. [Two miles] is/*are as far as they can walk.
In all of these examples with measure nouns, the plural subject combines with a singular verb.
An apparent conflict arises from the agreement features of the head noun. For proper agreement
inside the noun phrase, the head noun has to be plural, but for subject-verb agreement the
noun has to be singular. Consider the example in (34a). The nouns pounds and drops here are
morphologically plural and thus must select a plural determiner, as argued so far. But when
these nouns are anchored to the group as a whole – that is, conceptualized as referring to a
single measure, the index value has to be singular, as represented in (35).
(35)
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
pounds
SYN
_
¸
¸
¸
¸
¸
_
HEAD
_
_
POS noun
AGR 1 | NUM pl
_
_
SPR
_
DP
_
AGR 1
_
_
_
¸
¸
¸
¸
¸
_
SEM| IND | NUM sing
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
As indicated in the lexical entry (35), the morpho-syntactic number value of pounds is plural
whereas the index value is singular. In the present analysis, this would mean that pounds will
combine with a plural determiner but with a singular verb. This is possible, as noted earlier in
section 2, since the index value is anchored to a singular individual in the context of utterance.
The present analysis thus generates the following structure for the sentence (34a):
(36)
S
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
3 NP
_
_
AGR 1
IND i
_
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
VP
_
SPR 3 NP
i

_

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
Det
_
AGR 1 | NUM pl
_
N
_
_
AGR 1
IND i
_
_
V
_
SPR 3 NP
i

_
NP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
Five pounds is a lot of money
110
A similar mismatch is also found in cases with terms for social organizations or collections,
as in the following authentic examples:
6
(37) a. This Government have/has been more transparent in the way they have dealt with
public finances than any previous government.
b. In preparation for the return fixture this team have/has trained more efficiently
than they had in recent months.
The head noun government or team is singular so that it can combine with the singular de-
terminer this. But the conflicting fact is that the singular noun phrase can combine even with
a plural verb have as well as with a singular verb has. This is possible since the index value
of the subject can be anchored either to a singular or to plural kind of entity. More precisely,
we could represent the relevant information of the expressions participating in these agreement
relationships as in (38).
(38) a.
_
¸
¸
¸
_
this
HEAD
_
_
POS det
AGR| NUM sing
_
_
_
¸
¸
¸
_
b.
_
¸
¸
¸
¸
¸
¸
¸
_
team/government
SYN
_
_
HEAD
_
_
POS noun
AGR| NUM sing
_
_
_
_
SEM| IND | NUM pl
_
¸
¸
¸
¸
¸
¸
¸
_
As represented in (38a) and (38b), this and government agree each other in terms of the morpho-
syntactic agreement number value whereas the index value of government is what matters for
subject-verb agreement. This in turn means that when government refers to the individuals in
this given group, the whole NP this government carries a plural index value.
6.5 Partitive NPs and Agreement
6.5.1 Basic Properties
With regard to the NP-internal elements between which we may find instances of agreement,
there are two main types of NP in English: simple NPs and partitive NPs, shown in (39) and
(40) respectively.
(39) a. some objections
b. most students
c. all students
6
To some speakers, the plural verb does not go with the singular subject this team. To these people, the neutral
determiner the is much better.
111
d. much worry
e. many students
f. neither cars
(40) a. some of the objections
b. most of the students
c. all of the students
d. much of her worry
e. many of the students
f. neither of the cars
As in (40), the partitive phrases have a quantifier followed by an of -phrase, designating a set
with respect to which certain individuals are quantified. In terms of semantics, these partitive
NPs are different from simple NPs in several respects.
First, the lower NP in partitive phrases must be definite; and in the of -phrase, no quantifica-
tional NP is allowed, as shown in (41):
(41) a. Each student vs. each of the students vs. *each of students
b. Some problems vs. some of the problems vs. *some of many problems
Second, not all determiners with quantificational force can appear in partitive constructions.
As shown in (42), determiners such as the, every and no cannot occupy the first position:
(42) a. *the of the students vs. the students
b. *every of his ideas vs. every idea
c. *no of your books vs. no book(s)
Third, simple NPs and partitive NPs have different restrictions relative to the semantic head.
Observe the contrast between (43) and (44):
(43) a. She doesn’t believe much of that story.
b. We listened to as little of his speech as possible.
c. How much of the fresco did the flood damage?
d. I read some of the book.
(44) a. *She doesn’t believe much story.
b. *We listened to as little speech as possible.
c. *How much fresco did the flood damage?
d. *I read some book.
The partitive constructions in (43) allow a mass (non-count) quantifier such as much, little and
some to cooccur with a lower of -NP containing a singular count noun. But as we can see in
(44), the same elements serving as determiners cannot directly precede such nouns.
Another difference concerns lexical idiosyncrasies.
(45) a. One of the people was dying of thirst.
112
b. Many of the people were dying of thirst.
(46) a. *One people was dying of thirst.
b. Many people were dying of thirst.
The partitives can be headed by quantifiers like one and many, as shown in (45) and (46) but
unlike many, one cannot serve as a determiner when the head noun is collective as in (46a).
6.5.2 Two Types of Partitive NPs
We classify partitive NPs into two types based on the agreement facts, and call them Type I
and Type II. In Type I, the number value of the partitive phrase depends on the preceding head
noun whereas in Type II, the number value depends on the head noun inside the of -NP phrase.
Observe Type I examples.
(47) Type I:
a. Each of the suggestions is acceptable.
b. Neither of the cars has air conditioning.
c. None of these men wants to be president.
We can observe here that the verb’s number value is determined by the preceding expression
each, neither and none. Now see Type II:
(48) Type II:
a. Most of the fruit is rotten.
b. Most of the children are here.
c. Some of the soup needs more salt.
d. Some of the diners need menus.
e. All of the land belongs to the government.
f. All of these cars belong to me.
As shown in (48), when the NP following the preposition of is singular or uncountable, the main
verb is singular. When the NP is plural, the verb is also plural. From a semantic perspective, we
see that the class of quantificational indefinite pronouns including some, half, most and all may
combine either singular or plural verbs, depending upon the reference of the of -NP phrase. If
the meaning of these phrases is about how much of something is meant, the verb is singular; but
if the meaning is about how many of something is meant, the verb is plural. The expressions in
(49) also exhibit similar behavior in terms of agreement.
(49) half of, part of, the majority of, the rest of, two-thirds of, a number of (but not the
number of )
An effective way of capturing the relations between Type I and Type II constructions involves
the lexical properties of the quantifiers. First, Type I and Type II involve pronominal forms
serving as the head of the constructions, which select an of -NP inside which the NP is definite:
(50) a. *neither of students, *some of water
113
b. neither of the two linguists/some of the water
However, we know that the two types are different in terms of agreement: the pronouns in the
Type I construction are lexically specified to be singular whereas the number value for Type II
comes from inside the selected PP.
A slight digression is in order. It is easy to see that there are prepositions whose functions
are just grammatical markers.
(51) a. John is in the room.
b. I am fond of him.
The predicative preposition in here selects two arguments John and the room. Meanwhile, the
preposition of has no predicative meaning, but just functions as a marker to the argument of
fond. As for the PPs headed by these markers, as in the partitive construction, their semantic
features are identical with the prepositional object NP. There is no semantic difference (such as
definiteness effect represented as the feature DEF in the present system) between the PP of him
and the NP him.
Given this analysis in which the PP in the partitive construction has the identical AGR and
semantic features (e.g., DEF) with its inner NP, we can lexically encode the similarities and
differences between Type I and Type II in a simple manner:
(52) a.
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
neither
HEAD
_
_
POS noun
AGR| NUM sing
_
_
COMPS
_
PP
_
_
PFORM of
DEF +
_
_
_
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
b.
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
some
HEAD
_
_
POS noun
AGR| NUM 1
_
_
COMPS
_
PP
_
¸
¸
_
PFORM of
DEF +
AGR| NUM 1
_
¸
¸
_
_
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
(52) shows that both Type I neither and Type II some are lexically specified to require a PP
complement whose semantic value includes the definite (DEF) feature (with the value +). This
will account for the contrast in (50). However, the two types are different in terms of their
AGR’s NUM value. The NUM value of Type I neither is singular, whereas that of Type II is
identified with the PP’s NUM value which is actually coming from its prepositional object NP.
Showing these differences in the syntactic structures, we have the alternatives in (53):
114
(53) a.
NP[NUM sing]

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
N[NUM sing]
¸¸
PP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
neither P
NP

¸
¸
¸
¸
¸
¸
¸
of the students
b.
NP[NUM 1 ]

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
N[NUM 1 ]
¸
PP[NUM 1 ]

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
some P
NP[NUM 1 ]

¸
¸
¸
¸
¸
¸
¸
¸.
of the students
As shown in (53a), for Type I, it is neither which determines the NUM value of the whole
NP phrase. However, for Type II, it is the NP the students which determines the NUM value of
the whole NP.
We can check a few of the consequences of these different specifications in the two Types.
Consider the contrast in (54):
(54) a. many of the/those/her apples
b. *many of some/all/no apples
(54b) is ungrammatical since many requires an of -PP phrase whose DEF value is positive.
This system also offers a simple way of dealing with the fact that quantifiers like each affect
the NUM value as well as the countability of the of -NP phrase. One difference between Type I
and Type II is that Type I selects a plural of -NP phrase whereas Type II has no such restriction.
This is illustrated in (55) and (56).
(55) Type I:
a. one of the suggestions/*the suggestion/*his advice
b. each of the suggestions/*the suggestion/*his advice
c. neither of the students/*the student/*his advice
(56) Type II:
a. some of his advice/students
b. most of his advice/students
115
c. all of his advice/students
The only additional specification we need for Type I pronouns relates to the NUM value on the
PP’s complement as given in (57):
(57)
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
each
HEAD
_
_
POS noun
AGR| NUM sing
_
_
COMPS
_
PP
_
¸
¸
_
PFORM of
DEF +
NUM pl
_
¸
¸
_
_
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
We see that quantifiers like each select a PP complement whose NUM value is plural.
Type II pronouns do not have such a requirement on the PP complement – note that all the
examples in (58) are acceptable, in contrast to those in (59) (cf. Baker 1995):
(58) a. Most of John’s boat has been repainted.
b. Some of the record contains evidence of wrongdoing.
c. Much of that theory is unfounded.
(59) a. *Each of John’s boat has been repainted.
b. *Many of the record contained evidence of wrongdoing.
c. *One of the story has appeared in your newspaper.
The contrast here indicates that Type II pronouns can combine with a PP whose inner NP is
singular. This is simply predicted since our analysis allows the inner NP to be either plural or
singular (or uncountable).
We are also in a position now to understand some differences between simple NPs and par-
titive NPs. Consider the following examples:
(60) a. many dogs/*much dog/the dogs
b. much furniture/*many furniture/the furniture
(61) a. few dogs/*few dog/*little dogs/*little dog
b. little furniture/*little furnitures/*few furniture/*few furnitures
The data here indicate that in addition to the agreement features we have seen so far, common
nouns also place a restriction on the countability value of the selected specifier. Specifically, a
countable noun selects a countable determiner as its specifier. To capture this agreement restric-
tion, we introduce a new feature COUNT (COUNTABLE):
116
(62)
a.
_
¸
¸
_
dogs
HEAD| POS noun
SPR DP[COUNT +]
_
¸
¸
_
b.
_
¸
¸
_
furniture
HEAD| POS noun
SPR DP[COUNT –]
_
¸
¸
_
The lexical specification on a countable noun like dogs requires its specifier to be [COUNT +],
to prevent formations like *much dogs. This in turn means that determiners must also carry the
feature COUNT:
(63)
a.
_
¸
¸
¸
_
many
HEAD
_
_
POS det
COUNT +
_
_
_
¸
¸
¸
_
b.
_
¸
¸
¸
_
the
HEAD
_
_
POS det
COUNT +/–
_
_
_
¸
¸
¸
_
c.
_
¸
¸
¸
_
little
HEAD
_
_
POS det
COUNT –
_
_
_
¸
¸
¸
_
Notice here that some determiners such as the are not specified for a value for COUNT. Ef-
fectively, the value can be either + or −, licensing combination with either a countable or an
uncountable noun (the book or the furniture).
Now consider the following contrast:
(64) a. much advice vs. *many advice
b. *much story vs. many stories
(65) a. much of the advice vs. *many of the advice
b. much of the story vs. many of the stories
Due to the feature COUNT, we understand now the contrast between much advice and *many
advice or the contrast between *much story and many stories. The facts in partitive structures
are slightly different, as (65) shows, but the patterns in the data directly follow from these lexical
entries:
(66)
a.
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
many
HEAD| POS noun
COMPS
_
PP
_
¸
¸
_
PFORM of
NUM pl
DEF +
_
¸
¸
_
_
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
b.
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
much
HEAD| POS noun
COMPS
_
PP
_
¸
¸
_
PFORM of
NUM sing
DEF +
_
¸
¸
_
_
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
The pronoun many requires a PP complement whose inner NP is plural, whereas much does not.
117
6.5.3 Measure Noun Phrases
There are also so-called ‘measure noun phrase’ constructions, which are similar to partitive
constructions. Consider the following contrast:
(67) a. one pound of those beans
b. three feet of that wire
c. a quart of Bob’s cider
(68) a. one pound of beans
b. three feet of wire
c. a quart of cider
Notice here that (67) is a kind of partitive construction whereas (68) just measures the amount
of the NP after of . As the examples show, measure noun phrases do not require a definite article,
which is not an option for the true partitive constructions, repeated here:
(69) *many of beans, *some of wire, *much of cider, *none of yogurt, *one of straw-
berries
In addition, there are several more differences between partitive and measure noun phrases.
For example, measure nouns cannot occur in simple noun phrases. They obligatorily require an
of -NP phrase:
(70) a. *one pound beans vs. one pound of beans
b. *three feet wire vs. three feet of wire
c. *a quart cider vs. a quart of cider
Further, unlike partitive constructions, measure noun phrases require a numeral as their spec-
ifier:
(71) a. *one many of the books, *several much of the beer
b. one pound of beans, three feet of wire
As noted here, many or much in the partitive constructions cannot combine with numerals like
one or several; measure nouns pound and feet need to combine with a numeral like one or three.
Further complications arise due to the existence of defective measure noun phrases. Consider
the following examples:
(72) a. *a can tomatoes/a can of tomatoes/one can of tomatoes
b. a few suggestions/*a few of suggestions/*one few of suggestions
c. *a lot suggestions/a lot of suggestions/*one lot of suggestions
Expressions like few and lot actually behave quite differently. With respect to few, it appears
that a few acts like a complex word. However, lot acts more like a noun, but unlike can, it does
not allow its specifier to be a numeral.
In terms of agreement, measure noun phrases behave like Type I partitive constructions:
(73) a. A can of tomatoes is/*are added.
118
b. Two cans of tomatoes are/*is added.
We can see here that it is the head noun can or cans which determines the NUM value of the
whole NP. The inner NP in the PP does not affect the NUM value at all. These observations lead
us to posit the following lexical entry for a measure noun:
(74)
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
pound
HEAD
_
_
POS noun
NUM sing
_
_
SPR DP
COMPS
_
PP
_
PFORM of
_
_
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
That is, a measure noun like pound requires one obligatory SPR and a PP complement. Unlike
partitive constructions, there is no definiteness restriction on the PP complement.
Finally, there is one set of words whose behavior leaves them somewhere between quantity
words and measure nouns. These are words such as dozen, hundred, and thousand:
(75) a. three hundred of your friends
b. *three hundreds of your friends
c. *three hundreds of friends
d. three hundred friends
e. hundreds of friends/*hundreds friends
Consider the behavior of hundred and hundreds here. The singular hundred, when used as
noun, obligatorily requires a PP[of ] complement as well as a numeral specifier, as in (75a). The
plural hundreds requires no specifier although it also selects a PP complement. Not surprisingly,
similar behavior can be observed with thousand and thousands:
(76) a. several thousand of Bill’s supporters
b. *several thousands of Bill’s supporters
c. *several thousands of supporters
d. several thousand supporters
e. thousands of supporters/*thousands supporters
One way to capture these properties is to assign the following lexical specifications to hundred
and hundreds:
(77)
a.
_
¸
¸
¸
¸
¸
_
hundred
HEAD| POS noun
SPR DP
COMPS
_
PP[PFORM of]
_
_
¸
¸
¸
¸
¸
_
b.
_
¸
¸
¸
¸
¸
_
hundreds
HEAD| POS noun
SPR
COMPS
_
PP[PFOR of]
_
_
¸
¸
¸
¸
¸
_
119
Even though there may be some semantic reasons for all these different kinds of lexical spec-
ifications, for now, stating it all directly in the lexical entries will account at least for the data
given here.
6.6 Modifying an NP
6.6.1 Adjectives as Prenominal Modifiers
Adjectives are expressions commonly used to modify a noun. However, not all adjectives can
modify nouns. Even though most adjectives can be used either as in a modifying (attributive)
function or as a predicate (as in She is tall), certain adjectives are restricted to their usages.
Adjectives such alive, asleep, awake, afraid, ashamed, aware, can be used only predicatively,
whereas others such as wooden, drunken, golden, main and mere are only used attributively:
(78) a. He is alive.
b. He is afraid of foxes.
(79) a. It is a wooden desk.
b. It is a golden hair.
c. It is the main street.
(80) a. *It is an alive fish. (cf. living fish)
b. *They are afraid people. (cf. nervous people)
(81) a. *This objection is main. (cf. the main objection)
b. *This fact is key. (cf. a key fact)
The predicatively-used adjectives are specified with the feature PRD, and with a MOD value
being empty as default, as shown here:
7
(82)
_
¸
¸
¸
¸
¸
_
alive
HEAD
_
¸
¸
_
POS adj
PRD +
MOD
_
¸
¸
_
_
¸
¸
¸
¸
¸
_
This says that alive is used predicatively, and does not have a specification for a MOD value
(the value is empty). This lexical information will prevent predicative adjectives from also func-
tioning as noun modifiers.
8
In contrast to the predicative adjective, a modifying adjective will have the following lexical
entry:
7
All modifiers (adverbial element) will carry the head feature MOD.
8
In addition, all predicative expressions select one argument, their subject (SPR). This information is not shown
here.
120
(83)
_
¸
¸
¸
_
brave
HEAD
_
_
POS adj
MOD N

_
_
_
¸
¸
¸
_
This specifies an adjective which modifies any value whose POS is noun. This will license a
structure like the following:
(84)
NP

_
_
_
_
_
_
_
_
_
2 DP
N

_
SPR 2 DP
_
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
the
AP
_
MOD 1
_
1 N

_
SPR 2 DP
_
·
·
·
·
¸
¸
¸
¸
brave child
6.6.2 Postnominal Modifiers
Postnominal modifiers are basically the same as prenominal modifiers with respect to what they
are modifying. The only difference is that they come after what they modify. Various phrases
can function as such postnominal modifiers:
(85) a. [The boy [in the doorway]] waved to his father.
b. [The man [eager to start the meeting]] is John’s sister.
c. [The man [holding the bottle]] disappeared.
d. [The papers [removed from the safe]] have not been found.
e. [The money [that you gave me]] disappeared last night.
All these postnominal elements bear the feature MOD. Leaving aside detailed discussion of the
relative clause(-like) modifiers in b–e until Chapter 12, we can say that example (85a) will have
the following structure:
121
(86)
NP

_
_
_
_
_
_
_
_
_
2 DP
N

_
SPR 2 DP
_
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
the
1 N

_
SPR 2 DP
_
PP
_
MOD 1
_
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
boy in the doorway
These modifiers must modify an N

, but not a complete NP. This claim is consistent with the
examples above and with the (ungrammatical) examples in (87):
(87) a. *John in the doorway waved to his father.
b. *He in the doorway waved to his father.
A proper noun or a pronoun projects directly to the NP, with no complement or specifier. If it
were the case that post-nominal PP could modify any NP, these examples ought to be acceptable.
122
6.7 Exercises
1. Draw tree structures for the following and mark which expression determines the agree-
ment (AGR) and index values of the subject NP and the main verb.
(i) a. Neither of these men is worthy to lead Italy.
b. None of his customary excuses suffices Edgar now.
c. One of the problems was the robins.
d. All of the plant virus web sites have been conveniently collected in one
central location.
e. Some of the water from melted snow also goes into the ground for plants.
f. Most of the milk your baby ingests during breastfeeding is produced during
nursing.
g. All special rights of voting in the election were abolished.
h. One of major factors affecting the value of diamonds was their weight.
i. Each of these stones has to be cut and polished.
j. Most of her free time was spent attending concerts and plays or visiting
museums and art galleries.
2. Provide a detailed analysis for the following examples, focusing on subject-verb agree-
ment. In doing so, provide the correct AGR and IND value of the subject head noun and
the main verb.
(i) a. The committee were/*was unanimous in their decision.
b. The committee have/*has all now resigned.
c. The crew have/*has both agreed to change sponsor.
d. Her family are/*is all avid skiers.
e. A variety of styles has/*have been in vogue for the last year.
3. Compare the following examples and assign an appropriate structure to each. What kind
of lexical category can you assign to both and few? Can you provide arguments for your
decisions?
(i) a. Both of the workers will wear carnations.
b. Both the workers will wear carnations.
c. Both workers will wear carnations.
d. Both will wear carnations.
(ii) a. Few of the doctors approve of our remedy.
b. Few doctors approve of our remedy.
c. Few approve of our remedy.
4. Provide the correct VFORM value of the underlined verb and identify the noun that
semantically determines this VFORM value.
(i) a. An example of these substances be tobacco.
b. The effectiveness of teaching and learning depend on several factors.
123
c. One of the most serious problems that some students have be lack of moti-
vation.
d. Ten years be a long time to spend in prison.
e. Everyone of us be given a prize.
f. Some of the fruit be going bad.
g. All of his wealth come from real estate investments.
h. Do some of your relatives live nearby?
i. Fifty pounds seem like a lot of weight to lose in one year.
j. News of Persephone and Demeter reach the great gods and goddesses of
Olympus.
k. Half of the year be dark and wintry.
l. Some of the promoters of ostrich meat compare its taste to beef tenderloin.
5. Consider the following pairs of examples and explain the subject-verb and pronoun-
antecedent agreement relationships, and how they affect grammaticality:
(i) a. The committee
i
hasn’t yet made up its
i
/*their
i
mind.
b. The committee
i
haven’t yet made up their
i
/*its
i
mind.
(ii) a. That dog is so ferocious, it even tried to bite itself.
b. *That dog is so ferocious, it even tried to bite himself.
6. Consider the distribution of the reflexive pronouns (myself, yourself, himself, herself )
and simple pronouns (me, you, he, him, her), respectively. Provide rules which can ex-
plain their distribution, in terms of what the pronoun form must, may, or must not be
coreferential with. Please take the grammaticality judgements as given.
(i) a. *I washed me.
b. You washed me.
c. He washed me.
(ii) a. I washed myself.
b. *You washed myself.
c. *He washed myself.
(iii) a. I washed you.
b. *He washed him. (He and him referring to the same person.)
c. He kicked you.
(iv) a. *I washed yourself.
b. You washed yourself.
c. *He washed yourself.
Once you have your own hypothesis for the above data, now examine the following data,
and then determine whether your previous hypothesis can account for this extra data; and
if not, revise your hypothesis so that it can extend to these examples:
(v) a. Harry says that Sally dislikes him.
b. *Harry says that Sally dislikes himself.
124
(vi) a. Sally wishes that everyone would praise her.
b. *Sally wishes that everyone would praise herself.
(vii) a. Sally believes that she is brilliant.
b. *Sally believes that herself is brilliant.
7. Read the following passages and provide detailed lexical entries for the underlined ex-
pressions. For nouns, specify their AGR and IND values.
(i) The power of your mind and the power of your body have a tight connection.
If you have a strong body, your mind feels pumped and healthy, too. If you
have a strong mind, you can craft your body to accomplish amazing things.
I focus on constantly developing this double toughness. I train hard, play
hard, and when life snaps at me, I live hard. This philosophy gets me through
anything and everything.
(ii) One very important and highly productive feature of nouns in English is that
they can be put together to form a new phrase without our having to make
any structural changes to the grammar of either noun, as in tea cup, com-
puter screen, vacuum cleaner, chalk board, internet facility, garden fence,
etc. When two or more nouns combine like this, the first noun is said to mod-
ify the second. In a sense, the first noun is playing the role of an adjective,
which is what most people have in mind when we think about modification,
but nouns can do the job equally well. It is worth mentioning that not every
language has this facility, but native speakers of English are quite happy to
invent their own combinations of nouns in order to describe things, events
or ideas that they have not come across before; this is particularly true in
the workplace where we need constantly to refer to innovations and new
concepts.
9
9
Adopted from Waylink English at http://www.waylink.co.uk/.
125
7
Raising and Control Constructions
7.1 Raising and Control Predicates
As noted in Chapter 5, certain verbs select an infinitival VP as their complement. Compare the
following pairs of examples:
(1) a. John tries to fix the computer.
b. John seems to fix the computer.
(2) a. Mary persuaded John to fix the computer.
b. Mary expected John to fix the computer.
At first glance, these pairs are structurally isomorphic in terms of complements: both try and
tend select an infinitival VP, and expect and persuade select an NP and an infinitival VP. How-
ever, there are several significant differences which motivate two classes, known as control and
raising verbs:
(3) a. Control verbs and adjectives: try, hope, eager, persuade, promise, etc.
b. Raising verbs and adjectives: seem, appear, happen, likely, certain, believe, expect,
etc.
Verbs like try are called ‘control’ or ‘equi’ verbs, where subject is understood to be ‘equivalent’
to the unexpressed subject of the infinitival VP. In lingustic terminology, the subject of the verb
is said to ‘control’ the subject of the infinitival complement. Let us consider the ‘deep structure’
of (1a) representing unexpressed subject of the VP complement of tries:
1
(4) John tries [(for) John to fix the computer].
As shown here, in this sentence it is John who does the action of fixing the computer. In the
original tranformational grammar approach, this deep structure would be proposed and then
1
Deep structure, linked to surface structure, is a theoretical construct and abstract level of representation that seeks
to unify several related observed form and played an important role in the transformational grammar of the 20th century.
For example, the surface structures of both The cat chased the mouse and The mouse was chased by the cat are derived
from the identical deep structure similar to The cat chased the mouse.
127
undergo a rule of ‘Equivalent NP Deletion’ in which the second NP John would be deleted, to
produce the output sentence. This is why such verbs have the label of ‘equi-verbs’.
Meanwhile, verbs like seem are called ‘raising’ verbs. Consider the deep structure of (1b):
(5) seems [John to fix the computer].
In order to derive the ‘surface structure’ (1b), the subject John needs to be raised to the matrix
subject position marked by . This is why verbs like seem are called ‘raising’ verbs.
This chapter discusses the similarities and differences of these two types of verb, and shows
how we explain their respective properties in a systematic way.
7.2 Differences between Raising and Control Verbs
There are many differences between the two classes of verb, which we present here.
7.2.1 Subject Raising and Control
The semantic role of the subject: One clear difference between raising and control verbs is
the semantic role assigned to the subject. Let us compare the following examples:
(6) a. John tries to be honest.
b. John seems to be honest.
These might have paraphrases as follows:
(7) a. John makes efforts for himself to be honest.
b. It seems that John is honest.
As suggested by the paraphrase, the one who does the action of trying is John in (6a). How about
(6b)? Is it John who is involved in the situation of ‘seeming’? As represented in its paraphrase
(7b), the situation that the verb seem describes is not about the individual John, but is rather
about the proposition that John is honest. Due to this difference, we say that a control verb like
try assigns a semantic role to its subject (the ‘agent’ role), whereas a raising verb seem does not
assign any semantic role to its subject (this is what (5) is intended to represent).
Expletive subjects: Since the raising verb does not assign a semantic role to its subject, cer-
tain expressions which do not have a semantic role or any meaning may appear in the subject
position. Such items include the expletives it or there:
(8) a. It tends to be warm in September.
b. It seems to bother Kim that they resigned.
The situation is markedly different with control verbs:
(9) a. *It/*There tries to be warm in September.
b. *It/*There hopes to bother Kim that they resigned.
Since control verbs like try and hope require their subject to have an agent role, an expletive it
or there, which takes no semantic role, cannot function as their subject.
We can observe the same contrast with respect to raising and control adjectives:
128
(10) a. It/*John is easy to please Maja.
b. John/*It is eager to please Maja.
Since the raising adjective easy do not assign any semantic role to its subject, we can have it as
its subject. On the other hand, the control adjective eager assigns a role and thus does not allow
the expletive it as its subject.
Subcategorization: If we look into what determines the subject’s properties, we can see that
in raising constructions, it is not the raising verb or adjective, but the infinitival complement’s
predicate which determines the characteristic of the subject. In raising constructions, the subject
of the raising predicate is selected as the subject of the complement VP. Observe the following
contrast:
(11) a. Stephen seemed [to be intelligent].
b. It seems [to be easy to fool Ben].
c. There is likely [to be a letter in the mailbox].
d. Tabs are likely [to be kept on participants].
in the sense of: ‘The participants will be spied on.’
(12) a. *There seemed [to be intelligent].
b. *John seems [to be easy to fool Ben].
c. *John is likely [to be a letter in the mailbox].
d. *John is likely [to be kept on participants].
For example, the VP to be intelligent requires an animate subject, and this is why (11a) is fine
but (12a) is not. Meanwhile, the VP to be easy to fool Ben requires the expletive it as its subject.
This is why John cannot be the subject in (12b). The contrast in (c) and (d) is similar. The VP
[to be a letter in the mailbox] allows its subject to be there (cf. There is a letter in the mailbox)
but not John. The VP [to be kept on participants] requires a subject which must be the word
tabs in order to induce an idiomatic meaning.
In raising constructions, whatever category is required as the subject of the infinitival VP, is
also required as the subject by the higher VP – hence the intuition of ‘raising’: the requirement
for the subject passes up to the higher predicate.
However, for control verbs, there is no direct selectional relation between the subject of the
main verb and that of the infinitival VP. It is the control verb or adjective itself which fully
determines the properties of the subject:
(13) a. Sandy tried [to eat oysters].
b. *There tried [to be riots in Seoul].
c. *It tried [to bother me that Chris lied].
d. *Tabs try [to be kept on Bob by the FBI].
e. *That he is clever is eager [to be obvious].
129
Regardless of what the infinitival VP would require as its subject, a control predicate requires
its subject to be able to bear the semantic role of agent. For example, in (13b) and (13c), the
subject of the infinitival VP can be there and it, but these cannot function as the matrix subject
– because the matrix verb tried requires its own subject, a ‘trier’.
Selectional Restrictions: Closely related to the difference in selection for the type of subject,
we can observe a related similarity with regard to what are known as ‘selectional restrictions’.
The subcategorization frames, which we have represented in terms of VAL (valence) features,
are themselves syntactic, but verbs also impose semantic selectional restrictions on their sub-
jects or objects. For example, the verb thank requires a human subject and an object that is at
least animate:
(14) a. The king thanked the man.
b. #The king thanked the throne.
c.(?)The king thanked the deer.
d. #The castle thanked the deer.
And consider as well the following examples:
(15) a. The color red is his favorite color.
b. #The color red understands the important issues of the day.
Unlike the verb is, understands requires its subject to be sentient. This selectional restriction
then also explains the following contrast:
(16) a. The color red seems [to be his favorite color].
b. #The color red tried [to be his favorite color].
The occurrence of the raising verb seems does not change the selectional restriction on the
subject. However, tried is different: just like understand, the control verb tried requires its
subject to be sentient, at least. What we can observe here is that the subject of a raising verb
carries the selectional restrictions of the infinitival VP’s subject. This in turn means that the
subject of the infinitival VP is the subject of the raising verb.
Meaning preservation: We have seen that the subject of a raising predicate is that of the in-
finitival VP complement, and it has no semantic role at all coming from the raising predicate.
This implies that an idiom whose meaning is specially composed from its parts will still retain
its meaning even if part of it appears as the subject of a raising verb.
(17) a. The cat seems to be out of the bag.
in the sense of: ‘The secret is out’.
b. #The cat tries to be out of the bag.
In the raising example (17a), the meaning of the idiom The cat is out of the bag is retained.
However, since the control verb tries assigns a semantic role to its subject the cat, ‘the cat’ must
be the one doing the action of trying, and there is no idiomatic meaning.
130
This preservation of meaning also holds for examples like the following:
(18) a. The dentist is likely to examine Pat.
b. Pat is likely to be examined by the dentist.
(19) a. The dentist is eager to examine Pat.
b. Pat is eager to be examined by the dentist.
Since the raising predicate likely does not assign a semantic role to its subject, (18a) and (18b)
have more or less identical meanings – the proposition is about the dentist examining Pat, in
active or passive grammatical forms: the active subject is raised in (18a), and the passive subject
in (18b).
However, the control predicate eager assigns a semantic role to its subject, and this forces
(19a) and (19b) to differ semantically: in (19a), it is the dentist who is eager to examine Pat,
whereas in (19b), it is Pat who is eager to be examined by the dentist. Intuitively, if one of the
examples in (18) is true, so is the other, but this inference cannot be made in (19).
7.2.2 Object Raising and Control
Similar contrasts are found between what are know as object raising and control predicates:
(20) a. Stephen believed Ben to be careful.
b. Stephen persuaded Ben to be careful.
Once again, these two verbs look alike in terms of syntax: they both combine with an NP and
an infinitival VP complement. However, the two are different with respect to the properties of
the object NP in relation to the rest of the structure. Observe the differences between believe
and persuade in (21):
(21) a. Stephen believed it to be easy to please Maja.
b. *Stephen persuaded it to be easy to please Maja.
(22) a. Stephen believed there to be a fountain in the park.
b. *Stephen persuaded there to be a fountain in the park.
One thing we can see here is that unlike believe, persuade does not license an expletive object
(just like try does not license an expletive subject). And in this respect, the verb believe is similar
to seem in that it does not assign a semantic role (to its object). The differences show up again
in the preservation of idiomatic meaning:
(23) a. Stephen believed the cat to be out of the bag.
in the sense: ‘Stephen believed that the secret was out’.
b. *Stephen persuaded the cat to be out of the bag.
While the idiomatic reading is retained with the raising verb believed, it is lost with the control
verb persuaded.
Active-passive pairs show another contrast:
(24) a. The dentist was believed to have examined Pat.
131
b. Pat was believed to have been examined by the dentist.
(25) a. The dentist was persuaded to examine Pat.
b. Pat was persuaded to be examined by the dentist.
With the raising verb believe, there is no strong semantic difference in the examples in (24).
However, in (25), there is a clear difference in who is persuaded. In (25a), it is the dentist, but
in (25b), it is Pat who is persuaded. This is one more piece of evidence that believe is a raising
verb whereas persuade is a control verb, with respect to the object.
7.3 A Simple Transformational Approach
How then can we account for these differences between raising and control verbs or adjectives?
A simple traditional analysis, hinted at earlier, is to derive a surface structure via a derivational
process, for example, from (26a) to (26b):
(26) a. Deep structure: seems [Stephen to be irritating]
b. Surface structure: Stephen seems [t] to be irritating.
To derive (26b), the subject of the infinitival VP in (26a) moves to the matrix subject position,
as represented in the following tree structure:
(27)
S
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
NP
VP
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸

V S
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
seems NP VP[inf ]

¸
¸
¸
¸
¸
¸
¸
Stephen
¸¸
to be irritating
The movement of the subject Stephen to the higher subject position will correctly generate
(26b). This kind of movement to the subject position can be triggered by the requirement that
each English declarative have a surface subject (cf. Chomsky 1981). A similar movement pro-
cess can be applied to the object raising cases:
(28) a. Deep structure: Tom believes [Stephen to be irritating].
b. Surface structure: Tom believes Stephen to be irritating.
Here the embedded subject Stephen moves not to the matrix subject but to the matrix object
position:
132
(29)
S

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
NP
VP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
John
V
NP
S

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
believes NP VP[inf ]

¸
¸
¸
¸
¸
¸
¸
Mary
¸¸
to be irritating
Control constructions are different: there is no movement operation involved. Instead, it is
the lower subject position which has special properties. Consider the examples in (30):
(30) a. John tried to please Stephen.
b. John persuaded Stephen to be more careful.
Since try and persuade assign a semantic roles to their subject, and objects, an unfilled position
of the kind designated above by cannot be allowed. Instead, it is posited that there is an
unexpressed subject of the infinitival VP to please Stephen and to be more careful. This is
traditionally represented as the element called ‘PRO’ (a silent ‘pro’noun), and the examples
will have the following deep structures:
(31) a. John tried [PRO to please Stephen].
b. John persuaded Stephen [PRO to be more careful].
The final tree representations of these are as follows:
(32) a.
S

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
NP
VP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
John
i
V S

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
tried NP VP[inf ]

¸
¸
¸
¸
¸
¸
¸
¸
¸
PRO
i
to please Stephen
133
b.
S

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
NP
VP

_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
John
V
NP
S

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
persuaded Stephen
i
NP VP[inf ]

¸
¸
¸
¸
¸
¸
¸
¸
¸
PRO
i
to be more careful
An independent part of the theory of control links PRO in each case to its antecedent, marked
by coindexing. In (32a), PRO is coindexed with John whereas in (32b), it is coindexed with
Stephen.
These analyses which involve derivations on tree structures are driven by the assumption
that the mapping between semantics and syntax is very direct. For example, in (29), the verb
believe semantically selects an experiencer and a proposition, and this is reflected in the initial
structure. In some syntactic respects, though, believe acts like it has an NP object (separate from
the infintival complement), and the raising operation creates this object. In contrast, persuade
semantically selects an agent, a patient, and a proposition, and hence the structure in (32b)
reflects this: the object position is there all along, so to speak.
The classical transformational approach provides a useful graphical approach to understand-
ing the difference between raising and control. However, it requires assumptions about the na-
ture of grammar rather different from what we have made throughout this book. In the rest of
this chapter, we present a nontransformational account of control and raising.
7.4 A Nontransformational Approach
7.4.1 Identical Syntactic Structures
Instead of the movement approach in which movement operations and various kinds of empty
elements or positions play crucial roles, we simply focus directly on the surface structures of
raising and control constructions. Going back to seem and try, we can observe that both select
an infinitival VP, as in (33), giving the structures in (34):
(33) a.
_
¸
¸
¸
¸
_
seems
SPR NP
COMPS
_
VP
_
VFORM inf
_
_
_
¸
¸
¸
¸
_
134
b.
_
¸
¸
¸
_
tries
SPR NP
COMPS VP
_
VFORM inf
_
_
¸
¸
¸
_
(34) a.
S
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
NP
VP
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
John
V
VP[inf ]
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
seems V VP[bse]

¸
¸
¸
¸
¸
¸
to be honest
b.
S
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
NP
VP
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
John
V
VP[inf ]
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
tries V VP[bse]

¸
¸
¸
¸
¸
¸
to be honest
As shown here, seems and tries actually have identical structures.
The object raising verb expect and the control verb persuade also have identical valence
(SPR and COMPS) information:
(35) a.
_
¸
¸
¸
_
expects
SPR NP
COMPS NP, VP
_
VFORM inf
_

_
¸
¸
¸
_
b.
_
¸
¸
¸
_
persuaded
SPR NP
COMPS NP, VP
_
VFORM inf
_

_
¸
¸
¸
_
These two lexical entries will license the following structures:
135
(36) a.
S
.
.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
_
_
NP

¸
¸
¸
¸
¸
VP
.
.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
_
_
Kim V
NP
VP

¸
¸
¸
¸
¸
¸
¸
¸
expects it to rain tomorrow
b.
S
.
.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
_
_
NP

¸
¸
¸
¸
¸
VP
.
.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
_
_
Kim V
NP
VP

¸
¸
¸
¸
¸
¸
¸
¸
¸
persuaded Mary to leave tomorrow
As can be seen here, raising and control verbs are not different in terms of their subcategoriza-
tion or valence requirements, and so they project similar structures. The question is then how
we can capture the different properties of raising and control verbs. The answer is that their
differences follow from the other parts of the lexical information, in particular, the mapping
relations from syntax to semantics.
7.4.2 Differences in Subcategorization Information
We have observed that for raising predicates, whatever kind of category is required as subject
by the infinitival VP is also required as the subject of the predicate. Some of the key examples
are repeated here:
(37) a. Stephen/*It/*There seemed to be intelligent.
b. It seemed to rain.
c. There seemed to be a fountain in the park.
(38) a. Stephen/*It/*There tried to be intelligent.
b. *It tried to rain.
c. *There tried to be a fountain in the park.
While the subject of a raising predicate is identical to that of the infinitival VP complement., the
subjct of a control predicate has a different requirement. The subject of a control predicate is
coindexed with that of the infinitival VP complement. This difference can be represented in the
lexical information shown in (39). The raising verb involves shared subjects, while the control
verb only shares the index (of the subjects).
136
(39) a.
_
¸
¸
¸
¸
¸
¸
_
seemed
SPR 1
COMPS
_
VP
_
_
VFORM inf
SPR 1
_
_
_
_
¸
¸
¸
¸
¸
¸
_
b.
_
¸
¸
¸
¸
¸
¸
_
tried
SPR NP
i

COMPS
_
VP
_
_
VFORM inf
SPR NP
i

_
_
_
_
¸
¸
¸
¸
¸
¸
_
These two lexical entries represent the difference between seem and try: for seemed, the subject
of the VP complement is identical with its own subject (notated by 1 ) whereas for tried, only the
index value of its VP complement is identical to that of its subject. That is, the VP complement’s
understood subject refers to the same individual as the subject of tried. This index identity in
control constructions is clear when we consider examples like the following:
(40) Someone
i
tried NP
i
to leave the town.
The example here means that whoever someone might refer to, that same person left town.
The lexical entries in (39) generate following structures for the intransitive raising and control
sentences:
137
(41) a.
S
_
¸
_
HEAD 4 | POS verb
SPR
COMPS
_
¸
_

_
_
_
_
_
_
_
_
_
_
_
_
1 NP
VP
_
¸
_
HEAD 4
SPR 1
COMPS
_
¸
_
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
John
V
_
¸
¸
¸
_
HEAD 4
SPR 1
COMPS 2
_
¸
¸
¸
_
2 VP
_
SPR
_
1
_
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
seems V VP[bse]

¸
¸
¸
¸
¸
¸
to be honest
b.
S
_
¸
_
HEAD 3 | POS verb
SPR
COMPS
_
¸
_

_
_
_
_
_
_
_
_
_
_
_
_
1 NP
i
VP
_
¸
_
HEAD 3
SPR 1 NP
COMPS
_
¸
_
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
John
V
_
¸
¸
¸
_
HEAD 3
SPR 1 NP
COMPS 2
_
¸
¸
¸
_
2 VP
_
SPR
_
NP
i
_
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
tries V VP[bse]

¸
¸
¸
¸
¸
¸
to be honest
138
It is easy to verify that these structures conform to all the grammar rules (the Head-Specifier
Rule and Head-Complement Rule) and principles such as the HFP and the VALP.
Object raising and control predicates are no different. Raising verbs select a VP complement
whose subject is fully identical with the object. Control verbs select a VP complement whose
subject’s index value is identical with that of its object. The following lexical entries show these
properties:
(42) a.
_
¸
¸
¸
¸
¸
¸
_
expect
SPR 1 NP
i

COMPS
_
2 NP , VP
_
_
VFORM inf
SPR 2 NP
_
_
_
_
¸
¸
¸
¸
¸
¸
_
b.
_
¸
¸
¸
¸
¸
¸
_
persuade
SPR NP
COMPS
_
NP
i
, VP
_
_
VFORM inf
SPR NP
i

_
_
_
_
¸
¸
¸
¸
¸
¸
_
Let us look at the structures these lexical entries eventually project:
(43)
S
_
¸
_
HEAD 4 | POS verb
SPR
COMPS
_
¸
_

_
_
_
_
_
_
_
_
_
_
_
_
_
1 NP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
VP
_
¸
_
HEAD 4
SPR 1 NP
COMPS
_
¸
_
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
Kim
V
_
¸
¸
¸
_
HEAD 4
SPR 1 NP
COMPS 2 , 3
_
¸
¸
¸
_
2 NP
3 VP
_
SPR
_
2
_
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
expects it to rain tomorrow
139
(44)
S
_
¸
_
HEAD 4 | POS verb
SPR
COMPS
_
¸
_

_
_
_
_
_
_
_
_
_
_
_
_
_
_
1 NP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
VP
_
¸
_
HEAD 4
SPR 1
COMPS
_
¸
_
.
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
Kim
V
_
¸
¸
¸
_
HEAD 4
SPR 1
COMPS 2 NP, 3 VP
_
¸
¸
¸
_
2 NP
i
3 VP
_
SPR NP
i

_

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
persuaded Mary to be more careful
As represented here, the subject of to rain tomorrow is the NP object of expects, while the
subject of to be more careful is coindexed with the independent object of persuade.
7.4.3 Mismatch between Meaning and Structure
We have not yet addressed the issue of differences in the assignment of semantic roles. We
first need to introduce further semantic features, distinguished from syntactic features, for this
issue is closely related to the relationship between syntax and semantics. As we have seen in
Chapter 6, nouns and verbs have IND values. That is, a noun refers to an individual (e.g., i, j, k,
etc) whereas a verb denotes a situation (e.g., s0, s1, s2, etc). In addition, a predicate represents
a semantic property or relation. For example, the meaning of the verb hits in (45a) can be
represented in canonical first-order predicate logic as in (45b):
(45) a. John hits a ball.
b. hit

(j, b)
This shows that the verb hit takes two arguments with the predicate relation hit, with the

notation to indicate the semantic value. The relevant semantic properties can be represented in
a feature structure system as follows:
140
(46)
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
hit
SYN| VAL
_
_
SPR NP
i

COMPS NP
j

_
_
SEM
_
¸
¸
¸
¸
¸
¸
_
IND s0
RELS
_
_
¸
¸
_
PRED hit
AGENT i
PATIENT j
_
¸
¸
_
_
_
¸
¸
¸
¸
¸
¸
_
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
In terms of syntax, hit is a verb selecting a subject and a complement, as shown in the value
of the feature SYN(TAX). The semantic information of the verb is represented with the fea-
ture SEM(ANTICS). It first has the attribute IND(EX), representing what this expression refers
to; as a verb, hit refers to a situation s0 in which an individual i hits an individual j. The se-
mantic relation of hitting is represented using the feature for semantic relations (RELS). The
feature RELS has as its value a list of one feature structure, here with three further features,
PRED(ICATE), AGENT, and PATIENT. The predicate (PRED) relation is whatever the verb
denotes: in this case, hit takes two arguments. The AGENT argument in the SEM value is coin-
dexed with the SPR in the SYN value, while the the PATIENT is coindexed with COMPS. This
coindexing links the subcategorization information of hit with the arguments in its semantic
relation. Simply put, the lexical entry in (46) is the formal representation of the fact that in X
hits Y, X is the hitter and Y is the one hit.
Now we can use these extra parts of the representation for the semantic differences in raising
and control verbs. The subject of a raising verb like seem is not assigned any semantic role,
while that of a control verb like try is definitely linked to a semantic role. Assuming that ‘s0’
or ‘s1’ stands for situations denoted by an infinitival VP, seem and try will have the following
simplified meaning representations:
(47) a. seem

(s1) (‘s1 seems (to be the case”) = s0’)
b. try

(i, s1) (‘i tries to (make) s1 (be the case) = s0’)
These meaning differences are represented in terms of feature structures as follows:
141
(48) a.
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
seem
SYN| VAL
_
¸
¸
¸
¸
¸
¸
¸
_
SPR 1
COMPS
_
VP
_
¸
¸
¸
_
VFORM inf
SPR
_
1
_
IND s1
_
¸
¸
¸
_
_
_
¸
¸
¸
¸
¸
¸
¸
_
SEM
_
¸
¸
¸
¸
_
IND s0
RELS
__
PRED seem
SIT s1
__
_
¸
¸
¸
¸
_
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
b.
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
try
SYN| VAL
_
¸
¸
¸
¸
¸
¸
¸
_
SPR NP
i

COMPS
_
VP
_
¸
¸
¸
_
VFORM inf
SPR
_
NP
i
_
IND s1
_
¸
¸
¸
_
_
_
¸
¸
¸
¸
¸
¸
¸
_
SEM
_
¸
¸
¸
¸
¸
¸
¸
_
IND s0
RELS
_
_
¸
¸
_
PRED try
AGENT i
SIT s1
_
¸
¸
_
_
_
¸
¸
¸
¸
¸
¸
¸
_
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
We can see here that even though the verb seem selects two syntactic arguments, its meaning
relation has only one argument: note that the subject (SPR) is not coindexed with any argument
in the semantic relation. This means that the subject does not receive a semantic role (from
seem). Meanwhile, try is different. Its SPR is coindexed with the AGENT role in the semantics,
and the SPR is also coindexed with the VP complement’s SPR.
Now we look at object-related verbs like expect and persuade. Just like the contrast between
seem and try, the key difference lies in whether the object (y) receives a semantic role or not:
(49) a. expect

(x, s1)
b. persuade

(x, y, s1)
What one expects, as an ‘experiencer’, is a proposition denoted by the VP complement, whereas
what a person x persuades is not a proposition but rather, x persuades an individual y denoted
by the object to perform the proposition denoted by the VP complement. Once again, these
differences are more clearly represented in feature structures:
142
(50) a.
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
expect
SYN| VAL
_
¸
¸
¸
¸
¸
¸
_
SPR NP
i

COMPS
_
2 , VP
_
¸
¸
_
VFORM inf
SPR 2 NP
IND s1
_
¸
¸
_
_
_
¸
¸
¸
¸
¸
¸
_
SEM
_
¸
¸
¸
¸
¸
¸
¸
_
IND s0
RELS
_
_
¸
¸
_
PRED expect
EXPERIENCER i
SIT s1
_
¸
¸
_
_
_
¸
¸
¸
¸
¸
¸
¸
_
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
b.
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
persuade
SYN| VAL
_
¸
¸
¸
¸
¸
¸
_
SPR NP
i

COMPS
_
NP
j
, VP
_
¸
¸
_
VFORM inf
SPR NP
j

IND s1
_
¸
¸
_
_
_
¸
¸
¸
¸
¸
¸
_
SEM
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
IND s0
RELS
_
_
¸
¸
¸
¸
¸
_
PRED persuade
AGENT i
THEME j
SIT s1
_
¸
¸
¸
¸
¸
_
_
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
As seen in the lexical entries, expect has two semantic arguments, EXPERIENCER and SIT: the
object is not linked to a semantic argument of expect. In contrast, persuade has three semantic
arguments: AGENT, THEME, and SIT. We can thus conclude that raising predicates assign one
less semantic role in their argument structures than the number of syntactic dependents, while
with control predicates, there is a one-to-one correlation.
7.5 Explaining the Differences
7.5.1 Expletive Subject and Object
Recall that for raising verbs like seem and believe, the subject and object respectively is depen-
dent for its semantic properties solely upon the type of VP complement. This fact is borne out
by the examples in (51):
(51) a. There/*It/*John seems [to be a fountain in the park].
143
b. We believed there/*it/*John [to be a fountain in the park].
Control verbs are different, directly assigning a semantic role to the subject or object. Hence
expletives cannot appear (illustrated here for the subject of try):
(52) a. *There/*It/John tries to leave the country.
b. We believed *there/*it/John to try to leave the country.
7.5.2 Meaning Preservation
We noted above that in a raising example such as (53a), the idiomatic reading can be preserved,
but not in a control example like (53b):
(53) a. The cat seems to be out of the bag.
b. The cat tries to be out of the bag.
This is once again because the subject of seems does not have any semantic role: its subject is
identical with the subject of its VP complement to be out of the bag, whereas the subject of tries
has its own agent role.
Exactly the same explanation applies to the following contrast:
(54) a. The dentist is likely to examine Pat.
b. Pat is likely to be examined by the dentist.
Since likely is a raising predicate, as long as the expressions The dentist examines Pat and Pat
is examined by the dentist have roughly the same meaning, the two raising examples will also
have roughly the same meaning.
However, control examples are different:
(55) a. The dentist is eager to examine Pat.
b. Pat is eager to be examined by the dentist.
The control adjective eager assigns a semantic role to its subject independent of the VP com-
plement, as given in the following lexical entry:
(56)
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
eager
SYN| VAL
_
¸
¸
¸
¸
_
SPR NP
i

COMPS
_
VP
_
_
VFORM inf
IND s1
_
_
_
_
¸
¸
¸
¸
_
SEM
_
¸
¸
¸
¸
¸
¸
_
IND s0
RELS
_
_
¸
¸
_
PRED eager
EXPERIENCER i
SIT s1
_
¸
¸
_
_
_
¸
¸
¸
¸
¸
¸
_
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
144
This then means that (55a) and (55b) must differ in that in the former, it is the dentist who is
eager to perform the action denoted by the VP complement, whereas in the latter, it is Pat who
is eager.
7.5.3 Subject vs. Object Control Verbs
Consider finally the following two examples:
(57) a. They persuaded me to leave.
b. They promised me to leave.
Both persuaded and promised are control verbs since their object is assigned a semantic role
(and so is their subject). This in turn means that their object cannot be an expletive:
(58) a. *They persuaded it to rain.
b. *They promised it to rain.
However, the two are different with respect to the controller of the infinitival VP. Consider who
is understood as the unexpressed subject of the infinitival verb here. In (57a), it is the object
me which semantically functions as the subject of the infinitival VP. Yet, in (57b), it is the
subject they who will do the action of leaving. Due to this fact, verbs like promise are known as
‘subject control’ verbs, whereas those like persuade are ‘object control’ verbs. This difference
is straightforwardly represented in their lexical entries:
(59)
_
¸
¸
¸
¸
¸
¸
_
persuade
SPR NP
i

COMPS
_
NP
j
, VP
_
_
VFORM inf
SPR NP
j

_
_
_
_
¸
¸
¸
¸
¸
¸
_
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
promise
SPR NP
i

COMPS
_
NP
j
, VP
_
¸
¸
_
VFORM inf
SPR NP
i

IND s1
_
¸
¸
_
_
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
Based on world knowledge, we know that when one promises someone to do something, this
means that the person who makes the promise will do the action. Meanwhile, when one per-
suades someone to do something, the person who is persuaded will do the action. The lexical
entries here reflect this knowledge of the relations in the world.
In sum, the properties of rasing and control verbs presented here can be summarized as
follows:
145
.
Unlike control predicates, raising predicates are unusual in that they do not assign a semantic
role to their subject or object. The absence of a semantic role accounts for the possibility of
expletives it or there or parts of idioms as subject or object with raising predicates, and not
with control predicates.
.
With control predicates, the VP complement’s unexpressed subject is coindexed with one
of the syntactic dependents. With raising predicates, the entire syntactic-semantic value of
the subject of the infinitival VP is structure-shared with that of one of the dependents of
the predicate. This ensures that whatever category is required by the raising predicate’s VP
complement is the raising predicate’s subject (or object). Notice that even non-NPs can be
subject in certain kinds of example (see (60)).
(60) a. Under the bed is a fun place to hide.
b. Under the bed seems to be a fun place to hide.
c. *Under the bed wants to be a fun place to hide. (want is a control verb)
146
7.6 Exercises
1. Draw trees for the following sentences and provide the lexical entries for the italicized
verbs:
(i) a. Kim may have admitted to let Mary mow the lawn.
b. Gregory appears to have wanted to be loyal to the company.
c. Jones would prefer for it to be clear to Barry that the city plans to sue him.
d. John continues to avoid the conflict.
e. The captain ordered the troops to proceed.
f. He coaxed his brother to give him the candy.
g. Frank hopes to persuade Harry to make the cook wash the dishes.
h. John wants it to be clear to Ben that the city plans to honor him.
2. Explain why the following examples are ungrammatical, based on the lexical entries of
the predicates:
(i) a. *John seems to rain.
b. *John is likely to appear that he will win the game.
c. *Beth tried for Bill to ask a question.
d. *He believed there to be likely that he won the game.
e. *It is likely to seem to be arrogant.
f. *Sandy appears that Kim is happy.
g. *Dana would be unlikely for Pat to be called upon.
h. *Robin is nothing in the box.
i. *It said that Kim was happy.
j. *There preferred for Sandy to get the job.
3. In this chapter, we have learned that predicates (verbs and adjectives) can be classified
into two main groups, raising and control, as represented in the following simple table:
Raising predicates Control predicates
Intransitive seem, . . . try, . . .
Transitive believe, . . . persuade, . . .
Decide in which group each of the following lexical items belongs. In doing so, consider
the it, there, and idiom tests that this chapter has introduced:
(i) certain, anxious, lucky, sure, apt, liable, bound, careful, reluctant
(ii) tend, decide, manage, fail, happen, begin, hope, intend, refuse
4. As we have seen in Exercise 4 of Chapter 3, there is agreement between the copula be
and the postcopular NP in so-called ‘there’ constructions, as shown again here:
(i) a. There is/*are only one chemical substance involved in nerve transmission.
b. There *is/are more chemical substances involved in nerve transmission.
This kind of agreement relationship can be encoded as a property of the be used in this
construction:
147
(ii)
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
be
HEAD
_
_
POS verb
AUX +
_
_
SUBJ
_
NP
_
_
NFORM there
AGR 1
_
_
_
COMPS
_
NP[AGR 1 ], XP[PRED +]
_
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
This lexical information specifies that be selects there as its subject and two complements
(NP and XP). In this case, the complement NP’s agreement feature AGR is identical with
that of the subject there. This then will ensure that the verb agrees with the postcopular
NP.
Given this, how might we account for the following contrasts? Provide a structure for
each example and explain the rules or principles which are violated in the ungrammatical
versions:
(ii) a. There is/*are believed to be a sheep in the park.
b. There *is/are believed to be sheep in the park.
c. There seems/*seem to be no student absent.
d. There is/*are likely to be no student absent.
5. Discuss the similarities and differences among the following three examples; use the it,
there and idiom tests.
(i) a. Pat expected Leslie to be aggressive.
b. Pat persuaded Leslie to be aggressive.
c. Pat promised Leslie to be aggressive.
Also, state see the controller is of the infinitival VP in each case.
6. Consider the following data and discuss briefly what can be the antecedent of her and
herself .
(i) a. Kevin urged Anne to be loyal to her.
b. Kevin urged Anne to be loyal to herself.
Now consider the following data and discuss the binding conditions of ourselves and us.
In particular, determine the relevance of the ARG-ST list with respect to the possible and
impossible binding relations.
(ii) a. We
i
expect the dentist to examine us
i
.
b. *We
i
expect the dentist to examine ourselves
i
.
c. We expect them to examine themselves.
d. *We expect them
i
to examine them
i
.
(iii) a. We
i
persuaded the dentist to examine us
i
.
b. *We
i
persuaded the dentist to examine ourselves
i
.
148
c. We persuaded them
i
to examine themselves
i
.
d. *We persuaded them
i
to examine them
i
.
7. Read the following passage and provide a tree structure for each sentence, and lexical
entries for the underlined words.
(i) If you’ve ever tried to persuade other people to buy your product or service,
you also know that this can be one of the most discouraging and difficult
things to try to do as a business owner. In fact, this way of trying to get
business by trying to persuade other people is one of the factors that causes
most business owners to dislike or even hate the process of marketing and
selling. It’s very tough to try to convince other people to buy from you,
especially if it’s against their will. After all, if you try to persuade someone
to buy from you, you try to cause that person to do something. And usually
there’s always some kind of pressure involved in this process.
149
8
Auxiliary Constructions
8.1 Basic Issues
The English auxiliary system involves a relatively small number of elements interacting with
each other in complex and intriguing ways. This has been one of the main reasons that the
system has been one of the most extensively analyzed empirical domains in the literature on
generative syntax.
Ontological Issues: One of the main issues in the study of English auxiliary system concerns
ontological issues: is it necessary to posit ‘auxiliary’ as an independent part of speech or not?
Auxiliary verbs can be generally classified as follows:
.
modal auxiliary verbs such as will, shall, may, etc.: have only finite forms
.
have/be: have both finite & nonfinite forms
.
do: has a finite form only with vacuous semantic meaning
.
to: has a nonfinite form only with apparently vacuous semantic meaning
Such auxiliary verbs behave differently from main verbs in various respects. There have been
arguments for treating these auxiliary verbs as simply having the lexical category V, though
being different in terms of syntactic distribution and semantic contribution. For example, in
terms of similarities, both auxiliary and main verbs behave alike in carrying tense information
and participating in some identical syntactic constructions such as gapping, as shown in (1):
(1) a. John drank water and Bill wine.
b. John may drink water, and Bill drink beer.
Such phenomena provide apparent stumbling blocks to assigning a totally different lexical cat-
egory to English auxiliary verbs, compared to main verbs.
Distinctions between auxiliary and main verbs: One important issue that comes up in the
study of the English auxiliary system is that of which words function as auxiliary verbs, and how
we can differentiate them. Most reliable criteria for auxiliaryhood lie in syntactic phenomena
such as negation, inversion, contraction, and ellipsis (usually known as the ‘NICE’ properties):
151
1. Negation: Only auxiliary verbs can be followed by not marking sentential negation (in-
cluding have and be):
(2) a. Tom will not leave.
b. *Tom kicked not a ball.
2. Inversion: Only auxiliary verbs undergo subject-auxiliary inversion.
(3) a. Will Tom leave the party now?
b. *Left Tom the party already?
3. Contraction: Only auxiliary verbs have contracted forms with the suffix n’t.
(4) a. John couldn’t leave the party.
b. *John leftn’t the party early.
4. Ellipsis: The complement of an auxiliary verb, but not of a main verb, can be elided.
(5) a. If anybody is spoiling the children, John is .
b. *If anybody keeps spoiling the children, John keeps .
In addition to these NICE properties, tag questions are another criterion: an auxiliary verb can
appear in the tag part of a tag question, but not a main verb:
(6) a. You should leave, shouldn’t you?
b. *You didn’t leave, left you?
The position of adverbs or so-called floated quantifiers can also be adopted in differentiating
auxiliary verbs from main verbs. The difference can easily be seen in the following contrasts:
(7) a. She would never believe that story.
b. *She believed never his story.
(8) a. The boys will all be there.
b. *Our team played all well.
Adverbs such as never and floated quantifiers such as all can follow an auxiliary verb, but not a
main verb.
Ordering Restrictions: The third main issue in the syntax of auxiliaries centers on how
to capture the ordering restrictions among them. They are subject to restrictions which limit
the sequences in which they can occur, and the forms in which they can combine with other
auxiliary verbs. Observe the following examples:
(9) a. The children will have been being entertained.
b. He must have been being interrogated by the police at that very moment.
(10) a. *The house is been remodelling.
b. *Margaret has had already left.
c. *He has will seeing his children.
152
d. *He has been must being interrogated by the police at that very moment.
As shown here, when there are two or more auxiliary verbs, they must come in a certain order.
In addition, note that each auxiliary verb requires the immediately following to be in a particular
morphological form (e.g., has eaten vs. *has eating).
In the study of the English auxiliary system, we thus need to address at least the following
issues:
.
Should we posit an auxiliary category?
.
How can we distinguish main verbs from auxiliary verbs?
.
How can we account for phenomena (such as the NICE group) which are sensitive to
the presence of an auxiliary verb?
.
Howcan we capture the ordering and co-occurrence restrictions among auxiliary verbs?
This chapter provides answers to these fundamental questions related to the English auxiliary
system.
1
8.2 Transformational Analyses
The seminal work on the issues above is that of Chomsky (1957). His analysis, introducing the
rule in (11), directly stipulates the ordering relations among auxiliary verbs:
(11) Aux →Tense (Modal) (have + en) (be + ing)
The PS rule in (11) would generate sentences with or without auxiliary verbs as in (12):
(12) a. Mary solved the problem.
b. Mary would solve the problem.
c. Mary was solving the problem.
d. Mary would easily solve the problem.
For example, the following structure schematizes some examples in (12):
(13)
S

_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
NP

¸
¸
¸
¸
¸
¸
AUX

¸
¸
¸
¸
¸
¸
VP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
Mary Past will Adv VP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
easily V NP

¸
¸
¸
¸
¸
¸
¸
solve the problem
1
This chapter is based on Kim (2000).
153
To get to the surface structure, the famous ‘Affix Hopping’ rule of Chomsky (1957) ensures that
the affixal tense morpheme (Past) in Tense is hopped to M (Modal) (will), or over onto the main
verb (solve) if the Modal does not appear. If the modal is there, Past hops onto will and we get
Mary would (easily) solve the problem. If the modal is not there, the affix Past will hop onto the
main verb solve, giving Mary solved the problem.
In addition to the Affix Hopping Rule, typical transformational analyses introduce the En-
glish particular rule called ‘ do-support’ for dealing with the NICE properties in clauses that
otherwise have no auxiliary verb:
(14) a. *Mary not avoided Bill.
b. Mary did not avoid Bill.
The presence of not in a position like Adv in the tree (13) has been claimed to prevent the Tense
affix from hopping over to the verb (as not intervenes). As a last-resort option, the grammar
introduces the auxiliary verb do onto which the affix Tense is hopped. This would then generate
(14b). In other words, the position of do diagnoses the position of Tense in the structure.
The analysis lays bare the systematicity of the auxiliary system, but nevertheless it misses
several important points. For example, the constituent structure in (13) does not provide con-
stituent properties we find in coordinate structures, because the first auxiliary and the rest of the
sentence (the VP) do not form a constituent.
(15) a. Fred [must have been singing songs] and [probably was drinking beer].
b. Fred must both [have been singing songs] and [have been drinking beer].
c. Fred must have both [been singing songs] and [been drinking beer].
d. Fred must have been both [singing songs] and [drinking beer].
As we have seen earlier, identical phrasal constituents can be conjoined. The coordination ex-
amples here indicate that a VP with one auxiliary verb or more behaves just like the one without
any.
More recent analyses in this tradition (e.g., Chomsky 1986) use X

-theory to provide IP
and CP as categories for clausal syntax, which can deal with the coordination data just given.
Nevertheless, there are many problems which transformational analyses cannot easily overcome
(for a thorough review, see Kim 2000, Kim and Sag 2002).
8.3 A Lexicalist Analysis
In the approach we take in this book, ordering restrictions on auxiliay verbs will follow from
the correct specification of their lexical properties, interacting with the regular rules of syntactic
combination. The analysis requires no movement, either of whole words, or of affixes. In this
section, we discuss several different subtypes of auxiliary.
8.3.1 Modals
One main property of modal auxiliaries such as will, shall and must is that they place no seman-
tic restrictions on their subject, indicating their status as raising verbs (see Chapter 7).
154
(16) a. There might be a unicorn in the garden.
b. It will rain tomorrow.
c. John will leave the party earlier.
(17) a. *There hopes to finish the project.
b. *The bus hopes to be here at five.
As seen from the contrast, the type of subject in (16) depends on what kind of subject (there or
it or a regular NP) is required by the verb right after the modal. This is typical of raising verbs,
and different from what we see in examples with a control verb like hope in (17), which must
have a referential subject.
Modal verbs can only occur in finite (plain or past) forms. They cannot occur neither as
infinitives nor as participles.
2
(18) a. I hope *to would/*to can/to study in France.
b. *John stopped can/canning to sign in tune.
Modals do not show 3rd person inflection in the present tense, nor a transparent past tense form.
(19) a. *John musts/musted leave the party early.
b. *John wills leave the party early.
In terms of their own selectional properties, modal verbs select a base VP as their complement:
(20) a. John can [kick/*kicked/*kicking/*to kick the ball].
b. John will [kick/*kicked/*kicking/*to kick the ball].
Reflecting these basic lexical properties, a modal auxiliary will have at least the following lexi-
cal information:
(21)
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
must
HEAD
_
¸
¸
_
POS verb
VFORM fin
AUX +
_
¸
¸
_
VAL
_
¸
¸
¸
¸
_
SPR 1 NP
COMPS
_
VP
_
_
VFORM bse
SPR 1 NP
_
_
_
_
¸
¸
¸
¸
_
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
In the lexical information given here, we notice at least three things: first, auxiliary verbs have
the head feature AUX, which differentiates them from main verbs. In addition, the rule shows
that a modal verb selects a base VP as its complement. This subcategorization information will
rule out examples like the following, as well as the ungrammatical examples in (20):
2
As we have seen in 5.2.1, the VFORM value fin includes pres, pst, and pln whereas nonfin includes ing, en, inf ,
and bse.
155
(22) a. *Kim must [
VP[fin]
bakes a cake].
b. *Kim must [
VP[fin]
baked a cake].
c. *Kim must [
VP[fin]
will bake a cake].
The possible and impossible structures can be more clearly represented in tree format:
(23) a.
VP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
V
_
_
AUX +
COMPS 2 VP[bse]
_
_ 2 VP[bse]

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
must
V[bse]
NP

¸
¸
¸
¸
¸
¸
bake a cake
b.
*VP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
_
AUX +
COMPS VP[bse]
_
_ VP[fin]

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
must
V[fin]
NP

¸
¸
¸
¸
¸
¸
baked a cake
As can be easily seen here, the modal auxiliary must requires a VP[bse] as its complement:
The VP[fin] in (23b) cannot function as the complement of must.
The lexical entry in (21) also specifies that the VP’s subject is identical with the subject of the
modal auxiliary (indicated by the box 1 ). This specification, a crucial property of raising verbs
we have discussed in the previous chapter, rules out the ungrammatical versions of examples
like the following:
(24) a. It/*John will [
VP[bse]
rain tomorrow].
b. There/*It may [
VP[bse]
exist a man in the park].
The VP rain tomorrow requires the expletive subject it, not any other NP, such as John; and the
VP exist a man in the park requires there and nothing else its subject.
In addition, since modal verbs have the specification [VFORM fin], they cannot occur in any
environment where finite verbs are prohibited.
156
(25) a. *We expect there to [
VP[fin]
will rain].
b. *It is vital that we [
VP[fin]
will study everyday].
The simple lexical information of modal verbs given in (21), which is required in almost any
analysis, explains the main distributional possibilities of modal verbs.
8.3.2 Be and Have
The auxiliary verbs have and be are different from modal verbs. For example, unlike modals,
they have nonfinite forms (would have, would be, to have/to be); they have a 3rd person inflec-
tion form (has, is); they select not a base VP as their complement, but an inflected nonfinite
form. In addition, they are different from modals in that they also have uses as main verbs,
though in some cases with different syntax from when they are auxiliaries.
Consider the examples in (26):
(26) a. He is a fool.
b. He has a car.
On the assumption that every sentence has a main verb, be and have here are main verbs. How-
ever, a striking property of be is that it still shows the properties of an auxiliary: it exhibits all
of the NICE properties, as we will see below. The usage of be actually provides a strong reason
why the grammar should allow a verb categorized as ‘V’ to also have the feature specification
[AUX +]; be in (26a) is clearly a verb, yet it also behaves exactly like an auxiliary.
The verb be has three main uses: as a copula selecting an predicate XP, as an aspectual
auxiliary with a progressive VP following, and as an auxiliary as part of the passive construction:
(27) a. John is in the school.
b. John is running to the car.
c. John was found in the office.
There is no categorical or syntactic reason to distinguish these three, for they all have NICE
properties: they show identical behavior with subject-auxiliary inversion, their position relative
adverbs including floated quantifiers, and so forth.
(28) Subject-Aux Inversion:
a. Was the child in the school? (*Did the child be in the school?)
b. Was the child running to the car?
c. Was the child found?
(29) Position of an adverb:
a. The child (?never) was (never) crazy. (The child (never) became (*never) crazy.)
b. The child (?never) was (never) running to the car.
c. The child (?never) was (never) deceived.
Thus, all three uses have the lexical information given in (30) as their common denominator:
3
3
XP here is a variable over phrasal categories such as NP, VP, AP, and PP.
157
(30)
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
HEAD
_
_
POS verb
AUX +
_
_
VAL
_
¸
¸
¸
¸
_
SPR 1 NP
COMPS
_
XP
_
_
PRD +
SPR 1
_
_
_
_
¸
¸
¸
¸
_
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
All three bes bear the feature AUX with the + value, and select a predicative phrase whose
subject is identical with be’s subject. Every use of be thus has the properties of a raising verb.
The main difference between the three uses lies in the XP’s VFORM value:
4
(31) a. copula be:
_
COMPS XP
_
b. progressive be:
_
COMPS VP[VFORM ing]
_
c. passive be:
_
COMPS VP[VFORM pass]
_
As given here, the copula be needs no further specification: any phrase that can function as a
predicate can be its COMPS value. The progressive be requires its complement to be VP[ing],
and the passive be requires its complement to be VP[pass]. Hence, examples like those in (32)
are straightforwardly generated:
(32) a. John is [
AP
happy about the outcome].
b. John was [
VP[ing]
seeing his children].
c. The children are [
VP[pass]
seen in the yard].
Auxiliary have is rather similar in its properties to auxiliary be, and it selects a past participle
VP complement.
(33) a. John has not sung a song.
b. Has John sung a song?
c. John hasn’t been singing a song.
d. John has sung a song and Mary has , too.
Given facts like these, we can posit the following information in the lexical entry for auxiliary
have, part of the perfect aspect construction:
4
See Chapter 9 for the further discussion of passive constructions.
158
(34)
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
have
HEAD
_
_
POS verb
AUX +
_
_
SPR 1
COMPS
_
VP
_
_
VFORM en
SPR 1
_
_
_
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
The interaction of subcategorization and morphosyntactic information is enough to predict the
ordering restrictions among modals. For example, the auxiliaries have and be can follow a
modal since both have bse as its VFORM value:
(35) a. John can [
VP[bse]
have danced].
b. John can [
VP[bse]
be dancing].
In addition, we can predict the following ordering too:
(36) a. He has [seen his children].
b. He will [have [been [seeing his children]]].
c. He must [have [been [being interrogated by the police at that very moment]]].
(37) a. *Americans have [paying income tax ever since 1913].
b. *George has [went to America].
(37a) is ungrammatical since have requires a perfect participle VP. (37b) is out since the fol-
lowing VP is finite.
In some varieties of English, typically in British English, the main verb have also has the
specification [AUX +], as evidenced by the (b) examples below:
(38) a. You are a student.
b. You have not enough money.
(39) a. Are you a student?
b. Have you enough money?
The main verbs be and have show the NICE properties; even though they are main verbs, they
have the syntax of auxiliaries. This fact supports the idea that every sentence has a (main) verb
in it, at least, while the surface syntax of a verb is determined by whether it has the specification
[AUX +] or [AUX −].
8.3.3 Periphrastic do
Next we discuss the so-called “dummy” do, which is used as an auxiliary in the absence of other
(finite) auxiliaries. This do also exhibits the NICE properties:
(40) a. John does not like this town.
b. In no other circumstances does that distinction matter.
159
c. They didn’t leave any food.
d. Jane likes these apples even more than Mary does .
Like the modals, do does not appear in nonfinite clauses.
(41) a. *They expected us to do/should leave him.
b. I found myself needing/*doing need/*should needing sleep.
There are also some properties which distinguish do from other auxiliaries. First, unlike other
auxiliaries, do appears neither before nor after any other auxiliary:
(42) a. *He does be leaving.
b. *He does have been eating.
c. *They will do come.
Second, the verb do has no obvious intrinsic meaning to speak of. Except for carrying the
grammatical information about tense and agreement (in present tense), it has no semantic con-
tribution.
Third, if do is used in a positive statement, it needs to be emphatic (stressed). But in negative
statements and questions, no such requirement exists.
(43) a. *John does leave.
b. John DOES leave.
(44) a. John did not come.
b. John DID not come. (more likely in this case: John did NOT come.)
(45) a. Did John find the solution?
b. How long did it last?
The most economical way of representing these lexical properties is to give do the lexical
entry given in (46).
(46)
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
do
HEAD
_
¸
¸
_
POS verb
AUX +
VFORM fin
_
¸
¸
_
VAL
_
¸
¸
¸
¸
¸
¸
_
SPR 1 NP
COMPS
_
VP
_
¸
¸
_
AUX −
VFORM bse
SPR 1
_
¸
¸
_
_
_
¸
¸
¸
¸
¸
¸
_
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
Like other auxiliaries including modals, do is specified to be [AUX +], which ensures that do is
sensitive to negation, inversion, contraction, and ellipsis (NICE properties), just like the other
160
auxiliaries. Further, do selects a subject NP and a VP complement whose unrealized subject is
structure-shared with its subject ( 1 ). Treating do as a raising verb like other English auxiliaries
is based on typical properties of raising verbs, one of which is that raising verbs allow expletives
as their subject, as we have seen above:
(47) a. John may leave.
b. It may rain.
c. *John may rain.
(48) a. John did not leave.
b. It did not rain.
c. *John did not rain.
The [AUX +] specification and raising-verb treatment of do account for the similarities to other
auxiliaries and modals.
The differences stemfrom the lexical specifications on the feature values for HEAD|POS and
its complement VP. Unlike have and be, do is specified to be fin. This property then accounts
for why no auxiliary element can precede do, for only the first verb in a sequence may be finite.
(49) a. He might [have left].
b. *He might [do leave].
The requirement on the complement VP of the auxiliary do is [VFORM bse]. This feature
specification blocks modals from heading the VP following do, for modals are specified to be
[fin], predicting the ungrammaticality of the examples in (50):
(50) a. *He does [can leave here].
b. *He does [may leave here].
The lexical entry further specifies that the complement of do is a VP[AUX−]. This requirement
will correctly predict the ungrammaticality of examples in (51) and (52).
(51) a. *Jim [DOES [have supported the theory]].
b. *The proposal [DID [be endorsed by Clinton]].
(52) a. *I [do [not [have sung]]].
b. *I [do [not [be happy]]].
In (51) and (52), the VPs following the auxiliary do, stressed or not, bear the feature [AUX +]
inherited from the auxiliaries have and be. This explains their ungrammaticality.
5
5
There are special properties of do in imperatives, and different properties with don’t. Do in imperatives can occur
before another auxiliary like be and have.
(i) a. Do be honest!
b. Don’t be silly!
Do and don’t in imperatives also have one distinct property: only don’t allows the subject you to follow (try inserting
you in (ia) and (ib)). Their properties indicate that they have different lexical information from the verb forms used in
non-imperatives.
161
8.3.4 Infinitival Clause Marker to
The auxiliary verbs to and do, in addition to differing by just one phonological feature, voicing,
differ in an important syntactic property: do appears only in finite contexts, and to only in non-
finite contexts. The verb to is, of course, the actual marker of the infinitive in English. Even
though it has the form of a preposition, its syntactic behavior puts it in the class of auxiliary
verbs (cf. Gazdar et al. 1985):
(53) a. *John believed Kim to do not leave here.
b. John believes Kim not to leave here.
These verbs share the property that they obligatorily take bare verbal complements (hence, non-
base forms or modals cannot head the complement VP):
(54) a. *John believed Kim to leaving here.
b. *John did not leaving here.
c. *John expect to must leave.
d. *John did not may leave.
In terms of NICE properties, to also falls under the VP ellipsis criterion:
(55) a. Tom wanted to go home, but Peter didn’t want to .
b. Lee voted for Bill because his father told him to .
These properties indicate that to should have a lexical entry like this:
(56)
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
HEAD
_
¸
¸
_
POS verb
AUX +
VFORM inf
_
¸
¸
_
SPR
_
1 NP
_
COMPS
_
VP
_
_
VFORM bse
SPR 1
_
_
_
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
It is an infinitive auxiliary verb, whose complement must be headed by a V in the bse form.
8.4 Explaining the NICE Properties
In this section we discuss how we can account for the NICE properties, which are key diagnos-
tics for presence of auxiliary verbs.
8.4.1 Auxiliaries with Negation
The English negative adverb not leads a double life: one as a nonfinite VP modifier, marking
constituent negation, and the other as a complement of a finite auxiliary verb, marking sentential
negation. Constituent negation is the name for a construction where negation combines with
some constituent to its right, and negates exactly that constituent:
162
Constituent Negation: The properties of not as a nonfinite VP modifier can be supported
from its similarities with adverbs such as never in nonfinite clauses as given in (57):
(57) a. Kim regrets [never/not [having seen the movie]].
b. We asked him [never/not [to try to call us again]].
c. Duty made them [never/not [miss the weekly meetings]].
Taking not to modify a nonfinite VP, we can predict its various positional possibilities in nonfi-
nite clauses, via the following lexical entry:
(58)
_
¸
¸
¸
¸
¸
¸
_
not
HEAD
_
¸
¸
¸
_
POS adv
NEG +
MOD
_
VP[VFORM nonfin]
_
_
¸
¸
¸
_
_
¸
¸
¸
¸
¸
¸
_
The adverb not modifies any nonfinite VP:
(59) Constituent Negation:
VP[nonfin]
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
Adv
_
MOD 1 VP[nonfin]
_ 1 VP[nonfin]

¸
¸
¸
¸
¸
¸
¸
¸
not . . . . . .
In the grammatical examples in (60) and (61), not modifies a nonfinite VP; in the ungrammatical
examples the VP[nonfin] lexical constraint is violated.
(60) a. [Not [speaking English]] is a disadvantage.
b. *[Speaking not English] is a disadvantage.
c. *Lee likes not Kim.
(61) a. Lee is believed [not
VP[inf]
[to like Kim]].
b. Lee is believed to [not
VP[bse]
[like Kim].
c. *Lee is believed [to
VP[bse]
[like not Kim]].
Sentential Negation: In contrast to constituent negation, there is sentential negation, which
is the canonical expression of negation. One way to distinguish the two types of negation comes
from scope possibilities in an example like (62) (cf. Warner 2000).
(62) The president could not approve the bill.
Negation here could have the two different scope readings paraphrased in (63).
163
(63) a. It would be possible for the president not to approve the bill.
b. It would not be possible for the president to approve the bill.
The first interpretation is constituent negation; the second is sentential negation.
Sentential not may not modify a finite VP:
(64) a. Lee never/*not left. (cf. Lee did not leave.)
b. Lee will never/not leave.
This construction shows one clear difference between never and not: not can only modify a
nonfinite VP, a property further illustrated by the following examples:
(65) a. John could [not [leave the town]].
b. John wants [not [to leave the town]].
(66) a. *John [not [left the town]].
b. *John [not [could leave the town]].
Another difference between never and not is found in the VP ellipsis construction. Observe
the following contrast:
(67) a. Mary sang a song, but Lee never did .
b. *Mary sang a song, but Lee did never .
c. Mary sang a song, but Lee did not .
The data here indicate that not behaves differently from adverbs like never in finite contexts,
even though they all behave alike in nonfinite contexts. never is a true diagnostic for a VP-
modifier, and we use contrasts between it and not to reason what the properties of not must
be.
We have seen the lexical representation for constituent negation not above. Sentential not
appears linearly in the same position – following a finite auxiliary verb – but shows different
syntactic properties. The most economical way to differentiate sentential negation from con-
stituent negation is to assume that sentential negation is actually a syntactic complement of a
finite auxiliary verb (cf. Kim and Sag 1995, 2002). That is, we can assume that when not is
used as a marker of sentential negation, it is selected by the preceding finite auxiliary verb via a
lexical rule:
(68) Negative Auxiliary Verb Lexical Rule:
_
¸
¸
¸
_
HEAD
_
_
AUX +
VFORM fin
_
_
COMPS 1 XP
_
¸
¸
¸
_

_
¸
¸
¸
¸
¸
_
HEAD
_
¸
¸
_
AUX +
VFORM fin
NEG +
_
¸
¸
_
COMPS Adv[NEG +], 1 XP
_
¸
¸
¸
¸
¸
_
This lexical rule allows a finite auxiliary verb with a complement ( 1 ) to select an extra comple-
ment, marked [NEG +]. This rule gives a lexical entry which licenses the following structure
for sentential negation:
164
(69)
VP
_
¸
¸
¸
_
VFORM fin
AUX +
COMPS
_
¸
¸
¸
_
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
V
_
¸
¸
¸
_
VFORM fin
AUX +
COMPS 2 [NEG +], 3
_
¸
¸
¸
_
2 Adv
3 VP[bse]

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
could not leave the town
As shown here, the negative finite auxiliary verb could selects two complements, the adverb not
and the VP leave the town, and all together these produce a well-formed finite VP.
By treating not as an modifier meaning constituent negation and as a complement marking
sentential negation, we can account for the scope differences in (62) and various other phenom-
ena including VP Ellipsis (see 8.4.4). For example, the present analysis will assign two different
structures for the string (62):
(70) a.
VP[AUX +]
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
V[AUX +] VP[VFORM bse]
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
could
Adv
_
MOD 1 VP
_ 1 VP[VFORM bse]

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
not approve the bill
165
b.
VP
_
¸
¸
¸
_
VFORM fin
NEG +
COMPS
_
¸
¸
¸
_
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
V
_
¸
¸
¸
¸
¸
¸
_
VFORM fin
AUX +
NEG +
COMPS 2 [NEG +], 3 VP
_
¸
¸
¸
¸
¸
¸
_
2 Adv 3 VP[VFORM bse]

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
could not approve the bill
In the structure (70a), not modifies just a nonfinite VP, with scope narrower than could. Mean-
while, in (70b), not is at the same level in the syntax as could, and semantically not scopes over
could. In this case, the feature [NEG +] percolates up to the VP and then to the whole sentence.
The semantic consequence of this structural difference can be seen in the different tag questions
appropriate for each interpretation:
(71) a. The president [could [not [approve the bill]]], couldn’t/*could he?
b. The president [[could] [not] [approve the bill]], could/*couldn’t he?
The tag question forms show that (71a) is actually a positive statement, even though some part
of it is negative. On the other hand, (71b) is a negative statement.
8.4.2 Auxiliaries with Inversion
Questions in English are formed by structures which invert the subject and the auxiliary:
6
(72) a. Are you studying English syntax?
b. What are you studying nowadays?
The long-standing tranformational approach assumes that the auxiliary verb is moved from a
medial position to the clause-initial position (the node labels would typically differ in current
analyses, but the structure of the movement is what is relevant here):
6
For the analysis of wh-questions like (72b), see chapter 10.
166
(73)
S
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
V S
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
Are NP VP
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
you V VP
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
t
¸¸
studying English?
However, there are certain exceptions that present problems for the analysis of inverted auxil-
iaries involving a movement transformation. Observe the following contrast:
(74) a. I shall go downtown.
b. Shall I go downtown?
Here there is a semantic difference between the auxiliary verb shall in (74a) and the one in
(74b): the former conveys a sense of simple futurity – in the near future, I will go downtown
– whereas the latter example has a deontic sense, asking whether it is appropriate for me to go
downtown. If the inverted verb is simply moved from an initial medial position in (74b), it is
not clear how the grammar can represent this meaning difference.
English also has various interpretations for the subject-auxiliary inversion construction:
7
(75) a. Wish: May she live forever!
b. Matrix Polar Interrogative: Was I that stupid?
c. Negative Imperative: Don’t you even touch that!
d. Subjunctive: Had they been here now, we wouldn’t have this problem.
e. Exclamative: Boy, am I tired!
Each of these constructions has its own constraints which cannot fully be predicted from other
constructions. For example, in ‘wish’ constructions, only the modal auxiliary may is possible.
In negative imperatives, only don’t allows the subject to follow. These idiosyncratic properties
support a non-movement approach, in which auxiliaries can be specified to have particular uses
or meanings when inserted into particular positions in the syntax.
This in turn means that our grammar adds the following SAI grammar rule as a well-formed
condition:
7
See Fillmore (1999) for detailed discussion.
167
(76) Subject-Aux Inversion (SAI) Rule:
S
_
SPR
_
→ H
_
¸
¸
¸
¸
¸
¸
_
HEAD
_
_
INV +
AUX +
_
_
SPR A
COMPS B
_
¸
¸
¸
¸
¸
¸
_
, A, B
This rule thus licenses an an inverted, finite, auxiliary verb to combine with its subject (the SPR
value A) and complements (the COMPS value B ), forming a well-formed subject-auxiliary
inverted phrase like the following:
(77)
S
_
¸
¸
¸
¸
¸
¸
_
HEAD
_
_
AUX +
INV +
_
_
SPR
COMPS
_
¸
¸
¸
¸
¸
¸
_
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
V
_
¸
¸
¸
¸
¸
¸
_
HEAD
_
_
AUX +
INV +
_
_
SPR 1 NP
COMPS 2 VP
_
¸
¸
¸
¸
¸
¸
_
1 NP 2 VP[VFORM bse]

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
Will you study syntax?
As shown in the structure, the inverted will combines with the subject NP you and the VP
complement at the same level.
Even though most of the auxiliary verbs will be classified as [INV +/−], some are lexically
defined just as [INV −]. For example, consider better.
(78) a. You better not drink.
b. You can do it, but you better not .
c. *Better you not drink.
Even though the word better functions at least in spoken English as an auxiliary verb, it always
carries [INV −], as attested by the above contrast: it cannot be inverted.
168
8.4.3 Contracted Auxiliaries
Auxiliary verbs actually show two kinds of contraction: either with a preceding subject or with
a negation (but not both):
(79) a. They’ll be leaving.
b. They’d leave soon.
(80) a. They wouldn’t leave soon.
b. They shouldn’t leave soon.
Contracted negation forms show several lexical idiosyncrasies as in *willn’t, *amn’t, and
*mayn’t. It is common to analyze n’t as a kind of inflectional affix (cf. Zwicky and Pullum
(1983)). In the approach we adopt here, we would posit an inflectional rule applying to a
specific set of verbs, as in (81):
(81) N’t Inflection Lexical Rule:
_
¸
¸
¸
¸
¸
_
PHON 1
HEAD
_
¸
¸
_
POS verb
VFORM fin
AUX +
_
¸
¸
_
_
¸
¸
¸
¸
¸
_

_
¸
¸
¸
¸
¸
_
PHON 1 + n’t
HEAD
_
¸
¸
_
VFORM fin
AUX +
NEG +
_
¸
¸
_
_
¸
¸
¸
¸
¸
_
This means that a word like can will be mapped to can’t, gaining the NEG feature:
(82)
_
¸
¸
¸
¸
¸
_
can
HEAD
_
¸
¸
_
POS verb
VFORM fin
AUX +
_
¸
¸
_
_
¸
¸
¸
¸
¸
_

_
¸
¸
¸
¸
¸
_
can’t
HEAD
_
¸
¸
_
VFORM fin
AUX +
NEG +
_
¸
¸
_
_
¸
¸
¸
¸
¸
_
As we have seen earlier, the head feature NEG will play an important role in forming tag ques-
tions:
(83) a. They can do it, can’t they?
b. They can’t do it, can they?
c. *They can’t do it, can’t they?
d. *They can’t do it, can he?
The tag part of such a question has the opposite value for NEG compared to that in the main
part of the clause, and its subject needs to have the same index value as the matrix subject. For
this, we can introduce the feature XARG (external argument) for each predicate and link the
subject to this value.
8
This means that English has independently the following Tag Question
8
Traditionally, arguments are classified into external and internal ones in which the former usually refers to the
subject. The introduction of such a semantic feature is necessary if we want to make the subject value visible at the S
level. See Bender and Flickinger (1999) and Sag (2007).
169
Rule:
9
(84) Tag-Question Rule:
S →S
_
_
NEG 1
XARG i
_
_
, S
_
¸
¸
_
NEG ¬1
INV +
XARG i
_
¸
¸
_
This rule means that a tag part can be added when it has the opposite NEG value whose subject
index is identical with that of the matrix subject. This will then project the following structure
for (83a):
(85)
S
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸
S
_
_
NEG +
XARG i
_
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
S
_
¸
¸
¸
_
NEG −
INV +
XARG i
_
¸
¸
¸
_
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
1 NP
i
VP
_
¸
¸
¸
_
NEG +
XARG i
SPR 1 NP
i

_
¸
¸
¸
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
V
_
¸
¸
¸
¸
¸
¸
_
NEG −
INV +
XARG i
SPR 2 NP
i

_
¸
¸
¸
¸
¸
¸
_
2 NP
They
V
_
¸
¸
¸
_
NEG +
XARG i
SPR 1 NP
i

_
¸
¸
¸
_
VP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
can they
can’t do it
As represented here, the NEG feature of the matrix verb is passed up to the first S. The tag
question then needs to have the opposite NEG value in accordance with the rule in (84). The
semantic feature XARG identified with the subject starts from the auxiliary verb and then is
semantically composed into the meaning of S. The XARG value in a sense makes the subject’s
9
This rule is a simplified version. To be more precise, we need to ensure that the second S, corresponding to the tag
part, must have only the auxiliary and the subject. See Bender and Flickinger (1999) and Sag (2007).
170
index value visible at the top level of the sentence in question so that the tag subject can also
refer to this.
8.4.4 Auxiliaries with Ellipsis
The standard generalization of Verb Phrase Ellipsis (VPE) is that it is possible only after an
auxiliary verb, as shown in the contrast (86) and (87).
(86) a. Kim can dance, and Sandy can , too.
b. Kim has danced, and Sandy has , too.
c. Kim was dancing, and Sandy was , too.
(87) a. *Kim considered joining the navy, but I never considered .
b. *Kim got arrested by the CIA, and Sandy got , also.
c. *Kim wanted to go and Sandy wanted , too.
The VP complement of an auxiliary verb, but not a main verb, can undergo VP ellipsis as long
as the context provides enough information for its interpretation.
The syntactic part of this generalization can be succinctly stated in the form of lexical rule:
(88) VP Ellipsis Rule:
_
_
HEAD| AUX +
COMPS XP
_
_

_
_
HEAD| AUX +
COMPS
_
_
As the rule is stated to apply to any XP after a verb with the [AUX +] specification, it can apply
to more than just VPs, and to more than just the canonical auxiliary verbs, but also be and have
in their main verb uses. With be, non-VP complements can be elided:
(89) a. Kim is happy and Sandy is too.
b. When Kim was in China, I was too.
The main verb have is somewhat restricted, but the contrast in (90) is clear. Even though have
is a main verb in (90a), it can allow an elided complement, unlike the main verb bring in (90b):
(90) a. A: Have you anything to share with the group?
B: No. Have you ?
b. A: Have you brought anything to share with the group?
B: No. *Have you bought ?
Given the lexical rule (88) which specifies no change in the ARG-ST, a canonical auxiliary
verb like can will have a counterpart minus its phrasal complement on the COMPS list:
(91)
_
¸
¸
¸
¸
¸
_
can
SPR 1 NP
COMPS 2 VP[bse]
ARG-ST 1 , 2
_
¸
¸
¸
¸
¸
_

_
¸
¸
¸
¸
¸
_
can
SPR 1
COMPS
ARG-ST 1 , 2
_
¸
¸
¸
¸
¸
_
171
Notice here that even though the VP complement is elided in the output, the ARG-ST is intact.
In the first part of the example in (92), there are three auxiliary verbs:
(92) Kim must have been dancing and
_
¸
¸
_
¸
¸
_
a. Sandy must have been , too.
b. Sandy must have , too.
c. Sandy must , too.
_
¸
¸
_
¸
¸
_
There are therefore various options for an elided VP: the complement of been, or have, or must.
The analysis also immediately predicts that ellipsis is possible with the infinitival marker to,
for this is an auxiliary verb, too:
(93) a. Tom wanted to go home, but Peter didn’t want to .
b. Lee voted for Bill because his father told him to .
(94) a. Because John persuaded Sally to , he didn’t have to talk to the reporters.
b. Mary likes to tour art galleries, but Bill hates to .
Finally, the analysis given here will also account for the contrast shown above in (67); a
similar contrast is found in the following examples:
(95) a. *Mary sang a song, but Lee could never .
b. Mary sang a song, but Lee could not .
The negator not in (95b) is a marker of sentential negation and can be the complement of the
finite auxiliary verb could. This means we can apply the VPE lexical rule to could after the
negation lexical rule (68), as shown (96):
(96)
_
¸
¸
¸
¸
¸
_
could
SPR 1 NP
COMPS 2 Adv[NEG +], 3 VP[bse]
ARG-ST 1 , 2 , 3
_
¸
¸
¸
¸
¸
_

_
¸
¸
¸
¸
¸
_
could
SPR 1
COMPS 2
ARG-ST 1 , 2 , 3
_
¸
¸
¸
¸
¸
_
As shown here in the right-hand form, the VP complement of the auxiliary verb could is not
realized as a COMP element, though the negative adverb is. This form would then project a
syntactic structure in (97):
172
(97)
VP
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
V
_
¸
¸
¸
¸
¸
¸
_
AUX +
SPR 1
COMPS 2 Adv[NEG +]
ARG-ST 1 , 2 , 3 VP[bse]
_
¸
¸
¸
¸
¸
¸
_
2 Adv
_
NEG +
_
could not
As represented here, the auxiliary verb could forms a well-formed head-complement structure
with not.
Why is there a contrast in the examples in (95)? The reason is that not can ‘survive’ VPE
because it can be licensed in the syntax as a complement of an auxiliary, independent of the
following VP. However, an adverb like never is only licensed as a modifier of VP (it is adjoined
to VP to give another VP), and hence if the VP were elided, we would have a hypothetical
structure like this:
(98)
VP
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
V[AUX +] *VP
could Adv[MOD VP]
never
Here, the adverb never modifies a VP through the feature MOD, which guarantees that the
adverb selects the head VP that it modifies. In an ellipsis structure, the absence of such a VP
means that there is no VP for the adverb to modify. In other words, there is no well-formed
phrasal structure – predicting the ungrammaticality of *has never as opposed to has not.
10
10
As we have seen in 6.6.1, all modifiers carry the head feature MOD whose value is the expression they modify.
173
8.5 Exercises
1. Each of the following sentences contains an item (in the parentheses) which we might
want to call an auxiliary. In each case, construct relevant examples that will clarify
whether it actually is an auxiliary:
(i) a. John got sent to prison. (got)
b. He ought to leave his luggage here. (ought)
c. They needn’t take this exam. (need)
d. You better not leave it here. (better)
e. He dared not argue against his parents. (dared)
f. He used to go there very often. (used).
Explain your reasoning from the examples you provide.
2. Draw trees for the following sentences:
(i) a. The gardener must trim the rose bushes today.
b. This should be the beginning of a beautiful friendship.
c. I am removing the shovel from the shed.
d. The travelers have returned from their vacation.
e. Springfield would have built a police station with the federal grant.
f. Stringrays could have been cruising near the beach.
g. She seem to have given financial assistance to an important French art
dealer.
3. Provide an analysis of the grammaticality or ungrammaticality of the following examples
together, with a tree structure for each, and lexical entries for the words playing the
crucial roles in the determination of grammaticality.
(i) a. Ann may spend/*spending/*spends/*spent her vacation in Italy.
b. It has rained/*raining/*rains/*rain every day for the last week.
c. Tagalog is spoken/*speak/*speaks/*spoke in the Philippines.
d. The roof is leaking/*leaked/*leaks/*leak.
(ii) a. *Americans have musted pay income tax ever since 1913.
b. *George is having lived in Toledo for thirty years.
c. *The house is been remodeling.
d. *Margaret has had already left.
e. *A medal was been given to the mayor by the sewer commissioner.
f. *Does John have gone to the library?
g. *John seems fond of ice cream, and Bill seems, too.
(iii) a. Sam may have been being interrogated by the FBI.
b. *Sam may have been being interrogating by the FBI.
c. *Sam may be had been interrogating by the FBI.
174
4. Analyze the following sentences providing tree structures and lexical entries for the
verbs:
(i) a. Have social problems made police work difficult?
b. The senator should not have forgotten the concerns of her constituents.
c. Tokyo has not loosened trade restrictions.
d. They love to play golf, but I do not.
e. Did the doctor prescribe aspirin?
f. George has spent a lot of money, hasn’t he?
g. Sandy will read your reports, but Harold will not.
What grammar rules are needed for your structures? Are any lexical rules involved in
getting to the correct forms of the verbs?
5. English allows what is called ‘negative inversion’ as illustrated in (ii):
(i) a. He can hardly believe that it’s already over.
b. I could have little known that more trouble was just around the corner.
c. I have never been spoken to so rudely!
(ii) a. [Hardly] was there any rain falling.
b. [Little] did I know that more trouble was just around the corner.
c. [Never] have I been spoken to so rudely!
(iii) a. He had hardly collected the papers on his desk, had he/*hadn’t he?
b. He never achieved anything, did he/*didn’t he?
Draw tree structures for the sentences (ii) and provide the lexical entries for hardly, little
and never. The examples in (iii) indicate that these adverbs all involve some kind of
negation in the sentence in which they appear. In addition, think of how your analysis
can account for the unacceptable examples in (iv):
(iv) a. As a statesman, he scarcely could do anything worth mentioning.
b. As a statesman, scarcely could he do anything worth mentioning.
c. *As a statesman, scarcely he could do anything worth mentioning.
6. Observe the following contrast and state a rule that can describe the usage of words like
any. Does your rule involve negative words like not or hardly – if so, how? In addition,
construct examples replacing any with some and determine if there are any differences
between these two types of word.
(i) a. *Anyone isn’t sleeping in my bed.
b. *Any zebras can’t fly.
c. *Anything hasn’t happened to his optimism.
d. *Any of the citizens hardly ever say anything.
(ii) a. I didn’t find any bugs in my bed.
b. Nobody told them anything.
c. We never found any of the unicorns.
(iii) a. Never have I stolen from any members of your family.
175
b. Why haven’t any books been returned?
c. Hardly any of the citizens ever say anything.
7. After reading the following passage, provide lexical entries for the underlined words and
draw trees for the sentences which include them.
(i) This expanded role for auxiliaries in English has resulted in some curious
rules. One is that when a sentence is to be negated, the word not must
follow not the main verb (as used to be the case), but the auxiliary. This rule
creates an awkward dilemma in the occasional instance when the sentence
to be negated actually doesn’t have an auxiliary verb. Thus, if I wish to deny
the sentence, I walked home, I must add an entirely meaningless auxiliary
from the verb do just to stand as the prop for the word not. The result is
the sentence, I didn’t walk home. Now, do and did are often added to show
emphasis, but in those cases they are spoken with emphasis. Thus, there is
a difference between saying I didn’t walk home and saying I DIDN’T walk
home. The latter sentence expresses emphasis, but in the former sentence
the verb did expresses nothing at all; it is merely there to hang the not on.
If we tried to say, I walked not home, this would have an unacceptably odd
sound to it. It would, indeed, sound archaic. English literature is full of
such archaisms, since putting not after the main verb was still good usage
in the time of Shakespeare and a century or more later.
11
11
Adopted from ‘Creationism & Darwinism, Politics & Economics’ by Kelley L. Ross.
176
9
Passive Constructions
9.1 Introduction
One important aspect of syntax is how to capture systematic relations between related construc-
tions. For example, the following two sentences are similar in meaning:
(1) a. One of Korea’s most famous poets wrote these lines.
b. These lines were written by one of Korea’s most famous poets.
We recognize (1b) as the passive counterpart of the active sentence (1a). These two sentences are
truth-conditionally similar: they both describe the event of writing the lines by one Korean poet.
The only difference involves grammatical functions: in the active voice (1a), one of Korea’s most
famous poets is the subject, whereas in the passive voice (1b), these lines is the subject.
Observing these differences, the question that arises is: Why do we use different voices
for expressing or describing the same situation or proposition? It is generally accepted that
the passive construction is used for certain discourse-motivated reasons. For example, when
it is more important to draw our attention to the person or thing acted upon, we use passive.
Compare the following:
(2) a. Somebody apparently struck the unidentified victim during the early morning
hours.
b. The unidentified victim was apparently struck during the early morning hours.
We can easily see here that the passive in (2b) assigns more attention to the victim than the
active in (2a) does. In addition, when the actor in the situation is not important or specific, there
is often a preference to use the passive voice:
(3) a. Targets can be observed at any angle.
b. During the early evening, Saturn can be found in the north, while Jupiter rises in
the east.
Similarly, we use the passive voice in formal, scientific, or technical writing or reports to place
an emphasis or an objective presentation on the process or principle being described. For exam-
ple, compare the following pair:
177
(4) a. I poured 20cc of acid into the beaker.
b. About 20cc of acid was poured into the beaker.
It is clear that unlike the active sentence (4a), the passive sentence (4b) assigns a more objective
perspective to the process described.
In this chapter, leaving aside these discourse- or genre-motivated features of the use of pas-
sive constructions, we will look into the syntactic and semantic relationships between active
and passive as well as the properties of different passive constructions.
9.2 Relationships between Active and Passive
Consider the two canonical active and passive counterpart sentences:
(5) a. The executive committee approved the new policy.
b. The new policy was approved by the executive committee.
How do these construction types differ?
Grammatical Functions and Subcategorization: As briefly noted earlier, one of the main
differences we can observe between (5a) and (5b) is that the passive sentence has a ‘promoted’
object: the new policy is the passive sentence subject, while the notional subject the executive
committee is realized, optionally, in a PP (headed by by). By definition, a transitive verb form
such as taken or chosen must have an object:
(6) a. John has taken Bill to the library.
b. John has chosen Bill for the position.
(7) a. *John has taken to the library.
b. *John has chosen for the position.
Yet, with the passive construction of such verbs, the object NP is not present in post-verbal
position, and must not be:
(8) a. *The guide has been taken John to the library.
b. *The department has been chosen John for the position.
(9) a. John has been taken to the library.
b. John has been chosen for the position.
The absence of the object in the passive is due to the fact that the argument that would have
been the object of the active verb has been promoted to be the subject in the passive.
Apart from the direct object, other subcategorization requirement stays unchanged in a pas-
sive form. For example, the active form handed in (10) requires an NP and a PP[to] as its
complements, and the passive handed in (11) still requires the PP complement:
(10) a. Pat handed a book to Chris.
b. *Pat handed to Chris.
c. *Pat handed a book.
178
(11) a. A book was handed to Chris (by Pat).
b. *A book was handed (by Pat).
Other Selectional Properties: The third important property, following from the fact that the
active verb’s object is promoted to the passive construction’s subject, is that other selectional
properties of the verb are preserved. For example, if the usual postverbal constituent should
be an expletive form like it, this requirement is on the subject in the passive. Compare the
following:
(12) a. They believe it/*Stephen to be easy to annoy Ben.
b. They believe there to be a dragon in the wood.
(13) a. It/*Stephen is believed to be easy to annoy Ben.
b. There is believed to be a dragon in the wood.
If the active complement is itself a clause, so must the subject of the passive verb be a clause:
(14) a. No one believes/suspects [that he is a fool].
b. [That he is a fool] is believed/suspected by no one.
Finally, if the postverbal constituent can be understood as part of an idiom, so can the subject
in the passive:
(15) a. They believe the cat to be out of the bag.
b. The cat is believed to be out of the bag.
We thus can conclude that the subject of the passive form is the argument which corresponds to
the object of the active.
Morpho-syntactic changes: In addition to changes in argument realization, the passive con-
struction requires the auxiliary verb be in conjunction with the the passive form of the verb (a
subtype of the en form, see 5.2.1). In addition to ‘passive be’ italicized in the examples below,
there can be other auxiliary verbs, with the passive auxiliary last in the sequence:
(16) a. John drove the car. →The car was driven.
b. John was driving the car. →The car was being driven.
c. John will drive the car. →The car will be driven.
d. John has driven the car. →The car has been driven.
e. John has been driving the car. →The car has been being driven.
f. John will have been driving the car. →The car will have been being driven.
Semantics: In terms of meaning, as noted above, there is no change in the semantic role
assigned to the argument which is the subject in the passive. The agent argument of active verb
is expressed as an optional oblique argument of the PP headed by the preposition by in the
passive, or not at all:
179
(17) a. Pat handed Chris a note.
b. Chris was handed a note (by Pat).
(18) a. TV puts ideas into children’s heads.
b. Ideas are put into children’s heads (by TV).
The observations above mean that any grammar needs to capture the following basic properties
of passive:
.
Passive turns the active object into the passive subject;
.
Passive leaves other aspects of the COMPS value of the active verb unchanged;
.
Passive optionally allows the active subject to be the object in a PP headed by the preposition
by;
.
Passive makes the appropriate morphological change in the form of the main verb, and re-
quires the auxiliary be;
.
Passive leaves the semantics unchanged.
9.3 Approaches to Passive
There could be several ways to capture the systematic syntactic and semantic relationships
between active and passive forms. Given our discussion so far, one might think of relying on
grammatical categories in phrase structure (NP, VP, S, etc.), or on surface valence properties
(SPR and COMPS), often informally characterized as grammatical functions, or semantic roles
(agent, patient etc.). In what follows, we will see that we need to refer to all of these aspects of
the representation in a proper treatment of English passive constructions.
9.3.1 From Structural Description to Structural Change
Before we look into syntactic analyses for the formation of passive sentences, it is worth review-
ing Chomsky’s (1957) Passive Formation Rule formulated in terms of structural descriptions
(SD) and structural change (SC):
(19) Passive Formation Rule:
SD: X NP Y V NP Z
1

2

3

4

5
¸.
6
¸.
SC: 1 5 3 be 4+en 6 (by 2)
This rule means that if there is anything that fits the SD in (19), it will be changed into the given
SC: that is, if we have any string in the order of “X – NP – Y – V – NP – Z” (in which X, Y,
and Z are variables), the order can be changed into “X – NP – Y – be – V+en – Z – by NP”. For
example, consider one example:
180
(20)
SD: Yesterday, the child really kicked a monkey in the street.
X NP Y V NP Z
1

2

3

4

5
¸
6
¸.
SC: 1 5 3 be 4 + en 6 (by 2)
Yesterday, a monkey really was kicked in the street (by the child)
As noted here, the main change that occurs in the SC is that the first NP became an optional
PP whereas the second NP became the first NP. The rule also accompanies the addition of be
and the change of the main verb’s VFORM into the passive form. Even though this old SD-SC
style rule does not reflect constituenthood of the expressions in the given sentence and is not
satisfactory enough to account for all different types of passivization that we will see in the
following, this seminal work has influenced the development of subsequent transformational
analyses for English passive constructions.
9.3.2 A Transformational Approach
A typical transformational approach assuming movement for passive involves the operation
shown in (21) (Chomsky 1982):
(21)
IP
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
NP
¯
I
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
e I VP
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
Past V VP
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
be
¸¸
V NP
deceived Bill
¸¸
The operation moves the object Bill to the subject position and the verb be to I (Infl) position,
generating the output sentence Bill was deceived. This kind of movement analysis is based on
the following three basic assumptions:
.
Move α: Move a category.
.
Case Theory: NP needs Case. The subject receives NOM (nominative) case from tense, and
the object receives ACC (accusative) from an active transitive verb.
1
1
In English, case is morphologically visible only on pronouns: he is nominative, whereas him is accusative, for
example.
181
.
A passive participle does not license ACC case.
In the lower position inside VP, the NP Bill in (21) cannot receive ACC case, since by as-
sumption the passive participle form deceived cannot assign any case. Hence, was deceived
Bill would violate Case Theory, for every NP must be assigned case. If the NP is moved to the
subject position, where case is assigned by the tensed verb was, then Case Theory is satisfied.
Even though this kind of movement or derivational analysis can be appealing in capturing the
relationships between canonical active and passive examples, it leaves many facts unexplained.
In what follows, we will see more complicated types of passive construction in English, and
the need to refer to not only grammatical categories but also grammatical functions, as well as
semantic/pragmatic constraints on passive.
9.3.3 A Lexicalist Approach
If we look into more passive examples, we can see that we need to refer to lexical and semantic
properties of the transitive verbs. First, there are many exceptions to passive. For example,
transitive verbs like resemble or fit do not have any passive in some senses (see section 9.5
also):
(22) a. The model resembles Kim in nearly every detail.
b. *Kim is resembled by the model in nearly every detail.
(23) a. The coat does not fit you.
b. *You are not fitted by the coat.
Such a transitive verb presumably fits the tree structure in (21), but cannot be passivized.
In contrast, there are also verbs like rumor, say and repute which are used only in the passive,
as seen in the following contrast:
(24) a. I was born in 1970.
b. It is rumored that he is on his way out.
c. John is said to be rich.
d. He is reputed to be a good scholar.
(25) a. *My mother bore me in 1970.
b. *Everyone rumored that he was on his way out.
c. *They said him to be rich.
d. *They reputed him to be a good scholar.
Unlike verbs like resemble, these verbs are not used as active forms.
Such non-passive examples are hard to explain if we rely only on the assumption that passives
are derived from actives from configurational transformation rules. It seems that such lexical
idiosyncracies can be better treated in terms of a lexical process which allows us to refer to
the lexical and semantic properties of the verb in question. One way to capture these observed
lexical properties of passive is to posit a simplified lexical rule like the following:
182
(26) Passive Lexical Rule (to be revised):
_
¸
¸
_
trans-v
SPR 1 NP
COMPS 2 NP, . . .
_
¸
¸
_

_
¸
¸
¸
¸
¸
¸
_
pass-trans-v
HEAD| VFORM pass
SPR 2 NP
COMPS . . . ,
_
PP
_
by + 1 NP
_
_

_
¸
¸
¸
¸
¸
¸
_
This simple rule says that if there is a transitive verb (tran(sitive)-v) lexeme selecting one
SPR ( 1 ) and at least one COMPS element ( 2 and others), then there is a related passive verb
(pass-trans-v). This output verb selects the first element on the original COMPS list as its SPR
( 2 NP) and the SPR in the input as an optional PP(consisting of by and 1 NP) with the remain-
ing COMPS value unchanged ( . . . ). The lexical process also accompanies the change in the
VFORM value into pass.
2
As it stands, this lexical rule is not precise enough. For example, consider the following:
(27) a. He kicked the ball. vs. b. The ball was kicked by him.
c. John kicked him. vs. d. He was kicked by John.
We can observe that the case on the first argument of the predicate has changed from he in the
active to him in the passive. However, this difference is entirely predictable: all subjects of finite
clauses in English are nominative, and all objects of prepositions are accusative. So, rather than
stating the case changes directly, we can change the rule to refer only to the index value of the
subject and object:
3
2
As we noted in Chapter 5, in terms of the morphological form, the VFORM pass is a subtype of en.
3
Notice that the lexical rule given here can be represented in terms of the ARG-ST (Sag et al. 2002):
(i)
_
trans-v
ARG-ST XP
i
, XP
j
. . .
_

_
_
pass-trans-v
ARG-ST XP
j
, . . .
_
PP
i
[PFORM by]
_

_
_
This rule rearranges the elements of the input ARG-ST, also accompanying the change in the VFORM value into pass.
That is, the second element in the input ARG-ST becomes the first element in the ARG-ST of the output passive verb.
Whatever follows the second argument in the input is intact, but the first element in the ARG-ST becomes the object
of the optional PP. The Argument Realization Constraint in (4.4.3) will ensure that the first element in the ARG-ST is
realized as the SPR and the remaining elements as the COMPS value as given in (28). Given that there is no discrepancy
between the ARG-ST values and the VAL (SPR and COMPS) values, we can formulate the passive rule as in (28) too.
183
(28) Passive Lexical Rule (Final):
_
¸
¸
_
trans-v
SPR XP
i

COMPS XP
j
, . . .
_
¸
¸
_

_
¸
¸
¸
¸
¸
_
pass-v
HEAD| VFORM pass
SPR XP
j

COMPS . . . (PP
i
[PFORM by])
_
¸
¸
¸
¸
¸
_
With this revised lexical rule, the case of the various NPs will be predicted by general prin-
ciples of case marking in English clauses, and need not be mentioned in the rule, which now
refers to the index values of the SPR and COMPS expressions. Let us see how all this works,
concentrating on a simple example:
(29) a. John sent her to Seoul.
b. She was sent to Seoul.
The active verb send is turned into the passive verb sent by the Passive Lexical Rule in (28):
(30) _
¸
¸
¸
¸
¸
_
send
HEAD|POS verb
SPR NP
i

COMPS NP
j
, 2 PP[to]
_
¸
¸
¸
¸
¸
_

_
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
sent
HEAD
_
_
POS verb
VFORM pass
_
_
SPR 1 NP
j

COMPS 2 , (PP
i
[by])
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
As seen here in the output form, the passive sent takes a SPR whose index value is identical
to that of the first element of the COMPS list in the input. The passive sent also inherits the
PP[to] complement, tagged 2 , and selects an optional PP whose index value is identical to the
SPR (subject) of the input.
4
This output lexical entry will then license the following structure
for (29b):
4
As noted in Chapter 6.5.2, a preposition functioning as a marker rather than as a predicator with semantic content
does not contribute to the meaning of the head PP. This means that its index value is identical to that of its object NP.
184
(31)
S
_
VFORM fin
_

_
_
_
_
_
_
_
_
_
_
_
_
2 NP
VP
_
¸
_
VFORM fin
SPR 2
COMPS
_
¸
_
.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
_
N
V
_
¸
_
VFORM fin
SPR 2
COMPS 5
_
¸
_
5 VP
_
¸
_
VFORM pass
SPR 2
COMPS
_
¸
_
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
She was
V
_
¸
_
VFORM pass
SPR 2
COMPS 3
_
¸
_
3 PP

¸
¸
¸
¸
¸
¸
¸
¸
¸
sent to Seoul
As given in the structure, the passive sent combines with its PP[to] complement, forming a VP
that still requires a SPR. This VP functions as the complement of the auxiliary be (was). As
we saw in Chapter 8, be is a raising verb, with the repeated lexical entry in (32), whose subject
(SPR value) is identical to its VP complement’s subject She:
(32)
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
be
HEAD| POS verb
SPR 2
COMPS
_
VP
_
_
VFORM pass
SPR 2
_
_
_
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
The SPR requirement on be is passed up to the highest VP in accordance with the VALP that
regulates the value of SPR and COMPS (see Chapter 5.1). When this VP combines with the
subject she in accordance with the Head-Specifier Rule, the well-formed passive sentence is
complete.
The Passive Lexical Rule in (28) can be also applied to verbs which select for a CP comple-
ment. Consider the following examples:
(33) a. They widely believed that John was ill.
b. That John was ill was widely believed.
185
The application of the Passive Lexical Rule to the active believe will generate the passive output
shown in the following:
(34) _
¸
¸
¸
¸
¸
_
believe
HEAD|POS verb
SPR NP
i

COMPS CP
j

_
¸
¸
¸
¸
¸
_

_
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
believed
HEAD
_
_
POS verb
VFORM pass
_
_
SPR CP
j

COMPS (PP
i
)
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
The output passive verb believed then can license a structure like the following:
(35)
S

_
_
_
_
_
_
_
_
_
_
_
_
_
2 CP
·
·
·
·
·
·
·
·
·
¸
¸
¸
¸
¸
¸
¸
¸
¸
VP
_
SPR 2
_
.
.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
_
_
That John
was ill
V
_
SPR 2
COMPS 5
_
5 VP
_
VFORM pass
SPR 2
_

_
_
_
_
_
_
_
_
_
_
_
_
_
was Adv
VP
_
VFORM pass
SPR 2
_
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
widely
V
_
¸
_
VFORM pass
SPR 2
COMPS( 3 )
_
¸
_
3 PP

¸
¸
¸
¸
¸
¸
¸
¸
¸
believed (by them)
The passive verb believed first combines with its optional complement by them, and then with
the modifier widely. The resulting VP then combines with the raising verb be in accordance with
the Head-Complement Rule. This system, licensing each local structure by the defined grammar
rules and principles, thus links the CP subject of be to that of believed.
The same account also holds when the complement is an indirect question:
(36) a. They haven’t decided [which attorney will give the closing argument].
b. [Which attorney will give the closing argument] hasn’t been decided (by them).
186
The active decided selects an interrogative sentence as its complement, and the Passive Lexical
Rule (28) can apply to this verb:
5
(37) _
¸
¸
¸
¸
¸
_
decide
HEAD|POS verb
SPR NP
i

COMPS S
j
[QUE +]
_
¸
¸
¸
¸
¸
_

_
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
decided
HEAD
_
_
POS verb
VFORM pass
_
_
SPR S
j
[QUE +]
COMPS (PP
i
[by])
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
The output passive decided then will generate the following structure:
(38)
S

_
_
_
_
_
_
_
_
_
_
_
_
_
1 S[QUE +]

¸
¸
¸
¸
¸
¸
¸
¸
¸
VP
_
SPR 1
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
Which attorney
will give the
closing
argument
V
_
SPR 1
COMPS 4
_
4 VP
_
SPR 1
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
has
V
_
SPR 1
COMPS 3
_
3 VP
_
SPR 1
_
been
V
_
¸
_
VFORM pass
SPR 1 S[QUE +]
COMPS
_
¸
_
decided
The passive verb decided selects an optional PP complement as its complement and an indirect
question as its subject. The raising verb first combines with the first VP, the result of which
again combines with the auxiliary raising verb has. Notice that since be and have are raising
verbs, their VP complement has the same subject as their own. By these identifications, the
subject of has is identical to that of the passive verb decided.
5
We assume that indirect or direct questions are marked by the feature QUE(STION); see Chapter 10.
187
9.4 Prepositional Passives
In addition to the passivization of an active transitive verb, English also allows the so-called
‘prepositional verb’ to undergo passivization as illustrated in the following:
(39) a. You can rely on Ben.
b. Ben can be relied on.
(40) a. They talked about the scandal for days.
b. The scandal was talked about for days.
As we noted here, the object of the preposition in the active can function as the subject of the
passive sentence. Notice that such prepositional passives are possible with the verbs selecting a
PP with a specified preposition:
(41) a. The plan was approved of by my mother. (My mother approved of the plan.)
b. The issue was dealt with promptly. (They dealt with the issue promptly.)
c. That’s not what’s asked for. (That’s not what they asked for.)
d. This should be attended to immediately. (We should attend to this immediately.)
(42) a. *Boston was flown to. (They flew to/near/by Boston.)
b. *The capital was gathered near by a crowd of people. (A crowd of people gathered
near/at the capital.)
c. *The hot sun was played under by the children. (The children played under/near
the hot sun.)
The propositions in (41) are all selected by the main verbs (no other prepositions can replace
them). Meanwhile, each preposition in (42) is not selected by the main verb, since it can be
replaced by another one as noted in their active sentences.
One thing to observe is that there is a contrast between active and passive prepositional verbs
with respect to the appearance of an adverb (see Chomsky 1972, Bresnan 1982). Observe the
following:
(43) a. That’s something I would have paid twice for.
b. These are the books that we have gone most thoroughly over.
c. They look generally on John as selfish.
(44) a. *Everything was paid twice for.
b. *Your books were gone most thoroughly over.
c. *He is looked generally on as selfish.
The contrast here shows us that unlike the active, the passive does not allow any adverb to
intervene between the verb and the preposition.
There can be two possible structures that can capture these properties: ternary and reanalysis
structures. The ternary structure generates a flat structure like the following:
188
(45)
VP[pass]
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
V[pass] P NP

¸
¸
¸
¸
¸
¸
talked about the scandal
Unlike such a flat or ternary structure, there has been another possible structure assumed in the
literature as given in (46):
(46)
VP[pass]
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
V[pass]
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
NP

¸
¸
¸
¸
¸
¸
V[pass] P the scandal
talked about
This structure is different from (45) in the sense that the passive verb and the preposition is
‘reanalyzed’ as a verb again (in this sense this is often called ‘reanalysis’). Both (45) and (46)
can capture the coherence between the prepositional verb and the preposition itself. Even though
both have their own merits, we choose the structure (46), in which the passive verb and the
preposition form a unit. We can observe that there are environments where the passive verb (but
not active verb) forms a coherent lexical unit with the following preposition:
(47) a. Pavarotti relied on Loren and Bond on Hepburn.
b. *Pavarotti relied on Loren and Bond Hepburn.
c. Loren was relied on by Pavarotti and Hepburn by Bond.
d. *Loren was relied on by Pavarotti and Hepburn on by Bond.
What we can observe here is that unlike in the active verb, in the passive relied on acts like a
lexical unit in the gapping process: the passive relied alone cannot be gapped.
This contrast supports the reanalysis structure for the passive. In order for the grammar to
allow the passive V to be combined with the following P (which is defined to be ‘LIGHT’ in the
sense that it is not a prosodically heavy element), we introduce the following grammar rule:
6
(48) Head-Light Rule:
V →V, X[LIGHT +]
The rule allows a head V to combine with a LIGHT element such as a preposition in the prepo-
sitional passive verb construction.
7
6
See Abeill´ e and Godard (2000) for the motivation of introducing the feature LIGHT, with regard to French data.
7
In languages like Korean, German, and even French, such a syntactic combination is prevalent for the formation of
complex predicates (see Kim (2004b)). In English, particles can also be taken to carry a positive value for LIGHT.
189
Given this structural assumption, how then can we generate prepositional passives? We first
need to ensure that the object of the prepositional verb is promoted to the subject in the passive,
as represented in the following:
(49) Prepositional Passive Lexical Rule:
_
¸
¸
_
prep-v
SPR NP
i

COMPS PP
j
[PFORM 4 ]
_
¸
¸
_

_
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
pass-prep-v
VFORM pass
SPR NP
j

COMPS
_
P
_
_
LIGHT +
PFORM 4
_
_
, (PP
i
[PFORM by])
_
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
This rule ensures that a prepositional verb (prepositional-v) can have a counterpart passive
verb. This passive verb selects a SPR whose index value is identical to that of the input verb’s
PP complement (in other words, the object of the preposition). The output passive verb also
has two complements: a preposition with the same PFORM as the input and an optional PP
complement expressing the agent argument (see below for the function of the feature LIGHT).
Let’s see how the Prepositional Passive Rule and the Head-Light Rule combined together
can account for a prepositional passive:
(50) a. The lawyer looked into the document.
b. The document was looked into by the lawyer.
The active prepositional verb look can undergo the Prepositional Passive Lexical Rule as repre-
sented in the following:
(51)
_
¸
¸
_
look
SPR NP
i

COMPS PP
j
[into]
_
¸
¸
_

_
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
looked
VFORM pass
SPR NP
j

COMPS
_
P
_
_
LIGHT +
PFORM into
_
_
, (PP
i
[by])
_
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
The output passive verb selects one subject whose index value is identical to that of the input’s
PP complement. It also selects two complements: a preposition whose PFORM is identical with
that of the input PP and an optional PP[by] linked to the input subject. This output will then
license a structure like the following:
190
(52)
VP
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
V VP[pass]
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
was
V
_
_
VFORM pass
COMPS 3 PP
_
_
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
3 PP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
V
_
_
VFORM pass
COMPS 2 P, 3 PP
_
_
2 P
_
_
LIGHT +
PFORM into
_
_
by the lawyer
looked into
The Head-Light Rule in (48) allows the passive verb to combine with the preposition into first,
forming still a lexical element. This resulting lexical element then combines with its PP com-
plement by the lawyer in accordance with the Head-Complement Rule.
8
8
This means that in the Head-Complement Rule the complement that the head combines with is phrasal.
191
9.5 Exercises
1. Draw tree structures for each of the following sentences and then provide the lexical
entry for the italicized passive verb.
(i) a. Peter has been asked to resign.
b. I assume the matter to have been filed in the appropriate records.
c. Smith wants the picture to be removed from the office.
d. The events have been described well.
e. Over 120 different contaminants have been dumped into the river.
f. Heart disease is considered the leading cause of death in the United States.
g. The balloon is positioned in an area of blockage and is inflated.
h. There was believed to have been a riot in the kitchen.
i. Cancer is now thought to be unlikely to be caused by hot dogs.
2. Provide the active counterpart of the following examples and explain how we can gener-
ate each of them, together with tree structures, lexical entries, and grammar rules.
(i) a. That we should call the police was suggested by her son.
b. Whether this is feasible hasn’t yet been determined.
c. Paying taxes can’t be avoided.
Also see if there are any relationships between the above examples and the following
passives:
(ii) a. It was suggested by her son that we should call the police.
b. It hasn’t yet been determined whether this is feasible.
3. verbs like get and have can be used in so-called ‘pseudo-passives’:
(i) a. Frances has had the drapes cleaned.
b. Shirley seems to have Fred promoted.
(ii) a. Nina got Bill elected to the committee.
b. We got our car radio stolen twice on holiday.
In addition to these, have and get also allow constructions like the following:
(iii) a. Frances has had her clean the drapes.
b. Nina got them to elect Bill.
After drawing tree structures for the above examples, discuss the lexical properties of
have and get – for example, what are their ARG-STs like?
4. Consider the following prepositional passive examples and then analyze them as far as
you can with tree structures.
(i) a. Ricky can be relied on.
b. The news was dealt with carefully.
c. The plaza was come into by many people.
d. The tree was looked after by Kim.
In addition, consider the passive examples in (ii):
192
(i) a. We cannot put up with the noise anymore.
b. He will keep up with their expectations.
(ii) a. This noise cannot be put up with.
b. Their expectations will be kept up with.
Can our analysis given in this chapter account for such examples? Also observe the
following examples (in (iv)), which illustrate two different kinds of passive:
(iii) a. They paid a lot of attention to the matter.
b. The son took care of his parents.
(iv) a. The matter was paid a lot of attention to.
b. A lot of attention was paid to the matter.
Can you think of any way to account for such examples?
5. We have seen that when the verb does not select a specified preposition, it usually does
not undergo passivization. However, observe the following contrast:
(i) a. *New York was slept in.
b. The bed was slept in.
(ii) a. *The lake was camped beside by my sister.
b. The lake is not to be camped beside by anybody.
Why do we have such a contrast with the same type of prepositional verb? In answering
this, think about the following contrast too, with respect to semantic or pragmatic facotrs:
(iii) a. *Six inches were grown by the boy.
b. *A pound was weighed by the book.
c. *A mile to work was run by him.
(iv) a. The beans were grown by the gardener.
b. The plums were weighed by the greengrocer.
In addition, can your semantic or pragmatic constraints explain the following contrast
too? If not, what kind of generalization can you think of to account the contrast here?
(v) a. *San Francisco has been lived in by my brother.
b. The house has been lived in by several famous personages.
(vi) a. *Seoul was slept in by the businessman last night.
b. This bed was surely slept in by a huge guy last night.
6. In certain environments, passives allow the auxiliary verb part to be get instead of be:
(i) a. Rosie got struck by lightning.
b. I got phoned by a woman friend.
c. He got hit in the face with the tip of a surfboard.
d. John’s bike got fixed or got stolen.
Get passives usually convey the speaker’s personal involvement or reflect the speaker’s
opinion as to whether the event described is perceived as having favorable or unfavor-
able consequences. This is why it is rather unacceptable to use the get passive when the
193
predicate is stative or the subject-referent has no control over the process in question:
(ii) a. *The king got feared by everyone.
b. *The lesson got read by a choirboy.
c. *The letter got written by a poet.
d. *Tom got understood to have asked for a refund.
e. *Mary got heard to insult her parents.
Based on these observations, provide the lexical entries for get in passive and tree struc-
tures for (ia) and (ib).
7. Read the following passages and identify all the grammatical errors in the verbs’
VFORM values. In addition, provide the lexical information for the correct form.
(ii) Syntax is the discipline that examining the rules of a language that dictate
how the various parts of sentences gone together. While morphology looks
at how individual sounds formed into complete words, syntax looks at how
those words are put together for complete sentences. One part of syntax,
calling inflection, deals with how the end of a word might changed to tell
a listener or reader something about the role that word is playing. Regular
verbs in English, for example, change their ending based for the tense the
verb is representing in a sentence, so that when we see Robert danced, we
know the sentence is in the past tense, and when we see Robert is dancing,
we know it is not. As another example, regular nouns in English become
plural simply by adding an s to the end. Cues like these play a large role
for helping hearers understanding sentences.
9
9
From http://www.wisegeek.com/what-is-syntax.htm
194
10
Wh-Questions
10.1 Clausal Types and Interrogatives
Like other languages, English also distinguishes a set of clause types that are characteristically
used to perform different kinds of speech acts:
(1) a. Declarative: John is clever.
b. Interrogative: Is John clever? Who is clever?
c. Exclamative: How clever you are!
d. Imperative: Be very clever.
Each clause type in general has its own functions to represent a speech act. For example, a
declarative makes a statement, an interrogative asks a question, an exclamative represents an
exclamatory statement, and an imperative issues a directive. However, these correspondences
are not always one to one. For example, the declarative in (2a) represents not a statement but a
question, while the interrogative in (2b) actually indicates a directive:
(2) a. I ask you if this is what you want.
b. Would you mind taking out the garbage?
In this chapter, we will focus on the syntactic structure of interrogatives, putting aside the map-
ping relationships between form and function.
There are basically two types of interrogative: yes-no questions and wh-questions:
(3) a. Yes-No questions: Can the child read the book?
b. Wh-questions: What can the child read?
Yes-no questions are different from their declarative counterparts by having subject and auxil-
iary verb in an inverted order. As we have seen in Chapter 8, such yes-no questions are generated
through the combination of an inverted finite auxiliary verb with its subject as well as with its
complement in accordance with the SAI Rule:
195
(4)
S
.
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
V
_
¸
¸
¸
¸
¸
¸
_
INV +
AUX +
SPR 1
COMPS 2
_
¸
¸
¸
¸
¸
¸
_
1 NP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
2 VP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
Can the child read the book?
In addition to the subject-auxiliary inversion, wh-questions are introduced by one of the
interrogative words such as who, what, which or where. The wh-phrases formed from these wh-
words have a variety of functions in the clause. For example, they can be the subject, object,
oblique complement, or even an adjunct:
(5) a. [Who] called the police?
b. [Which version] did they recommend?
c. [With what] did the baby eat the food?
d. [How] did he eat the food?
The wh-phrase need not be an NP – it can be a PP, AP, or AdvP:
(6) a. [
NP
Which man] [did you talk to ]?
b. [
PP
To which man] [did you talk ]?
c. [
AP
How ill] [has Hobbs been ]?
d. [
AdvP
How frequently] [did Hobbs see Rhodes ]?
As seen in these examples, in terms of structure, each wh-question consists of two parts: a
wh-phrase and an inverted sentence with a missing phrase (indicated by the underline). The wh-
phrase (filler) and the missing phrase (gap) must have an identical syntactic category as a way
of ensuring their linkage:
(7) a. *[
NP
Which man] [did you talk [
PP
]]?
b. *[
PP
To which man] [did you talk to [
NP
]]?
Another important property is that the distance between the filler and the gap is not bound
within one sentence, and it can be ‘long-distance’ or ‘unbounded’:
(8) a. [[Who] do you think [Tom saw ]]?
b. [[Who] do you think [Mary said [Tom saw ]]]?
c. [[Who] do you think [[Hobbs imagined [Mary said [Tom saw ]]]]]?
196
As can be seen here, as long as the link between the filler and the gap is appropriate, the
distance between the two can be unbounded. This long-distance relationship gives wh-questions
and other similar constructions the family name of ‘long-distance dependency’.
10.2 Movement vs. Feature Percolation
Traditionally, there have been two different ways to link the filler wh-phrase with its missing
gap. One traditional way of linking the two is to assume that the filler wh-phrase is moved to the
sentence-initial position by movement operations as represented in (9) (cf. Chomsky 1981a):
(9)
CP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
NP C

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
who C S

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
will NP

¸
¸
¸
¸
VP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
they V VP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
t
¸¸
V NP
recommend t
¸¸
The wh-phrase who is originally in the object position of recommend and then moved to the
specifier position of the intermediate phrase C

. The auxiliary verb will is also moved from the
V position to the C.
This kind of movement operation can be appealing at first glance in capturing the linkage
between the filler and gap. However, the notion of moving an overt element to form a wh-
question immediately runs into a problem with examples like the following:
(10) a. Who did Kim work for and Sandy rely on ?
b. *Who did Kim work for and Sandy rely ?
c. *Who did Kim work for and Sandy rely on Mary?
If we adopt a movement process for (10a), there must be an operation in which the two NP gaps
(marked by the underlines) are collapsed into one NP and become who. We cannot simply move
one NP, because it will generate an ill-formed example like (10b).
There is also a class of so-called ‘movement paradox’ examples, provided by Bresnan (1991,
2001). First, consider the following:
(11) a. You can rely on [Edward’s help].
197
b. [Edward’s help], you can rely on
(12) a. We talked about [the fact that he was sick for days].
b. [The fact that he was sick for days], we talked about .
In a movement approach, both of the (b) examples are derived from the (a) examples by moving
the NPs to the sentence initial position. However, not every putatively ‘derived’ example has a
well-formed source:
(13) a. *You can rely on that he will help you.
b. [That he will help you], you can rely on .
(14) a. *We talked about [that he was sick for days].
b. [That he was sick], we talked about for days.
(15) a. *This theory captures that arrows don’t stop in midair.
b. [That arrows don’t stop in midair] is captured by this theory.
If we take the same rationale as for (12), it is difficult to explain how the putative source example
is ungrammatical while a derived form is grammatical. This inconsistency between the fronted
phrase and the putative missing phrase casts doubt on the existence of movement operations in
such examples.
Instead of postulating movement as an operation, we can assume that there is no movement
process at all, but instead a mechanism of communication through the tree, known as feature
percolation, to generate such wh-questions. For example, the information that an NP is missing
or gapped can be passed up in the tree until the gap meets its corresponding filler:
(16)
S

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
NP S/NP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
who V NP VP/NP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
.
did they V NP/NP
.
recommend t
The notations such as NP/NP (read as ‘NP slash NP’) or S/NP (‘S slash NP’) here mean that
the category to the left of the slash is incomplete and missing one NP. This missing information
is percolated up to the point where it meets its filler who. There is thus no notion of movement
here, but just a feature percolation up to the point where the missing gap meets its filler.
This kind of feature percolation system can account for the contrast given in (10a) and (10b).
Let us look at partial structures of these two examples:
198
(17) a.
S/NP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
S/NP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
and S/NP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
NP VP/NP

¸
¸
¸
¸
¸
¸
¸
NP VP/NP

¸
¸
¸
¸
¸
¸
Kim work for Sandy rely on
b.
*S/NP & S/PP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
S/NP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
and S/PP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
NP VP/NP

¸
¸
¸
¸
¸
¸
¸
NP VP/PP

¸
¸
¸
¸
¸
Kim work for Sandy rely
In (17a), the missing gaps are both NPs whereas in (17b), an NP and a PP are missing. Since
the mechanism of feature unification allows two nonconflicting phrases to be unified into one,
two S/NP phrases in (17a) are merged into one S/NP. Simply, the whole coordinate structure
is ‘missing an NP’ and this description is true of each internal conjunct. However, in (17b) we
cannot unify the two different phrases S/NP and S/PP into one since they have conflicting slash
values (See section 3.2 for the analysis of paradox examples).
10.3 Feature Percolation with No Abstract Elements
10.3.1 Basic Systems
As a more formal way of stating the feature percolation system, we can introduce the feature
attribute GAP for an empty phrase and pass this up to the point where the gap value is discharged
by its filler. However, even within such an approach, an issue remains of positing an empty
element. An empty element is an abstract entity introduced for a theoretical reason (for example,
the GAP feature may ‘start off’ at the bottom of the tree in virtue of an invisible element t of
category NP/NP, as in Gazdar et al. 1985). Though the introduction of an empty element with
no phonological value might be reasonable, examples like the following raise issues that are not
easily solved (Sag and Fodor 1994, Sag 2000):
(18) a. *Who did you see [
NP
[
NP
] and [
NP
a picture of [
NP
]]]?
b. *Who did you compare [
NP
[
NP
] and [
NP
]]?
On the assumption that empty elements are identical to canonical phrases except for the fact
that they have no phonological values at all, nothing would block us from coordinating two
199
empty phrases, leading to incorrect predictions. If we can avoid positing empty elements that
we cannot see or hear, it would be better in theoretical as well as empirical terms (cf. Pullum
1991).
One way to do without an abstract element is to encode the missing information in the lexical
head (Sag et al. 2003). For example, the verb recommend can be realized with different overt
complements:
(19) a. These qualities recommended him to Oliver.
b. The UN recommended an enlarged peacekeeping force.
(20) a. This is the book which the teacher recommended .
b. Who will they recommend ?
In (19), the verb recommend is realized in canonical uses whereas the one in (20) is not. That
is, in (19), the object of the verb is present as its sister, whereas in (20) the object is in a
nonlocal position. This difference can be represented as a difference in possibilities allowed by
the following revised ARC:
(21) Argument Realization Constraint (ARC, second approximation):
The first element on the ARG-ST list is realized as SPR, the rest as COMPS or
GAP in syntax.
This revised ARC will then allow the following realizations for recommend:
(22)
a.
_
¸
¸
¸
¸
¸
¸
_
recommend
VAL
_
¸
¸
¸
_
SPR 1
COMPS 2
GAP
_
¸
¸
¸
_
_
¸
¸
¸
¸
¸
¸
_
_
¸
¸
¸
_
verb-lxm
recommend
ARG-ST 1 , 2
_
¸
¸
¸
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸

b.
_
¸
¸
¸
¸
¸
¸
_
recommend
VAL
_
¸
¸
¸
_
SPR 1
COMPS
GAP 2
_
¸
¸
¸
_
_
¸
¸
¸
¸
¸
¸
_
In (22a), the two arguments of the verb are realized as the SPR and COMPS value respectively,
whereas in (22b) the second argument is realized not as a COMPS value but as a GAP value.
Each of these two different realizations will project the following structures for examples like
(19b) and (20b), respectively:
200
(23) a.
S
.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
_
NP

¸
¸
¸
¸
¸
VP[fin]
.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
_
The UN V[fin] NP
.
.
.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
_
_
_
recommended an enlarged peacekeeping force
b.
S[GAP ]
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
1 NP S[GAP 1 NP]
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
Who V NP VP[GAP 1 NP]
.
will they V[GAP 1 NP]
¸_
recommend
The main difference between the two is that in (23a), the object of recommend is its sister
whereas in (23b) it is not. That is, in the former the object is local to the verb whereas in the
latter it is nonlocal. In (23b), the verb contains a GAP value which is identified with the object.
This GAP value is passed up to the VP and then to the lower S. At this level, this GAP value is
discharged by the filler who, more specifically, by the following Head-Filler Rule:
(24) Head-Filler Rule:
S
_
GAP
_
→ 1 , S
_
GAP 1
_
This grammar rule says that when a head expression S containing a nonempty GAP value com-
bines with its filler value, the resulting phrase will form a grammatical head-filler phrase with
the GAP value discharged. This completes the ‘top’ of the long-distance or unbounded depen-
dency.
10.3.2 Non-subject Wh-questions
Let us see how the present system generates a non-subject wh-question, using the verb put for
illustration. This verb will select three arguments as given here:
(25)
_
_
put
ARG-ST NP, NP, PP
_
_
The ARC will ensure that of these three arguments the first one must be realized as a SPR
element, and the rest either as COMPS or as GAP elements. We thus will have at least the
201
following three realizations for the verb put:
1
(26)
a.
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
put
VAL
_
¸
¸
_
SPR 1
COMPS 2 , 3
GAP
_
¸
¸
_
ARG-ST
_
1 , 2 , 3
_
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
b.
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
put
VAL
_
¸
¸
_
SPR 1
COMPS 3
GAP 2
_
¸
¸
_
ARG-ST
_
1 , 2 , 3
_
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
c.
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
put
VAL
_
¸
¸
_
SPR 1
COMPS 2
GAP 3
_
¸
¸
_
ARG-ST
_
1 , 2 , 3
_
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
Each of these three lexical entries will then generate sentences like the following:
(27) a. John put the books in a box.
b. Which books did John put in the box?
c. Where did John put the books?
As we see here, the complements of the verb put are realized in three different ways. The
verb put in (26a) has the canonical realization of the verb’s arguments, generating an example
like (27a). Meanwhile, in (26b), the object NP argument is realized as a GAP as reflected in
(27b) whereas in (26c), the PP is realized as a GAP as shown in (27c). The following structure
represents how the lexical realization (26b) provides for examples like (27b) in more detail:
(28)
S[QUE +]

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
2 NP[QUE +]

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
S
_
GAP 2
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
Which book V
1 NP
VP
_
_
SPR 1
GAP 2
_
_
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
did John
V
_
¸
¸
¸
_
SPR 1
COMPS 3
GAP 2
_
¸
¸
¸
_
3 PP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
put in the box
1
The SPR value can be gapped too. See the next section.
202
Let us look at the structure working from bottom to top. At the bottom, the verb put has one
PP complement with its NP complement as a GAP value. This GAP information is passed all
the way up to the lower S where it is discharged by the filler NP which book. Each phrase is
licensed by the grammar rules: the verb put with this GAP information first combines with just
the necessary PP complement in the box, in accordance with the Head-Complement Rule. The
inverted auxiliary verb did combines with the subject NP John and the just-described VP in
accordance with the SAI rule. The resulting S is still incomplete because of the nonempty GAP
value (every sentence needs to have an empty GAP value). This GAP value is discharged when
the Head-Filler Rule in (24) combines the filler NP which book with the incomplete S.
2
This kind of feature percolation system, involving no empty elements, works well even for
long-distance dependency examples. Consider the following structure:
(29)
S
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
1 NP S[GAP 1 NP]
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
Who V NP VP[GAP 1 NP]
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
.
.
.
.
.
.
.
.
.
.
.
do you V S[GAP 1 NP]
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
.
think NP VP[GAP 1 NP]
.
Hobbs V[GAP 1 NP]
¸_
met
The GAP value starts from the lexical head met whose second argument is realized as a GAP
value. Since the complement of the verb met is realized as a GAP value, the verb met will not
look for its complement in the local domain (as its sister node). The GAP information will be
passed up to the embedded S, which is a nonhead daughter. It is the following NIP that ensures
the GAP value in the head daughter or nonhead daughter to be passed up through the structure
until it is discharged by the filler who by the Head-Filler Rule:
3
(30) Nonlocal Feature Inheritance Principle (NIP):
A phrase’s nonlocal feature such as GAP, QUE, and REL is the union of its daugh-
ters’ nonlocal feature values minus any bound nonlocal features.
The role of this principle is clear from the embedded S in (29): The principle allows the GAP
in this nonhead S to pass up to the VP.
2
Every wh-element in questions carries the feature [QUE +].
3
For the feature REL, see Chapter 11.
203
With this principle together, we can observe that the treatment of long distance dependency
within the feature percolation system involves three parts: top, middle, and bottom. The bottom
part introduces the GAP value according to the ARC. The middle part ensures the GAP value
is inherited ‘up’ to the mother in accordance with the NIP. Finally, the top level terminates the
GAP value by the filler in accordance with the Head-Filler Rule.
It is also easy to verify how this system accounts for examples like (31) in which the gap is
a non-NP:
(31) a. [In which box] did John put the book ?
b. [How happy] has John been ?
The Head-Filler Rule in (24) ensures that the categorial status of the filler is identical with that
of the gap. The structure of (31a) can be represented as following:
(32)
S
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
3 PP

¸
¸
¸
¸
¸
¸
¸
S[GAP 3 PP]
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
In which box V NP VP[GAP 3 PP]
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
did John
V
_
_
COMPS 2 NP
GAP 3 PP
_
_
2 NP

¸
¸
¸
¸
¸
¸
¸
¸
¸
put the book
In this structure, the missing phrase is a PP encoded in the GAP value. This value is percolated
up to the lower S and discharged by the filler in which box.
In addition, notice that in this approach we have a more clear account for the examples we
have seen in (10), which we repeat here:
(33) a. Who did Kim work for and Sandy rely on ?
b. *Who did Kim work for and Sandy rely on Mary?
As we have seen earlier in Chapter 2, English allows two identical phrases to be conjoined. This
then means that the GAP value in each conjunct also need to be identified:
(34) Coordination Rule:
XP →XP[GAP A] conj XP[GAP A]
This grammar rule then explains the contrast in (33), as represented in the simplified structures:
204
(35) a.
S
_
GAP 1 NP
_
.
.
.
.
.
.
.
.
.
.
.
.
¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸
S
_
GAP 1 NP
_

_
_
_
_
_
_
_
_
_
_
_
_
and
S
_
GAP 1 NP
_

_
_
_
_
_
_
_
_
_
_
_
_
NP
VP
_
GAP 1 NP
_

¸
¸
¸
¸
¸
¸
NP
VP
_
GAP 1 NP
_

¸
¸
¸
¸
¸
¸
Kim work for Sandy rely on
b.
*S
_
GAP 1 NP
_
.
.
.
.
.
.
.
.
.
.
.
.
¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸
S
_
GAP 1 NP
_

_
_
_
_
_
_
_
_
_
_
_
_
and
S
_
GAP
_

_
_
_
_
_
_
_
_
_
_
_
_
NP
VP
_
GAP 1 NP
_

¸
¸
¸
¸
¸
¸
NP
VP
_
GAP
_

¸
¸
¸
¸
¸
¸
¸
Kim work for Sandy rely on Mary
In (35a), the GAP value in the first conjunct is identical with that in the second conjunct, satis-
fying the Coordination Rule. The feature unification will allow these two identical GAP values
to be unified into one. However in (35b), the GAP values in the two conjuncts are different,
violating the Coordination Rule and thus cannot be unified.
Notice that this feature-based analysis can also offer a way of dealing with the movement
paradox examples we observed in (13), repeated here:
(36) a. *You can rely on that he will help you.
b. [That he will help you], you can rely on .
The introduction of a GAP value is a lexical realization process in the present system, implying
that we can assume that the complement of the preposition on in (37a) can be realized either as
an NP in (37b) or as a nominal GAP element as in (37c):
205
(37)
a.
_
¸
¸
_
prep-v-p
on
ARG-ST 1 XP
_
¸
¸
_
b.
_
¸
¸
_
on
COMPS NP
GAP
_
¸
¸
_
c.
_
¸
¸
_
on
COMPS
GAP XP[nominal]
_
¸
¸
_
This realization means that when the preposition on is serving as the part of a prepositional
verb (prep-v-p) like rely on, its prepositional complement can be either realized as an NP or as a
nominal GAP element. This means when the argument of on is realized as a COMPS element,
it must be an NP. However, when its argument is realized as a GAP, the GAP value can be either
a CP or an NP since as we have seen in Chapter 5 type nominal subsumes both comp and noun.
This lexical realization will then project a structure like the following:
(38)
S

_
_
_
_
_
_
_
_
_
_
_
_
_
1 CP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
S
_
GAP 1 [nominal]
_
.
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
That he will help you
NP
VP
_
GAP 1 [nominal]
_

_
_
_
_
_
_
_
_
_
_
_
_
_
you V
VP
_
GAP 1 [nominal]
_

¸
¸
¸
¸
¸
¸
¸
can V
PP
_
GAP 1 [nominal]
_
rely
P
_
GAP 1 [nominal]
_
on
The present system, allowing certain flexibility in the argument realization, thus can capture
movement paradox examples without resorting to any movement operations.
10.3.3 Subject Wh-Questions
Now consider examples in which the subject is the focus of the wh-question:
(39) a. Who put the book in the box?
206
b. Who DID put the book in the box?
c. Who can put the book in the box?
We can notice that when the subject who is questioned, the presence of an auxiliary verb is
optional. That is, the question in (39a) is well-formed, even though no auxiliary is present.
The related example (39b) is also well-formed, but is used only when there is emphasis on the
auxiliary.
As a first step to account for such examples, we can, adopting a similar structure like non-
subject wh-questions, posit a structure like (40) in which the subject is gapped:
(40) a. Who put the book in the box?
b. Who can put the book in the box?
In the current context, our grammar requires no additional mechanism other than slightly
revising the ARC:
(41) Argument Realization Constraint (ARC, final):
The first element on the ARG-ST list is realized as SPR or GAP, the rest as
COMPS or GAP in syntax.
This revised ARC eventually guarantees that the values of the ARG-ST is the sum of that of
SPR, COMPS, and GAP. The system then allows the following lexical realization for put, in
addition to those in (26):
(42)
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
put
VAL
_
¸
¸
_
SPR
COMPS 2 , 3
GAP 1
_
¸
¸
_
ARG-ST
_
1 , 2 , 3
_
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
This realization in which the subject is gapped then projects the following structure for (40a):
207
(43)
S
_
_
SPR
COMPS
_
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
1 NP[QUE +]
‘S’
_
¸
¸
¸
_
SPR
COMPS
GAP 1
_
¸
¸
¸
_
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
Who
V
_
¸
¸
¸
¸
¸
¸
_
SPR
COMPS 2 , 3
GAP 1
ARG-ST 1 , 2 , 3
_
¸
¸
¸
¸
¸
¸
_
2 NP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
3 PP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
put the book in the box
As shown in the structure, the subject of put is realized as the GAP value, passing up to the
mother node. This node looks like a VP, but in fact it is a projection of V with an empty SPR
list, and hence is effectively a kind of S (by definition, S is a projection of V which has an empty
SPR and COMPS list). This incomplete ‘S’ with the subject missing then can combine with the
filler who according to the Head-Filler Rule.
Even though the ‘S’ with the subject gapped cannot function as an independent sentence as
in *Visited him, it can function as the complement of a verb like think as in sentences like the
following:
(44) a. Who do you think [visited Seoul last year]?
b. That’s the UN delegate that the government thinks [visits Seoul last year].
The verb think can select either a finite S or a CP as in I think (that) she knows chorus. This
means that the verb can also combine with an ‘S’ with the subject being gapped:
208
(45)
S
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
NP S[GAP 1 NP]
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
1 Who
V NP VP[GAP 1 NP]
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
.
.
.
.
.
.
.
.
.
.
.
do you V ‘S’[GAP 1 NP]
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
think V[GAP 1 NP] NP
visited Seoul
This kind of analysis can provide a way of explaining the so-called ‘that-trace effect’ in (46)
as well as the ‘adverbial amelioration effect’ in (47):
(46) a. Who do you believe that Sara invited ?
b. Who do you believe invited Sara?
c. *Who do you believe that invited Sara?
(47) a. *Who do you think that would be nominated for the position?
b. This is the kind of person who I doubt that [under normal circumstances] would
have anything to do with such a scheme.
What we can notice in (46) is that when the complementizer that is present, we cannot have the
seemingly subject gap. However, the subject gap can be salvaged by the presence of an adverbial
phrase as given in (47). We can attribute the observed idiosyncratic behavior to the lexical
peculiarities of the complementizer that. Consider the possible and impossible structures:
4
(48) a.
CP[GAP 2 NP]

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
C S[GAP 2 NP]

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
that NP VP[GAP 2 NP]

¸
¸
¸
¸
¸
¸
Sara invited
4
The italicized hd-comps-ph and hd-mod-ph here indicates the type name of the phrase ‘S’.
209
b.
*CP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
C
‘S’[hd-comps-ph]

¸
¸
¸
¸
¸
¸
¸
that invited Sara
c.
CP[GAP 1 NP]

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
C ‘S’
_
_
hd-mod-ph
GAP 1 NP
_
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
that PP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
‘S’[GAP 1 NP]

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
under normal circumstances would have anything . . .
(48a) is a well-formed structure in which the embedded clause has an object gap. (48b) is out
in which the complementizer combines with the subject missing ‘S’. (48c) is fine even though
it also has the subject missing ‘S’. As we have noted throughout the chapters, English grammar
rules license well-formed phrases identified by the phrasal types such as hd-spr-ph, hd-comps-
ph, hd-mod-ph and so forth. As represented in the structures here, the complementizer that can
select either an S or ‘S’, but cannot combine with a head-comps-ph ‘S’ (a finite VP with its
subject gapped). This kind of negative constraint assigns a special status to the complementizer
usage of that.
10.4 Indirect Questions
10.4.1 Basic Structure
Among the verbs selecting a sentential or clausal complement (S or CP), there are also verbs
selecting an indirect question:
(49) a. John asks [whose book his son likes ].
b. John has forgotten [which player his son shouted at ].
c. He told me [how many employees Karen introduced to the visitors].
It is not the case that all verbs allow an indirect question as their complement:
(50) a. Tom denied [(that) he had been reading the article].
b. *Tom denied [which book he had been reading].
(51) a. Tom claimed [(that) he had spent five thousand dollars].
210
b. *Tom claimed [how much money she had spent].
Factive verbs like deny or claim cannot combine with an indirect question: only a finite declar-
ative clause can function as their complement. Verbs selecting an indirect question as their
complement can be in general classified by their meaning:
(52) a. interrogative verbs: ask, wonder, inquire, . . .
b. verbs of knowledge: know, learn, forget, . . .
c. verbs of increased knowledge: teach, tell, inform, . . .
d. decision verbs/verbs of concern: decide, care, . . .
We can easily see that the clausal complement of these verbs cannot be a canonical CP, and
must be an indirect question:
(53) a. *John inquired [that he should read it].
b. *Peter will decide [that we should review the book].
(54) a. John inquired [which book he should read].
b. Peter will decide [which book we should review].
A caution here is that there are some verbs, such as forget, tell and know, that select either a
[QUE +] or a [QUE −] complement.
(55) a. John told us that we should review the book.
b. John told us which book we should review.
There are thus at least three different types of verb which take clausal complements, in terms of
their semantic functions (e.g., whether the complement is interpreted as declarative or question),
reflected in the following lexical entries:
(56)
a.
_
¸
¸
¸
¸
¸
¸
_
wonder
HEAD| POS verb
SPR 1
COMPS
_
_
QUE +
_
_
_
¸
¸
¸
¸
¸
¸
_
b.
_
¸
¸
¸
¸
¸
¸
_
deny
HEAD| POS verb
SPR 1
COMPS
_
_
QUE −
_
_
_
¸
¸
¸
¸
¸
¸
_
c.
_
¸
¸
¸
¸
¸
¸
_
tell
HEAD| POS verb
SPR 1
COMPS
_
_
QUE ±
_
_
_
¸
¸
¸
¸
¸
¸
_
The feature QUE originates from a wh-word like who or which and is used to distinguish be-
tween indirect questions and declarative clauses. The difference in the QUE value of the verb’s
complement will ensure that each verb combines with an appropriate clausal complement. For
example, the verb wonder, requiring a [QUE +] clausal complement, will be licensed in a struc-
ture like the following:
211
(57)
VP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
V
_
COMPS 2 [QUE +]
_ 2 S[QUE +]

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
wonder 1 NP[QUE +]

¸
¸
¸
¸
¸
¸
¸
S[GAP 1 NP]

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
whose book NP

¸
¸
¸
¸
¸
¸
VP[GAP 1 NP]
his son V[GAP 1 NP]
likes
The GAP value of likes is passed up to the lower S and discharged by the filler whose book.
The wh-word whose carries the feature [QUE +] which, similar to the feature GAP, will pass up
to the point where it is required by a verb or to the highest position to indicate that the given
sentence is a question. For example, in (58), the feature QUE marks that the whole sentence is
a question, whereas in (59) it allows the embedded clause to be an indirect question:
(58) a. [
S[QUE +]
In which box did he put the book ]?
b. [
S[QUE +]
Which book by his father did he read ]?
(59) a. John asks [
S[QUE +]
in which box he put the book].
b. John asks [
S[QUE +]
which book by his father he read].
The percolation of the feature QUE upward from a wh-word can be ensured by the NIP that
guarantees nonlocal features like QUE to be passed up until they are bound off or selected by a
sister (like a filler phrase, or a selecting V). This principled constraint allows the QUE value to
pass up to the mother from a deeply embedded nonhead as illustrated in the following:
(60) a. Kim has wondered [[in which room] Gary stayed ].
b. Lee asked me [[how fond of chocolates] the monkeys are ]].
Let us consider the structure of (60a):
212
(61)
VP

_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
V S[QUE +]

_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
wondered
1 PP
_
QUE +
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
S[GAP 1 PP]
·
·
·
·
·
·
·
·
·
·
·
·
·
·
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
P
NP
_
QUE +
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
NP VP[GAP 1 PP]
in
Det
_
QUE +
_
N Gary V[GAP 1 PP]
which room stayed
Even though which is embedded in the PP and functions as the Det of the inner NP, its QUE
value will pass up to the S, granting it the status of an indirect question. The verb wonder then
combines with this S, satisfying its subcategorization requirement. If the verb combines with a
[QUE −] clausal complement, we would then have an ungrammatical structure:
(62) a. *Kim has wondered [
[QUE −]
that Gary stayed in the room].
b. *Kim asked me [
[QUE −]
that the monkeys are very fond of chocolates].
As we have seen above, the category of the missing phrase within the S must correspond to
that of the wh-phrase in the initial position. For example, the following structure is not licensed
simply because there is no Head-Filler Rule that allows the combination of the filler NP with an
S missing a PP:
213
(63)
VP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
V *S[QUE +]
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
wondered NP[QUE +]

¸
¸
¸
¸
¸
¸
¸
¸
S[GAP 1 PP]

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
which room NP

¸
¸
¸
¸
¸
¸
VP[GAP 1 PP]
Gary V[GAP 1 PP]
stayed
In a similar fashion, the present system also predicts the following contrast:
(64) a. John knows [whose book [Mary bought ] and [Tom borrowed from her]].
b. *John knows [whose book [Mary bought ] and [Tom talked ]].
The partial structure of these can be represented as following:
(65) a.
S[QUE +]
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
1 NP[QUE +]

¸
¸
¸
¸
¸
¸
¸
¸
S[GAP 1 NP]
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
whose book
S[GAP 1 NP ]

¸
¸
¸
¸
¸
¸
¸
¸
and
S[GAP 1 NP ]

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
Mary bought Tom borrowed from her
214
b.
*
S
_
_
QUE +
GAP
_
_
.
.
.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
_
_
_
1 NP[QUE +]

¸
¸
¸
¸
¸
¸
¸
*S[GAP 1 NP]
.
.
.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
_
_
_
whose book
S[GAP 1 NP ]

¸
¸
¸
¸
¸
¸
¸
and
S[GAP 2 PP ]

¸
¸
¸
¸
¸
¸
¸
John bought Tom talked
As long as the two GAP values are identical, we can unify the two into one as in (65a). However,
if the GAP values are different as in (65b), there is no way to unify them in the coordination
structure.
10.4.2 Non-Wh Indirect Questions
English also has indirect questions headed by the complementizer whether or if :
(66) a. I don’t know [whether/if I should agree].
b. She gets upset [whether/if I exclude her from anything].
c. I wonder [whether/if you’d be kind enough to give us information].
These indirect questions are all internally complete in the sense that there is no missing element.
This means that the complementizers whether and if will have at least the following lexical
information:
(67)
_
¸
¸
¸
¸
¸
_
whether
SYN
_
¸
¸
_
HEAD| POS comp
VAL| COMPS S
QUE +
_
¸
¸
_
_
¸
¸
¸
¸
¸
_
According to this lexical information, whether selects a finite S and provides a [QUE +] value,
licensing a structure like the following:
(68)
CP[QUE +]

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
C[QUE +]

¸
¸
¸
¸
¸
¸
S[fin]

¸
¸
¸
¸
¸
¸
¸
¸
whether/if I should agree
One thing to note here is that if and whether are slightly different even though they both
carry a positive QUE feature.
5
Just like indirect questions, the clauses headed by whether can
5
See Exercise 4 of Chapter 2.
215
serve as a prepositional object.
(69) a. I am not certain about [when he will come].
b. I am not certain about [whether he will go or not].
However, an if -clause cannot function as prepositional object:
(70) a. *I am not certain about [if he will come].
b. *I am not certain about [if he will go or not].
There is also a difference between if and whether in infinitival constructions:
(71) a. I don’t know [where to go].
b. I don’t know [what to do].
c. I don’t know [how to do it].
d. I don’t know [whether to agree with him or not].
(72) a. *I don’t know [if to agree with him].
b. *I don’t know [that to agree with him or not].
This means that whether and if can both bear the feature QUE (projecting an indirect question),
but different with respect to the fact that only whether behaves like a true wh-element.
6
10.4.3 Infinitival Indirect Questions
In addition to the finite indirect questions, English also has infinitival indirect questions:
(73) a. Fred knows [which politician to support].
b. Karen asked [where to put the chairs].
Just like the finite indirect questions, these constructions have the familiar bipartite struc-
tures: a wh-phrase and an infinitival clause missing one element.
Notice at this point that in English there exist at least four different ways for the subject to
be realized: as a canonical NP, gap, and PRO, or pro:
(74) a. The student protected him. (a canonical NP)
b. Who protected him? (a gap NP)
c. To protect him is not an easy task. (PRO)
d. Protect him! (pro)
In (74a), the subject is a ‘canonical’ NP whereas those in the others are ‘non-canonical’. In the
wh-question (74b), the subject is a GAP value; in (74c), the infinitival VP has an understood,
unexpressed subject, PRO whereas the imperative in (74d) the subject is an unexpressed one,
though understood as the second person subject you. Traditionally, the unexpressed pronoun
subject of a finite clause is called ‘small pro’ whereas that of an nonfinite clause is called ‘big
PRO’ (Chomsky 1982), as they have slightly different referential properties. In the terms we
6
One way to distinguish the wh-elements including whether from if is to have an extra feature WH to distinguish
them.
216
are using in this book, this means, we have ‘canonical’ pronouns like he and him as well as
‘non-canonical’ realizations of the pronouns. To allow a VP with the non-canonical subject to
be projected into a complete S, we can assume the following Head-Only Rule:
(75) Head-Only Rule:
S[SPR ] →VP[SPR NP[noncan-pro]]
The rule says a VP whose subject is either a pro or a PRO can be directly projected into a
complete sentence with the subject being discharged. A finite VP will, however, not be pro-
jected into an S, since it selects a canonical subject. The rule as given will license the following
structures:
(76)
a.
S[SPR ] b. S[SPR ]
VP[SPR NP[pro]]

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
VP[SPR NP[PRO]]

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
V
NP
·
·
·
·
·
·
¸
¸
¸
¸
¸
¸
V
VP

¸
¸
¸
¸
¸
¸
¸
¸
Protect yourself to protect yourself
The subject of the VP here is pro or PRO: either can be licensed, and this rule in (75) allows a
VP to form a complete sentence with no pronounced subject. With this new rule, we then can
have the following structure:
217
(77)
S
_
_
VFORM inf
QUE +
_
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
1 NP[QUE +]

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
S
_
_
VFORM inf
GAP 1
_
_
which politician
VP
_
¸
¸
¸
_
VFORM inf
SPR PRO
GAP 1
_
¸
¸
¸
_

¸
¸
¸
¸
¸
¸
¸
to support
Consider the structure from the bottom up. The verb support selects two arguments whose
second argument can be realized as a GAP:
(78)
_
¸
¸
¸
¸
¸
¸
¸
¸
_
support
VAL
_
¸
¸
_
SPR 1 NP[PRO]
COMPS
GAP 2 NP
_
¸
¸
_
ARG-ST 1 NP, 2 NP
_
¸
¸
¸
¸
¸
¸
¸
¸
_
The verb will then form a VP with the infinitival marker to. Since this VP’s subject is PRO, the
VP can be projected into an S with the accusative NP GAP value in accordance with the Head-
Only Rule in (75). The ‘S’ then forms a well-formed head-filler phrase with the filler which
politician. The QUE value on the phrase ensures the whole infinitival clause to function as an
indirect question which can be combined with the verb knows.
One constraint we observe in the infinitival wh-questions is that the infinitival part cannot
have its subject to be realized:
(79) a. *Fred knows [which politician for Karen/her to vote for].
b. *Karen asked [where for Jerry/him to put the chairs].
218
The data indicate that in infinitival indirect questions, the subject of the infinitival VP cannot
appear. If we look at the structure, we can easily see why this is not a legitimate structure:
7
(80)
VP
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
V *S
_
QUE +
_
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
knows 1 NP[QUE + ]

¸
¸
¸
¸
¸
¸
¸
¸
¸
CP
_
GAP 1 NP
_
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
which politician C S
_
GAP 1 NP
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
for Karen to vote for
The structure indicates that the Head-Filler Rule licenses the combination of an S with its filler,
not a CP with its filler.
10.4.4 Adjunct wh-questions
The main-clause wh-questions and indirect questions we have seen so far have a GAP value
originating from an argument position of a verb or preposition. How can the present system
account for examples like the following, in which wh-phrases are intuitively not arguments but
adjuncts?
(81) a. How carefully have you considered your future career?
b. When can we register for graduation?
c. Where do we go to register for graduation?
d. Why have you borrowed my pencil?
One way to deal with such examples is to take the adverbial wh-phrase to modify an inverted
question:
7
The grammar needs to block examples like (i) in which the infinitival VP combines with its subject:
(i) a. *Fed knows [
S
which politician [
S
her [to vote for]]].
b. *Karen asked [
S
where [
S
him [to vote for]]].
As in (77), the Head-Filler Rule allows an S (directly projected from an infinitival VP) to combine with its filler. As a
way of blocking such examples, we may assume an independent constraint that the infinitival subject can appear only
together with the complementizer for because the subject needs to get the accusative case from it. Cf. Chomsky (1982).
219
(82)
S
_
QUE +
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
AdvP
_
MOD 1 S
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
1 S
_
_
AUX +
INV +
_
_
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
How carefully have you considered your future career?
The structure indicates that the AdvP modifies the inverted S.
Matters become complicated when we consider questions with the interpretation of wh-
words (cf. Huang (1982)).
(83) a. When did he say that he was fired?
b. Where did he tell you that he met Mary?
c. How did you guess that he fixed the computer?
These sentences are ambiguous with respect to the local interpretation of the wh-words. (83a)
can question either the time he made the statement or the time he was fired; (83b) can question
the place he told you the fact or the place he met Mary. The same is true in (83c): the wh-word
can question the proposition of the main clause or that of the embedded clause.
These data indicate that in addition to a structure like (82) in which the adverbial wh-word
modifies the whole sentence, we need a structure where the adverbial wh-phrase linked to the
embedded clause. As a way of doing it, following Sag (2005) and others, we can assume that
English allows the extension of the ARG-ST to include a limited set of adverbial elements as
an argument. For example, we can extend the regular verb fix to include a locative adverbial as
its argument.
(84) Extended ARG-ST:
_
_
fix
ARG-ST 1 NP, 2 NP
_
_

_
_
fix
ARG-ST 1 NP, 2 NP, AdvP
_
_
This extended ARG-ST then can allow us its adverbial argument to be realized as a GAP value
according to the ARC (Argument realization Constraint):
(85)
_
¸
¸
¸
¸
¸
¸
¸
_
fix
SPR 1 NP
COMPS 2 NP
GAP 3 AdvP
ARG-ST 1 NP, 2 NP, 3 AdvP
_
¸
¸
¸
¸
¸
¸
¸
_
220
This lexical realization will then be able to project a structure like the following for the sentence
((83)c):
(86)
S

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
3 AdvP
S
_
GAP 3
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
How V NP
VP
_
GAP 3
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
did he V
S
_
GAP 3
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
guess NP
VP
_
GAP 3
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
he
V
_
GAP 3
_
NP
·
·
·
·
·
·
·
·
·
·
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
fixed the computer
This structure indicates that the wh-word how originates from the subordinate clause VP. We
have seen that the GAP value originates from the verb fixed whose arguments can include the
adverbial element. Note this does not mean that we can extend the ARG-ST randomly. For
example, the argument extension cannot be applied to examples like the following:
(87) a. Why do you wonder whether she will invite me?
b. How often did he ask when she will meet at the party?
In these examples we have only one interpretation where the wh-phrase modifies the matrix
verb wonder or ask. This means that the argument extension is limited, depending on context.
221
10.5 Exercises
1. Draw tree structures for the following sentences and indicate which grammar rules li-
cense the combination of each phrase.
(i) a. What causes students to select particular majors?
b. Who will John ask for information about summer courses?
c. Which textbook did the teacher use in the class last summer?
d. Whose car is blocking the entrance to the store?
(ii) a. When can we register for graduation?
b. Why do you think he left?
c. Where do we go to register for graduation?
(iii) a. Who do you guess will be here?
b. Who do you think borrowed my book?
c. Which city does Fred think that you believe that John lives in?
2. Draw tree structures for the following sentences involving indirect questions and provide
the lexical entries for the underlined words.
(i) a. I wonder on which shelf John will put the book?
b. What proof that he has implicated have you found?
c. Joseph has forgotten how many matches he has won.
d. Fred will warn Martha that she should claim that her brother is patriotic.
e. That Bill tried to discover which drawer Alice put the money in made us
realize that we should have left him in Seoul.
f. Jasper wonders which book he should attempt to persuade his students to
buy.
g. The committee knows whose efforts to achieve peace the world should
honor.
3. Explain why the following examples are ungrammatical.
(i) a. *I wonder if on which shelve John will put the book.
b. *Which house does your friend live?
c. *I wonder what city that Romans destroyed.
d. *John was wondering to whom he was referring to.
e. *Who do you think that has given the tickets to Bill?
f. *What city will Fred say that Mary thinks that John lives?
g. *On whom does Dana believe Chris knows Sandy trusts?
h. *The politician denied how the opponent was poisoned.
i. *Fred knows which book for the children to read during the summer vaca-
tion.
4. Look at the following data set and state the constraints on the usage of the ing verbs
(mending, investigating, restoring). In addition, draw trees for the a-examples together
with the lexical entries for the main and participle verbs.
222
(i) a. This needs mending.
b. *This needs mending the shoe.
c. *He mended.
d. He mended the shoe.
(ii) a. This needs investigating.
b. *This needs investigating the problem.
c. *They investigated.
d. They investigated the problem.
5. Provide the lexical entries for each of the underlined words and then draw structures for
those sentence that include the underlined word.
(i) Within grammar lies the power of expression. Understand grammar, and you
will understand just how amazing a language is. You uncover the magician’s
tricks, you find the inner workings of not only your own language, but you can
also see how it is different from the language you’re studying. You will find
that different languages are better for expressing different ideas, and you will
be able to make conscious decisions about how you communicate. Once you
know how to use each part of speech, you will be able to expand outside of the
box and express yourself in ways that no one has ever expressed themselves
before. A solid understanding of the grammar of a language gives you the
skeleton, and your words bring it to life. That is why we study grammar.
8
8
From ‘GRAMMAR (no, don’t run, I want to be your friend!)’ by Colin Suess
223
11
Relative Clause Constructions
11.1 Introduction
English relative clauses, modifying a preceding NP, are also another type of long distance de-
pendency constructions in the sense that the distance between the filler and the gap can be
unbounded:
(1) a. The video [which [you recommended ]] was really terrific.
b. The video [which [I thought [you recommended ]]] was really terrific.
c. The video [which [I thought [John told us [you recommended ]]]] was really
terrific.
Such English relative clauses can be classified according to several criteria. We can first classify
them by the type of missing element in the relative clause:
(2) a. the student who won the prize
b. the student who everyone likes
c. the baker from whom I bought these bagels
d. the person whom John gave the book to
e. the day when I met her
f. the place where we can relax
As seen here, the missing phrase can be subject, object, or oblique argument, prepositional
object, or even a temporal or locative adjunct.
Relative clauses can also be classified by the type of relative pronoun: in English we find
wh-relatives, that-relatives, and bare relatives.
1
(3) a. The president [who [Fred voted for]] has resigned.
b. The president [that [Fred voted for]] dislikes his opponents.
c. The president [ [Fred voted for]] has resigned.
1
We consider that in relative clauses to be a form of a relative pronoun.
225
wh-relatives like (3a) have a wh-type relative pronoun, (3b) has the relative pronoun that, while
(3c) has no relative pronoun at all.
Relative clauses can also be classified according to the finiteness of the clause. Unlike the
finite relative clauses in (1)–(3), the following examples illustrate infinitival relatives:
(4) a. He is the kind of person [with whom to consult ].
b. These are the things [for which to be thankful ].
c. We will invite volunteers [on whom to work ].
In addition, English allows so-called ‘reduced’ relative clauses. The examples in (5) are
‘reduced’ in the sense that the string ‘wh-phrase + be’ appears to be omitted, as indicated by the
parentheses:
(5) a. the person (who is) standing on my foot
b. the prophet (who was) descended from heaven
c. the bills (which were) passed by the House yesterday
d. the people (who are) in Rome
e. the people (who are) happy with the proposal
This chapter first reviews the basic properties these various types of English relative clauses,
and then provides analyses of their syntactic structures.
11.2 Non-subject Wh-Relative Clauses
Let us consider some canonical relative clauses, first:
(6) a. the senators [who [Fred met ]]
b. the apple [that [John ate ]]
c. the problem [ [you told us about ]]
Just like wh-questions, we can notice that relative clauses have bipartite structures: a wh-element
and a sentence with a missing element (S/XP):
(7) a. wh-element S/XP
b. that S/XP
c. [ ] S/XP
Assuming that relative wh-words carry the REL feature whose index value is identical with the
noun that the relative clause modifies, we can represent the structure of (6a) as following:
226
(8)
N

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
N

i
S
_
REL i
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
senators 1 NP[REL i] S[GAP 1 NP]

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
who NP VP[GAP 1 NP]
Fred V[GAP 1 NP]
met
As shown in the structure, the object of the verb met is realized as a GAP value, and the filler
who functions as its filler. The combination of the filler who and the gapped sentence Fred met
thus forms a well-formed head-filler phrase.This filler who has the nonlocal REL feature whose
value is an index. The REL value, identical with the antecedent noun, then percolates up to the
mother in accordance with the NIP (Nonlocal Inheritance Principle). This REL feature has to
be treated differently from the QUE feature due to contrasts like the following:
(9) a. This is the student [pictures of whom] appeared in the newspaper.
b. *[Pictures of whom] appeared in the newspaper?
The REL feature can percolate up within the phrase [pictures of whom], but the QUE feature
here can not.
One question that arises here is what mechanism ensures that the relative clause functions
as a modifier of a noun or noun phrase, carrying the MOD feature? In Chapter 6, we saw that
phrases such AP, nonfinite VP, and PP can modify an NP:
(10) a. the people [happy with the proposal]
b. the person [standing on my foot]
c. the bills [passed by the House yesterday]
d. the paper [to finish by tomorrow]
e. the student [in the classroom]
All these postnominal elements bear the feature MOD, originating from the head happy, stand-
ing, passed and to:
227
(11)
NP
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
DetP
N

_
_
_
_
_
_
_
_
_
the 1 N

VP
_
MOD 1 N

_
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
boy
V
_
MOD 1 N

_
PP

¸
¸
¸
¸
¸
¸
¸
¸
standing on the foot
Notice that not all phrases can function as postmodifiers. In particular, a base VP or finite
VP cannot be found in this environment:
(12) a. *the person [stand on my foot]
b. *the person [stood on my foot]
c. *the person [stands on my foot]
Also, a complete sentence cannot function as a postnominal modifier, either:
(13) a. *The student met the senator [John met Bill].
b. *The student met the senator [that John met Bill].
c. *The student met the senator [for John to meet Bill].
Only relative clauses with one missing element may serve as postnominal modifiers, indicating
that they also have the MOD feature. Since all the relative clauses (except bare relatives) are
introduced by a relative pronoun, it is reasonable to assume that a clause with the [REL i] feature
also bears the MOD feature as a constructional constraint, according to the following rule:
2
(14) Head-Rel Modifier Rule:
N

→ 1 N

i
, S
_
¸
¸
_
REL i
MOD 1
GAP
_
¸
¸
_
The rule, as a subtype of the Head-Modifier Rule, basically ensures that a clause marked with
the REL feature modifies a preceding noun with the identical index value. This grammar rule
now then ensures the presence of the MOD feature in the relative clause (8):
2
Following Sag (1997), one can develop an analysis in which the MOD value is introduced from a verb whose
argument contains a GAP value.
228
(15)
N

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
3 N

i
S
_
_
MOD 3 N

REL i
_
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
senators 2 NP[REL i]

¸
¸
¸
¸
¸
¸
¸
S
_
GAP 2
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
whom NP

¸
¸
¸
¸
¸
¸
¸
VP
_
GAP 2
_
Fred V
_
GAP 2
_
met
As shown here, the verb met realizes its object as a GAP value, which percolates up to the
S, where it is discharged by the relative pronoun whom. In addition, in accordance with the
Head-Rel Modifier Rule in (14), the relative clause, forming a head-filler phrase, now carries
the MOD feature as well.
Since the relative clause is a type of head-filler phrase, there must be a total syntactic identity
between the gap and the filler with a REL value:
(16) a. Jack is the person [[
NP
whom] [Jenny fell in love with [
NP
]]].
b. Jack is the person [[
PP
with whom] [Jenny fell in love [
PP
]]].
(17) a. *Jack is the person [[
NP
whom] [Jenny fell in love [
PP
]]] .
b. *Jack is the person [[
PP
with whom] [Jenny fell in love with [
NP
]]].
In (16a) and (16b), the gap and the filler are the same category, whereas those in (17) are not.
The putative gap in (17a) is a PP and the one in (17b) an NP, but the fillers are the non-matching
categories NP and PP, respectively.
In addition, the gap can be embedded in a deeper position as long as it finds the appropriate
filler:
229
(18)
N

_
_
_
_
_
_
_
_
_
_
_
_
_
2 N
i

S
_
_
REL i
MOD 2 N

_
_
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
video
1 NP
_
REL i
_
¸¸
S
_
GAP 1 NP
_

_
_
_
_
_
_
_
_
_
_
_
_
_
which NP
VP
_
GAP 1 NP
_

¸
¸
¸
¸
¸
¸
¸
¸
I V
S
_
GAP 1 NP
_

¸
¸
¸
¸
¸
¸
¸
¸¸
thought NP
VP
_
GAP 1 NP
_

¸
¸
¸
¸
¸
¸
¸
¸¸
you recommended
The GAP value starts from the verb of the embedded clause and passes up to the top S in
accordance with the NIP. The value is discharged by the filler wh-phrase which carrying the
REL feature. This nonlocal REL feature, in accordance with the NIP, is also passed up to the
top S to ensure that the clause functions as a modifier.
Just like the QUE feature, the nonlocal REL feature can also come from a deeper position
within the nonhead daughter of the relative clause:
(19) a. I met the critic [whose remarks [I wanted to object to ]].
b. This is the friend [for whose mother [Kim gave a party ]].
c. The teacher set us a problem[the answer to which [we can find in the textbook]].
The simplified structure of (19b) can easily illustrate this point:
230
(20)
N

_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
N
i

S[REL i]

_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
friend
1 PP
_
REL i
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
S[GAP 1 PP]
·
·
·
·
·
·
·
·
·
·
·
·
·
·
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
P
NP
_
REL i
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
NP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
VP[GAP 1 PP]

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
for
Det
_
REL i
_
N Kim gave a party
whose mother
The REL feature is embedded in the specifier of the inner NP, but the NIP guarantees that this
value is passed up to the top S so that it can function as a modifier to the antecedent friend.
11.3 Subject Relative Clauses
Subject relative clauses are not very much different from non-subject relatives clauses in terms
of modifying a nominal expression. One main difference is that the presence of a wh-relative
pronoun including that is obligatory, and bare relative clauses are ungrammatical:
(21) a. We called the senators [who] met Fred.
b. The kid picked up the apple [that] fell down on the ground.
(22) a. *[The student [ met John]] came.
b. *[The problem [ intrigued us]] bothered me.]
Subject relative clauses involve a missing subject – a [REL i] subject is gapped, represented like
this:
231
(23)
N

_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
3 N
i

S
_
¸
¸
¸
_
REL i
MOD 3 N
i

GAP
_
¸
¸
¸
_

_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
senators
1 NP[REL i]
‘S’
_
¸
¸
¸
_
SPR
COMPS
GAP 1 NP
_
¸
¸
¸
_
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
who
V
_
¸
¸
¸
¸
¸
¸
_
SPR
COMPS 2 NP
GAP 1 NP
ARG-ST 1 NP, 2 NP
_
¸
¸
¸
¸
¸
¸
_
2 NP
met Fred
As shown in the structure, the subject of met is realized as the GAP value which passes up to
the mother node. As noted in the previous chapter, this mother node is an ‘S’ with the empty
COMPS and SPR value though it looks like a VP. It is an S with a gap in it, and this combines
with the filler who by the Head-Filler Rule. The resulting S is a complete one carrying the REL
and MOD specifications that allows the resulting clause to modify senators in accordance with
the Head-Rel Modifier Rule.
Notice that this analysis then does not license bare subject relatives given like (22). The VP
with the missing subject met John cannot carry the MOD feature at all even if it can function as
an ‘S’ that can combine either with a wh-question phrase or a wh-relative phrase. However, the
analysis also predicts that the subject of an embedded clause can be gapped in sentences like
the following:
(24) a. He made a statement [which [
S
everyone thought [
S
was really interesting and
important]]].
232
b. They all agreed to include those matters [[
S
which [everyone believed [
S
had
been excluded from the Treaty]]]].
As we saw in Chapter 10, verbs like think and believe combine with a CP, an S, or even a ‘S’
with the subject being gapped:
(25)
N
i

_
_
_
_
_
_
_
_
_
_
_
_
_
3 N
i

S
_
_
REL i
MOD 3 N
i

_
_
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
statement
1 NP
_
REL i
_
S
_
GAP 1 NP
_

_
_
_
_
_
_
_
_
_
_
_
_
_
which NP
VP
_
GAP 1 NP
_

¸
¸
¸
¸
¸
¸
¸
everyone V
‘S’
_
GAP 1 NP
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
thought
V
_
GAP 1 NP
_
AP

¸
¸
¸
¸
¸
¸
¸
¸
¸
was interesting . . .
The VP was interesting here forms an ‘S’ with the subject being gapped. This ‘S’ combines
with the verb thought, forming an incomplete VP with the GAP information. This GAP value
percolates up to the lower S and then is discharged by the relative pronoun which which induces
the MOD value to the relative clause so that it can modify the antecedent statement.
11.4 That-relative clauses
As noted earlier, that can be used either as a complementizer or as a relative pronoun:
(26) Complementizer that:
a. Mary knows that John was elected.
b. That John was elected surprised Frank.
233
c. Mary told Bill that John was elected.
(27) Relative Pronoun that:
a. This is the book [that we had read].
b. The president abandoned the people [that voted for him].
c. It is an argument [that people think will never end in Egypt].
Coordination data indicate that that can also be used a relative pronoun:
(28) a. *Every essay [she’s written] and [that/which I’ve read] is on that pile.
b. Every essay [which she’s written] and [that I’ve read] is on that pile.
c. Every essay [that she’s written] and [which I’ve read] is on that pile.
The contrast here can easily be accounted for if we assign the REL feature to that. As we have
seen earlier, the Coordination Rule requires two identical phrases to be conjoined. In (28b) and
(28c), two [REL i] Ss are conjoined, whereas in the unacceptable example (28a), two different
phrases, S with a gap element and S with no gap, are conjoined. This means that that will appear
in the following two different environments:
(29)
a. CP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
Comp S

¸
¸
¸
¸
¸
¸
that . . .
b. S[REL i]

¸
¸
¸
¸
¸
¸
¸
¸
1 NP[REL i] S[GAP 1 NP]

¸
¸
¸
¸
¸
¸
that . . .
The relative pronoun that is different from wh-relative pronoun in several respects. For ex-
ample, the relative pronoun that disallows genitive and piped piping (Sag (1997)):
(30) a. the student whose turn it was
b. *the student that’s turn it was
(31) a. the pencil with which he is writing
b. *the pencil with that he is writing
In addition, that is used only in finite relative clauses:
(32) a. a pencil with which to write
b. *a pencil with that to write
One way to account for these differences from the other wh-relative pronouns is to assume
that the relative pronoun that has no accusative case, hence cannot be the complement of a
preposition that assigns accusative to it. The relative who, unlike relative pronouns like whose,
whom and which, has the same property in this respect:
(33) a. *The people [in who we placed our trust] . . . .
b. *The person [with who we were talking] . . . .
234
(34) a. The company [in which they have invested] . . .
b. The people [in whose house we stayed] . . .
c. The person [with whom he felt most comfortable] . . .
11.5 Infinitival and Bare Relative Clauses
An infinitival clause can also function as a modifier to the preceding noun. Infinitival relative
clauses in principle may contain a relative pronoun but need not:
(35) a. He bought a bench [on which to sit ].
b. He bought a refrigerator [in which to put the beer ].
(36) a. There is a book [(for you) to give to Alice].
b. There is a bench [(for you) to sit on].
Let us consider infinitival wh-relatives first. As we have seen in the previous chapter, an infini-
tival VP can be projected into an S when its subject is realized as the unrealized subject PRO.
This will then allow the following structure for (35a):
(37)
N

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
2 N

i
S
_
_
REL i
MOD 2 N

i

_
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
bench 1 PP[REL i]

¸
¸
¸
¸
¸
¸
¸
S
_
GAP 1 PP
_
on which VP
_
GAP 1 PP
_

¸
¸
¸
¸
¸
¸
to sit
As shown here, the VP to sit has a GAP value which functions as the complement of sit. The
infinitival VP, missing its PP complement, realizes its SPR as a PRO and thus can be projected
into an S in accordance with the Head-Only Rule. This S forms a head-filler phrase with the PP
on which. The resulting S also inherits the REL value from the relative pronoun which and thus
bears the MOD feature. Once again, we observe that every projection observes the grammar
rules as well as other general principles such as the HFP, the VALP, and the NIP.
Infinitival wh-relatives have an additional constraint on the realization of the subject.
(38) a. a bench on which (*for Jerry) to sit
b. a refrigerator in which (*for you) to put the beer
235
The examples indicate that wh-infinitival relatives cannot have an overt subject (such as (for)
Jerry) realized. We saw before that the same is true for infinitival wh-questions; the data are
repeated here:
(39) a. Fred knows [which politician (*for Karen) to vote for].
b. Karen asked [where (*for Washington) to put the chairs].
This tells us that both infinitival wh-relatives and infinitival wh-questions are subject to the same
constraint. The reason for the ungammaticality of an example like (38a) can be understood if
we look at its structure:
(40)
N

.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
N

*S
_
REL i
_
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
bench 1 PP[REL i]

¸
¸
¸
¸
¸
¸
¸
CP
_
GAP 1 PP
_
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
on which C S
_
GAP 1 PP
_

¸
¸
¸
¸
¸
¸
¸
for Jerry to sit
The Head-filler Rule in (24) does not allow the combination of a CP with a PP. The S here is ill-
formed: there is no rule that allows the combination of a CP/PP with a PP to form a head-filler
phrase.
How then can we deal with infinitival bare relative clauses like (41)?
(41) a. the paper [(for us) to read by tomorrow]
b. the paper [(for us) to finish by tomorrow]
Notice here that unlike infinitival wh-relative clauses, there is no relative pronoun. Given that
the infinitival VP can be projected into an S, we can assign the following structure to (41b)
when the subject is not overt:
236
(42)
NP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
Det N

[GAP ]

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
the 1 N

i
S
_
GAP NP
i
[acc]
_
paper
VP
_
_
SPR NP[PRO]
GAP NP
i
[acc]
_
_
·
·
·
·
·
·
¸
¸
¸
¸
¸
¸
to finish
The VP to finish has a GAP value for its object, and its subject is PRO. According to the Head-
only Rule in (75), this VP then will be projected into an incomplete ‘S’. There are two analytic
issues now: how to introduce the MOD feature and how to discharge the GAP value when there
is no filler. As we noted above, English also allows finite bare relatives with the gapped element
being accusative:
(43) a. the person [I met ]
b. the box [we put the books in ]
Note that unlike the traditional view, we can have bare relatives with the nominative subject
being gapped:
(44) a. He made a statement [everyone thought [ was interesting and important]].
b. They all agreed to include those matters [everyone believed [ had been excluded
from the Treaty]].
The subject gapped bare relative is only possible when it is combined by a matrix verb like
thought and believed, but not when it directly modifies its antecedent as in (22). This in turn
means that we have the following constructional differences for bare relatives:
237
(45) a.
*N

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
N

‘S’
_
GAP NP
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
statement was interesting
b.
N

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
N

S
_
GAP NP
_
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
statement everyone thought was interesting
As represented here, the simple ‘S’ (VP with the subject being gapped) cannot function as a
bare relative modifier, but the S whose embedded subject is gapped can well serve as a bare
relative clause. To reflect this observation and allow us to discharge the GAP value in such bare
relative clauses, we can assume the following rule for English:
(46) Bare Head-Rel Modifier Rule:
N

_
GAP
_
→ 1 N

i
, S| CP
_
_
MOD 1
GAP NP
i

_
_
This rule allows a finite or infinitival clause (S or CP, but not an ‘S’ or VP) bearing an NP
GAP value to function as a modifier of the preceding noun (the MOD value is added as a
constructional constraint). One specification in the rule is that the GAP value be discharged
even if there is no filler: the index of the head noun is identified with that of the discharged gap.
This rule will then allow examples like (44) as well as authentic examples like the following:
(47) I just know that the Big 12 South teams [everyone knew [ would win actually] won
the game].
11.6 Restrictive vs. Nonrestrictive Relative Clauses
In addition to the types of relative clause we seen before, there is an interpretive distinction
between ‘restrictive’ and ‘nonrestrictive’. Consider these examples:
(48) a. The person who John asked for help thinks he is foolish.
b. Mary, who John asked for help, thinks he is foolish.
238
The relative clause in (48a) semantically restricts the denotation of person whereas the one in
(48b) just adds extra information about Mary. Let us consider one more pair of examples:
(49) a. John has two sisters who became lawyers. (‘restrictive’)
b. John has two sisters, who became lawyers. (‘non-restrictive’)
The second example suggests that John has only two sisters, while the first means that two of
his sisters are lawyers, but leaves open the possibility that he has other sisters. The denotation
of the restrictive relative phrase two sisters who became lawyers is thus the intersection the set
of two sisters and the set of lawyers. There can be more than two sisters, but there are only two
who became lawyers. Meanwhile, the nonrestrictive phrase two sisters, who became lawyers
means that there are two sisters and they all became lawyers: there is no intersection meaning
here.
This meaning difference has given rise to the idea that the restrictive relative clause modifies
the meaning of N

– a noun phrase without a determiner – whereas the nonrestrictive relative
clause modifies a fully determined NP:
(50) Restrictive Relative Clause:
NP

¸
¸
¸
¸
¸
¸
¸
¸
DP N

¸
¸
¸
¸
¸
¸
¸
¸
the N

¸
¸
¸
¸
¸
¸
S

¸
¸
¸
¸
¸
¸
¸
¸
¸
man whom we respect
(51) Non-restrictive Relative Clause:
NP

¸
¸
¸
¸
¸
¸
¸
¸
NP
S

¸
¸
¸
¸
¸
¸
¸
¸
John, NP

¸
¸
¸
¸
¸
¸
S

¸
¸
¸
¸
¸
¸
¸
whom we respect
Together with the differences in meaning, this structural difference can explain why the restric-
tive relative clause cannot modify a pronoun or proper noun:
3
3
In certain expressions of English, who relative clause can modify the pronoun he:
239
(52) a. I met the man who grows peaches.
b. I met the lady from France who grows peaches.
(53) a. *I met John who grows peaches.
b. *I met her who grows peaches.
Given that the meanings of ‘John’ and ‘her’ refer to only single individuals, we expect that no
further modification or restriction is possible. Nonrestrictive relative clauses as (54) can modify
proper nouns or pronouns, simply because they just add information about the referent into the
discourse:
(54) a. In the classroom, the teacher praised John, whom I also respect.
b. Reagan, whom the Republicans nominated in 1980, lived most of his life in Cali-
fornia.
The relative clause whom I also respect modifies the proper noun John without restricting it,
and has the same interpretation as a conjoined clause like The teacher praised John, and I also
respect him.
Such a meaning difference also causes another difference: only a restrictive clause can mod-
ify a quantified NP like every N or no N:
(55) a. Every student who attended the party had a good time.
b. *Every student, who attended the party, had a good time.
(56) a. No student who scored 80 or more in the exam was ever failed.
b. *No student, who scored 80 or more in the exam, was ever failed.
Strictly speaking, phrases with no or every as determiners do not refer to an individual or given
set of individuals, and therefore cannot have their reference further elaborated by a nonrestric-
tive modifier (see Huddleston and Pullum (2002)).
Whether the syntax involves N

or NP has also been used to explain why a restrictive clause
must precede a nonrestrictive clause:
(57) a. The contestant who won the first prize, who is the judge’s brother-in-law, sang
dreadfully.
b. *The contestant, who is the judge’s brother-in-law, who won the first prize sang
dreadfully.
((57b) is interpretable as involving a sequence of two nonrestrictive clauses.) Partial structures
of these two can be represented as follows:
(i) a. He who laughs last laughs best.
b. He who is without sin among you, let him cast the first stone.
240
(58) a.
NP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
NP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
S

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
Det N

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
who is the judge’s . . .
the N

¸
¸
¸
¸
¸
¸
¸
S

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
contestant who won the first prize,
b.
*NP
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
NP
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
S

¸
¸
¸
¸
¸
¸
¸
¸
NP

¸
¸
¸
¸
¸
¸
¸
¸
S
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
who won . . .
the contestant, who is the judge’s brother-in-law,
Only in (58a) can the first relative clause be interpreted restrictively, as it is attached at the N

level. Strictly speaking, as noted above, (58b) is not ill-formed, but can only have an interpreta-
tion in which both relative clauses are nonrestrictive.
4
11.7 Constraints on the GAP
We have observed that in wh-interrogatives and relative clauses, the filler and the gap can be
in a long-distance relationship. Yet, there are constructions where this dependency seems to be
restricted in certain ways. Consider the following examples:
(59) a. [Who] did he believe [that he would one day meet ]?
b. [Which celebrity] did he mention [that he had run into ]?
(60) a. *[Who] did he believe [the claim that he had never met ]?
b. *[Which celebrity] did he mention [the fact that he had run into ]?
4
One additional difference between restrictive and nonrestrictive clauses is that that is used mainly in restrictive
clauses.
(i) a. The knife [which/that] he threw into the sea had a gold handle.
b. The knife, [which/??that] he threw into the sea had a gold handle.
241
Why do we have the contrast here? Let us compare the partial structures of (59a) with (60a):
(61) a.
VP[GAP NP]

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
V CP[GAP NP]

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
believe C S[GAP NP]

¸
¸
¸
¸
¸
¸
¸
¸
that he would meet
b.
VP[GAP NP]

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
V *NP[GAP NP]

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
believe Det N

[GAP NP]

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
the N CP[GAP NP]

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
claim that he has never met
As we can see in (61a), the syntax allows a CP with GAP value. However, the combination
in (61b) of an NP containing a CP cannot have a GAP value. This is traditionally known as
a ‘Complex NP’ (Ross 1967) and is considered an ‘island’ in the sense that it is effectively
isolated from the rest of the structure. That is, an element within this island cannot be extracted
from it, or linked to an expression outside. It is also usually assumed that English has island
constraints as follows:
• Coordinate Structure Constraint (CSC): In a coordinate structure, no element in one conjunct
alone can be wh-questioned or relativized.
(62) a. Bill cooked supper and washed the dishes.
b. *What did Bill cook and wash the dishes?
c. *What did Bill cook supper and wash ?
• Complex Noun Phrase Constraint (CNPC): No element within a CP or S dominated by an NP
can be wh-questioned or relativized.
242
(63) a. He refuted the proof that you cannot square it.
b. *What did he refute the proof that you cannot square ?
(64) a. They met someone [who knows the professor].
b. *[Which professor] did they meet someone who knows ?
• Sentential Subject Constraint (SSC): An element within a clausal subject cannot be wh-
questioned or relativized.
(65) a. [That he has met the professor] is extremely unlikely.
b. *Who is [that he has met ] extremely unlikely?
• Left-Branching Constraint (LBC): No NP that is the leftmost constituent of a larger NP can
be wh-questioned or relativized:
(66) a. She bought [John’s] book.
b. *[Whose] did she buy book?
• Adjunct Clause Constraint: An element within an adjunct cannot be questioned or relativized.
(67) a. Which topic did you choose without getting his approval?
b. *Which topic did you get bored [because Mary talked about ]?
• Indirect Wh-question Constraint: An NP that is within an indirect question cannot be ques-
tioned or relativized.
(68) a. Did John wonder who would win the game?
b. *What did John wonder who would win ?
Various attempts have been made to account for such island constraints. Among these, we
sketch an analysis within the present system that relies on licensing constraints on subtree struc-
tures. As we have seen in the previous chapters, the present analysis provides a straightforward
account for the CSC:
(69)
*VP
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
VP[ GAP NP ] Conj VP[ GAP ]

¸
¸
¸
¸
¸
¸
¸
¸
¸
V[GAP NP] and V NP
cook wash the dishes
Even though two VPs are coordinated, they are not identical in terms of the GAP values. This
violates the Coordination Rule which requires that only identical categories can be coordinated.
243
Except for this clear constraint, the existence of the other island constraints has been ques-
tioned through acceptable examples which actually violate the island constraints. For example,
the following examples are all acceptable even though they violate a claimed island constraint
(Sag et al. to appear):
(70) a. What did he get the impression that the problem really was ? (CNPC)
b. This is the paper that we really need to find the linguist who
understands . (CNPC)
In addition, observe the following examples (cf. Ross 1987, Kluender 2004):
(71) a. *Which rebel leader did you hear [Cheney’s rumor [that the CIA
assassinated ]]?
b.??Which rebel leader did you hear [the rumor [that the CIA assassinated ]]?
c. ?Which rebel leader did you hear [a rumor [that the CIA assassinated ]]?
d. Which rebel leader did you hear [rumors [that the CIA assassinated ]]?
All these examples have identical syntactic structures but are different in the degree of accept-
ability. The data indicate that it is not the syntactic structure, but the properties of the head of the
complex NP influence the acceptability grade of the sentences, implying that processing factors
closely interact with the grammar of filler-gap constructions (Sag et al. 2006).
244
11.8 Exercises
1. Find a grammatical error in each of the following sentences and then explain the nature
of the error.
(i) a. Students enter high-level educational institutions might face many prob-
lems relating to study habits.
b. A fellow student saw this felt sorry for Miss Kim and offered her his own
book.
c. Experts all agree that dreams cause great anxiety and stress are called
nightmares.
d. The victims of the earthquake their property was destroyed in the disaster
were given temporary housing by the government.
2. Draw tree structures for the following examples and discuss which grammar rules license
each phrase.
(i) a. This is the book which I need to read.
b. This is the very book that we need to talk about.
c. The person whom they intended to speak with agreed to reimburse us.
d. The motor that Martha thinks that Joe replaced costs thirty dollars.
(ii) a. The official to whom Smith loaned the money has been indicted.
b. The man on whose lap the puppet is sitting is ventriloquist.
c. The teacher set us a problem the answer to which we can find in the text-
book.
d. We just finished the final exam the result of which we can find out next
week.
3. Draw structures for the following ungrammatical examples and identify which island
constraints are violated.
(i) a. *What did Herb start to play only after he drank ?
b. *Who did Herb believe the claim that cheated?
c. *What did Herb like fruit punch and?
d. *What was that the Vikings ate a real surprise to you?
e. *What did you meet someone who understands?
4. Compare the following pairs of examples by considering the structure of each. In partic-
ular, consider whether the structure involves a relative clause or a CP complement.
(i) a. The fact that scientists have now established all the genes in the human
body is still not widely known.
b. The fact that the scientists used the latest technology to verify was reported
at the recent conference.
(ii) a. They ignored the suggestion that Lee made.
b. They ignored the suggestion that Lee lied.
245
(iii) a. They denied the claim that we had advanced by ourselves.
b. They denied the claim that they should report only to us.
5. English also allows adverbial relative clauses like those in (i). Can the analysis in this
chapter explain such examples? If it can, how? If it cannot, can you think of any possible
explanation?
(i) a. The hotel where Gloria stays is being remodelled.
b. The day when Jim got fired was a sad day for everyone.
c. That is the reason why he resigned.
6. Read the following passages and provide the correct verb form for the underlined expres-
sions and their lexical entries.
(i) Pied-piping (first identified by John R. Ross) describes the situation where a
phrase larger than a single wh-word occurs in the fronted position. In the case
where the wh-word is a determiner such as which or whose, pied-piping refer
to the fact that the wh-determiner appears sentence-initially along with its
complement. For instance, in the example Which car does he like?, the entire
phrase which car is moved. In the transformational analysis, the wh-word
which moves to the beginning of the sentence, taking car, its complement,
with it, much as the Pied Piper of Hamelin attract rats and children to follow
him, hence the term pied-piping.
5
(ii) Certain grammatical structures associate with corresponding functions, as in
the interrogative structure Do you drink tea the function is questioning. Such
a case can describe as a direct speech act. However, when the interrogative
structure is used to fulfill a different purpose as in Can you close the window?
where it clearly is not a question about ability, but a polite request, such a
situation is described as an indirect speech act.
6
5
Adapted from Wikipedia, http://en.wikipedia.org/wiki/Wh-movement
6
Adapted from http://www.tlumaczenia-angielski.info/linguistics/pragmatics.htm
246
12
Special Constructions
12.1 Introduction
English displays constructions illustrated in (1), respectively known as ‘tough movement’,
1
‘extraposition’, and ‘cleft’ constructions:
(1) a. John is tough to persuade . (‘Tough’ movement)
b. It bothers me that John snores. (Subject Extraposition)
c. John made it clear that he would finish it on time. (Object Extraposition)
d. It is John that I met last night in the park. (Cleft)
Though these constructions each involve some kind of nonlocal dependency, they are different
from wh-question or relative clause constructions in several respects. This chapter looks into
the main properties of these new nonlocal dependency constructions.
As we have seen in the previous two chapters, in these wh-questions and relative clauses, the
gap must match its filler in terms of the syntactic category:
(2) a. I wonder [whom [Sandy loves ]]. (Wh-question)
b. This is the politician [on whom [Sandy relies ]]. (Wh-relative clause)
In addition, we can observe that the filler whom and on whom here are not in the core clause
position (subject or object) but are in the adjoined filler position.
Now compare these properties with other examples of the easy type:
(3) a. He is hard to love .
b. This car is easy to drive
The gap in (3a) would correspond to an accusative NP (him) whereas the apparent filler in
the core position is a nominative subject he. The filler and the gap here are thus not exactly
identical in terms of their syntactic information, though they are understood as referring to the
same individual. Due to the lack of syntactic identity, the dependency between the filler and the
gap is considered ‘weaker’ than the one in wh-questions or wh-relatives (cf. Pollard and Sag
(1997)).
1
The construction is named after adjectives which appear in it, such at tough, easy, difficult, etc.
247
The extraposition and cleft constructions in (1b–d) are also different from wh-questions, as
well as from easy construction examples. In clefts, we have a gap and a corresponding filler, but
in extrapositions we have a long-distance relationship between the extraposed clause and the
expletive pronoun it. In due course, we will see the differences of these constructions in detail.
12.2 ‘Easy’ Constructions
12.2.1 Basic Properties
Adjectives like easy, tough, difficult, and so on can appear in three seemlingly-related construc-
tions:
(4) a. To please John is easy/tough.
b. It is easy/tough to please John.
c. John is easy/tough to please.
Superficially, quite similar predicates such as eager and ready do not allow all these three:
(5) a. *To please John is eager/ready.
b. *It is eager/ready to please John.
c. John is eager/ready to please.
Even though (4c) and (5c) are grammatical and look similar in terms of structure, they are
significantly different if we look in detail at their properties. Consider the following contrast:
(6) a. Kim is easy to please.
b. Kim is eager to please.
One obvious difference between these two examples lies in the semantic roles of Kim: in (6a),
Kim is the object of please whereas Kim in (6b) is the subject of please. More specifically,
the verb please in (6a) is used as a transitive verb whose object is identified with the subject
Kim. Meanwhile, the verb please in (6b) is used intransitively, not requiring any object. This
difference can be shown clearly by the following examples:
(7) a. *Kim is easy [to please Tom].
b. Kim is eager [to please Tom].
The VP complement of the adjective easy cannot have a surface object whereas eager has no
such restriction. This means that the VP complement of easy has to be incomplete in the sense
that it is missing an object, and this is so with other easy adjectives as well:
(8) a. This doll is hard [to see ].
b. The child is impossible [to teach ].
c. The problem is easy [to solve ].
(9) a. *This doll is [hard to see it].
b. *The child is impossible [to teach him].
c. *The problem is easy [to solve the question].
248
In all these examples, there must be a gap in the VP complement. Meanwhile, eager places no
such a restriction on its VP complement which should be internally complete:
(10) a. John is eager [to examine the patient].
b. John is eager [to find a new home].
(11) a. *John is eager [to examine ].
b. *John is eager [to find ].
These observations lead us to the following descriptive generalization:
(12) Unlike eager-type adjectives, easy-type adjectives select an infinitival VP com-
plement which has one missing element semantically linked to their subject.
12.2.2 Transformational Analyses
Let us consider two related examples first:
(13) a. It is easy to please John.
b. John is easy to please.
Traditional movement analyses have assumed the following deep structure for (13a):
(14) [
S
is easy [
CP
[
S
PRO to please John]]]
The expletive it is introduced at S-structure in the matrix subject position to generate (13a).
One might assume direct movement of John to the subject position for (13b), but an issue
immediately arises with examples like (15):
(15) He
i
is easy to please
i
.
The problematic aspect is the status of the subject He: how can a direct movement approach
move him into the subject position and then change the form into he?
2
As a solution, Chomsky
(1986) proposes an empty operator (Op) movement operation, represented as shown here:
2
In more technical terms, this will violate the ‘Case Filter’ of Government-Binding Theory, for he receives two
cases: accusative from the original object position and nominative from the subject position.
249
(16)
S

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
NP VP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
He
i
V AP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
is A CP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
easy C IP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
Op
i
NP VP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
NP[PRO] V NP
please t
i
¸¸
The subject he is base-generated in the matrix subject position, while the null operator Op
i
moves to the intermediate position from its original object position, leaving the trace (t
i
). At an
interpretive level, this operator is coindexed with the subject, indirectly linking the gap with the
filler even though the two have different case markings.
12.2.3 A Lexicalist Analysis
As we have seen earlier, unlike eager-type adjectives, easy-type adjectives require an incom-
plete VP complement as a lexical property. This subcategorization restriction appears to be a
lexical fact for a family of adjectives and verbs. In addition to adjectives like easy, verbs like
take and cost also select an infinitival VP containing an accusative NP gap coindexed with the
subject:
(17) a. This theorem will take only five minutes to prove .
b. This theorem will take only five minutes to establish that he proved in 1930.
(18) a. This scratch will cost Kim $500 to fix .
b. This $500 bribe will cost the government $500,000 to prove that Senator Jones
accepted .
Meanwhile, as we have noted in the previous section, eager-type adjectives do not have such a
subcategorization restriction.
We can represent this lexical difference in terms of lexical information. Let us begin with the
easy-type which selects a VP complement with one NP missing:
250
(19) easy-type adjectives
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
HEAD| POS adj
VAL
_
¸
¸
¸
¸
_
SPR NP
i

COMPS
_
_
_
VFORM inf
GAP 1 NP
i
[acc]
_
_
_
_
¸
¸
¸
¸
_
TO-BIND| GAP 1 NP
i

_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
The lexical entry in (19) specifies that the infinitival complement (VP or CP) of adjectives like
easy contains a GAP value (NP
i
) which is coindexed with the subject. This coindexation will
ensure the semantic linkage between the matrix subject and the gapped NP. Notice that unlike
canonical filler-gap constructions in which the GAP value is discharged when it meets the filler
(by the Head-Filler Rule), the feature TO-BIND is introduced to lexically discharge the GAP
value in the VP complement. This lexical information will then project the following structure
for (6a):
(20)
S
.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
_
NP
i VP

_
_
_
_
_
_
_
_
_
_
_
_
Kim
V
AP
_
GAP
_
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
is
A
_
_
COMPS 2 VP[inf ]
TB-GAP 1 NP
_
_
2 VP
_
GAP 1 NP
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
easy V
VP
_
GAP 1
_
to
V
_
GAP 1 NP
i
[acc]
_
please
251
As shown in the tree, the transitive verb please introduces its object as the GAP value; hence the
mother infinitival VP is incomplete. The adjective easy selects this VP and then lexically dis-
charges through the TO-BIND|GAP (TB-GAP) value in accordance with the following revised
NIP:
(21) Nonlocal Feature Inheritance Principle (NIP, final):
A phrase’s nonlocal feature such as GAP and QUE is the union of its daughters’
nonlocal feature values minus either the lexically or grammatically bound nonlo-
cal features.
Meanwhile, the lexical information for eager-type adjectives is very simple:
(22) eager-type adjectives
_
¸
¸
¸
¸
¸
_
HEAD| POS adj
VAL
_
¸
_
SPR NP
COMPS
_
VP
_
VFORM inf
_
_
_
¸
_
_
¸
¸
¸
¸
¸
_
These adjectives select a complete infinitival VP with no missing element, eventually generating
a simple canonical head-complement structure like the following:
(23)
S

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
NP
VP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
Kim
V AP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
is
A
_
COMPS 2 VP
_
2 VP
_
GAP
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
eager V
VP
to V
please
The adjective eager thus places no restriction on its VP complement, and so can legitimately
combine with the fully saturated VP complement. When its VP complement has a GAP value,
252
it must be later discharged by a filler as seen in the following contrast:
(24) a. *Kim is eager to recommend .
b. Who is Kim eager to recommend ?
Notice that the present analysis can straightforwardly account for examples in which the VP
complement includes more than one GAP element. Compare the following pair of examples:
(25) a. This sonata is easy to play on this piano.
b. Which piano is this sonata easy to play on ?
The structure of (25a) is similar to that of (20):
(26)
S
.
.
.
.
.
.
.
.
.
.
.
.
¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸
1 NP
i

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
VP
_
SPR 1 NP
_

¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸
This sonata V
AP
_
_
SPR 1 NP
GAP
_
_
...........
_
_
_
_
_
_
_
_
_
_
_
_
is
A
_
¸
¸
¸
_
SPR 1 NP
COMPS 3 VP
TB-GAP 2 NP
_
¸
¸
¸
_
VP
_
_
SPR 1 NP
GAP 2 NP
_
_

_
_
_
_
_
_
_
_
_
_
_
_
easy V
3 VP
_
GAP 2 NP
i

_
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
to play on this piano
Just like the structure in (20), here the adjective easy combines with an incomplete VP whose
missing GAP value is lexically discharged and coindexed with the matrix subject. Now consider
the structure of (25b) in which the object is linked to the subject this sonata whereas the object
of on is linked to the wh-phrase which piano:
253
(27)
S
.
.
.
.
.
.
.
.
.
.
.
.
.
¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸
2 NP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
S
_
GAP 2
_

¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸ ¸
Which piano V
1 NP
i

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
AP
_
_
SPR 1
GAP 2
_
_
............
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
is this sonata
A
_
¸
¸
¸
_
SPR 1
COMPS 3
TB-GAP 4
_
¸
¸
¸
_
3 VP
_
GAP 4 NP
i
, 2 NP
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
easy to play on
In the structure above, the VP complement of easy has two GAP values: one is the object ( 4 NP)
and the other is the object ( 2 ) of on. The first GAP value coindexed with the subject is lexically
bound by easy through the feature TB-GAP. The remaining GAP value ( 2 NP) is passed up to
the second higher S where it is discharged by its filler, which piano, through the Head-Filler
Rule.
12.3 Extraposition
12.3.1 Basic Properties
English employs an extraposition process that places a heavy constituent such as a that-clause,
wh-clause, or infinitival clause at the end of the sentence:
(28) a. [That dogs bark] annoys people.
b. It annoys people [that dogs bark].
(29) a. [Why she told him] is unclear.
b. It is unclear [why she told him].
(30) a. [(For you) to leave so soon] would be inconvenience.
b. It would be inconvenience [(for you) to leave so soon].
This kind of alternation is quite systematic: given sentences like (31a), English speakers have
an intuition that (31b) is possible:
(31) a. That the Dalai Lama claims Tibet independence discomfits the Chinese govern-
ment.
254
b. It discomfits the Chinese government that the Dalai Lama claims Tibet indepen-
dence.
The extraposition process can also be applicable to a clausal complement:
(32) a. I believe the problem to be obvious.
b. *I believe [that the problem is not easy] to be obvious.
c. I believe it to be obvious [that the problem is not easy].
As seen in (32b–c), when a clausal complement is followed by infinitival VP complement, the
former is much more preferably extraposed to sentence-final position. In addition to a finite CP,
as in (32c), extraposition applies also to an infinitival CP/VP, a simple S, or even a gerundive
phrase:
(33) a. I do not think it unreasonable [to ask for the return of my subscription].
b. He made it clear [he would continue to co-operate with the United Nations].
c. They’re not finding it a stress [being in the same office].
12.3.2 Transformational Analysis
In terms of movement operations, there have been two main ideas to capture the systematic
relationships between examples such as the following:
(34) a. [That you came early] surprised me.
b. It surprised me [that you came early].
One approach assumes that the surface structure of a subject extraposition like (34b) is gener-
ated from (34a) as represented in the following (Rosenbaum (1967)):
3
(35)
S
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
NP

¸
¸
¸
¸
¸
¸
¸
VP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
It [t]
¸
VP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
S

¸
¸
¸
¸
¸
¸
¸
¸
¸
V NP
you came early
surprised me
3
The notation [t] means a trace left after a movement.
255
The extraposition rule moves the underlying sentence you came early to a sentence-final posi-
tion. This movement process also introduces the insertion of that, generating (34b). To generate
nonextraposed sentences like (34a) the system posits a process of deleting it in (34a) and then
adding the complementizer that.
A slightly different analysis has also been suggested with the opposite direction of movement
(Emonds 1970, Chomsky (1981), Groat (1995)). That is, instead of extraposing the clause from
the subject, the clause is assumed to be already in the extraposed position as in (36a):
(36) a. [[ ] [
VP
surprised [me] [
CP
that you came early]]].
b. [[It] [
VP
surprised me that you came early]].
The insertion of the expletive it in the subject position in (36a) would then account for (36b).
When the CP clause is moved to the subject position, the result is the nonextraposed sentence
(34a).
Most current movement approaches follow this second line of thought. Though such deriva-
tional analyses can capture certain aspects of English subject extraposition, they are not speci-
fied in enough detail to predict lexical idiosyncrasies as well as non-local properties of extrapo-
sition (see Kim and Sag (2005) for further discussion).
12.3.3 A Lexicalist Analysis
As we have seen, English exhibits a systematic alternation between pairs of non-extraposed and
extraposed sentences like the following:
(37) a. [That Chris knew the answer] occurred to Pat.
b. It occurred to Pat [that Chris knew the answer].
This alternation relation is quite productive. For example, as English acquires new expressions,
e.g. freak out, weird out, or bite, it acquires both extraposed and non-extraposed sentence types
(cf. Jackendoff 2002):
(38) a. It really freaks/weirds me out that we invaded Iraq.
b. That we invaded Iraq really freaks/weirds me out.
(39) a. It really bites that we invaded Iraq.
b. That we invaded Iraq really bites.
The simple generalization about the process of extraposition is that it applies to a verbal element
(CP, VP, and S). Adopting Sag et al. (2003), Kim and Sag (2006), we then can assume that the
extraposition process also refers to the verbal category whose subtypes include both comp and
verb (see Chapter 5.4.2). In particular, we can adopt the following lexical rule to capture the
systematic relationship in extraposition:
(40) Extraposition Lexical Rule (ELR):
_
ARG-ST . . . , 1 XP[verbal], . . .
_

_
_
ARG-ST . . . , NP[NFORM it], . . .
EXTRA 1 XP
_
_
256
What this rule says is that if a predicative element (actually, adjective or verb) selects a verbal
argument (either CP or S), this verbal element can be realized as the value of the feature EXTRA
together with the introduction of it as an additional argument.
For example, consider the following data set:
(41) a. Fido’s barking annoys me.
b. That Fido barks annoys me.
c. It annoys me that Fido barks.
As shown here, the verb annoys can take either a CP or an NP as its subject. When the verb
annoys selects a verbal argument (CP), it can undergo the Extraposition Lexical Rule in (40) as
follows:
(42)
_
_
annoys
ARG-ST 1 [nominal] , 2 NP
_
_

_
_
ARG-ST NP[NFORM it], 2 NP
EXTRA 1 CP
_
_
Since the verb annoys selects a nominal (CP or NP since its subtypes are noun and comp) as
one of its arguments, it can undergo the ELR when this argument is realized as a CP (comp is a
subtype of verbal). As shown here, the output annoy now selects the expletive it as its subject
while its original CP now serves as the value of the EXTRA. The two arguments in the output
ARG-ST, in accordance with the ARC, will be realized as the SPR and COMPS value, with the
EXTRA value intact. This realization will allow us to generate a structure like the following:
257
(43)
S

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
3 NP
VP
_
_
SPR 3
EXTRA
_
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
It
VP
_
_
SPR 3
EXTRA 1
_
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
1 CP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
V
_
¸
¸
¸
¸
¸
¸
_
SPR 3
COMPS 2
ARG-ST 3 NP[it], 2
EXTRA 1
_
¸
¸
¸
¸
¸
¸
_
2 NP
that Fido barks.
annoys me
As shown in the tree, the two arguments of the verb annoys are realized as SPR and COMPS
respectively. When it combines with the NP me, it forms a VP with a nonempty EXTRA value.
This VP then combines with the extraposed clause CP in accordance with the following Head-
Extra Rule:
(44) Head-Extra Rule:
_
EXTRA
_
→H
_
EXTRA 1
_
, 1
As given here, the rule also discharges the feature EXTRA passed up to the head position.
This grammar rule reflects the fact that English independently allows a phrase in which a head
element combines with an extraposed element:
(45)
_
EXTRA
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
H
_
EXTRA 1
_
1
258
We can observe that English freely employs this kind of well-formed phrase condition even in
the extraposition of an adjunct element.
(46) a. [[A man came into the room] [that no one knew]].
b. [[A man came into the room] [with blond hair]].
c. I [read a book during the vacation [which was written by Chomsky]].
All of these examples are licensed by the Head-Extra Rule which allows the combination of a
head element with an extraposed element.
Object extraposition is no different – consider the following examples:
(47) a. Ray found the outcome frustrating.
b. Ray found it frustrating [that his policies made little impact on poverty].
The data indicate that the lexical entry for find selects three arguments including a CP and thus
can undergo the ELR Rule:
(48)
_
_
find
ARG-ST 1 NP, 2 [nominal], 3 AP
_
_

_
_
ARG-ST 1 NP, NP[it], 3 AP
EXTRA 2 [comp]
_
_
Since the type comp is a subtype of nominal and verbal at the same time, the verb can undergo
the ELR. The output introduces a new element it together with the EXTRA value. The three
arguments in the output will then be realized as the SPR and COMPS values, projecting a
structure like the following:
(49)
VP
_
EXTRA
_
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
VP
_
EXTRA 2
_
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
2 CP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
V
_
_
ARG-ST NP, 4 NP[it], 3 AP
EXTRA 2
_
_
4 NP 3 AP
that his policies . . .
found it frustrating
The verb find requires an expletive object and an AP as its complement. It also has a clausal
element as its EXTRA element. The first VP thus has a nonempty EXTRA value projected from
259
the verb, and this VP forms a well-formed phrase with the extraposed CP clause.
One main difference between subject and object extraposition is that the latter is obligatory:
(50) a. *I made [to settle the matter] my objective.
b. I made it [my objective] to settle the matter.
c. I made [the settlement of the matter] my objective.
(51) a. *I owe [that the jury acquitted me] to you.
b. I owe it [to you] that the jury acquitted me.
c. I owe [my acquittal] to you.
This contrast is due to a general constraint which disprefers any element within VP from occur-
ring after a CP:
(52) a. I believe strongly [that the world is round].
b. *I believe [that the world is round] strongly.
In the present context this means that there is no predicative expression (verbs and adjectives)
whose COMPS list contains an element on the list after a CP complement (see Kim and Sag
(2005)).
12.4 Cleft constructions
12.4.1 Basic Properties
The examples in (53) represent the canonical types of three kinds of cleft construction: it-cleft,
wh-cleft, and inverted wh-cleft in English:
(53) a. It’s their teaching material that we’re using. (it-cleft)
b. What we’re using is their teaching material. (wh-cleft)
c. Their teaching material is what we are using. (inverted wh-cleft)
These three types of clefts all denote the same proposition as the following simple declarative
sentence:
(54) We are using their teaching material.
The immediate question that follows is then what is the extra function of the cleft struc-
ture instead of the simple sentence (54)? It is commonly accepted that clefts share identical
information-structure properties given in (55), for the example in question:
(55) a. Presupposition (Background): We are using X.
b. Highlighted (Foreground or focus): their teaching material
c. Assertion: X is their teaching material.
In terms of the structures, the three types of cleft all consist of a matrix clause headed by a copula
and a relative-like cleft clause whose relativized argument is coindexed with the predicative
argument of the copula. The only difference is where the highlighted (focused) expression is
placed.
260
12.4.2 Distributional Properties of the Three clefts
It-clefts: As noted before, the it-cleft construction consists of the pronoun it as the subject
of the matrix verb be, the highlighted (or focused) phrase XP, and a remaining cleft clause.
The pronoun it here functions as a place holder, though it is similar in form to the referential
pronoun it. For example, it is hard to claim that the pronoun it in the following dialogue has any
referential property:
(56) A: I share your view but I just wonder why you think that’s good.
B: Well I suppose it’s the writer that gets you so involved.
As for the type of highlighted XP, we observe that only certain types of phrase can be used:
(57) a. It was [
NP
the man] that bought the articles from him.
b. It was [
AdvP
then] that he felt a sharp pain.
c. It was [
PP
to the student] that the teacher gave the best advice.
d. It was [
S
not until I was perhaps twenty-five or thirty] that I read and enjoyed
them.
Phrases such as an infinitival VP, AP, or CP cannot function as the XP:
(58) a. *It was [
VP
to finish the homework] that John tried.
b. *It is [
AP
fond of Bill] that John seems to be.
c. *It is [
CP
that Bill is honest] that John believes.
Also notice that in addition to that, wh-words like who and which can also introduce a cleft
clause:
(59) a. It’s the second Monday [that] we get back from Easter holiday.
b. It was the girl [who] kicked the ball.
c. It’s mainly his attitude [which] convinced the teacher.
Wh-clefts: Unlike the it-cleft, the wh-cleft construction places a cleft clause in the subject
position followed by the highlighted XP in the postcopular position. This gives a wide range of
highlighted phrases. As shown in (60), almost all the phrasal types can serve as the highlighted
XP:
(60) a. What you want is [
NP
a little greenhouse].
b. What’s actually happening in London at the moment is [
AP
immensely exciting].
c. What is to come is [
PP
in this document].
d. What I’ve always tended to do is [
VP
to do my own stretches at home].
e. What I meant was [
CP
that you have done it really well].
Different from it-cleft, the wh-cleft allows AP, base VP, and clause (CP, simple S, and wh-
clause) to serve as the highlighted XP:
(61) a. What you do is [
VP
wear it like that].
261
b. What happened is [
S
they caught her without a license].
c. What the gentleman seemed to be asking is [
S
how policy would have differed].
Inverted wh-clefts: Though the inverted wh-cleft construction is similar to the wh-cleft, the
possible types of highlighted phrase are in fact different:
(62) a. [
NP
That] is what they’re trying to do.
b. [
AP
Insensitive] is how I would describe him.
c. [
PP
In the early morning] is when I do my best research.
(63) a. *[
VP
Wear it like that] is what you do.
b. *[
S
They caught her without a license] is what happened.
c. *[
CP
That you have done it really well] is what I meant.
In general, all wh-words except which are possible in inverted wh-clefts:
(64) a. That’s [when] I read.
b. That was [why] she looked so nice.
c. That’s [how] they do it.
d. That’s [who] I played with over Christmas.
e. *That was [which] I decided to buy.
12.4.3 Syntactic Structures of the Three Types of Cleft: Movement Analyses
There have been two main directions in movement analyses to deal with English it-cleft con-
structions: an extraposition analysis and an expletive analysis. The extraposition analysis as-
sumes a direct syntactic or semantic relation between the cleft pronoun it and the cleft clause
through extraposition (Akmajian 1970, Gundel 1977, Hedberg 2000).
(65) a. [What you heard] was an explosion. (wh-cleft)
b. It was an explosion, [what you heard]. (right-dislocated)
c. It was an explosion [that you heard]. (it-cleft)
For example, in Gundel (1977), the wh-cleft clause in (65a) is first right dislocated as in (65b)
which then can generate the it-cleft (65c) with the replacement of what into that. Analyses of
this view basically take the cleft clause to be extraposed to the end of the sentence.
Meanwhile, the expletive analysis (Chomsky 1977,
´
E. Kiss 1998, Lambrecht 2001) takes the
pronoun it to be an expletive expression generated in place, while the cleft clause is semantically
linked to the clefted constituent by a ‘predication’ relation.
(66) It was [
PRED
John + who heard an explosion].
An elaborated analysis within this view has been proposed by
´
E. Kiss (1998):
262
(67)
IP
.
.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
_
_
NP I

.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
It I FP
.
.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
_
_
was
j
NP F

.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
John
i
F CP

¸
¸
¸
¸
¸
¸
¸
¸
t
j
¸¸
who
i
IP
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
t
i
heard an explosion
¸
As shown here, the clefted phrase John, functioning as focus, is assumed to occupy the specifier
of the FP (focus phrase) while the copula is the head of the FP and the cleft clause is the
complement of F. The highlighted focus phrase John and the cleft clause are thus in a predication
relation.
Even though the wh-cleft and it-cleft are identical in presenting ‘salient’ discourse informa-
tion for emphasis, they have different syntactic properties which make it hard to derive one
from the other (cf. Pavey 2004). It is because there are many cases where we can observe clear
differences among the three types of clefts. For example, one noticeable difference lies in the
fact that only wh-clefts allow a base VP as the highlighted XP phrase:
(68) a. What you should do is [
VP
order one first].
b. *It is [
VP
order one first] that you should do first.
c. *[
VP
Order one first] is what you should do.
The three are different as well with respect to the occurrence of an adverbial subordinate clause:
(69) a. It was not until I was perhaps twenty-five or thirty that I read them and enjoyed
them.
b. *When I read them and enjoyed them was not until I was perhaps twenty-five.
c. *Not until I was perhaps twenty-five was when I read them and enjoyed them.
As seen here, the not until adverbial clause appears only in it-clefts.
It is not difficult to find imperfect relationships among the three types of cleft. For example,
neither wh-clefts nor inverted wh-clefts allow the cleft clause part to be headed by that:
(70) a. It’s the writer [that gets you so involved].
b. *[That gets you so involved] is the writer.
263
c. *The writer is [that gets you so involved].
In addition, the head of the cleft clause in the it-cleft can be a PP, but not in the wh-cleft or
inverted wh-cleft:
(71) a. And it was this matter [[on which] I consulted with the chairman of the Select
Committee].
b. *[[On which] I consulted with the chairman of the Select Committee] was this
matter.
c. *This matter was [[on which] I consulted with the chairman of the Select Commit-
tee].
These facts suggest that the different types of cleft cannot be put in direct derivational rela-
tionships with each other. Though we cannot provide detailed analyses for them, we sketch out
possible directions here.
12.4.4 Lexically-Based Analyses
Wh-clefts: Let us first consider wh-clefts:
(72) a. [What I ate] is an apple.
b. [What we are using] is their teaching material.
Before getting to the specific syntactic structures there are two things to note here: the role of
the copula be and the cleft clause. The copula in the cleft construction has a ‘specificational’
use, not a ‘predicational’ one. In examples like (73a), the copula is predicational, whereas in
examples like (73b), the copula is specificational.
(73) a. The student who got A in the class was very happy.
b. The one who broke the window was Mr. Kim.
One main difference is that in the former the postcopular element denotes a property of the
subject whereas in the latter the postcopular element specifies the same individual as the subject.
In the wh-cleft too, the postcopular expression specifies the same individual as the subject.
As for the properties of the cleft part itself, we can observe that it behaves just like a free
relative clause. Not all wh-words can occur in free relatives:
(74) a. He got what he wanted.
b. He put the money where Lee told him to put it.
c. The concert started when the bell rang.
(75) a. *Lee wants to meet who Kim hired.
b. *Lee bought which car Kim wanted to sell to him.
c. *Lee solved the puzzle how Kim solved it.
In the examples in (74), what, where and when can head a free relative clause in the sense that
they are interpreted as ‘the thing that, the place where, and the time when’. However, this kind
264
of interpretation is not possible with who, which or how. As in free relatives, neither who nor
which can appear in wh-clefts, for example:
(76) a. *Who achieved the best result was Angela.
b. *Which book he read the book was that one.
Also note that the syntactic distribution of a free relative clause is as an NP, not as a clause of
some kind. For example, the object of eat is a diagnostic environment:
(77) a. I ate [what John ate].
b. I ate [an apple].
Since the verb ate requires only an NP as its complement, the only possible structure is as
follows:
(78)
S

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
NP
VP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
I
V
NP
_
GAP
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
ate 1 NP
S
_
GAP 1
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
what
NP
VP
_
GAP 1
_
John
V
_
GAP 1 NP
_
ate
Even though the filler what and the head phrase John ate form a constituent, the result cannot be
an S since ate can combine with only an NP. This kind of free relative structure, rather unusual in
the sense that the non-head filler what is the syntactic head, is formed by the following grammar
265
rule (Pullum 1991):
4
(79) Free-Relative Phrase Rule:
NP[GAP ] → 1 NP[FREL i], S[GAP 1 NP]
This rule ensures that when a free relative pronoun combines with a sentence missing one
phrase, the resulting expression is not an S but a complete NP.
On the assumption that the cleft clause in the wh-cleft is a free relative, we then can assign
the following structure to (72b):
(80)
S

_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
NP
_
FREL i
_

¸
¸
¸
¸
¸
¸
¸
¸
VP

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
NP[FREL i] S/NP

¸
¸
¸
¸
¸
¸
¸
¸
V NP[FOC +]
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
what we are using is their teaching material
As shown here, the cleft clause is formed by the combination of what with an S missing an
NP. The index of the free relative is identified with that of the postcopular NP their teaching
material.
Taking wh-clefts as a type of free-relative clause construction headed by an NP, we can
understand the ungrammaticality of examples like the following:
(81) a. *[To whom I gave the cake] is John.
b. *[That brought the letter] is Bill.
The subjects here are not headed by NPs, and therefore cannot be free relatives.
Inverted Wh-clefts The inverted wh-cleft is motivated by a different information structure per-
spective. In particular, the inverted cleft highlights the phrase in subject position:
4
The feature FREL is assigned to wh-words like what, where, and when, but not to how and why, to categorize which
can head free relatives and which cannot. See Kim (2001b).
266
(82)
S

_
_
_
_
_
_
_
_
_
_
_
_
_
NP
i

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
VP

_
_
_
_
_
_
_
_
_
_
_
_
_
Their teaching material V
NP
i
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
is
NP
S/NP

¸
¸
¸
¸
¸
¸
¸
¸
what we are using
In these structures, the cleft clause has no FREL value to allow almost all wh-words to head the
relative clause:
(83) a. This is how he did it.
b. This is why he came early.
In other words, while the subject position of a wh-cleft is restricted to NPs, the postcopular
complement position in the inverted cleft is not restricted – simply due to properties of the
copula be.
It-clefts: There are two different types of it-cleft. Compare the following:
5
(84) a. Type A: It is [
PP
on Bill] [
CP
that [John relies [
PP
]]].
b. Type B: It is [
NP
Bill] [
S
[
PP
on whom] [John relies [
PP
]]].
In (84a), the cleft clause contains a gap matching the filler PP on Bill. If we treat that as a
relativizer, the PP gap cannot be discharged by that because of the category mismatch, for that
is an NP. However, in (84b) the cleft clause has two parts: one with a missing gap John relies
and the other with the wh-phrase on whom functioning as the filler. This second example is
similar to ones where the highlighted element is an adverbial:
(85) a. It was [then] [when we all went to bed].
b. It was [only gradually] [that I came to realize how stupid I was].
As noted in the literature, even though a cleft clause is similar in structure to a restrictive relative
clause, there are some differences. For example, consider the following:
(86) a. It is John that we are looking for.
b. *John that we are looking for showed up.
We can notice here that, unlike a cleft, a canonical restrictive relative clause does not allow a
pronoun to function as the antecedent of the relative clause.
5
See Gazdar et al. (1985) recognizing two different it-cleft constructions.
267
Most wh-phrases can freely occur in the it-cleft, such as who, whose, or no wh-word if that
is present:
(87) a. It’s the second Monday [that] we get back from Easter.
b. It was the peasant girl [who] got it.
c. It was in 1997 [when] the INS introduced the alien registration receipt card.
d. It is Uncle John [whose address] I lost.
To capture these two different types and restrictions on the type of wh-phrases, we can assume
that the copula be in the it-cleft has its own lexical information:
(88) Copula be for Type A it-Cleft:
_
¸
¸
¸
¸
¸
¸
¸
_
be
SPR NP[it]
COMPS
_
2 YP
i
[FOCUS +], CP
_
GAP 2
i

_
_
TO-BIND| GAP 2
i

_
¸
¸
¸
¸
¸
¸
¸
_
(89) Copula be for Type B it-Cleft:
_
¸
¸
¸
¸
¸
¸
_
be
SPR NP[it]
COMPS
_
2 YP[FOCUS +], S
_
_
MOD 2
GAP
_
_
_
_
¸
¸
¸
¸
¸
¸
_
In both constructions, the contrastive focus (marked with the feature FOCUS functions as the
most salient contextual information. In Type A, the second complement, functioning as the
background, is a CP with a GAP value. The lexical head be also binds off the GAP inside the
CP complement. Notice that the index of the GAP value is identical to that of the focus YP,
providing a strict semantic linkage between these two.
In contrast, in Type B, the second COMPS element is a saturated S which modifies the
focused element. Let us consider the structure that (88) licenses:
268
(90)
S

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
NP
VP
_
GAP
_
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
It
V
_
_
COMPS 2 , 3
TB-GAP 2
_
_
2 PP
i
_
FOCUS +
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
3 CP
_
GAP 2 PP
i

_

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
is on Bill C
S
_
GAP 2 PP
i

_

¸
¸
¸
¸
¸
¸
that John relies
This is a Type A cleft sentence: the copula be selects two complements: PP and CP. The cleft
clause CP has a PP gap whose GAP value is linked to the focus PP. This GAP value is lexically
discharged by the TO-BIND feature. Meanwhile, the lexical information in (89) will license
structures like the following:
(91)
S

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
NP
VP
_
GAP
_
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
It
V
_
COMPS 2 , 3
_
2 NP
_
FOCUS +
_
3 S
_
_
GAP
MOD 2
_
_

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
is Tom
5 PP

¸
¸
¸
¸
¸
¸
¸
¸
S
_
GAP 5
_

¸
¸
¸
¸
¸
¸
on whom Bill relies
269
(92)
S

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
NP
VP
_
GAP
_
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
It
V
_
COMPS 2 , 3
_
2 AdvP
_
FOCUS +
_
3 S
_
MOD 2
_
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
was then when we all went to bed
These two structures involve the Type B copula in the sense that the cleft clause contains no
GAP element and modifies the highlighted phrase.
6
Our analysis explains why the following examples are unacceptable:
(93) a. *It is [Kim] [[on whom] [that Sandy relies ]].
b. *It is [Kim] [[on whom] [Sandy relies on ]].
c. *It is [Kim] [[whom] [Sandy relies ]].
The example (93a) is ruled out since the combination of on whom and that Sandy relies is not
a well-formed S, even though it could be a CP; (93b) is not allowed because of the mismatch
between the gap (NP) and the filler (PP); and (93c) is ruled out similarly. The fragment Sandy
relies requires a PP but the filler is an NP (whom).
Within the present system where the missing element in the cleft clause is taken to be a GAP
element, we also expect to see an unbounded dependency relation:
(94) a. It was the director that she wants to meet .
b. It was the director that she said she wants to meet .
c. It was the director that I think she said she wants to meet .
In addition, our analysis licenses examples like the following:
(95) a. I wonder who it was who saw you.
b. I wonder who it was you saw .
c. I wonder in which pocket it was that Kim had hidden the jewels.
Let us look at the structure of (95a), as our system generates it:
6
The MOD feature here originates from the subordinator conjunction when.
270
(96)
VP
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
V
S
_
QUE +
_
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
wonder 2 NP i[QUE +]

¸
¸
¸
¸
¸
¸
¸
S
_
GAP 2 NP i
_
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
who 1 NP VP
_
GAP 2 NP
_
.
.
.
.
.
.
.
.
.
.
.
.
.
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
it V
_
¸
¸
¸
_
SPR 1
COMPS 3
GAP 2 NP
_
¸
¸
¸
_
3 S

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
was who saw you
As shown here, the first COMPS value of the cleft copula be is realized as a GAP element. This
GAP value is passed up to the point where it is discharged by the wh-element who. This induces
an interrogative meaning to the complement clause of the verb wonder.
Even though the present system allows the focus phrase (complement of the copula) to be
indirectly gapped, a GAP value originating in the cleft clause cannot pass up further:
(97) a. Who do you think it is that Mary met ?
b. *To whom do you think it is the book that Mary gave ?
The ‘real’ gap in (97a) is the one immediately after be, which is indirectly connected through
the lexical information of be to the second one. Notice that in (88) the first GAP value (linked
to the book originating in the cleft clause is lexically terminated at the level sister to CP, by the
copula be. Notice that in the lexical entry (89), the cleft clause S does not contain any GAP.
Neither realizations of be can sanction an example like (97b).
271
12.5 Exercises
1. Explain the relationship among the following sentences.
(i) a. It is difficult for me to concentrate on calculus.
b. For me to concentrate on calculus is difficult.
c. Calculus is difficult for me to concentrate on.
2. Draw structures for the following sentences and show which grammar rules are involved
in generating them.
(i) a. This problem will be difficult for the students to solve.
b. Being lovely to look at has its advantages.
c. This toy isn’t easy to try to hand to the baby.
d. That kind of person is hard to find anyone to look after.
e. Letters to Grandma are easy to help the children to write.
(ii) a. It was to Boston that they decided to take the patient.
b. It was with a great deal of regret that I vetoed your proposal.
c. It was Tom who spilled beer on this couch.
d. It is Martha whose work critics will praise.
e. It was John on whom the sheriff placed the blame.
f. I wondered who it was you saw.
g. I was wondering in which pocket it was that Kim had hidden the jewels.
3. Explain why the following examples are ungrammatical, referring to the analysis pre-
sented in this chapter.
(i) a. *It is Kim on whom that Sandy relies.
b. *It is Kim on whom Sandy relies on.
c. *It is Kim whom Sandy relies.
d. *It is on Kim on whom Sandy relies.
Further, consider the following examples in (ii) and (iii), draw structures for them and
show which grammar rules and principles are involved in their generation.
(ii) a. I wonder who it was who saw you.
b. I wonder who it was you saw.
c. I wonder in which pocket it was that Kim had hidden the jewels.
(iii) a. Was it for this that we suffered and toiled?
b. Who was it who interviewed you?
4. Analyze the following raising examples and show clearly how the cleft and raising con-
structions interact.
(i) a. I believe it to be her father who was primarily responsible.
b. I believe it to be the switch that is defective.
5. Consider the following set of examples, all of which contain the expression what Mary
offered to him. Explain whether the phrase functions as an indirect question or an NP and
272
support your explanations by drawing the syntactic structures:
(i) a. Tom ate [what Mary offered to him].
b. I wonder [what Mary offered to him].
c. [What Mary offered to him] is unclear.
6. Read the following passage and then provide the correct form of the underlined expres-
sions together with their lexical entries:
(i) The misfortunes of human beings may divide into two classes: First, those
inflicted by the non-human environment and, second, those inflicted by
other people. As mankind have progressed in knowledge and technique,
the second class has become a continually increasing percentage of the
total. In old times, famine, for example, was due to natural causes, and
although people did their best to combat it, large numbers of them died of
starvation. At the present moment large parts of the world be faced with
the threat of famine, but although natural causes have contribute to the
situation, the principal causes are human. For six years the civilized nations
of the world devoted all their best energies to killing each other, and they
find it difficult suddenly to switch over to keeping each other alive. Having
destroyed harvests, dismantled agricultural machinery, and disorganized
shipping, they find it no easy matter relieve the shortage of crops in one
place by means of a superabundance in another, as would easily be done
if the economic system were in normal working order. As this illustration
shows, it be now man that is man’s worst enemy. Nature, it is true, still see
to it that we are mortal, but with the progress in medicine it will become
more and more common for people to live until they have had their fill of
life.
7
7
From ‘Ideas that Have harmed Mankind’ by Bertrand Russell.
273
References
Aarts, Bas. 1997. English Syntax and Argumentation. Basingstoke and London: Macmillan.
Abney, Steven. 1987. The English Noun Phrase in its Sentential Aspect. Ph.D. dissertation, MIT.
Akmajian, Adrian. 1970. On deriving cleft sentences from pseudo-cleft sentences. Language 1:
149–168.
Akmajian, Adrian, and Frank Heny. 1975. Introduction to the Principles of Transformational
Syntax. Cambridge, MA: MIT Press.
Akmajian, Adrian, Susan Steele, and Thomas Wasow. 1979. The category AUX in Universal
Grammar. Linguistic Inquiry 10: 1-64.
Akmajian, Adrian, and Thomas Wasow. 1974. The constituent structure of VP and AUX and
the position of verb BE. Linguistic Analysis 1: 205-245.
Bach, Emmon. 1974. Syntactic Theory. New York: Holt, Rinehart and Winston.
Bach, Emmon. 1979. Control in Montague Grammar. Linguistic Inquiry 10: 515-31.
Baker, C. L. 1970. Double Negatives. Linguistic Inquiry 1: 169-186.
Baker, C.L. 1978. Introduction to Generative-transformational Syntax. Englewood Cliffs, N.J.:
Prentice-Hall.
Baker, C.L. 1979. Syntactic Theory and the Projection Problem. Linguistic Inquiry 10: 533-581.
Baker, C.L. 1991. The Syntax of English not: The Limits of Core Grammar. Linguistic Inquiry
22: 387-429.
Baker, C.L. 1997. English Syntax. Mass.: MIT Press.
Baker, Mark. 2001. The Atoms of Language: The Mind’s Hidden Rules of Grammar. New York:
Basic Books.
Bender, Emily, and Dan Flickinger. 1999. Peripheral constructions and core phenomena: Agree-
ment in tag questions. In G. Webelhuth, J.-P. Koening, and A. Kathol (Eds.), Lexical and
Constructional Aspects of Linguistic Explanation, 199-214. Stanford, CA: CSLI.
Blake, Barry J. 1990. Relational Grammar. London: Routledge.
Bloomfield, Leonard. 1933. Language. New York: H. Holt and Company.
275
Borsley, R.D., 1989a. Phrase structure grammar and the Barriers conception of clause structure.
Linguistics 27: 843 863.
Borsley, R.D., 1989b. An HPSG approach to Welsh. Journal of Linguistics 25: 333 354.
Borsley, Bob. 1991. Syntactic Theory: A Unified Approach. Cambridge: Arnold.
Borsley, Bob. 1996. Modern Phrase Structure Grammar. Cambridge: Blackwell.
Bouma, Gosse, Rob Malouf, and Ivan A. Sag. 2001 Satisfying constraints on extraction and
adjunction. Natural Language and Linguistic Theory 19: 1-65.
Brame, Michael K. 1979. Essays Toward Realistic Syntax. Seattle: Noit Amrofer.
Bresnan, Joan. 1978. A Realistic Transformational Grammar. In M. Halle, J. Bresnan, and G. A.
Miller (Eds.), Linguistic Theory and Psychological Reality. Cambridge, MA: MIT Press.
Bresnan, Joan. 1982a. Control and Complementation. In The Mental Representation of Gram-
matical Relations (Bresnan 1982c).
Bresnan, Joan. 1982b. The passive in lexical theory. In The Mental Representation of Grammat-
ical Relations (Bresnan 1982c).
Bresnan, Joan. (Ed). 1982c. The Mental Representation of Grammatical Relations. Cambridge,
MA: MIT Press.
Bresnan, Joan. 1994. Locative inversion and the architecture of universal grammar. Language
70: 1-52.
Bresnan, Joan. 2001. Lexical-Functional Syntax. Oxford and Cambridge, MA: Blackwell.
Briscoe, Edward, and Ann Copestake. 1999. Lexical rules in constraint-based grammar. Com-
putational Linguistics 25(4):487-526.
Briscoe, Edward, Ann Copestake, and Valeria de Paiva (Eds.). 1993. Inheritance, Defaults, and
the Lexicon. Cambridge: Cambridge University Press.
Brody, Michael. 1995. Lexico-Logical Form: A Radically Minimalist Theory. Cambridge, MA:
MIT Press.
Burton-Roberts, N. 1997. Analysing Sentences: An Introduction to English Syntax. 2nd Edition.
Longman. pp. 7-23
Carnie, Andrew. 2002. Syntax: A Generative Introduction. Oxford: Blackwell. pp. 51-53
Carpenter, Bob. 1992. The Logic of Typed Feature Structures: with Applications to Unification
Grammars, Logic Programs, and Constraint Resolution. Cambridge: Cambridge Univer-
sity Press.
Chierchia, Gennaro, and Sally McConnell-Ginet. 1990. Meaning and Grammar: An Introduc-
tion to Semantics. Cambridge, MA: MIT Press.
Chomsky, Noam. 1957. Syntactic Structures. The Hague: Mouton.
Chomsky, Noam. 1963. Formal properties of grammars. In R. D. Luce, R. Bush, and E. Galanter
(Eds.), Handbook of Mathematical Psychology, Vol. Volume II. New York: Wiley.
276
Chomsky, Noam. 1965. Aspects of the Theory of Syntax. Cambridge, MA: MIT Press
Chomsky, Noam. 1969. Remarks on Nominalization. In R. Jacobs and P.S. Rosenbaum(eds),
Readings in English Transformational Grammar, 184-221. Waltham, MA: Ginn.
Chomsky, Noam. 1973. Conditions on Transformations. In S. Anderson and P. Kiparsky (eds.),
A Festschrift for Morris Halle. New York: Holt, Rinehart and Winston.
Chomsky, Noam. 1975. The Logical Structure of Linguistic Theory. Chicago: University of
Chicago Press.
Chomsky, Noam. 1977. On Wh-movement. P. Culicover, A. Akmajian, and T. Wasow (eds.),
Formal Syntax, 71-132. New York: Academic Press.
Chomsky, Noam. 1980. Rules and Representations. New York: Columbia University Press.
Chomsky, Noam. 1981a. Lectures on Government and Binding. Dordrecht: Foris.
Chomsky, Noam. 1981b. Principles and Parameters in Syntactic Theory. In Hornstein and
Lightfoot 1981, 32-75.
Chomsky, Noam. 1982. Some Concepts and Consequences of the Theory of Government and
Binding. Cambridge, MA: MIT Press.
Chomsky, Noam. 1986. Barriers. Cambridge, MA: MIT Press.
Chomsky, Noam. 1995. The Minimalist Program. Cambridge, MA: MIT Press.
Chomsky, Noam, and Howard Lasnik. 1977. Filters and control. Linguistic Inquiry 8:4 25-504.
Clark, Eve V., and Herbert H. Clark. 1979. When nouns surface as verbs. Language 55: 767-
811.
Copestake, Ann. 1992. The Representation of Lexical Semantic Information. PhD thesis, Uni-
versity of Sussex. Published as Cognitive Science Research Paper CSRP 280, 1993.
Copestake, Ann 2002. Implementing Typed Feature Structures Grammars. Stanford: CSLI Pub-
lications.
Copestake, Ann, Dan Flickinger, Carl Pollard, and Ivan A. Sag. 2006. Minimal Recursion Se-
mantics: an Introduction. Research on Language and Computation 3.4: 281–332.
Cowper, Elizabeth A. 1992. A Concise Introduction to Syntactic Theory: The Government-
Binding Approach. University of Chicago Press.
Crystal, David. 1985. A Dictionary of Linguistics and Phonetics. London: B. Blackwell in as-
sociation with A. Deutsch.
Culicover, Peter, Adrain Akmajian, and Thomas Wasow(eds.). 1977. Formal Syntax. NewYork:
Academic Press.
Culicover, Peter. 1993. Evidence against ECP accounts of the that-t effect. Linguistic Inquiry
24: 557–561.
Dalrymple, Mary. 2001. Lexical Functional Grammar. (Syntax and Semantics, Volume 34).
New York: Academic Press.
277
Dalrymple, Mary, Annie Zaenen, John Maxwell III, and Ronald M. Kaplan (eds.) 1995. Formal
Issues in Lexical-Functional Grammar. Stanford: CSLI Publications.
Davidson, Donald. 1980. Essays on Actions and Events. Oxford: Clarendon Press; New York:
Oxford University Press.
Davis, Anthony. 2001. Linking by Types in the Hierarchical Lexicon. Stanford: CSLI Publica-
tions.
Dowty, David, Robert Wall, and Stanley Peters. 1981. Introduction to Montague Semantics.
Dordrecht: D. Reidel.
Dowty, David. 1982. Grammatical Relations and Montague Grammar. In P. Jacobson and G.
Pullum (eds.), The Nature of Syntactic Representation, 79-130. Dordrecht: Reidel.
Dowty, David. 1989. On the Semantic Content of the Notion of Thematic Role. In G. Chierchia,
B. Partee, and R. Turner (eds.), Properties, Types, and Meanings, Volume 2, pp. 69–129.
Dordrecht: Kluwer Academic Publishers.
Dubinksy, Stanley and William Davies. 2004. The Grammar of Raising and Control: A Course
in Syntactic Argumentation. Oxford: Blackwell Publishers.
´
E. Kiss, Katalin. 1998. Identificational focus versus information focus. Language 74: 245–273.
Emonds, Joseph. 1970. Root and structure-preserving transformations. Doctoral dissertation,
MIT.
Emonds, Joseph. 1975. A Transformational Approach to Syntax. New York: Academic Press.
Fillmore, Charles. 1963. The Position of Embedding Transformations in A Grammar. Word 19:
208-231.
Fillmore, Charles. 1999. Inversion and Constructional Inheritance. In G. Webelhuth, J.P Koenig,
and A. Kathol (eds.), Lexical and Constructional Aspects of Linguistics Explanation, 113–
128. Stanford: CSLI Publications.
Fillmore, Charles J., Paul Kay, and Mary Catherine O’Connor. 1988. Regularity and idiomatic-
ity in grammatical constructions: The case of let alone. Language 64(3): 501-538.
Fodor, Jerry A. 1983. The Modularity of Mind. Cambridge, MA: MIT Press
Fodor, Jerry A., and Jerrold J. Katz, (eds). 1964. The structure of language. Englewood Cliffs,
NJ: Prentice-Hall.
Fraser, Bruce. 1970. Idioms within a transformational grammar. Foundations of Language 6:
22-42.
Gazdar, Gerald. 1981. Unbounded dependencies and coordinate structure. Linguistic inquiry
12: 155-184.
Gazdar, Gerald. 1982. Phrase structure grammar. In P. Jacobson and G. K. Pullum (eds.), The
nature of Syntactic Representation. Dordrecht: Reidel.
Gazdar, Gerald, Ewan Klein, Geoffrey K. Pullum, and Ivan A. Sag. 1985. Generalized Phrase
Structure Grammar. Cambridge, MA; Havard University Press and Oxford; Basil Black-
278
well.
Gazdar, Gerald, and Geoffrey K. Pullum. 1981. Subcategorization, constituent order, and the
notion ’head’. In M. Moortgat, H. van der Hulst, and T. Hoekstra (eds.), The Scope of
Lexical Rules. Dordrecht: Foris.
Gazdar, Gerald, Geoffrey K. Pullum, and Ivan A. Sag. 1982. Auxiliaries and related phenomena
in a restrictive theory of grammar. Language 58: 591-638.
Ginzburg, Jonathan, and Ivan A. Sag. 2000. Interrogative Investigations: The Form, Meaning
and Use of English Interrogatives. Stanford: CSLI Publications.
Goldberg, Adele E. 1995. A Construction Grammar Approach to Argument structure. Chicago:
University of Chicago Press. Green,
Grimshaw, Jane. 1997. Projection, Heads, and Optimality. Linguistic Inquiry, 28: 373–422.
Green, Georgia M. 1976. Main Clause Phenomena in Subordinate Clause. Language 52: 382-
397.
Green, Georgia M. 1981. Pragmatics and Syntactic Description. Studies in the Linguistic Sci-
ences 11.1: 27-37.
Greenbaum, Sidney. 1996. The Oxford English Grammar. Oxford: Oxford University Press.
Groat, Erich M. 1995. English Expletives: A Minimalist Approach. Linguistic Inquiry 26.2:
354ł365.
Grosu, Alexander. 1972. The Strategic Content of Island Constraints. Ohio State University
Working Papers in Linguistics 13: 1-225.
Grosu, Alexander. 1974. On the Nature of the Left Branch Constraint. Linguistic Inquiry 5:
308-319.
Grosu, Alexander. 1975. On the Status of Positionally-Defined Constraints in Syntax. Theoret-
ical Linguistics 2: 159-201.
Grice, H. Paul. 1989. Studies in the Way of Words. Cambridge, MA: Harvard University Press.
Haegeman, Liliance. 1994. Introduction to Government and Binding Theory. Oxford and Cam-
bridge, MA: Basil Blackwell.
Harman, Gilbert. 1963. Generative grammar without transformation rules: A defense of phrase
structure. Language 39: 597-616.
Harris, Randy Allen. 1993. The Linguistic Wars. Oxford: Oxford University Press.
Harris, Zellig S. 1970. Papers in Structural and Transformational Linguistics. Dordrecht: Rei-
del.
Hooper, Joan, and Sandra Thompson. 1973. On the Applicability of Root Transformations.
Linguistic Inquiry 4: 465-497.
Hornstein, Norbert, and Daivd Lightfoot, (eds.) 1981. Explanation in linguistics: The logical
problem of language acquisition. London: Longman.
279
Huddleston, Rodney, and Geoffrey K. Pullum. 2002. The Cambridge Grammar of the English
Language. Cambridge University Press.
Hudson, Richard. 1984. Word Grammar. Oxford: Blackwell.
Hudson, Richard. 1990. English Word Grammar. Oxford: Blackwell.
Hudson, Richard. 1998. Word Grammar. In V. Agel, et al (eds.), Dependency and Valency: An
International Handbook of Contemporary Research. Berlin: Walter de Gruyter.
Huang, James. 1982. Logical Relations in Chinese and the Theory of Grammar. Ph.D. Disser-
tation, MIT.
Jackendoff, Ray. 1972. Semantic Interpretation in Generative Grammar. Cambridge, MA: MIT
Press.
Jackendorff, Ray. 1975. Morphological and semantic regularities in the lexicon. Language 51:
639-671.
Jackendorff, Ray. 1994. Patterns in the Mind. New York: Basic Books.
Jackendoff, Ray. 1977. X

-syntax. Cambridge, MA: MIT Press.
Jackendorff, Ray. 2002. Foundation of Language: Brian, Meaning, Grammar, Evolution. Ox-
ford: Oxford University Press.
Jacobs, Roderick. 1995. English Syntax: A Grammar for English Language Professionals. Ox-
ford University Press.
Johnson, David, and Paul Postal. 1980. Arc-Pair Grammar. Princeton: Princeton University
Press.
Johnson, David, and Shalom Lapin. 1999. Local Constrains vs. Economy. Stanford: CSLI Pub-
lication.
Kager, Rene. 1999. Optimality Theory. Cambridge: Cambridge University Press.
Kaplan. Ronald M., and Annie Zaenen. 1989. Long-distance Dependencies, Constituent Struc-
ture and Functional Uncertainty. In M. R. Baltin and A. S. Kroch (eds.), Alternative Con-
ceptions of Phrase Structure, 17-42. University of Chicago Press.
Katz, Jerrold J., and Paul M. Postal. 1964. An Integrated Theory of Linguistic Descriptions.
Cambridge, MA: MIT Press.
Katz, Jerrold J., and Paul M. Postal. 1991. Realism versus Conceptualism in Linguistics. Lin-
guistics and Philosophy 14: 515-554.
Kay, Paul. 1995. Construction grammar. In J. Verschueren, J.-O. Ostman, and J. Blommaert
(Eds.), Handbook of Pragmatics. Amsterdam and Philadelphia: John Benjamins.
Kay, Paul, and Charles J. Fillmore. 1999. Grammatical Constructions and Linguistic General-
izations: The What’s x Doing y Construction. Language 75.1: 1-33.
Kayne, Richard, and Jean-Yves Pollock. 1978. Stylistic Inversion, Successive Cyclicity, and
Move NP in French. Linguistic Inquiry 9: 595-621.
280
Keenan, Deward. 1975. Some Universals of Passive in Relational Grammar. In Robin E. Gross-
man, L. James San, and Timothy J. Vance (eds.), Papers from the 11th Regional Meeting,
Chicago Linguistic Society, 340-352. Chicago: Chicago Linguistic Society.
Keenan, Edward, and Bernard Comrie. 1977. Noun phrase accessibility and universal grammar.
Linguistic Inquiry 8: 63-99.
Kim, Jong-Bok. 2001a. On the Types of Prepositions and Their Projections in Syntax. Studies
in Modern Grammar. 26. 1-22.
Kim, Jong-Bok. 2001b. Constructional Constraints in English Free Relative Constructions: Lan-
guage and Information 5.1: 35–53.
Kim, Jong-Bok. 2002a. On the Structure of English Partitive NPs and Agreement. Studies in
Generative Grammar 12.2: 309–338.
Kim, Jong-Bok. 2002b. English Auxiliary Constructions and Related Phenomena: From a
Constraint-based Perspective. Language Research, 38.4 1037-1076
Kim, Jong-Bok. 2002c. The Grammar of Negation: A Constraint-Based Approach. Stanford:
CSLI Publications.
Kim, Jong-Bok. 2003. Similarities and Differences between English VP Ellipsis and VP
Fronting: An HPSG Analysis. Studies in Generative Grammar. 13.3
Kim, Jong-Bok. 2004a. Hybrid English Agreement. Linguistics. Linguistics 42.6: 1105-1128.
Kim, Jong-Bok. 2004b. Korean Phrase Structure Grammar (In Korean). Seoul: Hankwuk Pub-
lishing.
Kim, Jong-Bok, and Ivan Sag. 1995. The Parametric Variation of French and English Negation.
In Jose Camacho, Lina Choueiri, and Maki Watanabe (eds), Proceedings of the Fourteenth
West Coast Conference on Formal Linguistics(WCCFL), 303-317. Stanford: SLA CSLI
Publications.
Kim, Jong-Bok and Ivan A. Sag. 2002. Negation Without Movement. Natural Language and
Linguistic Theory, Volume 20, Number 2, 339-412.
Kim, Jong-Bok, and Ivan A. Sag. 2006. English Object Extraposition: A Constraint-Based Ap-
proach. Proceedings of the HPSG05 Conference, University of Lisbon, Stefan Mueller
(Editor). CSLI Publications, Pp. 192–212.
King, Paul J. 1989. A Logical Formalism for Head-Driven Phrase Structure Grammar. PhD
thesis, University of Manchester.
Koenig, Jean-Pierre. 1999. Lexical Relations. Stanford: CSLI Publications.
Kornai, Andras, and Geoffrey K. Pullum. 1990. The X-bar Theory of Phrase Structure. Lan-
guage 66: 24-50.
Koster. Jan. 1987. Domains and Dynasties, the Radical autonomy of Syntax. Dorarecht: Foris.
Langacker, Ronald. 1987. Foundations of Cognitive Grammar. Stanford, CA: Stanford Univer-
sity Press.
281
Lappin, Shalom, Robert Levine, and David Johnson. 2000. The structure of unscientific revolu-
tions. Natural Language and Linguistic Theory 18: 665-671.
Lasnik, Howard, Marcela Depiante, and Arthur Setpanov. 2000. Syntactic Structures Revisited:
Contemporary Lectures on Classic Transformational Theory. Cambridge, MA: MIT Press.
Levin, Beth. 1993. English verb Classes and Alternations: A Preliminary Investigation. Chicago:
University of Chicago Press.
Levin, B. and M. Rappaport Hovav. 2005. Argument Realization, Research Surveys in Linguis-
tics Series. Cambridge: Cambridge University Press.
Lees, Robert B. and Edward S. Klima. 1963. Rules for English Pronominalization. Language
39: 17-28.
Li, Charles, and Sandra A. Thompson. 1976. Subject and Topic: ANewTypology of Languages.
In Li, Charles N. (ed.) Subject and Topic, New York/San Francisco/London: Academic
Press, 457-490.
Lambrecht, Knud. 1994. Information Structure and Sentence Form. Cambridge: Cambridge
University Press.
Malouf, Rob. 2000. Mixed Categories in the Hierarchical Lexicon. Stanford: CSLI Publications.
McCawley, James D. 1968. Concerning the Base Component of a Transformational Grammar.
Foundations of Language 4: 243-269.
McCloskey, James. 1988. Syntactic theory. In Frederick J. Newmeyer (ed.), Linguistics: The
Cambridge Survey 18-59. Cambridge: Cambridge University Press.
McCawley, James D. 1988. The Syntactic Phenomena of English. Chicago: University of
Chicago Press.
Michaelis, Laura, and Knud Lambecht. 1996. Toward a construction-based theory of language
function: The case of nominal extraposition. Language 72: 215-248.
Montague, Richard. 1973. The Proper Theory of Quantification. Approaches to Natural Lan-
guage, ed. by J. Hintikka, J. Moravcsik, and P. Suppes. Dordrecht: Reidel.
Moortgat, Michael. 1988. Categorial Investigations. Dordrecht: Foris.
Nunberg, Geoffrey, Ivan A. Sag, and Thomas Wasow. 1994. Idioms. Language 70: 491-538.
Nunberg, Geoffrey. 1995. Transfer of Meaning. Journal of Semantics, 12.2: 109–132.
Pavey, Emma. 2004. The English it-cleft construction: A Role and Reference Grammar Analysis.
Ph.D. dissertation, SUNY.
Perlmutter, David M., (ed.) 1983. Studies in Relational Grammar 1. Chicago: University of
Chicago Press.
Perlmutter, David M., and Carol Rosen. 1984. Studies in Relational Grammar 2. Chicago: Uni-
versity of Chicago Press.
Perlumutter, David, and Paul Postal. 1977. Toward a universal characterization of passivization.
In Proceedings of the 3rd Annual Meeting of the Berkeley Linguistics Society, Berkeley.
282
University of California, Berkeley. Reprinted in Perlmutter (1983).
Perlmutter, David, Scott Soames. 1979. Syntactic Argumentation and the Structure of English.
Berkeley: University of California Press.
Pinker, Steven. 1994. The Language Instinct. New York: Morrow.
Pollard, Carl, and Ivan A. Sag. 1987. Information-Based Syntax and Semantics, Volume 1: Fun-
damentals. Stanford: CSLI Publication.
Pollard, Carl, andIvan A. Sag. 1992. Anaphors in English and the scope of binding theory.
Linguistic Inquiry 23:261-303.
Pollard, Carl, and Ivan A. Sag. 1994. Head-Driven Phrase Structure Grammar. Chicago: Uni-
versity of Chicago Press.
Pollock, Jean-Yves. 1989. Verb movement, Universal Grammar, and the structure of IP. Lin-
guistic Inquiry 20: 365-422.
Postal, Paul M. 1971. Crossover Phenomena. New York: Holt, Rinehart, and Winston.
Postal, Paul. 1974. On Raising. Cambridge, MA: MIT Press.
Postal, Paul. 1986. Studies of Passive Clause. Albany: SUNY Press.
Postal, Paul, and Brian Joseph (eds.). 1990. Studies in Relational Grammar 3. Chicago: Univer-
sity of Chicago Press.
Postal, Paul, and Geoffrey K. Pullum. 1998. Expletive Noun Phrases in Subcategorized Posi-
tions. Linguistic Inquiry 19: 635-670.
Pullum, Geoffrey K., Gerald Gazdar. 1982. Natural languages and context-free languages. Lin-
guistics and Philosophy 4: 471-504.
Pullum, Geoffrey. 1979. Rule Interaction and the Organization of a Grammar. New York: Gar-
land.
Pullum, Geoffrey. 1991. English Nominal Gerund Phrases as Noun Phrases with Verb-Phrase
Heads. Linguistics 29: 763–799.
Pullum, Geoffrey K. and Barbara C. Scholz. 2002. Empirical Assessment of Stimulus Poverty
Arguments. The Linguistic Review 19, 9-50.
Prince, Alan, and Paul Smolensky. 1993. Optimality Theory: Constraint Interaction in Genera-
tive Grammar. Tech Report RuCC-TR-2. ROA-537: Rutgers University Center for Cogni-
tive Science.
Quirk, Randoph, Sidney Greenbaum, Geoffrey Leech, and Jan Svartvik. 1972. A Grammar of
Contemporary English. London and New York: Longman.
Quirk, Randoph, Sidney Greenbaum, Geoffrey Leech, and Jan Svartvik. 1985. A Comprehensive
Grammar of the English Language. London and New York: Longman.
Radford, Andrew. 1981. Transformational syntax: A Student’s Guide to Chomsky’s Extended
Standard Theory. Cambridge: Cambridge University Press.
283
Radford, Andrew. 1988. Transformation grammar. Cambridge: Cambridge University Press.
Radford, Andrew. 1997. Syntactic Theory and the Structure of English. New York and Cam-
bridge: Cambridge University Press. Richter.
Radford, Andrew. 2004. English Syntax: An Introduction. Cambridge: Cambridge University
Press.
Riemsdijk, Henk van, and Edwin Williams. 1986. Introduction to the Theory of Grammar. Cam-
bridge, Mass.: MIT Press.
Rosenbaum, Peter S. 1967. The Grammar of English Predicate Complement Constructions.
Cambridge, Mass.: MIT Press.
Ross, John R. 1967. Constraints on Variables in Syntax. PhD thesis, MIT. Published as Infinitive
Syntax. Norwood, NJ: Ablex, 1986.
Ross. John R. 1972. Doubl-ing. Linguistic Inquiry 3: 61-86.
Ross. John. R. 1969. Auxiliaries as main verbs, In W. Todd(Ed.), Studies in Philosophical Lin-
guistics 1. Evanston, Ill.: Great Expectations Press.
Sag, Ivan A. to appear. Rules and exceptions in the English auxiliary system. Journal of Lin-
guistics.
Sag, Ivan A. 2005. Remarks on Locality. Ms. Stanford University.
Sag, Ivan A. 2000. Another Argument Against Wh-Trace. Jorge Hankamer Webfest.
Sag, Ivan A. 1997. English relative clause constructions. Journal of Linguistics 33(2): 431-484.
Sag, Ivan A., and Janet D. Fodor. 1994. Extraction Without Traces. In Proceedings of the Thir-
teenth Annual Meeting of the West Coast Conference on Formal Linguistics. Stanford:
CSLI Publications. Pp 365–384.
Sag, Ivan A., Philip Hofmeister, and Neal Snider. to appear. Processing Complexity in Subja-
cency Violations: The Complex Noun Phrase Constraint. Proceedings of the 43rd Annual
Meeting of the Chicago Linguistic Society. Chicago: CLS.
Sag, Ivan A., and Carl Pollard. 1991. An Integrated Theory of Complement Control. Language
67: 63-113.
Sag, Ivan A. Sag and Thomas Wasow and Emily M. Bender. 2003. Syntactic Theory: A Formal
Introduction. Stanford: CSLI Publications.
Saussure, Ferdinand de. 1916. Course of General Linguistics.
Savitch, Walter J., Emmon bach, William Marsh, and Gila Safran-Naveh. 1987. The Formal
Complexity of Natural Language. Dordrecht: D. Reidel.
Schutze, Carson T. 1996. The Empirical Base of Linguistics. Chicago: University of Chicago
Press.
Sells, Peter. 1985. Lectures on Contemporary Syntactic Theories. Stanford: CSLI Publications.
284
Sells, Peter (ed.). 2001. Formal and Empirical issues in Optimality Theoretic Syntax. Stanford:
CSLI Publications.
Shieber, Stuart. 1986. An Introduction to Unification-based Approaches to Grammar. Stanford:
CSLI publications.
Skinner, B. F. 1957. Verbal Behavior. New York: Appleton-Century-Crofts.
Smith, Jeffrey D. 1999. English Number Names in HPSG. In G. Webelhuth, J.-P. Koenig, and
A. Kathol (eds.), Lexical and Constructional Aspects of Linguistic Explanation, 145-160.
Stanford: CSLI Publications.
Steedman, Mark. 1996. Surface Structure and Interpretation. Cambridge, MA: MIT Press.
Steedman, Mark. 2000. The Syntactic Process. Cambridge, MA: MIT Press/Bradford Books.
Steele, Susan.1981. An Encyclopedia of AUX. Cambridge, MA: MIT Press.
Stockwell, Robert P., Paul Schachter, and Barbara H. Partee. 1973. The major Syntactic Struc-
tures of English. New York: Holt, Rinehart and Winston.
Trask, Robert Lawrence. 1993. A Dictionary of Grammatical Terms in Linguistics. London and
New York: Routledge.
Ward, Gregory. 1985. The Semantics and Pragmatics of Preposing. Ph.D. Dissertation., Uni-
versity of Pennsylvania.
Wasow, Thomas. 1977. Transformations and the lexicon. In Formal Syntax (Culicover et al.
1977).
Wasow, Thomas. 1989. Grammatical Theory. In Foundations of Cognitive Science (Posner
1989).
Webelhuth, Gert (ed.). 1995. Government and Binding Theory and the Minimalist Program.
Oxford: Basil Blackwell.
Zwicky, Arnold, and Geoffrey K. Pullum. 1983. Cliticiziation vs. inflection: English n’t. Lan-
guage 59: 502-13.
285
Index
nominal, 89, 90, 206, 256, 257, 259
verbal, 89, 256
acceptability grade, 244
accusative, 181, 183, 218, 234, 237, 247, 250
active, 131, 177, 178, 192
adjective, 3, 12, 13, 25, 43, 81, 91, 93, 102,
120, 250
attributive, 120
control, 127, 128, 144
predicative, 120, 257
raising, 128, 129
adjunct, 41, 43, 54, 196, 219, 243, 259
Adjunct Clause Constraint (ACC), 243
adverb, 12, 24–26, 41, 42, 152, 163, 164, 173
adverbial, 41, 219, 221, 263, 267
adverbial amelioration effect, 209
Affix Hopping Rule, 154
agent, 36, 37, 44–46, 65, 141, 143, 179
AGR (agreement), 65, 103–108
agreement, 18, 31, 37, 102, 160
index, 107
mismatch, 109
morphosyntactic, 107, 111
noun-determiner, 103
pronoun-antecedent, 105
subject-verb, 7, 43, 56, 105, 110
ambiguity, 19, 33
structural, 8, 18
anomalous, 66
semantically, 27
antecedent, 102, 105, 109, 124, 233, 267
ARG-ST (argument-structure), 65–67, 73, 90,
148
Argument Realization Constraint (ARC), 66,
67, 200, 202, 207, 220
arguments, 62, 65, 66
article, 5, 14, 58, 118
atomic, 62
attribute, 62–64, 77
attribute-value matrix (AVM), 62
autonomous, 18
auxiliary verb, 15, 16, 24, 28, 37, 179, 187,
195, 207
Baker, C.L., 4
Bare Head-Rel Modifier Rule, 238
benefactive, 39, 44
Bresnan, Joan, 188, 197
British English, 159
Case Filter, 249
Case Theory, 181
Chomsky, Noam, 1, 3, 36, 153, 154, 180, 188,
197, 217, 219, 249, 262
clausal
complement, 15, 93, 95, 211, 255
subject, 90, 243
clause, 11
complement, 85
embedded, 203, 210, 220, 230, 232
finite, 15
infinitival, 15, 16, 91, 127, 238
infinitval, 87
287
subordinate, 11, 14, 221, 263
cleft, 19, 247, 248, 260
it-cleft, 260, 261
wh-cleft, 260, 261
inverted wh-cleft, 260, 262
coindex, 134, 136, 141
combinatory
possibilities, 1
properties, 49
requirement, 49
rules, 2
common noun, 82
competence, 2–4
complement, 15, 41, 50–52, 54, 55, 59
clausal, 15, 84
infinitival, 129
oblique, 40, 44, 50, 51, 196
predicative, 39, 40, 50, 51
VP, 16
complementation pattern, 56, 70
complementizer, 14, 15, 30, 85, 87, 88, 256
Complex Noun Phrase Constraint (CNPC), 242
complex NP, 242, 244
COMPS (complements), 41, 66, 67
conjunction, 14, 22
coordinating, 14
subordinate, 270
subordinating, 14
constituency, 31
constituent, 11, 18–22, 25, 35, 49, 88, 179, 254
constituent question, 20
constituenthood, 32, 181
context dependent, 107
context free grammar, 26
contraction, 151, 152, 161, 169
contrastive focus, 268
convention, 1
Coordinate Structure Constraint (CSC), 242,
243
coordination, 22, 88, 154, 200, 215
Coordination Rule, 30, 204, 205, 234, 243
copula, 82, 157, 158, 260, 263
core clause, 247
COUNT, 117
COUNT (countable), 116
count nouns, 4, 6, 7, 100
creativity, 2
declarative, 77, 79, 195
deep structure, 127, 128, 132
DEF (definite), 114
definite, 113, 118
definiteness, 119
demonstrative, 14, 58
dependency
strong, 247
weak, 247
descriptive rules, 4
determiner, 14, 57, 73, 99, 100, 102, 103, 107,
110, 112, 117
Determiner Phrase (DP), 101
direct object, 50
directive, 195
discharge, 199, 201, 203
discourse, 177, 240
distributional criteria, 13
distributional crtierion, 12
Do-so Replacement Condition, 53
do-so test, 53
Do-support, 154
Dowty, David, 44
DP (Determiner Phrase), 58, 59, 239
Emonds, Joseph, 256
empty element, 199, 208
empty operator, 249
endocentricity, 55, 57, 61
English Declarative Sentence Rule, 51, 77
exclamative, 167, 195
experiencer, 44, 46, 143, 144
expletive, 83, 128, 131, 145, 161, 249, 261
external syntax, 49
EXTRA, 257, 258, 260
extraction, 207
extraposition, 247, 248, 255, 262
Extraposition Lexical Rule, 257
feature, 61, 62
288
name, 63
specification, 24
structure, 62–64
system, 61
unification, 64
feature percolation, 197
femine, 105
filler, 196–199, 204, 225
filler-gap, 244
Fillmore, Charles, 167
finite, 2, 15, 23, 62, 74, 76–78, 90, 155, 164,
165, 172
VP, 51
finite set, 2
Flickinger, Dan, 170
floated quantifier, 152, 157
form, 12, 195
fragment, 20, 24
free relative, 265
FREL, 266
function, 12, 195
GAP, 200, 202, 233, 238, 242, 271
gap, 196–198
gapping, 151
Gazdar, Gerald, 78, 267
GEN (gender), 105
generate, 3
generative grammar, 4
generative linguistics, 1
generative syntax, 151
generativity, 27
gerundive, 50, 78, 255
goal, 39, 44, 69
Godard, Daniel, 189
grammatical category, 12
grammatical competence, 3
grammatical function, 35, 42, 49, 177, 178, 180
head, 7, 15, 50, 51
head feature, 73, 74, 78
Head Feature Principle (HFP), 74–76, 79, 82,
235
Head-Complement Rule, 59, 73, 74, 76, 80, 82,
186, 191
Head-Extra Rule, 258, 259
Head-Filler Rule, 75, 201, 203, 208, 213, 219
Head-Light Rule, 190, 191
Head-Modifier Rule, 59, 60, 73, 74
Head-Only Rule, 217, 235
Head-Rel Modifier Rule, 229, 232
Head-Specifier Rule, 59, 61, 73, 74, 76, 92,
103, 185
headedness, 50, 55, 61
heavy element, 189
hierarchical structure, 29, 59
HPSG, 62
Huddleston, Rodney, 47, 240
hypothesis, 4, 6
imperative, 161, 167, 195
IND (index), 107
indirect question, 211, 215, 216, 218, 243
Indirect Wh-question Constraint (IWC), 243
infinite use, 1
infinitival
CP, 91
marker, 15, 82
S, 88
VFORM, 77
VP, 88, 127, 129, 130, 132, 136, 145, 216
wh-relative, 235
information
argument, 62
phonological, 62
semantic information, 62
syntactic , 62
instrument, 36, 45
interability, 52
intermediate category, 57, 59, 71, 197
internal syntax, 49
interrogative, 99, 187, 195
inverted sentence, 196
islands, 243
Jackendoff, Ray, 46, 256
Kim, J.-B., 89, 153, 154, 164, 189, 256, 260
289
Kleene Star, 18
Kleene Star Operator, 18
Lambrecht, Knud, 1, 262
language specific, 59
Left-Branching Constraint (LBC), 243
Levin, Beth, 45
lexeme, 76
lexical category, 11, 17
lexical idiosyncrasy, 112, 182
lexicalist, 182
lexically discharge, 251
lexicon, 17, 46, 62
LFG, 62
LIGHT, 189
location, 41, 45
locative, 36
locative adjunct, 225
long-distance, 196, 201, 225, 241
long-distance dependency, 197
main verb, 151
manner, 41
masculine, 105
maximal phrase, 51
meaning criterion, 12
meaning preservation, 130, 144
memorize, 3
minimal phrase, 51
mismatch, 140
MOD, 227, 233, 270
modals, 88
modifier, 41, 51, 52, 54, 55, 61, 73, 120, 121,
225
postnominal, 228
morphological criteria, 13
morphological criterion, 12
movement, 197, 256
movement paradox, 197
multiple gap, 254
N-bar, 57–59
negation, 15, 151, 152, 161–163, 169
constituent, 163
sentential, 164
neuter, 105
NFORM, 83, 84, 256, 257
NICE properties, 152
nominative, 181
non-count nouns, 4
nonfinite, 76
nonhead daughter, 230
nonlocal dependency, 247
Nonlocal Inheritance Principle (NIP), 203, 204,
227, 231, 235, 252
nonlocal position, 200
nontransformational, 134, 144, 182
noun, 12
collective, 111
common, 99, 100
countable, 99
measure, 118
partitive, 112
pronoun, 99, 102
proper, 99, 102
uncountable, 99
NUM (number), 61, 103–108
Numberg, Geoffery, 107
object, 35
obligatory, 50, 52
ontological issue, 151
ordering restriction, 152
particle, 16, 20, 30, 31, 189, 190
partitive, 111, 117
parts of speech, 12
passive, 131
prepositional, 188
Passive Lexical Rule, 183, 187
passivization, 38, 39, 131, 188
past, 12, 15, 76–78, 154, 158
past participle, 77
patient, 36, 44, 46, 141
PER (person), 105, 106, 109
performance, 3, 4
personal pronoun, 105
PFORM, 81, 96, 114, 119, 190
PHON (phonology), 65
290
phrasal category, 19
phrasal verb, 30
phrase, 11
Phrase Structure Rule (PS) Rule, 22
plural, 12
Pollard, Carl, 52
POS (part-of-speech), 61, 73, 106–108, 110,
117, 119
position
of adverb, 157
possessive, 14, 58, 100, 101
postcopular, 147, 148, 264
pragmatics, 182
PRD (predicate), 68, 120
pre-terminal, 17
predicate, 24, 35, 36, 40, 41, 57, 66, 70, 120,
129, 131, 248
predication, 262
predicational, 264
preposition, 8, 13, 20, 81, 94, 113, 114, 188,
190
prepositional object, 225
Prepositional Passive Rule, 190
prepositional verb, 30, 188, 190
prescriptive rules, 4
present, 77, 78
present participle, 77
principles, 1
PRO, 133, 216–218, 235
projection, 50, 55
promoted, 178
pronoun, 5, 21, 37, 83, 99, 113, 181
proper noun, 99
proposition, 142, 177
PS rules, 24, 27, 29, 30, 35, 55, 56, 59, 61, 153
Pullum, Geoffrey, 47, 240
quantificational, 112, 113
quantified NP, 240
quantifier, 14
QUE (question), 187, 213, 227
question, 195
raising properties, 179
Rappaport Hovav, 45
reanalysis, 188
reason, 41
recipient, 39
recursive application, 28
redundancy, 56, 57
reflexive, 102
REL, 226, 229
relative
pronoun, 225, 233
relative clause
bare, 231, 232, 236–238
infinitival, 226, 235
nonrestrictive, 238
reduced, 226
restrictive, 238
relativizer, 267
Rosenbaum, Peter, 255
rules, 1, 4
Sag, Ivan A., 3, 46, 52, 55, 62, 78, 89, 164, 228,
256, 260
SAI Rule, 167, 195, 203
salient contextual information, 268
salient discourse information, 263
Saussure, Ferdinand de, 1
selectional restriction, 130, 179
SEM (semantics), 65, 107
semantic constancy, 53
semantic criteria, 12
semantic restriction, 53
semantic role, 36, 128, 141, 180, 248
semantics, 179, 182
sentence, 11
sentential negation, 152
Sentential Subject Constraint (SSC), 243
situation, 177
slash, 198
sounds and meanings, 1
source, 44
specificational, 264
specifier (SPR), 57
speech acts, 195
291
SPR (specifier), 57, 66, 67, 69, 70, 73, 76, 79,
83, 84, 101, 103, 109, 119, 136
stand-alone test, 20
statement, 195
structural
change, 180
description, 180
difference, 54, 239
position, 14
structure sharing, 63, 106
subcategorization, 56, 68, 129, 130, 141, 159,
178, 213, 250
subject, 7, 35
embedded, 132
subject-auxiliary inversion, 37, 38, 152
subjecthood tests, 37
substitution, 21, 60
subsumption, 64
sucategorization, 136
surface structure, 132, 134, 255
SYN (syntax), 65, 75, 109, 140, 142
syntactic
category, 35, 49
syntactic function, 12
syntactic knowledge, 9
tag question, 37, 152, 166, 169
temporal adjunct, 225
Tense, 154
tense, 13, 15, 23, 76, 77, 153, 155, 181
ternary structure, 188
theme, 44–46, 65, 69, 143
TO-BIND, 251, 252
topicalization, 1
tough, 247
trace, 199, 250, 255
transformation, 132, 153, 167, 180–182, 255
transformational, 128, 181
unbounded, 196, 201
underlying structure, 256
underspecification, 88, 104
ungrammatical, 2
unification, 64, 65, 199
unify, 199
universal, 59
VAL (valence), 66, 67, 130
Valence Principle (VALP), 76, 185, 235
verb, 12
complex transitive, 69
diransitive, 40
ditransitive, 56, 69
equi, 127
intransitive, 45, 56, 67, 248
linking, 68, 70
transitive, 56, 68, 90, 181–183, 188
VFORM, 51, 65, 74, 76–80, 190
VP
finite, 51
infinitival, 43, 129
nonfinite, 162, 163
VP ellipsis, 16, 151, 152, 161, 162, 164, 165,
171
VP Ellipsis Rule, 171
well-formed, 3
wh-question, 20, 206, 207, 226, 236, 242
wh-relative pronoun, 231
word, 11
word order, 1
X

rules, 55, 61, 73
X

theory, 59
yes-no question, 195
292
This new textbook, focusing on the descriptive facts of English, provides a systematic introduc-
tion to English syntax for the students with no prior knowledge of English grammar or syntactic
analysis. The textbook aims to help students to appreciate the various sentence patterns avail-
able in English, understand insights into core data of English syntax, develop analytic abilities
to further explore the patterns of English, and learn precise ways of formalizing syntactic analy-
sis for a variety of English data and major English constructions such as agreement, raising and
control, the auxiliary system, passives, wh-questions, relative clauses, extraposition, and clefts.
Jong-Bok Kim is Associate Professor of School of English at Kyung Hee University, Seoul,
Korea. Peter Sells is Professor of Linguistics at Stanford University.
293

Contents

1

Some Basic Properties of English Syntax 1.1 1.2 1.3 1.4 How We Discover Rules Exercises 9 11 4

1 1 7

Some Remarks on the Essence of Human Language Why Do We Study Syntax and What Is It Good for?

2

From Words to Major Phrase Types 2.1 2.2 2.3 2.4 2.5 Introduction 11 12 Lexical Categories

2.2.1 Determining the Lexical Categories

12

Grammar with Lexical Categories Phrasal Categories
2.5.1 NP: Noun Phrase 2.5.2 VP: Verb Phrase

17

19 22
22 23 25 25 26

Phrase Structure Rules

2.5.3 AP: Adjective Phrase 2.5.4 AdvP: Adverb Phrase 2.5.5 PP: Preposition Phrase

2.6 2.7 3

Grammar with Phrases Exercises 32

26

Syntactic Forms, Grammatical Functions, and Semantic Roles 3.1 3.2 Introduction
3.2.1 Subjects

35

35 36
38 3 36

Grammatical Functions

3.2.2 Direct and Indirect Objects

3.2.3 Predicative Complements 3.2.4 Oblique Complements 3.2.5 Modifiers 41 40

39

3.3 3.4 3.5 4

Form and Function Together Semantic Roles Exercises 46 43

41

Head, Complements, and Modifiers 4.1
4.1.1 Internal vs. External Syntax 49

49 49
50

Projections from Lexical Heads to Phrases

4.1.2 Notion of Head, Complements, and Modifiers

4.2 4.3 4.4

Differences between Complements and Modifiers PS Rules, X -Rules, and Features Lexicon and Feature Structures 55 62
62 65

52

4.4.1 Feature Structures and Basic Operations 4.4.2 Feature Structures for Linguistic Entities 4.4.3 Argument Realization 66 67 4.4.4 Verb Types and Argument Structure

4.5 5

Exercises

71 73 73 76

More on Subjects and Complements 5.1 5.2 Grammar Rules and Principles
5.2.1 5.2.2 5.2.3 Complements of Verbs 76 80

Feature Specifications on the Complement Values
Complements of Adjectives

Complements of Common Nouns

82

5.3 5.4

Feature Specifications for the Subject Clausal Complement or Subject
5.4.1 5.4.2 5.4.3 5.4.4 5.4.5

83
84 91 94

84
90 93

Verbs Selecting a Clausal Complement Verbs Selecting a Clausal Subject

Adjectives Selecting a Clausal Complement Nouns Selecting a Clausal Complement Prepositions Selecting a Clausal Complement

5.5 6

Exercises

95 99 99 100
4

Noun Phrases and Agreement 6.1 6.2 Classification of Nouns Syntactic Structures

6.2.1 Common Nouns 6.2.2 Pronouns 102 6.2.3 Proper Nouns

100 102

6.3

Agreement Types and Morpho-syntactic Features
6.3.1 Noun-Determiner Agreement 6.3.2 Pronoun-Antecedent Agreement 6.3.3 Subject-Verb Agreement 105 103 105

103

6.4 6.5

Semantic Agreement Features Partitive NPs and Agreement
6.5.1 Basic Properties 111 6.5.2 Two Types of Partitive NPs 6.5.3 Measure Noun Phrases 118

107 111
113

6.6

Modifying an NP

120
120 121

6.6.1 Adjectives as Prenominal Modifiers 6.6.2 Postnominal Modifiers

6.7 7

Exercises

123 127 127 128
128 131

Raising and Control Constructions 7.1 7.2 Raising and Control Predicates
7.2.1 Subject Raising and Control 7.2.2 Object Raising and Control

Differences between Raising and Control Verbs

7.3 7.4

A Simple Transformational Approach A Nontransformational Approach
7.4.1 Identical Syntactic Structures

132

134
134 136 140

7.4.2 Differences in Subcategorization Information 7.4.3 Mismatch between Meaning and Structure

7.5

Explaining the Differences
7.5.2 Meaning Preservation

143
143 145 144

7.5.1 Expletive Subject and Object

7.5.3 Subject vs. Object Control Verbs

7.6 8

Exercises

147 151 153

Auxiliary Constructions 8.1 8.2 8.3 Basic Issues 151

Transformational Analyses A Lexicalist Analysis
8.3.1 Modals 154

154

5

8.3.2 8.3.3

Be and Have Periphrastic do

157 159 162

8.3.4 Infinitival Clause Marker to

8.4

Explaining the NICE Properties
8.4.1 Auxiliaries with Negation 8.4.2 Auxiliaries with Inversion 8.4.3 Contracted Auxiliaries 8.4.4 Auxiliaries with Ellipsis

162
162 166

169 171

8.5 9

Exercises

174 177 178
180

Passive Constructions 9.1 9.2 9.3 Introduction 177

Relationships between Active and Passive Approaches to Passive
9.3.1 9.3.2 9.3.3

180
181

From Structural Description to Structural Change A Transformational Approach A Lexicalist Approach 182

9.4 9.5

Prepositional Passives Exercises 192 195

188

10 Wh-Questions

10.1 Clausal Types and Interrogatives 10.2 Movement vs. Feature Percolation
10.3.1 Basic Systems 199

195 197 199

10.3 Feature Percolation with No Abstract Elements
10.3.2 Non-subject Wh-questions 10.3.3 Subject Wh-Questions 206 201

10.4 Indirect Questions
10.4.1 Basic Structure

210
210 215 216

10.4.2 Non-Wh Indirect Questions 10.4.3 Infinitival Indirect Questions 10.4.4 Adjunct wh-questions 219

10.5 Exercises

222 225 226

11 Relative Clause Constructions 11.1 Introduction 225

11.2 Non-subject Wh-Relative Clauses 11.3 Subject Relative Clauses 231
6

4 Cleft constructions 12.5 Exercises Index 287 272 7 .7 Constraints on the GAP 11.2 Transformational Analysis 12.3.2.3.1 Basic Properties 12.6 Restrictive vs.11.4. Nonrestrictive Relative Clauses 241 12 Special Constructions 12.4.8 Exercises 245 247 248 248 249 250 11.4.4 That-relative clauses 233 235 238 11.3 A Lexicalist Analysis 12.5 Infinitival and Bare Relative Clauses 11.2 Transformational Analyses 12.1 Introduction 247 12.2 Distributional Properties of the Three clefts 12.1 Basic Properties 260 260 261 262 12.2.3 A Lexicalist Analysis 12.2.3 Extraposition 254 254 255 256 12.2 ‘Easy’ Constructions 12.3.1 Basic Properties 12.4.3 Syntactic Structures of the Three Types of Cleft: Movement Analyses 12.4 Lexically-Based Analyses 264 12.

.

the textbook introduces the feature structure system of Head-Driven Phrase Structure Grammar (HPSG. In Chapter 6. (2003). We understand that one important goal of teaching English syntax to undergraduate students is to help students enhance their understanding of the structure of English in a systematic and scientific way. clauses. complements. 2004). ultimately.Preface One important aspect of teaching English syntax (to native and nonnative undergraduate students alike) involves the balance in the overall approach between facts and theory. The book focuses primarily on the descriptive facts of English syntax. Sag et al. in which they will be able to perform linguistic analyses for simple as well as complex English phenomena. much of the content. 1997. their grammatical functions. Borsley (1991. 1996). Radford (1988. In particular. We acknowledge our debt to these works. Baker (1997). and then rules for combining these words to form well-formed phrases and. Pollard and Sag (1994). We owe a great deal of intellectual debt to the previous textbooks and literature on English syntax. reason. phrasal types. From Chapter 6 on. their semantic roles. which have set the course for teaching syntax over the years. This new introductory textbook has been developed with this goal in mind. (2003)) which places strong emphasis on the role of lexical properties and the interactions among grammatical components. has been inspired by and adopted from renowned textbooks such as Aarts (1997). theoretical concepts of generative grammar from which students can develop abilities to think. as well as our exercises. Chapters 1 to 5 cover the fundamental notions of English grammar. We start with the basic properties of English words. It then proceeds with the basic. In Chapter 4. and overall aspects of clausal meaning. These chapters guide students through the basic concepts of syntactic analysis such as lexical categories. we introduce English subject verb agreement. Within this book. to list just a few. and concentrate on interrelationships i . the book discusses major constructions of English within a holistic view of grammar allowing interactions of various grammatical properties including syntactic forms. presented in a way that encourages students to develop keen insights into the English data. Basic knowledge of this kind is essential for students to move on the next stages. Sag et al. heads. as a way of formalizing the observed generalizations. and modifiers. and analyze English sentences from linguistic points of view.

but different from the constructions in chapters 10 and 11.. which we would otherwise be missed. These two chapters present the basic properties of these constructions and show how the mechanism of feature percolation is a crucial part of a systematic account for them.among different grammatical components which play crucial interacting roles in English agreement phenomena. Director of CSLI Publications... Each chapter also contains exercises ranging from simple to challenging. for his patience and support.. which students will then have in hand.... Studying this chapter... within a principled theory.. We thank for their comments in various places. Chapters 10 and 11 deal with wh-questions and relative clause constructions. and Juwon Lee for their administrative help... often called non-local or long-distance dependency constructions. Numerous people have helped us in writing this textbook. students will be able to fully understand why. We have tried to make each chapter maximally accessible. how... the textbook discusses how to capture systematic relations between related constructions. The goal of all these chapters is the present a groundwork of facts... We also thank Kaunghi Un for helping us with ii . or even pragmatics. but different in terms of semantics. These constructions are also based on long-distance dependencies. aiming to promote deeper understanding of the factual and theoretical contents of each chapter. as well as Lauri Kanerva for his help in matters of production.. The theoretical notions are kept as simple yet precise as possible so that students can apply and use them in analyzing English sentences. in the sense that a gap and its filler are in a potentially long-distance relationship. Chapter 9 deals with the relationships between active and passive voice clauses. extraposition. We provide clear. we can provide good solutions for some puzzling English agreement phenomena. This allows us to express generalizations among auxiliary-sensitive phenomena such as negation. semantics... and cleft constructions. Chapter 8 deals with the English auxiliary system. in order to consider theoretical accounts which apply in precise ways. Chapter 7 covers raising and control phenomena. This chapter assigns the precise lexical information to auxiliary verbs and constructional constraints sensitive to the presence of an auxiliary verb.. and ellipsis. The final chapter of the book covers the so-called ‘tough’ constructions... itself remarkable in that a relatively small number of elements interact with each other in complicated and intriguing ways.. which are famously rather similar in terms of syntactic structures. in various ways.. We also owe out thanks to Dikran Karagueuzian.. Dongjun Lee. In particular. help and interest in our textbook: [. and when to choose between canonical and passive constructions.. inversion.] We also thank teachers and colleagues in Kyung Hee University and Stanford University for their constant encouragement over the years.. this chapter shows that once we allow morphological information to interface with the system of syntax. We also thank Jinyoung Kim.. From Chapter 9 through Chapter 12... simple tree diagrams which will help students understand the structures of English and develop analytic skills to English syntax... Our gratitude also goes to undergraduate and graduate students at Kyung Hee University who used the draft of this as the textbook and raised so many questions that help us reshape its structure as well as contents. and provides insights into the properties of the two different constructions. contraction...

Lastly. We dedicate this book to our beloved ones who with true love and refreshing and comforting words have lead us to think ‘wise and syntactic’ when we are spiritually and physically down. but not the least. we also truly thank our close friends and family members who gave us unconditional love and support in every possible regard. iii .LaTex problems.

.

The first well-known property (as emphasized by Saussure 1916) is that there is no motivated relationship between sounds and meanings.1 Some Basic Properties of English Syntax 1. casa). or how to provide a convincing argument. hamburger. 1 Examples 1 . The second important feature of language. is to convey various kinds of information from the everyday to the highly academic. For words such as hotdog. and one more central to syntax. the observation of which led the development of generative linguistics in the 20th century (cf. is that language makes infinite use of finite set of rules or principles. See Lambrecht (1994) and references therein. the. only 6 form grammatical English sentences:1 (1) a. a. the word hotdog has no relationship with a dog which is or feels hot. are all of these combinations grammatical sentences? Mathematically. equalling 120 instances. One piece of evidence of the system can be observed in word-order restrictions. This is simply observed in the fact that the same meaning is usually expressed by a different sounding-word in a different language (think of house. sweetbread. how many possible combinations can we have from these five words? More importantly. how to understand English grammar. the number of possible combinations of 5 words is 5! (factorial). But among these 120 possible combinations. cranberry. The man kicked a ball.1 Some Remarks on the Essence of Human Language One of the crucial functions of any human language. maison. A language is a system for combining its parts in infinitely many ways. For example. how to remove cherry stains. their meanings have nothing to do with their shapes. such as English or Korean. is placed in the sentence initial position. and kicked. Language provides a means for us to describe how to cook. like (1e) and (1f) are called ‘topicalization’ sentences in which the topic expression (the ball and the man). We commonly consider certain properties of language to be key essential features from which the basic study of linguistics starts. There is just an arbitrary relationship between the word’s sound and its meaning: this relationship is decided by the convention of the community the speakers belong to. Chomsky 1965). bike. already mentioned and understood in the given context. ball. desk. dog. If a sentence is an arrangement of words and we have 5 words such as man.

A ball kicked the man. (2) a. *The man a ball kicked. The ball. a few of which are given in (2). a man kicked. a ball kicked. (3) a. b. seen. Consider the examples in (4): (4) a. we will see these combinatory rules. For example. Such combinatory rules also play important roles in our understanding of the syntax of an example like (3a).2 Whatever these rules are. Kim lives in the house Lee sold to her. All the other 114 combinations.000 times more luminous than our Sun. These rules constrain which words can be combined together or how they may be ordered. *Kim fond of Lee. It is clear that there are certain rules in English for combining words. More natural evidence of the ‘finite set of rules and principles’ idea can be found in cognitive. The requirement of such combinatory knowledge also provides an argument for the assumption that we use just a finite set of resources in producing grammatical sentences. and have no difficulties in producing or understanding sentences which they have never heard. a dull star in an obscure constellation suddenly became 600. A man kicked the ball. Kim is fond of Lee. Even though it is not difficult to understand the meaning of (4a). even though we may well not have seen the following sentence before. 2 Starting in Chapter 2. c. an example which is judged ungrammatical by native speakers even though the intended meaning of the speaker is relatively clear and understandable. Speakers are unconscious of the rules which they use all the time. 2 . or talked about before. *Man the ball kicked the. are unacceptable to native speakers of English. English has a structural requirement for the verb is as in (4b). The ball kicked a man. The man. b. with respect to each other. and that we do not just rely on the meaning of words involved. We use the notation * to indicate that a hypothesized example is ungrammatical. we can understand its meaning if we have a linguistic competence in English: (5) In January 2002. sometimes in groups.b. *Kim lives in the house Lee sold it to her. c. e. f. b. they should give a different status to (3b). creative abilities. d. *Kicked the man the ball. temporarily making it the brightest star in our galaxy.

there would be a dead-end to this process. tall. d.A related part of this competence is that a language speaker can produce an infinite number of grammatical sentences. . To this sentence. b. 3 Think of a simple analogy: what is the longest number? Yet. Competence refers to speakers’ internalized knowledge of their language. and in fact we do not (Pullum and Scholz 2002). we can make a more complex one like (6b) by adding the adjective tall. . These considerations imply that a more appropriate hypothesis would be something like (8):4 (8) All native speakers have a grammatical competence which can generate an infinite set of grammatical sentences from a finite set of resources. how many numbers do you know? The second question only makes sense if the answer is 0–9 (ten digits). 4 The notion of ‘competence’ is often compared with that of ‘performance’ (Chomsky 1965). tall man kicked the ball. To this resulting sentence (7c). and it would be implausible to think that we store an infinite number of sentences and retrieve whenever we need to do so. Sag et al. theoretically enabling us to generate an infinitive number of sentences: (6) a. The handsome. in principle. One might argue that since the number of English adjectives could be limited. e. Some sentences can go on and on and on and on. The handsome. producing a longer one (7b). no one would find themselves lost for another way to keep the process going (cf. We know that this could not be true. we can again add another adjective handsome as in (6c). 3 . 2003): (7) a. c. We could in principle go on adding without stopping: this is enough to prove that we could make an infinite number of well-formed English sentences. d. To (7a). we once again add and on. b. we add the string and on. The man kicked the ball. . whereas performance refers to actual usage of this abstract knowledge of language. In addition. how then can we explain the fact that we can produce or understand an infinite number of grammatical sentences that we have never heard or seen before? It seems implausible to consider that we somehow memorize every example.3 Given these observations. there is limit to the amount of information our brain can keep track of. Some sentences can go on and on and on. . . in particular when we consider that native speakers can generate an infinite number of infinitely long sentences. However. Some sentences can go on. e. nice man kicked the ball. The tall man kicked the ball. Some sentences can go on and on. c. For example. given the simple sentence (6a). . We could continue adding adjectives.

but are not consciously accessible. Step IV: Revise the hypothesis. ‘rules’ refers not to ‘prescriptive rules’ but to ‘descriptive rules’. Step II: Make a hypothesis to cover the first set of data. c. Avoid split infinitives. these prescribe forms which are generally in use. Generative grammar. Prescriptive rules are those which disfavor or even discredit certain usages. the first thing we need to do is to check out grammatical and ungrammatical variants of the expression in question. in particular. this grammatical competence is hypothesized to characterize a generative grammar. or generative syntax. Avoid double negatives. 4 . 5 In generative syntax. (i) a. . C. if asked to do so. 1. . if necessary. Hence we discover the rules indirectly. Meanwhile. the rule for distinguishing count and non-count nouns:6 [Step I: Observing Data] To discover a grammar rule.2 How We Discover Rules How can we then find out what the generative rules of English syntax are? These rules are present in the speakers’ minds. Do not end a sentence with a preposition. moral. speakers cannot articulate their content. (1995). Let us see how these steps work for discovering one of the grammar rules in English. For example. or intellectual judgement. descriptive rules are meant to characterize whatever forms speakers actually use. hypotheses based on the observed data of the given language are perhaps the most reliable. thus aims to define these rules which will characterize all of the sentences which native speakers will accept as well-formed and grammatical. and has been taken as the subject matter of syntactic theory. which we then can define as follows (for English. Step III: Check the hypothesis with more data. 6 Much of the discussion and data in this section are adopted from Baker.L. . in this instance): (9) Generative Grammar: An English generative grammar is the one that can generate an infinite set of wellformed English sentences from a finite set of rules or principles. and of the several methods for inferring these hidden rules. These data can come from speakers’ judgments – known as intuitions – or from collected data sets – often called corpora.5 These rules tell us how words are put together to form grammatical phrases and sentences. Linguistics is in one sense an empirical science as it places a strong emphasis on investigating the data underlying a phenomenon of study. The spoken performance of most English speakers will often contain examples which violate such prescriptive rules. The canonical steps for doing empirical research can be summarized as follows: . as in (i).This hypothesis has been generally accepted by most linguists. In terms of grammar. with any social. Step I: Data collection and observation. The job of syntax is thus to discover and formulate these rules or principles. b.

evidence cannot be used with the indefinite article a(n). The teacher gave John some good tools for the purpose. It is usually necessary to find contrastive examples to understand the range of a given observation. c. evidence cannot be used in the plural. we can easily find more words that behave like evidence: (12) Data Set 2: equipment a. *The evidence that John found was more helpful than the one that Smith found. but we only had enough money to get an equipment every two weeks. b. The tool that Jones got was more helpful than the one that Smith got.let us look at the usage of the word evidence: (10) Data Set 1: evidence a. evidence cannot be referred to by the pronoun one. c. b. What can you tell from these examples? We can make the following observations: (11) Observation 1: a. b. c. different from Observation 2: (16) Observation 3: 5 . *The professor was hoping for a strong evidence. However. *The equipment we bought last year was more expensive than the one we bought this year. The student was hoping for a good clue. b. We thus can add Observation 3. We thus extend Observation 1 a little bit further: (13) Observation 2: a. b. For instance. evidence/equipment cannot be referred to by the pronoun one. words like clue and tool act differently: (14) Data Set 3: clue a. c. The professor gave John some good clues for the question. evidence/equipment cannot be used in the plural. The student was hoping for a tool. The clue that John got was more helpful than the one that Smith got. Unlike equipment and evidence. *The professor found some strong evidences of water on Mars. evidence/equipment cannot be used with the indefinite article a(n). c. b. the nouns clue and tool can be used in the test linguistic contexts we set up. *We had hoped to get three new equipments every month. (15) Data Set 4: tool a. In any scientific research one example is not enough to draw any conclusion.

The mayor gave John some good cakes. few clues. *many evidence. little information b. and also see if it can bring other analytical consequences. *much bags (19) a. much equipment. the indefinite article. Similar behavior can be observed with a noun like beer. consider the usage of much and many: (18) a. little evidence. *little clue. The cake that Jones got was more delicious than the one that Smith got. many tools. Given these data. *few furniture. it appears that the two-way distinction is quite plausible and persuasive.a. little advice. yet few cannot. and the pronoun one. [Step II: Forming a Hypothesis] From the data and observations we have made so far. *many information. many armchairs As observed here. few armchairs The word little can occur with non-count nouns like evidence. little equipment. *many furniture. b. *much clue. clue/tool can be used with the indefinite article a(n). many clues. Similar support can be found from the usage of little and few: (20) a. much furniture. *few equipment. can we make any hypothesis about the English grammar rule in question? One hypothesis that we can make is something like the following: (17) First Hypothesis: English has at least two groups of nouns. much evidence. much advice b. few occurs only with count nouns. The president was hoping for a good cake. information. whereas non-count nouns can combine with much. Consider the following examples with cake: (22) a. diagnosed by tests of plurality. *many equipment. clue/tool can be referred to by the pronoun one. *few evidence. count nouns can occur only with many. clue/tool can be used in the plural. *little suggestion. *few information b. For example. *few advice. *many advice b. *much tool. few tools. we need to check out if it is true of other data. b. c. suggestions. Meanwhile. c. too: 6 . many suggestions. *little tool. Group I (count nouns) and Group II (non-count nouns). *little armchair (21) a. We can now ask if this distinction into just two groups is really enough for the classification of nouns. *much armchair. A little further thought allows us to find support for the two-way distinction for nouns. [Step III: Checking the Hypothesis] Once we have formed such a hypothesis.

1. observe the following: (24) a. (25) a. *The recent strike by pilots have cost the country a great deal of money from tourism and so on. and revising this hypothesis with more data. We can expect that context will determine whether a Group 3 noun is used as count or as noncount.7 (28) a. To enable us to analyze the structure of English sentences in a systematic and explicit way. The data mean that cake and beer can also be used as non-count nouns since that can be used with less or much. Group 2 (non-count nouns). As we have observed so far. b. the main verb agrees with the head element of the subject. drawing generalizations.(23) a. b. This leads us to revise the hypothesis in (17) as following: (26) Revised Hypothesis: There are at least three groups of nouns: Group 1 (count nouns). which refers to the essential element within a phrase. People now drink less beer. and Group 3 (count and non-count). [Step IV: Revising the Hypothesis] The examples in (24) and (25) imply that there is another group of nouns that can be used as both count and non-count nouns. This informal rule can pinpoint what is wrong with the following two examples: For example. b. let us consider how we could use the syntactic notion of head. (27) In English. We recommend to eat less cake and pastry. The students drank too much beer last night. My pastor says I ate too much cake. No one knows how to tell from a good beer to a bad one. 7 The notion of ‘subject’ is further discussed in Chapter 3 and that of ‘head’ in Chapter 4. 7 . These data show us that cake and beer may be classified as count nouns. The bartender gave John some good beers. To help us to illustrate the patterns of English more effectively and clearly. The following is a short and informal rule for English subject-verb agreement. the process of finding finite grammar rules crucially hinges on finding data. However.3 Why Do We Study Syntax and What Is It Good for? There are many reasons for studying syntax. making a hypothesis. from general humanistic or behavioral motivations to much more specific goals such as those in the following: . .

and further that canonical English prepositions combine only with noun phrases. and this is the head. It would not do to simply talk about ‘the noun’ in the subject in the examples in (28).b. This is why the main verb should be has but not have to observe the basic agreement rule in (27). b. *[The recent strike by pilots] have cost the country a great deal of money from tourism and so on. violating this rule. which I saw. b. Here is another example which requires certain syntactic knowledge: (32) He said that that ‘that’ that that man used was wrong. or that the relatives visiting us are themselves boring. The first one can mean either that the event of seeing our relatives is a boring activity. Can you analyze it? What are the differences among these five thats? Structural (or syntactic) knowledge 8 . Visiting relatives can be boring. If one knows English syntax. b. A more subtle instance can be found in the following: (31) a. These examples each have more than one interpretation. *Despite of his limited educational opportunities. as there is more than one. Now let us look at some slightly different cases. The second example can either mean that a specific can containing gas exploded. notwithstanding on many applications in favor of him. if one understands the syntactic structure of these English sentences. we first need to recognize that the words despite and notwithstanding are prepositions. and thus the main verb vary needs to agree with this singular noun. To understand these examples. Abraham Lincoln became one of the greatest intellectuals in the world. that is. these prepositions combine with prepositional phrases again (headed by of and on respectively). This is the kind of sentence one can play with when starting to learn English grammar. it is easy to see that the essential element of the subject in (28a) is not pilots but strike. so is the whole phrase. it is easy to identify these different meanings. The head of the subject and the verb (in the incorrect form) are indicated in (29): (29) a. *The average age at which people begin to need eyeglasses vary considerably. the head is the noun age. I saw that gas can explode. Once we have structural knowledge of such sentences. or it can mean that I observed that gas has a possibility of exploding. in (28b). If the head is singular. In (30). Either example can be made into a grammatical version by pluralizing the head noun of the subject. Meanwhile. and similarly for plural. *A pastor was executed. *[The average age at which people begin to need eyeglasses] vary considerably. We need to be able to talk about the one which gives its character to the phrase. Can you explain why the following examples are unacceptable? (30) a.

religion. 9 . knowledge of English syntax can be a great help. g. water. provide at least one (informal) reason for its ungrammaticality. i. according to your intuitions or ideas. pronoun one. e. indefinite article. There is no difference in principle between the kinds of examples we have presented above and (33): (33) The government’s plan. Apart from having more words than the examples we have introduced above. progress. When it comes to understanding a rather complex sentence. art. Strings have been pulled many times to get students into that university. which was elaborated in a document released by the Treasury yesterday.can be used to diagnose the differences. He washed himself. d. In doing so. difficulty. c. h. focusing on the usage of ‘self’ reflexive pronouns and personal pronouns: (i) a. life 2. decide if it can be used as a count or as a non-count (mass) noun. sculpture. f. suitcase. b. few/little. nothing in this example is particularly complex. For each ungrammatical one. I have never put the book. 1. Kim and Sandy is looking for a new bicycle. What is John eager to do? What is John easy to do? Is the boy who holding the plate can see the girl? Which chemical did you mix the hydrogen peroxide and? There seem to be a good feeling developing among the students. paper. many/much tests) we have discussed in this chapter. discussion. Chris must liking syntax. is the formal outcome of the Government commitment at the Madrid summit last year to put forward its ideas about integration. There is eager to be fifty students in this class. cheese. For each of the following nouns. Syntactic or structural knowledge helps us to understand simple as well as complex English sentences in a systematic way. 3. cheese. luggage. Check or find out whether each of the following examples is grammatical or ungrammatical. research. (i) activity. experience. j. baggage. The boat floated down the river sank.4 Exercises 1. and how it contributes to a whole sentence. (i) a. Part of our study of syntax involves making clear exactly how each word is categorized. Consider the following set of data. construct acceptable and unacceptable examples using the tests (plurality. k.

c. b. Read the following passage and identify all the grammatical errors. Also consider the following imperative examples: (iii) a. we can putting sentences together even as children–we can all do grammar. *Wash myself. *Wash you! b. when we and our students closely reading the sentences in poetry and stories. If you can. try to put pronouns in the unrealized subject position. But knowing about grammar also helps us understood what makes sentences and paragraphs clearly and interesting and precise. *Wash himself. He washed us. (‘he’ and ‘him’ referring to the same person) He washed me. *He washed myself. Wash yourselves. discuss the relevant grammar rules that you can think of. c. Wash yourself. Grammar naming the types of words and word groups that make up sentences not only in English but in any language.8 8 Adapted from “Why is Grammar Important?” by The Assembly for the Teaching of English Grammar 10 . d. pay attention to what the pronouns can refer to. As human beings. d. b. d. *He washed ourselves. Wash him! Can you explain why we can use yourself and yourselves but not you as the object of the imperatives here? In answering this. And knowing about grammar means finding out that all language and all dialect follow grammatical patterns. Grammar can be part of literature discussions. (iv) a. Can you make a generalization about the usage of ‘self’ pronouns and personal pronouns like he here? In answering this question. Wash me! c. c. (i) Grammar is important because it is the language that make it possible for us to talk about language. (ii) a. 4. He washed her. People associate grammar for errors and correctness. *He washed herself. *He washed him.b.

or ‘lexical’ categories. These rules can be inferred from observations about the English data.1 Introduction In Chapter 1. Each of the sentences in (2b)–(2d) contains more than one clause.2 From Words to Major Phrase Types 2. rvv rr vvv rr vv rr . d. clause . . . . we start from words. . . These lexical categories then form a larger constituent ‘phrase’. A clause either is. a well-formed sentence: (1) sentence rvv rr vvv rr vv rr . . and phrases go together to form a ‘clause’. and sentences. b. The birds are singing because [the weather is lovely today]. we observed that the study of English syntax is the study of rules which generate an infinite number of grammatical sentences. . The weather is lovely today. clauses. but which may be part of another clause. c. 11 . If [the weather is lovely today] then we will go out.. . word . with one clause embedded inside another: (2) a. . This chapter deals with what kind of combinatorial rules English employs in forming these phrases. in particular. or is part of. I am hoping that [the weather is lovely today].. phrase . . One simple mechanism we recognize is that in forming grammatical sentences. rvv rr vvv rr vv rr . as a subordinate or adverbial clause. Typically we use the term ‘clause’ to refer to a complete sentence-like unit.

For example. and syntactic function.1 Lexical Categories Determining the Lexical Categories The basic units of syntax are words. A: f.2. The first question is then what kinds of words (also known as parts of speech. why do we categorize book as a noun. For example. At first glance. bother. we could have the following rough semantic criteria for N (noun). how do we know that virtue is a noun. but kick as a verb? To make it more difficult. N: c. and maybe a few others? Most of us would not be able to come up with simple definitions to explain the categorization of words. happiness. N: b. but they are always nouns. For example. Let us check what each of these criteria means. words like sincerity. preposition. but do not involve any action. nouns allow the plural marking suffix -(e)s to be attached. and exist are verbs. location. Absence and loss are even harder cases. or grammatical categories) does English have? Are they simply noun. words like assassination and construction may refer to an action rather than an individual. Words like remain. that without is a preposition.2. A: referring to a property d. these notional definitions leave a great number of words unaccounted for. Adv: referring to the manner. it seems that words can be classified depending on their meaning. There are also many words whose semantic properties do not match the lexical category that they belong to. V: e. N: referring to an individual or entity b. time or frequency of an action Though such semantic bases can be used for many words. V: d. For instance. verb. A more reliable approach is to characterize words in terms of their forms and functions. The ‘form-based’ criteria look at the morphological form of the word in question: (4) a. or the possessive ’s. and pain do not simply denote any individual or entity. V: referring to an action c. V (verb). and how reliable each one is. whereas verbs can have the past tense -ed or the 3rd singular form -(e)s. Adjectives can take comparative and superlative 12 . A: + plural morpheme -(e)s + possessive ’s + past tense -ed or 3rd singular -(e)s + 3rd singular -(e)s + -er/est (or more/most) + -ly (to create an adverb) According to these frames. A (adjective).2 2. and that well is an adverb (in one meaning)? Words can be classified into different lexical categories according to three criteria: meaning. adjective. morphological form. and Adv (adverb): (3) a. or lexical categories. appear. adverb. where the word in question goes in the place indicated by .

Adv: fully. d. They have no b. devours. book. Adv. roughly only one lexical category can appear in each position: (7) a. As shown here. He walked right *very/*happy/*the wall. In addition to nouns like information and furniture that we presented in Chapter 1. They can . fullest. A. (11) a. etc. He treats John very *kind/*shame/*under. There are adjectives (such as absent and circular) that do not have comparative -er or superlative -est forms. He walked right . They read the big/new/interesting/scientific book. It turns out. N: trains. He treats John very nicely/badly/kindly. V: devoured. c. They can *happy/*down/*door/*very. diligently. b. b. (8) a. d. (10) a. verbs cannot have plural marking. (9) a. there are many nouns such as love and pain that do not have a plural form. The most reliable criterion in judging the lexical category of a word is based on its syntactic function or distributional possibilities. They read the *sing/*under/*very book. rooms. The morphological (form-based) criterion. They can sing/run/smile/stay/cry. b. that these morphological criteria are also only of limited value. They have no *went/*in/*old/*very/*and. c. They read the e. we can then assign a specific lexical category to these elements: 13 . etc. b. only a restricted set of lexical categories can occur in each position. He walked right into/on the wall. b. etc. He treats John very The categories that can go in the blanks are N. most careful. man’s. more careful.endings -er or -est. Let us try to determine what kind of lexical categories can occur in the following environments: (6) a. b. sister’s. . They have no TV/car/information/friend. etc. carefully. and P (preposition). or combine with the suffix -ly. however. The morphological properties of each lexical category cannot be overridden. laughed. nor can adjectives have tense marking. As can be seen in the data in (7). (5) shows some examples derived from these frames: (5) a. though reliable in many cases. due to their meanings. actors. laughs. V. A: fuller. clearly. the wall. is not a necessary and sufficient condition for determining the type of lexical categories.

14 . (17) a. I think that [learning English is not all that easy]. Here we see that only words like the. quantifiers. *[Some my jobs] are in jeopardy. and so forth. 1 These conjunctions are ‘coordinating conjunctions’ different from ‘subordinating conjunctions’ like when. . a. d. does English have other lexical categories? There are a few more. (13b) provides a frame for conjunctions (Conj) such as and. . but. Mary played the piano. . and so forth can occur here. John thinks The only words that can occur in the open slot in (13a) are words like the. Consider the following syntactic environments: (13) a. .(12) a. One clear piece of evidence for grouping these elements as the same category comes from the fact that they cannot occupy the same position at the same time: (15) a. friend. since. and for this reason. . *[The his jobs] are in jeopardy. I doubt learning English is not easy at all. kindly. . . so. e. into. Bill is honest. my. c. possessives. . which are determiner (Det). on. that. . Now consider the following examples: (16) a. Words like my and these or some and my cannot occur together. information. and so forth. Consider the following: (14) We found out that very lucrative jobs were in jeopardy. stay. . scientific. . . I am anxious Once again. car. and demonstratives all ‘determine’ the referential properties of jobs here. b. interesting. *[My these jobs] are in jeopardy. smile. run.1 In (13c). or. this. these. . N: TV. b. for. some. the department decided not to open that course this year. we can have the category we call ‘complementizer’. yet. here the word that – we return to these in (17) below. they are called determiners. the possible words that can occur in the specific slot in (17) are strictly limited. c. indicating that they compete with each other for just one structural position. . b. I think b. if. In addition to these basic lexical categories. These articles. new. Adv: nicely. Can we find any supporting evidence for such lexical categorizations? It is not so difficult to construct environments in which only these lexical elements appear. cry. those. A: big. you to study English grammar hard. The former conjoins two identical phrasal elements whereas the latter introduces a subordinating clause as in [Though students wanted to study English syntax]. c. under. c. badly. you can help me in understanding this. over. student hits the ball. . though. his. V: sing. P: in. John sang a song.

c. shall and must. there is clear evidence that these verbs are different from main verbs. I am anxious for [you to study English grammar hard]. 15 . b.b. Students objected to the teacher. to is a preposition. The term ‘complement’ refers to an obligatory dependent clause or phrase relative to a head. b. Note the change in the main verb form in (20b) when the negation is added: (20) a. too. which we call the ‘base’ form. drink beer last night. John c. and Mary The words that can appear in the blanks are neither main verbs nor adjectives. *I am anxious for [you should study English grammar hard]. For example. . Students intended to surprise the teacher. b. He did not leave. There is also one type of to which is auxiliary-like. They introduce a complement clause. and may be sensitive to the tense of that clause. In English. if . marked above by the square brackets. that and if introduce or combine with a tensed sentence (present or past tense). The italicized words here are different from the other lexical categories that we have seen so far. and we call them auxiliary verbs (Aux). Consider the examples in (21) and (22): (21) a. The auxiliary verb appears in front of the main verb. but rather words like will. John will study syntax. whereas for requires an infinitival clause marked with to. There are only a few complementizers in English (that. can. not leave. I doubt if [you can help me in understanding this]. *I doubt if [you to help me in understanding this]. We cannot disturb these relationships: (18) a. John leave for Seoul tomorrow? d. but nevertheless they have their own lexical category. and whether). Students sent letters to the teacher. we treat the infinitival marker 2 See Chapter 4 for a fuller discussion of ‘head’ and ‘complement’. Students wanted to write a letter. But how about the infinitival marker to in (21). *I think that [learning English to be not all that easy]. for. Now consider the following environments: (19) a. (22) a. c. as opposed to an infinitive. followed by a base verb form? What lexical category does it belong to? Though the detailed properties of auxiliary verbs will not be discussed until Chapter 8. It is easy to see that in (22). which is typically in its citation form. He left. b. A tensed clause is known as a ‘finite’ clause. John b.2 The italicized elements in (18) introduce a clausal complement and are consequently known as ‘complementizers’ (abbreviated as ‘C’).

(preposition) 3 See Chapter 8 for detailed discussion on the ellipsis. She thought it was likely [that everyone *to/might/would fit into the car]. but I know I should. She thought it was easy [for everyone to/*might/*would fit into the car]. b. *The umpire fell the deck off . b. I don’t really want to [go to the dentist’s]. The umpire fell off it. *The umpire fell it off . The umpire called the game off . b. they both can license the ellipsis of its VP complement. It is crucial for John to show an interest. (particle) b. b. In (24). *The students looked the high stairs up (from the floor).to as an auxiliary verb.3 Another property to shares with other auxiliary verbs like will is that it requires a base verb to follow. b. Most auxiliary verbs are actually finite (tensed) forms which therefore pattern with that in a finite clause. Such distributional possibilities cannot be observed with true prepositions: (28) a. We can also find differences between particles and prepositions in combination with an object pronoun: (30) a. The umpire called off the game. (31) a. to and should introduce the clause and determines the tenseness of the clause. b. The two boys looked the word up. The two boys looked up the word. The umpire called it off . For example. while the infinitival clause introduced by for is only compatible with to: (25) a. (24) a. The two boys looked up the high stairs (from the floor). b. b. Words like off and up here behave differently from prepositions. In (23). The umpire fell off the deck. 16 . (29) a. but I just don’t want to. I know I should [go to the dentist’s]. in that they can occur after the object. the ‘particles’ (Part). (27) a. *The umpire called off it. we can observe that to behaves like an auxiliary verb should: (23) a. It is crucial that John should show an interest. illustrated in (26): (26) a. Finally. there is one remaining category we need to consider.

. 17 . possibly with an A in front of either N. string. d. b. and N. student. . the. . . man. We assume a lexicon.. . piano. sang. The happy student played the piano. the main goal of syntax is building a grammar that can generate an infinite set of well-formed. grammatical English sentences.. By inserting lexical items into the appropriate pre-terminal nodes in the structure. So this rule characterizes any sentence which consists of a Det. in that order. .. Det: a. that.. Such contrasts between prepositions and particles give us ample reason to introduce another lexical category Part (particle) which is differentiated from P (preposition). kind. A: handsome.. not all of which describe a possible real-world situation: (36) a. but not the particle in (30b). . A tall boy threw the ball. . That ball hit a student. d. a list of categorized words. V: kicked.. c.3 Grammar with Lexical Categories As noted in Chapter 1. . . To start off. . A man kicked the ball. happy. 2. we will use the examples in (32): (32) a. . this. tall. b. played. Let us see what kind of grammar we can develop now that we have lexical categories. V. . N. . N: ball. The cat chased the long string. to be part of the grammar along with the rule in (33): (35) a. threw. we can set up a grammar rule for sentence (S) like the following: (33) S → Det (A) N V Det (A) N The rule tells us what S can consist of: it must contain the items mentioned. .. where the labels above .. we can generate grammatical examples like those (32) as well as those like the following. Det. . long. Given only the lexical categories that we have identified so far. . c.The pronoun it can naturally follow the preposition as in (31b). hit. we will see more tests to differentiate these two types of word. In the next section. are. except that those which are in parentheses are optional... We can represent the core items in a tree structure as in (34): (34) S ‚ yi‚ lll i ‚‚ llyyy iii‚‚‚‚ ll y ii ‚‚‚ lll yy ‚‚ i lll y Det N V Det N . .

the mother. A grammar using only lexical categories can be specified to generate an infinite number of well-formed English sentences. Exercise 9 of this chapter and the discussion of ‘selectional restrictions’ in Chapter 4. The tall man kicked the ball. [The mother of the boy] and [the girl] are arriving soon. handsome man kicked the ball. b. thereby generating sentences like (38). Note that any anomalous example can be preceded by the statement “Now. Note that the parentheses around ‘A’ in (34) are no longer necessary in this instance.4 Notice that even this simple grammar rule can easily extend to generate an infinite number of English sentences by allowing iteration of the A:5 (37) S → Det A∗ N V Det A∗ N The operator allows us to repeat any number of As. b. The tall. implying that syntax is rather ‘autonomous’ from semantics. That ball sang a student. indicating ungrammaticality. kind. (38) a. Such examples are all syntactically well-formed. happy. Now.b. happy man sang a song. It should not be confused with the * prefixed to a linguistic example. The mother of the boy and the girl are arriving soon. . here’s something hard to imagine: . The mother of the boy and the girl is arriving soon. even if semantically in some cases. 5 This 4 See 18 . The tall. The piano played a song. The different groupings shown by the brackets indicate who is arriving: in (41a). Why do the verbs in these two sentences have different agreement patterns? Our intuitions tell us that the answer lies in two different possibilities for grouping the words: (41) a. handsome man kicked the ball. consider the following examples: (42) a. c. d. and is a notation meaning ‘zero to infinitely many’ occurrences. John saw the man with a telescope. The piano kicked a student. [The mother of [the boy and the girl]] is arriving soon. iteration operator ∗ is called the ‘Kleene Star Operator’. c. ”. happy. . but it nevertheless misses a great deal of basic properties that we can observe. this simple grammar cannot capture the agreement facts seen in examples like the following: (40) a. for the Kleene Star operator means any number including zero. b. b. I like chocolate cakes and pies. For example. One could even generate a sentence like (39): (39) The happy. The grouping of words into larger phrasal units which we call constituents provides the first step in understanding the agreement facts in (41). happy. happy. while in (41b) it is both the mother and the girl.

c. and need to consider how phrases may be formed. . in addition to semantic coherence. With no difficulty. (46) The policeman met several young students in the park last night. ’. [The student] [enjoyed [his English syntax class last semester]]. It was [the policeman] that met several young students in the park last night. c. These sentences have different meanings depending on how we group the words. but not those in (45b) or (45c): (45) a. We need more intelligent leaders. It was [in the park] that the policeman met several young students last night. forms the basis for our intuitions of constituency? Are there clear syntactic or distributional tests which demonstrate the appropriate grouping of words or specific constituencies? There are certain salient syntactic phenomena which refer directly to constituents or phrases. what constituents would you come up with? (44) The student enjoyed his English syntax class last semester. 19 . can provide us with simple evidence for the existence of phrasal units. If you were asked to group the words in (44) into phrases. b. Cleft: The cleft construction. (John used the telescope) Even these very cursory observations indicate that a grammar with only lexical categories is not adequate for describing syntax. For instance. John [[saw the man] with a telescope]. [The student] [[enjoyed his English] [syntax class last semester]]. . It was [last night] that the policeman met several young students in the park. For example. It was [several young students] that the policeman met in the park last night.4 Phrasal Categories In addition to the agreement and ambiguity facts. 2. which places an emphasized or focused element in the X position in the pattern ‘It is/was X that . d. (the man had the telescope) b. we can cleft almost all the constituents we can get from the above sentence: (47) a. b. In addition.c. we need a notion of ‘constituent’. think about how many different cleft sentences we can form from (46). [The] [student enjoyed] [his English syntax class] [last semester]. (42a) will have the following two different constituent structures: (43) a. Perhaps most of us would intuitively assign the structure given in (45a). our intuitions may also lead us to hypothesize constituency. What kind of knowledge. John saw [the man with a telescope]. grouping certain words together.

and that put old books in the box is also a (larger) constituent. (50) A: Who(m) did the policeman meet in the park? B: Several young students. 20 . . B: *Old books in the box. the answer can either be a full sentence or a fragment. B: *Put in the box. when. . John looked up the meaning of ‘chanson’. This stand-alone fragment is a constituent: (49) A: Where did the policeman meet several young students? B: In the park. Consider the similar-looking examples in (55). we cannot cleft sequences that not form constituents:6 (48) a. night here cannot be clefted for independent reasons (see Chapter 12). including looked and up: (55) a. how. B: *Old books in the box. c. John looked up the inside of the chimney. b. which involve a wh-word such as who. *It was [in the park last night] that the policeman met several young students. the tests here will show that old books and in the box are constituents. we will illustrate with example (51): (51) John put old books in the box. B: Put old books in the box. Constituent Questions and Stand-Alone Test: Further support for the existence of phrasal categories can be found in the answers to ‘constituent questions’. where. The test is also sensitive to the difference between particles and prepositions. 6 The verb phrase constituent met . b. For any given wh-question. Overall. This kind of test can be of use in determining constituents. (54) A: What did you do? B: *Put old books. *It was [several young students in] that the policeman met the park last night. (53) A: Where did you put the book? B: In the box. Are either old books in the box or put old books in the box a constituent? Are there smaller constituents? The wh-question tests can provide some answers: (52) A: What did you put in your box? B: Old books.However. *It was [the policeman met] that several young students in the park last night.

John might [go home]. 21 . so. like most languages. We can again apply the wh-question test: (56) A: What did he look up? B: The inside of the chimney. the man who is standing by the door in (59a) can be ‘substituted’ by the pronoun he in (59b). referring to a VP. B: The meaning of ‘chanson’. c. *John asked me to put the clothes in the cupboard. (59) a. What the contrasts here show is that up forms a constituent with the inside of the chimney in (55a) whereas it does not with the meaning of ‘chanson’ in (55b). but I did so [=put the clothes] in the suitcase. so might Bill. and to annoy him I really stuffed them there [there=in the cupboard]. For instance. (60) a. B: *Up the meaning of ‘chanson’. What do you think the man who is standing by the door is doing now? b.The examples differ. John might [pass the exam]. as. (58) A: Up what did he look? B: The inside of the chimney. John asked me to put the clothes in the cupboard. has a system for referring back to individuals or entities mentioned by the use of pronouns. so in (61c). refers only part of a constituent put the clothes. John asked me to put the clothes in the cupboard. however. d. (57) A: Where did he look? B: Up the inside of the chimney. making it unacceptable. and to annoy him I stuffed them there [them=the clothes]. If John can [speak French fluently] – which we all know he can – we will have no problems. Both the pronoun there and them refer to a constituent. which also refer back to other constituents. Substitution by a Pronoun: English. b. as to whether up forms a constituent with the following material or not. A pronoun cannot be used to refer back to something that is not a constituent: (61) a. What do you think he is doing now? There are other pronouns such as there. However. and as might Bill. c. Have you been [to Seoul]? I have never been there. b. B: *The meaning of ‘chanson’. and which.

Even though such syntactic constituent tests are limited in certain cases. 2. the students who came from Seoul. VP. an obligatory N. the students from Seoul.Coordination: Another commonly-used test is coordination. *Lee went [to the store] and [crazy]. This leads us to posit the following rule:7 (66) NP → (Det) A* N (PP/S) This rule characterizes a phrase. The children were neither [in their rooms] nor [on the porch]. The rule indicates that an NP can consist of an optional Det. we need to make A* as AP*. She was [poor] but [quite happy]. the tall students. The girls [played in the water] and [swam under the bridge]. we just use the former in the rule. If we try to coordinate unlike constituents.5. PP. we can see that each includes at least an N and forms an NP (noun phrase). etc. Many people drink [beer] or [wine]. we use distributional evidence to classify each type. As before. The kinds of expression that do appear here include: (65) Mary. b. The expressions that can occur in the blank position here are once again limited. These options in the NP rule can be represented in a tree structure: relative clause who came from Seoul is kind of modifying sentence (S). If we look into the sub-constituents of these expressions. and hence we have phrases such as NP. and so on. the students. any number of optional A. you. license an example like the very tall man. the results are typically ungrammatical.5 Phrase Structure Rules We have seen evidence for the existence of phrasal categories. they are often adopted in determining the constituent of given expressions. I.1 NP: Noun Phrase Consider (64): (64) [liked ice cream]. See Chapter 11.8 The slash indicates different options for the same place in the linear order. For simplicity. and is one instance of a phrase structure rule (PS rule). students. 2. and then specify rules to account for the distributions we have observed. c. We say that phrases are projected from lexical categories. and each conjunct is typically the same kind of constituent as the other conjuncts: (62) a. 8 To 7 The 22 . (63) a. *Mary waited [for the bus] and [to go home]. and then an optional PP or a modifying S. Words and phrases can be coordinated by conjunctions. d. b.

lifted 50 pounds. V projects a VP. These phrases all have a V as their head – as projections of V. (69) The student . b. . Hence the subject of the sentence She sings will be an NP. they form VP. *the leave meeting. loved music. even though that NP consists only of a pronoun. Consider the following examples: (73) a. warned us that storms were coming. a single noun such as Mary.... etc. . and the rule will not generate the following NPs: (68) *the whistle tune.. and the VP must be finite (present or past tense). The rule thus does not generate ill-formed VPs such as these: (72) *leave the meeting sing. *leave on time the meeting. we will have well-formed NPs. (70) snored. *The monkeys proud of their leader. (70) lists just a few of the possible phrases that can occur in the underlined position. . We can also observe that the presence of a VP is essential in forming a grammatical S. b. The monkeys approved of their leader. (74) a. . . *the my dog. The men practice medicine. . ran. which can be represented as a PS rule: 23 . you. 2.5. Once we insert appropriate expressions into the pre-terminal nodes..2 VP: Verb Phrase Just as N projects an NP. *The monkey eager to leave the meeting. or students can constitute an NP by itself.(67) NP i lll lyyyy iiii l lllyy ii ll y i lll y (Det) A* N (PP/S) . sang. .. (75) a. b... . thought Tom is honest. *the easily student. The monkey wants to leave the meeting. A simple test environment for VP is given in (69). *The men doctors of medicine. . VP can be characterized by the rule in (71). These examples show us that an English well-formed sentence consists of an NP and a (finite) VP. to a first level of analysis: (71) VP → V (NP) (PP/S) This simple VP rule says that a VP can consist of an obligatory V followed by an optional NP and then any number of PPs or an S. One important point is that as only N is obligatory in NP. walked the dog through the park.

the adverb loudly and the PP in the class are modifying the preceding VP. c. to feel happy.(76) S → NP VP We thus have the rule that English sentences are composed of an NP and a VP. the parts after the auxiliary verbs in (78b) and (78c) are themselves regular VPs. . One more aspect to the structure of VP involves the presence of auxiliary verbs. . . but how about those in (78b) and (78c)? These are also VPs. . b. . John [[read the book] loudly]. to run. In the full grammar we will consider to and can and so on as auxiliary verbs. b. can run. to study English syntax. We have seen that the expressions in (78a) all form VPs. use a triangle when we need not represent the internal structure of a phrase. (78) a. the phrases in (78a) and (78b) can occur in (77a) whereas those in (78c) can appear in (77b). For example. will feel happy. Then all auxiliary verbs are simply introduced by a second VP rule:9 (79) VP → V[AUX +] VP One more important VP structure involves the VP modified by an adverb or a PP: (80) a. feel happy. Think of continuations for the fragments in (77): (77) a. we need the PS rule in (81): (81) VP → VP Adv/PP This rule. . run. The students want . . together with (76) will allow the following structure for (80b):10 (82) S kƒƒƒƒƒ ƒƒƒƒ kkkk kk ƒƒ kkkk NP VP kƒƒƒƒƒ xpp ƒƒƒƒ kkkk xx ppp x kkk ƒƒ p kkk xx The teacher VP PP w i yy iii qqq wwwww qq ii yy ww yy qqq met his students in the class 9 The 10 We detailed discussion of English auxiliary verbs is found in Chapter 8. The students . b. In fact. . 24 . . the precise structural counterpart of the traditional ideas of a sentence being ‘a subject and predicate’ or ‘a noun and a verb’. must study English syntax. with a feature specification [AUX +] to distinguish them from regular verbs. To form such VPs. The teacher [[met his students] in the class]. which happen to contain more than one V. In such examples. study English syntax.

etc. Looking into the constituents of these. b. an NP the student or AP happy cannot occur in these syntactic positions. carefully. but not with a VP. etc.3 AP: Adjective Phrase The most common environment where an adjective phrase (AP) occurs is in ‘linking verb’ constructions as in (83): (83) John feels . For example.4 AdvP: Adverb Phrase Another phrasal syntactic category is adverb phrase (AdvP). proud to be his student. the AdvP rule can be given as follows: (91) AdvP → (AdvP) Adv 25 . uncomfortable. Since these all include an adjective (A). *The monkeys seem [want to leave the meeting]. He treated her very . The verb sounded requires an AP to be followed. (89) soundly. adjectives. These examples tell us that the verb seem combines with an AP. proud that he passed the exam. but in (86c) we have no AP. we can formulate the following simple PS rule for the AP: (85) AP → A (PP/VP/S) This simple AP rule can easily explain the following: (86) a. b. John seems [certain about the bananas]. c. . Expressions like those in (84) can occur in the blank space here: (84) happy. very soundly. The monkeys seem [eager to leave the meeting].5.2. In addition. clearly. and adverbs themselves. b. *John seems [know about the bananas]. and they can all occur in principle in the following environments: (90) a. . (88) a. proud of her. terrified. very slowly. we can safely conclude that they all form an AP. as exemplified in (89). sad. He behaved very c. John sounded happy/uncomfortable/terrified/proud of her. Based on what we have seen so far. They worded the sentence very Phrases other than an AdvP cannot appear here.5. John sounded *happily/*very/*the student/*in the park. almost certainly. These phrases are often used to modify verbs. b. observe the contrasts in the following examples: (87) a. John felt proud that his son won the game. 2. extremely. well.

NP → (Det) A* N (PP/S) c. (92) from Seoul. From the examples in (92). The intensifiers straight and right can occur neither with an AP nor with an AdvP: (95) a. S → NP VP b. some of which we have already seen. . The squirrel ran straight/right up the tree. . *into sing a song. Another direction is to take the qualifier categorically as an adverb carrying the feature QUAL while allowing only such adverbs to modify a PP. *The squirrel ran straight/right quickly. How much further do we get with a grammar which includes phrases? A set of PS rules. *with happily. They stayed in the hotel. *The squirrel is straight/right angry. . is given in (98). 2. John came from Seoul.12 (98) a. b. 11 Depending 26 . b. d. in the hotel. One clear case in which only a PP can appear is the following: (94) The squirrel ran straight/right . The fly fell into the soup.6 Grammar with Phrases We have seen earlier that the grammar with just lexical categories is not adequate for capturing the basic properties of the language. regardless of any other contextual restrictions. They put the book in the box. We cannot construct unacceptable PPs like the following: (97) *in angry. we may need to extend this PP rule as “PP → (Qual) P NP” so that the P may be preceded by an optional qualifier like right or straight. 12 The grammar consisting of such form of rules is often called a ‘Context Free Grammar’. this means that we need to introduce another lexical category ‘Qual’. under the table.2. PPs like those in (92). c. as each rule may apply any time its environment is satisfied. However. we can deduce the following general rule for forming a PP:11 (96) PP → P NP The rule states that a PP consists of a P followed by an NP.5 PP: Preposition Phrase Another major phrasal category is preposition phrase (PP). VP → V (NP) (PP/S/VP) d.5. generally consist of a preposition plus an NP. into the soup. c. with John and his dog. etc. AP → A (PP/S) on how we treat the qualifier straight and right. These PPs can appear in a wide range of environments: (93) a. in the box.

. b. any. . we need to refer to the notion of ‘selectional restrictions’ (see Chapter 7).. his. c. yellow. let us assume that we have the following lexical entries: (100) a. . hat. this.. etc.. her. chased. . 27 . no.e. With the structural possibilities shown here.. There are also other ways of generating an infinite number of grammatical sentences.. Det ttt tt tttt tt t tt . sang. etc. any number of As.. cat. man. Det: a. AdvP → (AdvP) Adv f. met..... There are several ways to generate an infinite number of sentences with this kind of grammar. friend. an. some.. big. Look at the following two PS rules from (98) again: 13 The grammar still generates semantically anomalous examples like#The desk believed a man or#A man sang her hat. This handsome man chased a dog. which. Of the possible tree structures that these rules can generate. and an NP can be made up of a Det. . . For such semantically distorted examples. A man kicked that ball. we are able to generate various sentences like those in (101):13 (101) a. d. . ball. woman. . PP → P NP The rules say that a sentence is the combination of NP and VP. the following is one example: (99) S jjj„„„„„„„ „„„„ jjjj j „ jjjj VP NP jˆˆˆˆˆˆˆ ttt ˆˆˆˆˆ jjjj tt tttt ˆˆˆˆˆ tt jjjj t tt jjj Det A N V NP PP t t tt tttt tt tttt tt tt tt tt t t tt tt N P NP . tall. etc.. an obligatory N. A: handsome. fat. Det N ... dog. thinks. d.. large. believed. assumes etc.g. Inserting these elements in the appropriate pre-terminal nodes (the places with dots) in (99). adjective) infinitely. b. N: book. imagines. one simple way is to repeat a category (e. and any number of PPs. dirty. and so on. That tall woman chased a cat. John. As we have seen before. His friend kicked a ball. that. . V: kicked. . c. .

and then vice versa: (103) 7654 0123 S v rr vvv r vv rr v rr 89:. Jane imagines Bill claims John believes Mary thinks Tom is honest. S → NP VP b. Bill claims John believes Mary thinks Tom is honest. the maximum number of auxiliaries in a single English clause is 3. we can ‘recursively’ apply the two rules. 28 . John believes NP VP rvv rr vvv r r vv rr V S N ttt tt tttt tt t tt Mary thinks Tom is honest It is not difficult to expand this sentence by applying the two rules again and again: (104) a. b. As noted before in (79). in the sense that one can feed the other. There is no limit to this kind of recursive application of PS rules: it proves that this kind of grammar can generate an infinite number of grammatical sentences. VP → V S As we show in the following tree structure. an auxiliary verb forms a larger VP after combining with a VP: (105) S v rr vvv rr vv r v rr 0123 7654 NP VP rvv rr vvv rr vv rr 0123 7654 N V[AUX +] VP v rr vvv rr vv r v rr They will V NP study English syntax This means that we will also have a recursive structure like the following:14 14 Due to the limited number of auxiliary verbs. ?>=< NP VP rvv rr vvv rr vv rr 0123 7654 N V S rvv rr vvv rr vv rr ?>=< 89:. One structure which can be also recursive involves sentences involving auxiliary verbs. and restrictions on their cooccurrence. See Chapter 8.(102) a.

(106) S € nnn €€€€€ n €€€ nnn nn 89:. One merit of such hierarchical structural properties is that they enable us to represent the structural ambiguities of sentences we have seen earlier in (42). we will have different hierarchical tree structures. 15 One can draw a slight different structure for (108b) with the introduction of the rule ‘NP → NP PP’. Chocolate cakes and pies are my favorite desserts. 29 . b. ?>=< NP VP € nnn €€€€€ nn €€€ nn nn 0123 7654 N V [AUX +] VP € nnn €€€€€ nn €€€ nn nn 89:. for the sentences here. Depending on which PS rules we apply. VP n‡‡‡‡‡‡‡ ‡‡‡‡‡ nnn nn ‡ n VP PP € rrvvvvv nnn €€€€€ v rrr € nnn n r V NP with the toy h zz hhh h zz z hit the child VP jj„„„„„„ „„„ jjjj jjj V NP j„„ jjjj „„„„„„„ jjj j hit Det N PP v rrr vvvvv rr r the child with the toy b. Consider the possible partial structures of (107a) which the grammar can generate:15 (108) a. The little boy hit the child with a toy. Let us look at more examples: (107) a. where parts are assembled into sub-structures of the whole. ?>=< They will V[AUX +] VP € nnn €€€€€ nn €€€ nn nn have V[AUX +] VP kkƒƒƒƒƒƒ ƒƒƒƒ kkkk k ƒ kkkk been studying English syntax Another important property that PS rules bring us is the ability to make reference to hierarchical structures within given sentences.

16 Different from the Kleene star operator ∗ .g. By altering the position of off . for phrasal conjunction. we can determine that off in (112a) is a preposition whereas off in (112b) is a particle: (113) a. b. The structural differences induced by the PS rules directly represent these meaning differences. This in turn means that off in (112a) is a preposition. where XP is any phrase in the grammar. PP ggg‡‡‡‡‡‡‡‡ ‡‡‡‡‡ ggggg ggggg PP Conj PP y w ooo yyyyy qqq wwwww oo qq y oo q in their rooms or on the porch *PP gggg‡‡‡‡‡‡‡‡‡ ‡‡‡‡ ggggg gggg Conj AP PP tttt ×UU ttt ttt ×× UU t × t to the store and crazy b. it modifies the whole VP phrase whereas (108b) modifies just the noun child. whereas off in (112b) is a particle that forms no constituent with the following NP the customers. We have noted that English allows two alike categories to be coordinated. we can easily show why examples like the following are not grammatical: (109) a. John suddenly got off the bus. call off ) constructions and prepositional verb constructions (e. we will then allow (111a) but not (111b): (111) a. This can be written as a PS rule. *Lee went to the store and crazy. *The children were in their rooms or happily. b. rely on). some of whose properties we have seen earlier. Unlike categories such as PP and AP may not be coordinated. 30 . In addition. Consider a representative pair of contrasting examples: (112) a. *John suddenly got the bus off. b.16 (110) XP → XP+ Conj XP The ‘coordination’ rule says two identical XP categories can be coordinated and form the same category XP.g. Applying this PS rule.The structures clearly indicate what with the toy modifies: in (108a). forming a PP with the following NP. John suddenly put the customers off. John suddenly put off the customers. The PS rules further allow us to represent the difference between phrasal verb (e. the plus operator + here means the XP here occurs at least once.

This in turn means that in addition to the PP formation rule. we have not discussed the treatment of agreement with PS rules. VP j„„ jjjj „„„„„„„ jjj j V PP j„„ jjjj „„„„„„„ jj jj get P NP c c  cc  c  off the bus VP … iiii …………………… iiii …… iii V Part NP ppxxxxx xx ppp pp put off the customers VP iii…………………… iiii ……… iiii V NP Part pxx ppp xxxxx ppp put the customers off b. the particle does not form a constituent with the following or preceding NP whereas the preposition does form a constituent with it.17 This motivates the introduction of phrases into the grammar. we then can easily represent the differences of these grammatical sentences (112a). (112b) and (113b) in tree structures: (115) a. but also account for some fundamental properties. 17 In this chapter. 31 . we have seen that a grammar with lexical categories can not only generate an infinite number of grammatical English sentences. VP → V PP Equipped with these rules. Chapter 6 discusses the subjectverb agreement in detail. As represented here. In summary. such as agreement and constituency. VP → V Part NP b. VP → V NP Part c. c. the grammar needs to introduce the following VP rules: (114) a.

Jane stood aside to let her pass. d. Price quotes on selected categories will be sent out upon request.7 Exercises 1. She turned down his offer. (i) a. (ii) a. Consider the following data carefully and describe the similarities and differences among that. g. b. f. 3. k. I forgot to return the book that I borrowed from the teacher. (ii) a.g. b. He intended to will the large amount of money to Frank. 32 . 2. (iii) a. first compare that and for and then see how these two are different from if and whether. He was arrested for being drunk. If students study hard. semantic. Check if the italic parts form a constituent or not. (vi) a. (i) a. John put a book on the table. using at least two constituenthood tests (e. We believe that he is quite reasonable. h. you arrange to have it. In so doing. (iii) a. c. I am anxious for you to arrive on time. His second book came out earlier this year and became an instant bestseller. I wonder whether/if you’d be kind enough to give us information. and syntactic) to provide the evidence for your answer and state which one is the most reliable one. e. John bought a book on the table. for..). He has a rail pass that’s right for you. Also provide tree structures for each of the following examples. He looked at a book about swimming. He talked to a girl about swimming. *I am anxious for you should arrive on time. j. b.2. b. When you book something such as a hotel room. if and whether. i. No doubt that he was forced to leave his family against his will. I don’t know whether/if I should agree. most teachers expect their students to study hard. I am anxious that you should arrive on time. pronoun substitution. In doing so. (i) a. *I am anxious that you to arrive on time. It is important for us to spend time with children. stand-alone. cleft. Determine the lexical category of the italicized words in the following. b. Whether they say it or not. l. I think that person we met last week is insane. She turned down the side street b. b. etc. teachers will be happy. use the three criteria (morphological. b.

Provide tree structures for each of the following sentences and see if there are any new PS rules that we need to add. e. *Lee went to the store and crazy.. John sang a song and danced to the music. 7. by the park. d. 6. b. As part of the exercise. I don’t know the people present. (i) Different languages may have different lexical categories. or they might associate different properties to the same one. John wants to study linguistics in near future. I saw him leaving the main building. He took Masako to the school by the park. presented earlier in this chapter. d. d. please use triangles. Japanese. pay attention to the position of modifiers like proudly. first draw structure for each example and try to determine the applicability of the the PS rules such as the coordination rule in (110). 5. b. (i) a. Korean. e. (i) a. e. 18 For 33 . (iv) a. consider related potential interpretations like Please put that book on my desk. Provide a tree structure for each of the following sentences and suggest what kind of VP rules are necessary. *Could you turn off the fire and on the light? *A nuclear explosion would wipe out plant life and out animal life. I consider Telma the best candidate. *I know the truth and that you are innocent. For example. f. Time flies like an arrow. to help tease out the ambiguity. If there are any places you cannot assign structures. 4.c. to supplement those we covered in this chapter. b. i-e. Japanese has two classes of adjectives whereas English has one. and That report on our webpage alleges that it does not function well. c. *He ran down the road and down the President. c. Spanish uses adjectives almost interchangeably as nouns while English cannot. He talked with a girl about swimming. h. That Louise had abandoned the project surprised everyone. In doing so. I know you like the back of my hand. I forgot how good beer tastes. Represent the structural ambiguities by providing different tree structures for each string of words. c. John called the president a fool. John refused the offer proudly. and so forth.18 (i) a. Explain why the examples in (i) are ungrammatical. I need to have that report on our webpage by tomorrow. g. b. I saw that gas can explode. They told Angelica to arrive early for the award. Each of the following sentences is structurally ambiguous – it has at least two structures.

Many linguists argue that the formal distinctions between parts of speech must be made within the framework of a specific language or language family. etc. many languages don’t have a distinction between adjectives and adverbs. and should not be carried over to other languages or language families.19 19 Adapted from http://en.and Chinese have measure words while European languages have nothing resembling them.wikipedia. or adjectives and nouns.org/wiki/Part_of_speech 34 .

and Semantic Roles 3. Constituents can be considered in terms of conceptual notions of semantic roles such as agent. For example. we consider the entire VP to be a PRED (predicate) which describes a property of the subject. the monkey is an NP in terms of its syntactic form. V.1 Introduction In the previous chapter. Grammatical Functions. . [S [NP The monkey] [VP scratched [NP a boy] [PP on Monday]]]. . The NP a boy is the OBJ (object) while the verb scratched functions as a predicator. NP. There are other dimensions of the analysis of sentences. 35 . b. AP. we analyzed English sentences with PS rules. VP. Syntactic categories: N. For example. but is the SUBJ (subject) in terms of its grammatical function.3 Syntactic Forms. MOD (Modifier). . [S [SUBJ The monkey] [PRED scratched [OBJ a boy] [MOD on Monday]]]. location. PRED (Predicate). . As shown here. one such way is using the notion of grammatical functions such as subject and object: (1) a. P. . Grammatical functions: SUBJ (Subject). OBJ (Object). such PS rules allow us to represent the constituent structure of a given sentence in terms of lexical and phrasal syntactic categories. but serves as a MOD (modifier) here. A. . On Monday is a PP in terms of its syntactic category. the PS rule ‘S → NP VP’ represents the basic rule for forming well-formed English sentences. We also can represent sentence structure in terms of semantic roles. consider one simple sentence: (2) The monkey scratched a boy on Monday. b. The notions such as SUBJ. As we have seen. patient. More importantly. OBJ and PRED represent the grammatical function each constituent plays in the given sentence. This sentence can be structurally represented in terms of either syntactic categories or grammatical functions as in the following: (3) a.

and semantic role).2 Grammatical Functions How can we identify the grammatical function of a given constituent? Several tests can be used to determine grammatical function. the subject is the cat. 1986). However. b.instrument. Consider the semantic roles of the NPs in the following two sentences:1 (4) a. For now. [[pat The monkey] [pred was tagged [loc in the wood] [agt by John]]]. [The rat] [devoured [the cat]]. and the object is the rat in (6a) but the cat in (6b).1 Subjects Consider the following pair of examples: (6) a. we will see many ways in which the three levels are not simply co-extensive. As noted here. b. We thus can assign the following semantic roles to each constituent of the examples: (5) a. in addition to agent and patient. and a patient is an object and an NP. 3. John is the agent and the monkey is the patient of the tagging event. This in turn means that in both cases. b. whereas the monkey has the semantic role of patient (pat). Both of these sentences describe a situation in which someone named John tagged a particular monkey. These two sentences have exactly the same words and have the same predicator devoured. Throughout this book we will see that English grammar refers to these three different levels of information (syntactic category. we have predicate (pred) and locative (loc) which also express the semantic role that each phrase performs in the described situation. A semantic role denotes the underlying relationship that a participant has with the relation of the clause. this is not always so: 1 Semantic roles are also often called ‘thematic roles’ or ‘θ-roles (“theta roles”) in generative grammar (Chomsky 1982. even though their grammatical functions are different. In (6a). it may appear that they are equivalent classifications: for example. The most common structure for a sentence seems to be one in which the NP subject is the one who performs the action denoted by the verb (thus having the semantic role of agent). as we show here. as we get further into the details of the grammar. The monkey was tagged in the forest by John. 3. grammatical function. and the like. whereas in (6b) it is the rat. an agent is a subject and an NP. However.2. [[agt John] [pred tagged [pat the monkey] [loc in the forest]]]. John has the semantic role of agent (agt). and the main difference comes from what serves as subject or object with respect to the predicator. expressed by the main verb. John tagged the monkey in the forest. 36 . [The cat] [devoured [the rat]]. In this situation. and they interact with each other. Yet they are significantly different in meaning.

including all the chapters in the first section. a short question tagged onto the end of an utterance. Such facts show that we cannot rely on the semantic roles of agent for determining subjecthood. My brother wears a green overcoat. but not necessarily to the closest NP. number. She never writes/*write home. It rains. This teacher is a genius. The book. which applies only to the subject. and tornadoes respectively. The committee disliked her proposal. stinking. Agreement: The main verb of a sentence agrees with the subject in English: (8) a. 37 . c. The tornadoes that tear through this county every spring *is/are more than just a nuisance. b. they indicate stative descriptions or situations. isn’t she/*isn’t he? b. The kids have arrived safely. tag questions. More reliable tests for subjecthood come from syntactic tests such as agreement. The subject in each example is book. (11) a. The lady singing with a boy is a genius. rather. Our neighbor takes/*take his children to school in his car. or disliking one’s proposal are not agentive activities. d. is also a reliable subjecthood test: (10) a. simply being closer to the main verb does not entail subjecthood: (9) a. The pronoun she in (10a) shows us that lady is the head of the subject NP in that example. agreement shows us what the subject of the sentence is. even though there are nouns closer to the main verb. As we noted in Chapter 1. effectiveness. nor to the most topical one. Subject-auxiliary inversion: In forming questions and other sentence-types. Tag questions: A tag question. The effectiveness of teaching and learning *depend/depends on several factors. b. and gender – it refers back to the subject. the kids have arrived safely. and subject-auxiliary inversion. b. With their teacher. c. is/*are very interesting. This indicates that it is not simply the linear position of the NP that determines agreement.(7) a. English has subject-auxiliary inversion. The generalization is that a tag question must contain a pronoun which identifies the subject of the clause to which the tag is attached. raining. This car stinks. b. haven’t they/ *hasn’t he? The pronoun in the tag question agrees with the subject in person. c. and they in (10b) leads us to assign the same property to kids. Wearing a green overcoat. These books *saddens/sadden me.

38 . b. (12) a.2 Direct and Indirect Objects A direct object (DO) is canonically an NP. *A good friend is remained to me (by him). The teapot was broken by the child by accident. It could be more detrimental. In (15a) the dog is experiencing something. the formation of ‘Yes/No questions’ such as these involves the first tensed auxiliary verb moving across the subject: more formally. in which a notional direct object appears as subject. The sentences in (16) can be turned into passive sentences in (17): (16) a. His girlfriend bought this computer. undergoing the process denoted by the verb: (14) a. this is not a solid generalization. He remained a good friend to me. That silly fool broke the teapot. (17) a. b. *The student is belonged to by this item. This computer was bought for him by his girlfriend. Have the kids arrived safely? c.2. Once again. This item belongs to the student. The test comes from the fact that non-object NPs cannot be promoted to the subject: (18) a. What we can notice here is that the objects in (16) are ‘promoted’ to subject in the passive sentences. the data show us that we cannot identify the object based on semantic roles. His girlfriend bought this computer for him.c. Could it be more detrimental? As seen here. the auxiliary verb is inverted with respect to the subject. b. b. *Have in our class the kids arrived safely? Subject-auxiliary inversion provides another reliable subjecthood test. (19) a. The kids in our class have arrived safely. and in (15b) the thunder is somehow causing some feeling in the dog: (15) a. b. The dog fears [thunder]. However. This is not possible with a non-subject: (13) a. b. b. 3. hence the term ‘subject-auxiliary inversion’. The child broke the teapot by accident. Is this teacher a genius? b. A much more firm criterion is the syntactic construction of passivization. Thunder frightens [the dog]. The objects (OBJ) in (15a) and (15b) are not really affected by the action.

c. 39 . distinct from indirect objects. She was sent a review copy of the book (by the publisher). They elected Graham chairman. An indirect object (IO) is one which precedes a direct object (DO). recipient. or benefactive: (20) a. In this kind of example. This is my ultimate goal. *The best writer was considered Andrew. My mother baked [me] [a birthday cake]. My mother baked a cake for me. The CDs were given to the boys by John. b. IOs are NPs and have the semantic roles of goal. Michelle became an architect. b. even though they are NPs. (IO = goal) b. Note that sentences with the IO-DO order are different from those where the semantic role of the IO is expressed as an oblique PP. I threw [the puppy] [the ball].The objects that undergo passivization are direct objects. The boys were given the CDs (by John). Consider the following sentences: (24) a. b. passive has the property of making the IO into the subject. The difference between objects and predicative complements can also be seen in the following contrast: (27) a. *Chairman was elected Graham. (25) a.3 Predicative Complements There also are NPs which follow a verb but which do not behave as DOs or IOs. b. (IO = benefactive) In examples like (20). 3. (IO = recipient) c. The publisher sent a review copy of the book to her.2. b. b. (21) a. A review copy of the book was sent to her by the publisher. giving examples like the following: (23) a. c. I consider Andrew the best writer The italicized elements here are traditionally called ‘predicative complements’ in the sense that they function as the predicate of the subject or the object. as in (20). John gave [the boys] [the CDs]. John gave the CDs to the boys. following the DO (such a process is sometimes called dative alteration): (22) a. However. John made Kim a great doll. they do not passivize: (26) a. it is once again the DO which can be passivized. This nice cake was baked for me by my mother.

2. They would inform Mary of any success they have made. (28) a. The message was that you should come on time. John gave a book to the student. c. The PPs here. Since their presence is obligatory. The italicized complements function to predicate a property of the subject in (29) and of the object in (30). I consider him immoral. as in John made a great doll for Kim. 40 . If we think of part of the meaning informally. d. I regard Andrew as the best writer. John put books in the box. They spoil their kids rotten. John talked to Bill about the exam. c. The situation became terrible. John bought a book for the student. b. This map is what he wants. as oblique complements. I made Kim angry.b. 3. These italicized expressions are neither objects nor predicative complements. b. c. Even though the italicized expressions here are both NPs. Roughly speaking. b. which cannot be objects since they are not NPs. (30) a. for syntactic well-formedness. they are called oblique complements. and oblique phrases are typically expressed as PPs in English. only in the second example would we say that the final NP describes the NP Kim. (27)b: Kim = a great doctor In addition. whereas the NP a great doctor in (27b) cannot be an object: it serves as the predicate of the object Kim. most ditransitive verbs can also take oblique complements: (32) a. The NP a great doll in (27a) is the direct object. John made Kim a great doctor. She reminded him of the last time they met. b. ‘oblique’ contrasts with the ‘direct’ functions of subject and object. also do not serve as predicate of the subject or object – they relate directly to the verb. (27)a: Kim = a great doll b. As we have seen before. they function differently.4 Oblique Complements Consider now the italicized expressions in (31): (31) a. phrases other than NPs can serve as predicative complements: (29) a. d.

whereas the two complements a book and a record in (34a) cannot. and why) phrases. d. Unlike these COMPS. *Today was visited by my uncle. b. Of course. predicative complement. these are called complements (COMPS). and oblique complement all have one common property: they are all selected by the verb. time. (35) a. location. and typically they cannot be omitted. where. of the situations expressed by the given sentences – informally. suggesting that they cannot serve as objects. temporal adjuncts like several times and last year can be repeated. (34) a. among many other properties. c. and we view them as being present to ‘complement’ the verb to form a legitimate VP. One additional characteristic of modifiers is that they can be stacked up. there are expressions which do not complement the predicate in the same way. Let us see how we can analyze a simple sentence along these two dimensions: (36) S q†††††† †††† qqq q †††† qqq VP: PRED NP: SUBJ qww qqwwwww qqq wwwww www qqq q w qqq qqq A N V NP: w OBJ Det qqq wwwww qq ww qqq The little cat devoured Det N a 41 mouse . or reason. The bus stopped suddenly. IO. The italicized expressions here are all optional and function as modifiers (also called ‘adjuncts’ or ‘adverbial’ expressions). temporal adjuncts do not become the subject of a passive sentence.2. and which are truly optional: (33) a. They went to the theater in London. *John gave Tom [a book] [a record]. I saw this film [several times] [last year] [during the summer]. they are the (how.5 Modifiers The functions of DO. when. b. whereas complements cannot. Shakespeare wrote his plays a long time ago.3. As shown here. 3. He failed chemistry because he can’t understand it. These modifiers specify the manner. My uncle visited today. Hence. b.3 Form and Function Together We now can analyse each sentence in terms of grammatical functions as well as the structural constituents.

we will discuss in detail the properties of each phrase type here.As represented here.3 Every category in a given sentence has a grammatical function. 4 In due course. b. SUBJ and OBJ. 3 The phrase CP is headed by the complementizer that. 2A 42 . d. but there is no one-to-one mapping between category such as NP or CP and its possible grammatical function(s). but only a limited set of words are adverbs. They wondered [S what she did yesterday]. In English almost any kind of phrasal category can function as an adverbial element. (39) a. The VP as a whole functions as the predicate of the sentence. [PP Under the bed] is a safe place to hide. [VP Being honest] is not an easy task. [VP To finish this work on time] is almost unexpected. c. whereas ‘adverb’ only designates a part of speech. [NP The termites] destroyed the sand castle. whereas the cat is the subject of the lower clause. I sent [NP a surprise present] to John. The following data set shows us how different phrase types can function as SUBJ or OBJ:4 (38) a.2 Assigning grammatical functions within complex sentences is no different: (37) S iii………………… iiii ………… i … iiii NP: SUBJ VP: PRED it iiii tttt tt iiii iii N V CP: OBJ ii……………… iiii ………… iii …… iii S John believes C iitttt iiii tt ii t iiii that NP: SUBJ VP: PRED ttt ØTT tt tttt ØØ TTT tt t ØØ tt the cat V NP: OBJ devoured a mouse Each clause has its own SUBJ and PRED: John is the subject of the higher clause. the expressions the little cat and a mouse are both NPs. but they have different grammatical functions. We also can notice that there are two OBJs: the CP is the object of the higher clause whereas the NP is that of the lower clause. word of caution is in order here. We should not confuse the functional term ‘adverbial’ with the category term ‘adverb’. They believed [CP that everybody would pass the test]. The term ‘adverbial’ is used interchangeably with ‘adjunct’ or ‘modifier’. b. e. c. describing the property of the subject. [CP That John passed] surprised her.

4 Semantic Roles As noted before. didn’t it? b. or PP can function as a modifier: (42) a. VP. subject-verb agreement. As we go through this book. semantic roles were introduced as a way of classifying the arguments of predicators (mostly verbs and adjectives) into a closed set of participant types. The sentence (42a) will have the following structure: (43) S † hhhh ††††††† †††† hhhh hhh †† hhh VP: PRED NP: SUBJ w †† qqq †††††††† qqq wwwww †††† ww qqq qqq qq qq A N VP: PRED NP: MOD Det qww ~d ~ dd qqq wwwww ~ dd ~ qq d ~ w ~ qq The little cat V NP: ` OBJ last night ÒÒ ``` ` ÒÒ Ò devoured a mouse Here the expression last night is an adverbial NP in the sense that it is categorically an NP but functions as a modifier (adjunct) to the VP. we will see that the distinction between grammatical functions and categorical types is crucial in the understanding of English syntax. d. 3. Are you going on holiday before or after Easter? I prefer [PP after Easter]. c. [[To delay the march] and [to go ahead with it]] have/*has been argued by different people at different times. b. (41) a. [[That the march should go ahead] and [that it should be cancelled]] have/*has been argued by different people at different times. [To finish it on time] made quite a statement. As the examples in (38) and (39) show. not only NPs but also infinitival VPs and CPs can also function as SUBJ and OBJ. John studied hard [VP to pass the exam]. S. The little cat devoured a mouse [NP last night]. [That John passed] surprised her. but also phrases such as NP. John left [AdvP very early]. Even though we 43 . Not only AdvP. The following tag-question. (40) a. didn’t it? b. She disappeared [S when the main party arrived].d. as noted before. The same goes for MOD. and subjecthood tests show us that an infinitival VP and CP can function as the subject. e. John has been at Stanford [PP for four years].

John bought a lot of books for his sons. kick. (46) a. hear. John ate his noodle quietly. b. hit. b. smell. The student heard a strange sound.6 (45) a. Examples: object of kick. Examples: subject of promise. or as being in a state or position. • Theme: A participant which is characterized as changing its position or condition. free. die.cannot make any absolute generalizations about the relationship between grammatical functions and semantic roles. (44) a. buy. John promised Bill to leave tomorrow morning. b. hand. • Patient: A participant which the verb characterizes as having something happen to it. etc. indirect object of tell. Examples: oblique complement of make. 6 Patient 5 The 44 . • Goal: The one to which motion proceeds. etc. etc. John died last night. A boy hit the ball. possibly intentionally. cure. A list of the most relevant thematic roles and their associated properties is given below. etc. John made a doll for his son. hammer. A boy hit the ball. happen. b. give. Examples: subject of receive. etc. etc. subject of come. • Experiencer: A participant who is characterized as aware of something. John deprived his sons of game cards. • Source: The one from which motion proceeds. b. John told the rumor to his friend. John gave a book to the students. Examples: subject of perception verbs like feel. b. buy. hammer. (50) a. and as being affected by what happens to it. Examples: subject of eat.5 • Agent: A participant which the meaning of the verb specifies as doing or causing something. A smith hammered the metal. A smith hammered the metal. object of deprive. (49) a. • Benefactive: The entity that benefits from the action or event denoted by the predicator. b. The students felt comfortable in the class. the properties of semantic roles do interact in regular ways with certain grammatical constructions. definition of semantic roles given here is adopted from Dowty (1989). hit. Mary received an award from the department. Examples: direct object of give. etc. (47) a. c. see. (48) a. and theme are often unified into ‘undergoer’ in the sense that both the patient and theme individual undergo the action in question.

*There sang a man with a pipe. b. The problem is illustrated by the following simple examples: (56) a. *There dances a man with an umbrella. An important advantage of having such semantic roles available to us is that it allows us to capture the relationship between two related sentences. (52) a. wipe. own. b. c.7 For example. b. Even though the above two sentences have different syntactic structures. whereas verbs like sing. and the mouse is the patient. We thus can conjecture that there-constructions do not accept the verb whose subject carries an agent semantic role. (51) a. [agt The cat] chased [pat the mouse]. b. etc. The semantic roles also allow us to classify verbs into more fine-grained groups. John hit the ball with a bat. The government kept all the money. The difference can come from the semantic role of the postverbal NP. There arrived a tall. etc. John resembles his mother. hammer. as we have already seen. locative PPs. b. as assigned by the main verb. For example. we should point out that the theoretical status of semantic roles is still unresolved. While semantic roles provide very useful ways of describing properties across different constructions. As another example. there is no agreement about exactly which and how many semantic roles are needed. (55) a. [pat The mouse] was chased by [agt the cat]. John wiped the window with a towel. red-haired and incredibly well dressed man. John put his books in the attic. consider the following pair: (53) a. There lived a man with his grandson. There still remains an issue to be solved. remain. 7 See Levin and Rappaport Hovav (2005) for further discussion of this issue. retain. b. the cat is the agent. 45 . All the verbs are intransitive. A is similar to B. but not all are acceptable in the there-construction. • Instrument: The medium by which the action or event denoted by the predicator is carried out. Verbs like arrive. they have essentially identical interpretations. Examples: oblique complement of hit. dance assign an ‘agent’ role. consider the following examples: (54) a. The reason is that the same semantic roles assigned to the NPs: in both examples. Examples: subject of keep. Different grammatical uses of verbs may express the same semantic roles in different arrays.• Location: The thematic role associated with the NP expressing the location in a sentence with a verb of location. live are taken to assign the semantic role of ‘theme’ (see the list of roles above).

In the subsequent chapters. there are also many advantages of using them. adjunct adjunct. f. predicator. d. predicative complement subject. predicator. while an NP following the word from is serving as the ‘source’. oblique complement subject. subject. 3. 8 Jackendoff 46 . so. b. indirect object. i. adjunct subject.What kind of semantic roles do the arguments here have? Both participants seem to be playing the same role in these examples – they both cannot be either agent or patient or theme. direct object subject. predicator.5 Exercises 1. (1987) develops an account of thematic roles in which agency and motion are two separate dimensions. g. the subject Mary in (57b) can either be an experiencer or an agent depending on her intention – one can just look at the sky with no purpose at all. adjunct 2. direct object subject. See Pollard and Sag (1987). b. j. They are also cases where we might not be able to pin down the exact semantic role: (57) a. subject. direct object. h. predicator. John runs into the house. predicative complement. in fact. direct object subject. For example. direct object. direct object. predicator. has two individualized semantic roles ‘kicker’ and ‘kicked’. e. c. we will refer to semantic roles in various places. one can assume that each predicator has its own (individual) semantic roles. subject. Give the grammatical function of the italicized phrases in the following examples: (i) a. predicative complement subject. semantic roles are also recognized as the standard concepts used for organizing predicate-argument structures for predicates within the lexicon. a single NP can be agent and theme.8 Also. predicator. Construct sentences containing the following grammatical functions: (i) a. All of his conversation was reported to me. predicator. We can make generalizations about the grammar of the language: typically the ‘agent’ takes the subject position. As we will see in the next chapter. predicative complement. the verb kick. predicator. instead of having an agent and a patient. predicator. predicator adjunct. 9 To overcome the problem of assigning the correct semantic role to an argument. The subject John in (57a) is both agent and theme: it is agent since it initiates and sustains the movement but also theme since it is the object that moves. Mary looked at the sky.9 Even though there are theoretical issues involved in adopting semantic roles in the grammar.

This week will be a difficult one for us. b. e. providing at least one syntactic test we have discussed in the chapter. 47 . d. 5. The teacher made students happy. 4. They helped us edit the script. d. This proved my hypothesis. f. what can we infer from these data? Try out more subjecthood tests such as the tag-question test to determine the grammatical function of there in these examples. That they have completed the course is amazing. Consider the following examples: (i) a. Everyone hoped that she would sing. e. (ii) a. With respect to the grammatical function of there. They elected him America’s 31st President. Draw tree structures for the following sentences and then assign an appropriate grammatical function to each phrase. k. (i) a. c. b. 3. j. She was the nicest teacher in the Senior School. i. Doing syntax is not easy. Last year I saw this film several times. j. The school awarded a few of the girls in Miss Kim’s class scholarships. He baked Tom the bread last night. f. h. try to decide what is the subject in the following so-called ‘locative inversion’ examples and provide at least three different locative inversion examples that you can find from naturally-occurring material. This is the place to go to. There *is/are more chemical substances involved in nerve transmission. He saw the man with the stick. Sandy removed her ballet shoes. Benny worked in a shoe factory when he was a student. Determine the grammatical function of the italiczed phrase.b. In addition. h. k. g. c. b. b. 10 This exercise is adopted from Huddleston and Pullum (2002). They parted the best of friends. We reminded him of the agreement. We need to finish the project this week. The gang robbed her of her necklace. Among the guests was/*were sitting my friend Louise.10 (i) a. This proved a decisive factor. There is/*are only one chemical substance involved in nerve transmission. In the garden stands/*stand a statue. In the summer we always go to France. The next morning we set out for Seoul. i. g.

John’s mother sent a letter to Mary. To see whether this dip in learning was caused by the same kind of pre-coffee fog that many people feel in the morning. however. d. John came from Seoul. They also tried keeping the birds from singing during the day. The ice melted. Feb 2. h. Frank threw himself into the sofa. but performed badly in the mornings. John smelled the freshly baked bread. d. John is afraid of Bill. The vacuum cleaner terrifies the child. (i) Scientists found that the birds sang well in the evenings. c. j. Assign a semantic role to each argument in the following sentences. A big green insect flew into the soup. f. b. and they used a chemical called melatonin to make the birds nap at odd times. We placed the cheese in the refrigerator. g. e. 2007 48 . The scientist made her a robot. 6. The researchers concluded that their study supports the idea that sleep helps birds learn.11 11 From Science News Online. f. e. (i) a. Studies of other animals have also suggested that sleep improves learning. the researchers prevented the birds from practicing first thing in the morning.c. The crocodile devoured the doughnut. The students all worked that summer. i. The students all enjoyed that summer. After being awake several hours. 7. the young males regained their mastery of the material and then improved on the previous day’s accomplishments. The students called me a teacher. Determine the grammatical functions for the underlined expressions in the following text.

complement. both syntactic categories (NP. this condition is not fulfilled. (cf. Some of the unacceptable examples in (1) can be legitimate expressions if they occur in the proper (syntactic) context.4 Head. PP.1 Internal syntax deals with how a given phrase itself is constructed in a well-formed manner whereas external syntax is concerned with how a phrase can be used in a larger construction. and is not related to any external properties of the VP. e. and Modifiers 4. *John [put his gold to be under the bathtub]. *John [put his gold].) and grammatical functions (subject. This is the box in which John [put his gold].1 4. AP. because only it satisfies the condition that the verb put selects an NP and a PP as its complements. In the other examples. (cf. and modifier) play important roles in the analysis of English sentences. In contrast. *John [put his gold safe]. The combinatory properties of words and phrases involve two aspects of syntax: internal and external syntax. (2) a. *John [put under the bathtub]. (1a)) b.1. d.1 Projections from Lexical Heads to Phrases Internal vs. VP. John [put his gold under the bathtub]. This is the gold that John [put under the bathtub]. 49 . b. External Syntax As we have seen in the previous chapters. Observe the following examples: (1) a. Complements. etc. This combinatory requirement starts from the internal (or lexical) properties of the verb put. (1b)) 1 The terms ‘internal’ and ‘external’ syntax are from Baker (1995). the external syntax is concerned with the external environment in which a phrase occurs. and it combines with them in the VP. Why is only (1e) acceptable? Simply. We have also observed that the grammatical function and form of each constituent depend on where it occurs or what it combines with. c.

1. Why are the examples in (7b) and (8b) ungrammatical? 2 See section 1. The teacher handed the student a book. . *The defendant denied. VP NP v rvv rr vvv rr vvv r vv rr rr vv v rr rr 0123 7654 7654 0123 V . b. and the head of an AP is A. The notion of headedness plays an important role in the grammar. b. b.Meanwhile. The complements include direct object. consider frame induced by the governing verb kept in (3): (3) a.. The king kept [putting his gold under the bathtub]. depending on external contexts. obligatory element within the given phrase. but cannot occur in (3a) since this is not the environment where such a finite VP occurs. That is. N c.. *The king kept [put his gold under the bathtub]. there is one obligatory element in each phrase.. b. in this use. Complements. 4. The properties of the head become properties of the whole phrase. NP and PP. The properties of the head verb itself determine what kind of elements it will combine with. *The teacher handed the student. each phrase has one essential element as represented in the diagrams in (4): (4) a. For example. We call this essential element the head of the phrase. the head of a VP is V. and Modifiers One important property we observe in English phrase-internal syntax is that in building up any phrase. (6) a. For example. The defendant denied the accusation. indirect object. the verb kept requires the presence of a gerundive VP like putting his gold under the bathtub. dictates what it must combine with – two complements. The verb denied here requires an NP object whereas handed requires two NP complements. the verb put. the well-formed VP in (1e) can be unacceptable. and therefore imposes an external constraint on VPs.2 Notion of Head. predicative complement. 50 ..3 in Chapter 1 also. As noted in the previous chapter. the elements which a head verb should combine with are called complements. The VP put his gold under the bathtub is a well-formed phrase. functioning as the head of a VP.. That is. AP rvv rr vvv rr vv rr 0123 7654 A . The head of an NP is thus N. and oblique complement since these are all potentially required by some verb or other. Consider other examples: (5) a. The circled element here is the essential..2 The head of each phrase thus determines its ‘projection’ into a larger phrasal constituent.

The unacceptable examples lack a finite (tensed) VP as the bracketed part. (8) a. The senators [know that the president is telling a lie]. Complement: A phrasal element that a head must combine with or a head select. 3 See Chapter 5. Each finite VP is headed by a finite verb. The PPs in his office or with love here provide further information about the action described by the verb. respectively. *The senators [certain that the president is telling a lie]. predicative complement. and oblique complement. a phrase may also contain modifiers (or also called adjuncts): (11) a. Tom [VP [VP offered advice to his students] with love]. as informally represented as in (9): (9) English Declarative Sentence Rule: Each declarative sentence must contain a finite VP as its head. b. b. and they function to augment the minimal phrase projected from the head verb offered. *They [(to) be eager to leave the meeting]. The examples in (7b) and (8b) are unacceptable because of the absence of the required head.2 for the detailed discussion of English verb form (VFORM) values including finite and nonfinite. Head: A lexical or phrasal element that is essential in forming a phrase. and the outer VP is the ‘maximal’ VP which includes optional modifiers. one which is ultimately headed by a finite or nonfinite verb. *The senators [(to) be certain that the president is telling a lie]. b. What we have seen can be summarized as follows: (12) a. Tom [VP [VP offered advice to his students] in his office]. they are still ungrammatical: (10) a. we can talk about a finite or non-finite sentence. In this way. 51 . *They [eager to leave the meeting]. We might say that the inner VP here forms a ‘minimal’ VP which includes all the ‘minimally’ required complements. Minimal Phrase: A minimal phrase is the phrase including this head and all of its complements. but we know that English sentences require a finite VP as their immediate constituent. but are not required as such by the verb. indirect object. The VP is considered to be the (immediate) head of the sentence.(7) a. These include direct object. c. They [want to leave the meeting]. d. Modifier: A phrasal element not selected by the verb functions as a modifier to the head phrase. with the verb itself as the head of the VP. If we amend the ungrammatical examples above to include a verb but not a finite one. b.3 In addition to the complements of a head. b. The VP which includes this kind of modifier forms a maximal phrase. These phrases are optional and function as modifiers.

b. *John kept behind the counter. b. *John stayed him busy. In (16b). b. John kept him behind the garage. John stayed behind the counter. John deposited some money in the bank on Friday. 4 Most of the criteria we discuss here are adopted from Pollard and Sag (1987). 4. Kim and Sandy met [in Seoul] [in the lobby of the Lotte Hotel] in March. serving as a modifier. 52 . Their presence is not required by the grammar: (16) a. Maximal Phrase: A maximal phrase is the phrase that includes complements as well as modifiers. In contrast. and stayed a PP. John placed Kim behind the garage. modifiers are optional. the question that follows is then how we can distinguish between complements and modifiers. c.e. complements are strictly-required phrases whereas modifiers are not. *John placed him busy. c. (15) a. (14) a. but this is impossible for complements. Yet in Seoul is a modifier and we can repeatedly have the same type of PP. Iterability: The possibility of iterating identical types of phrase can also distinguish between complements and modifiers.4 Obligatoriness: As hinted at already. John deposited some money in the bank. kept an NP and a PP or an AP. (13) a. There are several tests to determine whether a phrase is a complement or a modifier. John kept him busy. In general two or more instances of the same modifier type can occur with the same head. c. b. The examples in (13)–(15) show that the verb placed requires an NP and a PP as its complements. In (17a) on humans is a complement and thus the same type of PP on natural causes cannot co-occur. b. *The UN blamed global warming [on humans] [on natural causes]. *John stayed Kim behind the garage. (17) a. *John placed behind the counter. the PP on Friday is optional here.2 Differences between Complements and Modifiers Given these notions of complements and modifiers.

What we can observe in (18b) is that the VP did the same thing can replace either the minimal phrase deposited some money in the checking account or the maximal phrase including the modifier on Friday. This gives us the following informal generalization: (21) Do-so Replacement Condition: The phrase do so or do the same thing can replace a verb phrase which includes at least any complements of the verb. The unacceptability of the examples in (22) also supports this generalization about English grammar: (22) a. leaving out the modifier. *John [gave a present to the student] and Mary did the same thing to the teacher. 53 . b. *John [deposited some money in the checking account] and Mary did the same thing in the savings account. This in turn means that this ‘replacement’ VP cannot be understood to leave out any complement(s). John deposited some money in the checking account on Friday and Mary did the same thing (too). Notice that this VP can replace only the minimal phrase. b. then it is either a minimal or a maximal phrase. Note the following contrast: (23) a. we can draw the conclusion that if something can be replaced by do the same thing. *John locked Fido in the garage and Mary did so in the room. This condition explains why we cannot have another locative complement phrase in the savings account or to the teacher in (20). we can use do the same thing to avoid repetition of an identical VP expression: (18) a. As shown in (18). Constancy of semantic contribution: An adjunct can cooccur with a relatively broad range of heads whereas a complement is typically limited in its distribution. This can be verified with more data: (20) a. From these observations. John deposited some money in the checking account and Mary did the same thing (too). b. Kim camps/jogs/mediates on the hill. *John ate a carrot and Mary did so a radish.Do-so Test: Another reliable test often used to distinguish complements from modifiers is the do so or do the same thing test. Here the PPs in the checking account and to the student are both complements. (19) John deposited some money in the checking account on Friday and Mary did the same thing on Monday. and thus they should be included in the do the same thing phrase.

b. Kim jogs on the hill/under the hill/over the hill. (24) a. Kim depends/relies on Sandy. b. Kim depends on Sandy/*at Sandy/*for Sandy. The semantic contribution of the adjunct on the hill in (23a) is independent of the head whereas that of the complement on Sandy is idiosyncratically dependent upon the head.

Structural Difference: We could distinguish complements and modifiers by tree structures, too: complements combine with a lexical head (not a phrase) to form a minimal phrase whereas modifiers combine with a phrase to form a maximal phrase. This means that we have structures of the following forms: (25) XP rvv rr vvv rr vv rr Modifier XP rvv rr vvv rr vv rr X Complement(s)

As represented in the tree structures, complements are sisters of the lexical head X, whereas modifiers are sisters of a phrasal head. This structural difference between complements and modifiers provides a clean explanation for the contrast in do-so test. Given that the verb ate takes only an NP complement whereas put takes an NP and a PP complement, we will have the difference in the two structures shown in (26): (26) a. VP nn‡‡‡‡‡‡‡‡‡ ‡‡‡‡ nnn nn VP PP n€€ suu nnn €€€€€ sss uuuu s nn n s V NP in the room sss uuu sss uuu ate some food VP gg‡‡‡‡‡‡‡ ‡‡‡‡‡ ggggg g ‡ ggggg V NP PP ss u sss uuuuu uuu ssss uuu sss the money in the room

b.

put

In this way, we represent the difference between complements and modifiers.

54

Ordering Difference: Another difference that follows from the structural distinction between complements and modifiers is an ordering difference. As a complement needs to combine with a lexical head first, modifiers follow complements: (27) a. John met [a student] [in the park]. b. *John met [in the park] [a student]. A similar contrast can be observed in the following contrast: (28) a. the student [of linguistics] [with long hair] b. *the student [with long hair] [of linguistics] The PP with long hair is a modifier whereas the of linguistics is the complement of student. This is why with long hair cannot occur between the head student and its complement of linguistics.5 As such, observed ordering restrictions can provide more evidence for the distinction between complements and modifiers.

4.3

PS Rules, X -Rules, and Features

We have seen in Chapter 2 that PS rules can describe how English sentences are formed. However, two main issues arise with the content of PS rules.6 The first is related to the headedness of each phrase, often called the ‘endocentricity’ property of each phrase. Let us consider the PS rules that we saw in the previous chapters. We have seen that PS rules such as those in (29) can characterize well-formed phrases in English, together with an appropriate lexicon: (29) a. S → NP VP b. NP → Det AdjP∗ N c. VP → V (NP) (VP) d. VP → V NP AP e. VP → V NP NP f. VP → V S g. AP → A VP h. PP → P NP i. VP → Adv VP One common property of all these rules is, as we have discussed, that every phrase has its own head. In this sense, each phrase is the projection of a head, and thereby has the endocentricity. However, we can ask the theoretical question of whether or not we can have rules like the following, in which the phrase has no head at all: (30) a. VP → P NP
5 See 6 The

(49) for the structural differences between with long hair and of linguistics. discussion of this section is based on Sag et al. (2003).

55

b. NP → PP S Nothing in the grammar makes such PS rules unusual, or different in any way from the set in (29). Yet, if we allow such ‘non-endocentric’ PS rules in which a phrase does not have a lexical head, the grammar would then be too powerful to generate only the grammatical sentences of the language. Another limit that we can find from the simple PS rules concerns an issue of redundancy. Observe the following: (31) a. *The problem disappeared the accusation. b. The problem disappeared. (32) a. *The defendant denied. b. The defendant denied the accusation. (33) a. *The boy gave the book. b. The boy gave the baby the book. What these examples show is that each verb has its own requirement for its complement(s). For example, deny requires an NP, whereas disappear does not, and gave requires two NPs as its complements. The different patterns of complementation are said to define different subcategories of the type verb. The specific pattern of complements is known as the ‘subcategorization’ requirement of each verb, which can be represented as following (IV: intransitive, TV: transitive, DTV: ditransitive): (34) a. disappear: IV, b. deny: TV, c. give: DTV, NP NP NP

In addition, in order to license the grammatical sentences in (31)–(33), we need to have the following three VP rules: (35) a. VP → IV b. VP → TV NP c. VP → DTV NP NP We can see here that in each VP rule, only the appropriate verb can occur. That is, a DTV cannot form a VP with the rules in (35a) or (35b): It forms a VP only according to the last PS rule. Each VP rule thus also needs to specify the kind of verb that can serve as its head. Taking these all together, we see that a grammar of the type just suggested must redundantly encode the subcategorization information both in the lexical type of each verb (e.g., DTV) and in the PS rule for that type of verb. A similar issue of redundancy arises in accounting for subject-verb agreement: (36) a. The bird devours the worm. b. The birds devour the worm.
56

To capture the fact that the subject NP agrees with the predicate VP, we need to differentiate the S rule into the following two: (37) a. S → NPsing VPsing (for (36)a) b. S → NPpl VPpl (for (36)b) The two PS rules ensures that the singular (sing) subject combines with a singular VP whereas the plural (pl) subject NP with a plural VP. Descriptively, there is no problem with a grammar with many specific parts. From a theoretical perspective, though, we have a concern about the the endocentricity and redundancy issues. A more particular related question is that of how many PS rules English has. For example, how many PS rules do we need to characterize English VPs?—Presumably there are as many rules as there are subcategories of verb. We need to investigate the abstract content of PS rules, in order to develop a theoretical view of them. For example, it seems to be the case that each PS rule must have a ‘head’. This will disallow many possible PS rules which we can write using the rule format, from being actual rules of any language. In order to understand more about the structures that rules describe, we need two more notions, ‘intermediate category/phrase’ and ‘specifier (SPR)’. We motivate the idea of the intermediate category, and then specifier is a counterpart of it. Consider the examples in (38): (38) a. Every photo of Max and sketch by his students appeared in the magazine. b. No photo of Max and sketch by his students appeared in the magazine. What are the structures of these two sentences? Do the phrases every photo of Max and sketch by his students form NPs? It is not difficult to see sketch by his students is not a full NP by itself, for if it was, it should be able to appear as subject by itself: (39) *Sketch by his students appeared in the magazine.

In terms of the semantic units, we can assign the following structures to the above sentences, in which every and no operate over the meaning of the rest of the phrase: (40) a. [Every [[photo of Max] and [sketch by his students]]] appeared in the magazine. b. [No [[photo of Max] and [sketch by his students]]] appeared in the magazine. The expression photo of Max and sketch by his students are phrasal elements but not full NPs — so what are they? We call these ‘intermediate phrases’, notationally represented as N-bar or N . The phrase N is thus intuitively bigger than a noun, but smaller than a full NP, in the sense that it still requires a determiner from the class the, every, no, some, and the like. The complementary notion that we introduce at this point is ‘specifier’ (SPR), which can include the words just mentioned as well as phrases, as we illustrate in (41): (41) a. [the enemy’s] [N destruction of the city] b. [The enemy] [VP destroyed the city].

57

The phrase the enemy’s in (41a) and the subject the enemy in (41b) are semantically similar in the sense that they complete the specification of the event denoted by the predicate. These phrases are treated as the specifiers of N and of VP, respectively. As for the possible specifiers of N , observe the following: (42) a. a little dog, the little dogs (indefinite or definite article) b. this little dog, those little dogs (demonstrative) c. my little dogs, their little dog (possessive adjective) d. every little dog, each little dog, some little dog, either dog, no dog (quantifying) e. my friend’s little dog, the Queen of England’s little dog (possessive phrase) The italicized expressions here all function as the specifier of N . However, notice that though most of these specifiers are determiners, some consist of several words as in (42e) (my friend’s, the Queen of England’s) . This motivates us to introduce the new phrase type DP (determiner phrase) that includes the possessive phrase (NP + ’s) as well as determiners. This new phrase then will give us the generalization that the specifier of N is a DP.7 Now let us compare the syntactic structures of (41a) and (41b): (43) NP w qqq wwwww www qqq q w qqq N DP cc qqwwwww  ccc www qqq qq cc  ww qqq  the enemy’s N PP ÔXX ÔÔ XXX Ô XX Ô ÔÔ destruction of the city S qww qqq wwwww www qqq qqq NP VP qww Ñaa qqq wwwww ÑÑ aaa Ñ qq www q aa Ñ qqq ÑÑ The enemy V NP W QQ ÕÕ WWW Õ  QQQ WW  ÕÕ Q W ÕÕ  destroyed the city

(44)

Even though the NP and S are different phrases, we can notice several similarities. In the NP structure, the head N destruction combines with its complement and forms an intermediate
7 Some analyses take the whole expression in (43) to be a DP (e.g., a little dog, my little dogs) in which expressions like little dog is not an N but an NP.

58

phrase N which in turn combines with the specifier DP the enemy’s. In the S structure, the head V combines with its complement the city and forms a VP. This resulting VP then combines with the subject the enemy, which is also a specifier. In a sense, the VP is an intermediate phrase that requires a subject in order to be a full and complete S. Given these similarities between NP and S structures, we can generalize over them as in (45), where X is a variable over categories such as N, V, P, and other grammatical categories:8 (45) XP rvv rr vvv rr vv rr Specifier X rvv rr vvv rr vv rr X Complement(s)

This structure in turn means the grammar now includes the following two rules:9 (46) a. XP → Specifier, X (Head-Specifier Rule) b. XP → X, YP∗ (Head-Complement Rule) These Head-Specifier and Head-Complement Rules, which form the central part of ‘X -theory’, account for the core structure of NP as well as that of S. In fact, these two general rules can also represent most of the PS rules we have seen so far. In addition to these two, we just need one more rule:10 (47) XP → Modifier, X (Head-Modifier Rule)

This Head-Modifier Rule allows a modifier to combine with its head as in the PS rule VP → VP Adv/PP, as presented in the following: (48) XP rvv rr vvv r r vv rr Specifier X rvv rr vvv r vv rr r X Modifier v rr vvv rr vv r v rr X Complement(s)

can assume that the head of S is VP and that VP is an intermediate phrase in the sense that it still requires a subject as its specifier. 9 Unlike the PS rules we have seen so far, the rules here are further abstracted, indicated by the comma notation between daughters on the right-hand side. We assume that the relative linear order of a head and complements etc. is determined by a combination of general and language-specific ordering principles, while the hierarchical X -structures themselves are universal. 10 The comma indicates that the modifier can appear either before the head or after the head as in always read books or read books always.

8 We

59

We can observe in (50b) that the combination of king with with a hat forms an N . b. *the king [with a hat] [of Rock and Roll] The badness of (49b) is due to the fact that the modifier with a hat is combined with the head king first. the king [of Rock and Roll] [with a hat] b.One thing to notice in the Head-Complement Rule is that the head must be a lexical element. The present king of country music is more popular than the last one. Why do we have the contrast here? One simple answer is that the pronoun one here replaces an N but not an N or an NP. NP‘ h‘ ‘‘‘‘‘‘‘‘ ‘‘‘‘‘‘‘‘ hhhhh ‘‘‘‘ hhhh hh DP N hhhh…………………… hhhhh ………… hhh the PP N c  ccc hhhh……………………  hhh cc  ………… hhhhh  PP with a hat N oyy ooo yyyyy o y ooo king of Rock and Roll NP‘ h‘ ‘‘‘‘‘‘‘‘ ‘‘‘‘‘‘‘‘ hhhhh ‘‘‘‘ hhhh hh DP *N hhhh…………………… hhhhh ………… hhh the PP N ww rrr wwww hhhh…………………… r hhh r ………… ww hhhhh rrr PP of Rock and Roll N {gg {{ ggg { g {{ king with a hat b. but the combination of the complement of Rock and Roll with this N will not satisfy the Head-Complement Rule. can be further supported from the pronoun substitution examples in (51): (51) a. too: 60 . The existence and role of the intermediate phrase N . *The king of Rock and Roll is more popular than the one of country music. This will also account for the following contrast. This in turn means that we cannot apply the Head-Modifier Rule first and then the Head-Complement Rule. which is a larger than a lexical category but still not a fully-fledged phrase. (50) a. This explains the following contrast: (49) a.

However. with values singular and plural.(52) A: Which student were you talking about? B: The one with long hair. or a head and its specifier. The structure (53) shows how the values of the features in different parts of a structure are related: (53) XP[POS 1 ] rvv rr vvv rr vv rr Specifier X [POS 1 ] rvv rr vvv rr vv rr X [POS 1 ] Modifier rvv rr vvv rr vv rr X[POS 1 ] Complement(s) The notation 1 shows that whatever value the feature has in one place in the structure. That is. we can add a detail to the Head-Specifier Rule as following: (54) XP → Specifier[NUMBER 1 ]. verb. for they refer to ‘Head’. The phrase of linguistics is the complement of student. 61 . The grammar rules can account for the same structures as all the PS rules we have seen so far: with those rules we can identify phrases whose daughters are a head and its complement(s). N . We provide more details of the formal feature system in the following section. This is a representational tag. The two rules in (37) are both represented in (54). The three X rules thereby greatly minimize the number of PS rules that need to characterize well-formed English sentences. We can formalize this more precisely by introducing the feature POS (part of speech). and NP structures. these X rules directly solve the endocentricity issue. This means the N-bar pronoun one should include this. NUMBER. or a head and its modifier. capturing the headedness of each phrase: the grammar just does not allow any phrase without a head. it has the same value somewhere else. The redundancy issue mentioned above for agreement is now a matter of introducing another feature. B: *The one of linguistics with long hair. So (53) indicates that the phrase’s POS value is identical to its head daughter. the modifier with long hair cannot be within the N . X [NUMBER 1 ] The rule states that the subject’s NUMBER value is identical with that of the predicate VP’s NUMBER value. with the new feature NUMBER. Assume that X is N. then we will have N. in which the number 1 has no significance: it could as easily be 7 or 437 . There are several more welcome consequences that the three X rules that bring to us. adjective. In addition. which has values such as noun.

the word puts will have at least the following information: (55) Minimal Lexical Information for puts: a. syntactic. These indexes refer to the participants denoted by the arguments. phonological information: <puts> b..4. or a feature structure: 11 In particular. 3rd singular c.11 We present here a formal and explicit way of representing it with feature structures. semantic information: put (i. in terms of complements. we have seen that the properties of a lexical head determine the components of the minimal phrase. The theory developed in this textbook heavily relies upon the feature structure system of HPSG. The value of each attribute can be an atomic element. 4... j and k. to indicate the participants that are minimally involved in the event expressed by the verb. (2003). for each word in the lexicon. a set.j. These lexical entries can be represented in a more systematic and precise way with the system of feature structures. The argument structure represents the number of arguments which the verb selects. 62 .k) The phonological information is the information about how the word is pronounced. something j being put in a place. and that other properties of the head are directly properties of the phrase. orthographic).4.. the syntactic information indicates that this particular word is a verb and is in the 3rd singular present (finite) form. For example. This argument information is linked to its more precise meaning as indicated by the indexes i. Every lexical entry at least includes phonological (but in practice. finite. grammars such as Head-driven Phrase Structure Grammar (HPSG) and Lexical Functional Grammar (LFG) are couched upon mathematically well-defined feature structure systems. a list. which we now introduce. . and semantic information.4 Lexicon and Feature Structures In the previous section. locationk > d. syntactic information: verb. and someplace k it is put in. Finally. See Sag et al. Each feature structure is an attribute-value matrix (AVM):   Attribute1 value1 (56) Attribute2 value2     Attribute3 value3 .1 Feature Structures and Basic Operations Most modern grammars rely on a representation of lexical information in terms of features and their values. This information is encoded in a lexical entry. themej . argument information: <agenti . morphological. the semantic structure represents that the verb’s meaning relates three participants – someone i who is doing the action of putting.

 NAME LOCATION seoul  b. showing that there is no significance in the relative ordering. including its name and location. jogging. * university   kyunghee univ. we might declare that TENSE is appropriate only for verb. . . Here. . . The upper left declaration in italics is the type of the feature structure:  (58) a. university   kyunghee univ. is inappropriate. Richard  SONS    HOBBIES swimming. but having a MAJOR . 12 Even 63 . whereas the value of SONS is a list which represents something relative about the two values. So. the value of the feature ADVANCED-DEGREE is a feature structure which in turn has three attributes.        FIELD linguistics      ADVANCED-DEGREE AREA syntax-semantics YEAR 1996 This illustrates the different types of values that attributes (feature names) may have. . in this case that one is older than the other. Now consider the following example of a typed feature structure. which we have already seen above in terms of the 1 notation (see (53)). This is to represent cases where two features (or though every feature structure is typed in the present grammar. each feature structure is relevant only for a given type. we will not specify the type of each feature structure unless it is necessary for the discussion. One important property of every feature structure is that it is typed. One useful notion in the feature structure is structure-sharing. for example. the value of the attribute NAME is atomic. as a subject of study. In the linguistic realm. for example ‘youngest son’ would be the right-most element in the list value of SONS. information about one author of this book:   author (59)  NAME kim     Edward. the value of HOBBIES is a set. NAME MAJOR linguistics The type university may have many properties.12 That is. Meanwhile. reading. A simple illustration can show why feature structure needs to be typed.(57)   Attribute1 atomic   Attribute2        Attribute3      Attribute4 . Finally.

the symbol represents subsumption. However. it will be more than enough just to represent the particular or general aspects of the facts concerned. and their TEL attribute has a value which is the same as the value of his TEL attribute. In such a case.   (61) individual individual   A: B: NAME kim  NAME kim TEL 961-0892 In (62). a feature structure A subsumes another feature structure B if A is not more informative than B. NAME 1 TEL For the type individual. Feature unification means that two compatible feature structures are unified. represented by . Put differently. whatever the value actually is. A represents more general information than B. conveying more coherent and rich information. Consider the feature structures in (62). The most crucial operation in feature structures is unification. for in fact we cannot represent the total information describing all possible worlds or states of affairs. the general one subsumes the specific one. In addition to this. the notion of subsumption is also important in the theoretical use of feature structures.attributes) have an identical value:  individual (60) NAME kim   1 TEL    individual    SONS NAME  TEL            edward  1  individual  richard. they cannot be unified: 64 . the first two may unify to give the third: (62) individual NAME kim  individual  NAME TEL individual TEL   kim  961-0892 961-0892 → The two feature structures are unified. The subsumption relation concerns the relationship between a feature structure with general information and one with more specific information. Each small component of feature structure will provide partial information. In describing a given phenomenon. This kind of subsumption relation is used to represent ‘partial’ information. (60) represents a situation in which the particular individual kim has two sons. attributes such as NAME and TEL and SONS are appropriate. if two feature structures have incompatible feature values. the different pieces of information are put together. resulting in a feature structure with both NAME and TEL information. and as the structure is built up.

Unification will make sure that information is consistent as it is built up in the analysis of a phrase or sentence.4. can have a feature structure like the following:13   (64) verb    PHON puts         POS verb      SYN   VFORM fin        ARG-ST [agt]i . [loc]k       PRED put-relation      AGENT   i   SEM    THEME   j     LOCATION k This feature structure has roughly the same information as the informal representation in (55). One thing to note here is that since there are some cases where we have difficulties in assigning a specific named semantic role to a selected argument discussed in Chapter 3. and k. that it has a finite verbal inflectional form value (VFORM). [th]j . location). argument structure (ARG-ST). whose general type is verb. The SYN attribute indicates that the POS (part of speech) value is verb. we will not represent the PHON and SEM values unless relevant to the discussion at hand. 65 . For example. syntactic (SYN). 4. The SEM feature represents the information this verb denotes the predicate relation put-relation. which will be realized as the subject and two complements in the full analysis. The verb puts.The ARG-ST attribute indicates that the verb selects for three arguments (with thematic roles agent. they cannot be unified.2 Feature Structures for Linguistic Entities Any individual or entity including a linguistic expression can be represented by a feature structure.(63) individual individual NAME edward NAME   individual   *NAME edward NAME richard richard → Since the two smaller feature structures have different NAME values. whose three participants are linked to the elements in the ARG-ST via the indexing values i. and semantic (SEM) information. the word puts. j. we typically just indicate the number of arguments each predicate is selecting in ARG-ST: we underspecify 13 Later on. like any verb. theme. has its own phonological (PHON) value.

3 ARG-ST The boxed tags show the different identities in the overall structure. The again.3 Argument Realization Each element on the ARG-ST list is realized as SPR (specifier) or COMPS (complements). 3 PP  COMPS    1. b. This realization is obligatory in English. which are combined in the notion of valence (VAL) features.16 More formally. d. 16 The term ‘valence’ refers to the number of arguments that a lexical item can combine with. and to other parts of the representation of the grammatical properties. through one of the rules in (46). two-place predicates like devour take two arguments. for example. the three arguments of put are realized as subject and complements. It is inspired by the notion of valence as used in atomic theory in chemistry. John put the book in the box. b. respectively: (65) a. c. to make a syntactically well-formed sentence. verbs like smile. often along with a description of the categories of those constituents. c. first approximation): The first element on the ARG-ST list is realized as SPR. the arguments selected by each predicate are linked to grammatical functions. ensuring that the two are identical. but semantically anomalous. the basic pattern is that the first element on the list is realized as subject and the rest as complements: (66) Argument Realization Constraint (ARC. *In the box put John the book. 2. *John put in the box. the rest as COMPS in syntax. remember that the term SPR includes subject as well as the noun’s specifier. to the core semantic properties.4. [ ]. So. For example. notation # indicates that the structure is technically well-formed from a syntactic perspective. for example. We see that the arguments selected by a lexical head should be all realized as SPR and COMPS.14 In general. and three-place predicates take three arguments. we can represent this constraint as applied to put as the following feature structure:    (68) 1 NP SPR VAL     2 NP. 15 The 14 Once 66 . devour and give will have the following ARG-ST representations. Eventually. 4. [ ] One-place predicates like smile select for just one argument. [ ] ]. with the putter (agent) as subject:15 (67) a.the information unless it is necessary to show more details. the first element of ARG-ST and of SPR have the boxed tag 1 . ARG-ST [ ARG-ST [ ARG-ST [ ] ]. #The book put John in the box.

but the first argument can be realized either as an NP subject as in (70a) or a CP subject as in (70b). 67 . the arguments can be realized in a different phrase. and P. SPR 1 NP VAL     COMPS 2 NP     1. Notice that the arguments can be realized into different categories. As the book goes on. Verb types can be differentiated by looking only at the COMPS value since every verb will have one SPR (subject) element. 2 ARG-ST Though there is no difference in terms of the number of arguments that surprise select.general ARC constraint blocks examples like (67c) in which the location argument is realized as the subject. as shown in (69):    (69) * 3 PP SPR VAL     COMPS 1 NP. *John disappeared Bill. This difference in the argument realization can be represented as the following. 2. we know that every lexical element has ARG-ST information which will be realized in surface form through the SPR and COMPS values.4 Verb Types and Argument Structure As mentioned earlier. respectively:    (71) a. b. This is exactly the way that verbs are differentiated using the traditional notion of subcategorization. John disappeared. A. which requires the first element of ARG-ST be realized as the SPR (the subject of a verb or the specifier of a noun). SPR 1 CP VAL     COMPS 2 NP     1. we will see how the argument realization is further constrained by the lexical properties of the verb in question or by other grammatical components. Intransitive: This is a type of verb that does not have any COMPS: (72) a. That he won the election surprised everybody. depending on the properties of the given verb: (70) a. 3 ARG-ST This violates the ARC. b. N. 4. select one or more complement(s) to form a minimal phrase. The data indicate that verbs like surprise will take two arguments. 2 NP     1. The election results surprised everybody. With the construct of ARG-ST.4. 2 ARG-ST    b. lexical elements in the classes V.

John saw Fred. John became a success. b. The president looked [weary]. Such a verb will have just one argument that is realized as subject:   (74) disappear       1 NP  SPR VAL     COMPS     1 ARG-ST Linking verbs: Verbs such as look. Transitive verbs: Unlike linking verbs.) This subcategorization requirement can be represented as follows:   (77) become        1 NP SPR VAL     2 AP/NP[PRD +]  COMPS    1. c. The teacher became [tired of the students]. c. and feel require different complements that are typically of category AP: (75) a. 2 ARG-ST This kind of verb selects two arguments: one is canonically an NP to be realized as the subject and the other is any phrase (XP) that can function as a predicate (PRD +) (see also the examples in (84)). they all at least select a predicative complement. d. John remained [somewhat calm]. b. where a property is ascribed to the subject. These verbs also can select other phrases (here. pure transitive verbs select a referential. The jury seemed [ready to leave]. functioning as direct object: (78) a. (Compare John remained a student and John revived a student. The lasagna tasted [scrumptious]. Though each verb may select different types of phrases. John seemed a fool. *John sneezed the money. Of course. These verbs have no COMPS element—the list is necessarily empty. NP): (76) a. e. which we simply assume here.(73) a. remain. John remained a student. seem. b. John sneezed. b. 68 . this presupposes an accurate characterization of which phrases can be [PRD +]. non-predicative NP as their complement. Alice typed the letter.

or VP). Ad agencies call young people Generation X-ers. AP. respectively:   (81) teach        1 NP SPR VAL     2 NP. one functioning as a direct object and the other as a predicative phrase (NP. The parents bought non-fiction novels for the children. 2 [goal]. The parents bought the children non-fiction novels. The school board leader asked the students a question.   (83) teach        1 NP SPR VAL     2 NP. b. c. d. 3 [theme] ARG-ST These verbs typically have a related realization with the goal argument realized as a dative PP complement: (82) a. Clinton supported the health care bill. The sexual revolution makes some people uncomfortable. Raccoons destroyed the garden. 3 PP  COMPS    1 . 2 [theme]. c. 2 ARG-ST Ditransitive: There are also ‘ditransitive’ verbs that require IO and DO: (80) a. John taught new students English Syntax. 3 [goal] ARG-ST Complex Transitive: There is another type of transitive verb which selects two complements. b. 3 NP  COMPS    1 . The school board leader asked a question of the students. John regards Bill as a good friend. describing the object: (84) a.c. Such verbs have three arguments: one subject and two complement NPs functioning as indirect and direct object. 69 . In this realization. the second argument has the theme role while the third one has the goal role. b. John taught English Syntax to new students. c. Such verbs will have the following lexical information:   (79) destroy       1 NP  SPR VAL     2 NP  COMPS    1.

the predicative phrase is an infinitive VP. there are other verbs that do not fit into these classes. the use of the verb carry in (86).d. 70 . we will see yet more different types of verbs. the predicative PP as a good friend follows the object Bill. for instance. The examples in (86) illustrate that carried requires an NP and a PP. In (84a). 3 PP COMPS     1 [agt]. Historians believe FDR to be our most effective president. Just like linking verbs. there are various other verb types that we have not described here. John carried her on his back. the NP Generation X-ers is the predicative phrase. Even though these five types of verb that we have seen so far represent many English verb types. In (84c). the AP uncomfortable serves as a predicate phrase of the preceding object some people. In (84d). in terms of complementation patterns. 3 [PRD +] ARG-ST This means that the verbs in (84) all select an object NP and an XP phrase that function as a predicate. (86) a. b. As the book goes on. in (84b). it denotes the location to which John carries her. these verbs require a predicative ([PRD +]) XP as complement:   (85) call        1 NP SPR VAL     2 NP. 2 . Of course. 3 [loc] ARG-ST The PP here cannot be said to be predicate of the object her. *John carried to the door. as represented in the feature structure:   (87) carry         1 NP SPR VAL     2 NP. 2 [th]. 3 XP  COMPS    1 .

*Oliver mentioned Charles the problem. e. Provide tree structures for the following pairs of sentences while checking the grammatical function of each phrase with valid distributional tests. Tom hid the manuscript in the cupboard. I wonder if you will come back tomorrow. write down the example and provide its lexical entry including the COMPS value. 3. *Oliver fined ten pounds to the prisoner. i. tell. borrow. (i) appear. c. *Oliver ascribed his longevity there. f. owe. Fred hired Sharon to change the oil. f. In doing so. *Oliver addressed the king the letter. draw its structure and then give the lexical entry (SPR and COMPS value) of the main verb. award. We have not confirmed whether the flight had been booked. Tom keeps asking Karen’s sister to buy the car. provide detailed NP structures using the intermediate phrase N : (i) a. (i) a. d. In addition. George mailed the attorney his photograph of the accident. consider. Jane left the book on the table. say 2. *Oliver drove me a lunatic. transcribing. The verbs in the following examples are used incorrectly. For each example below. introduce. e. b. You would have a reply if you come back tomorrow. They confined his remarks to the matter under discussion. 4. The students of English from Seoul faced many issues in the process of interpreting. state what differences we need to represent for the underlined expressions. b.4. Correct the errors or replace the verb with another one. 5. b. In particular. Tom locked Fido in the garage. *Oliver absented his brother from the meeting. explain. Tom played it under the table. use it in a grammatical example. b. c. j. Tom placed it under the table. Draw tree structures for the following two sentences. (iii) a. g. (ii) a. Tom bathed Fido in the garage. Frank hopes to persuade Harry to make the cook wash the dishes. (i) a. d.5 Exercises 1. and editing the poems. discuss. and write out each new example. provide the COMPS value for each verb (in its use in your grammatical examples). b. They pushed the prisoners into the truck. We saw him beaten by the champion. (i) a. h. For each of the verbs below. 71 .

Teachers cannot learn the language for their students. The museum displayed no painting by Miro or drawing by Klee. roads. this ethic would be interested in solving the problem of human mobility in a way that would not require the disruption of highways.unitedearth. c. They can set their students on the road. Rather than being concerned about how to raise automobile production. Read the following texts and provide the ARG-ST. Then they must wait on the sidelines. 6. helping them to develop confidence in their own learning powers. every dog and cat in the area has to be spayed or neutered. It rebels against an industrialist world view: “Before it is possessed and used. the use of a field by an African tribe to grow grain for survival is an example of a vital need whereas the conversion of a swamp to an exclusive golf course would not. Rest assured that much of the mining.”17 (ii) 17 From http://www. By law. and development of our technological age would not meet the requirement of this principle.au/deepecology. every plant is a weed and every mineral is just another rock. and parking lots.com. SPR and COMPS values of the underlined expressions. The love of my life and father of my children would never do such a thing. (i) Learning to use a language freely and fully is a lengthy and effortful process. d. ready to encourage and assist. Thus. harvesting.html 72 . Deep ecologists put a reign on human exploitation of natural “resources” except to satisfy vital needs.b.

g. (1c) allows a combination of a head with its modifier. The X rules control their combination with a relevant head: (1) a.. we need to introduce the category variable X and the feature POS: (2) a. as shown in (3): (3) VP[POS verb] y M ooo yyyyy yyy ooo oo yy ooo V[POS verb] PP 73 . in order to guarantee that the head’s POS (part of speech) value is identical with its mother phrase. XP POS 1 The POS feature is thus a head feature which passes up to a ‘mother’ phrase from its head ‘daughter’. Head-Complement Rule: XP POS 1 → X[POS 1 ]. Head-Modifier Rule: XP POS 1 → Modifier. Head The rule (1a) represents the case where a head combines with its specifier (e. whereas (1b) says that a head combines with its complement(s) and forms a phrase. Head b.5 More on Subjects and Complements 5. Head-Specifier Rule: XP POS 1 → Specifier. XP → Specifier.1 Grammar Rules and Principles As we have seen in the previous chapter. X [POS 1 ] b. a VP with its subject and an N with its determiner). Complement(s) c. Complement(s) c. As noted earlier. the arguments in ARG-ST are realized as the syntactic elements SPR (subject of a verb and determiner of a noun) and COMPS. XP → Head. XP → Modifier.

c.This percolation from a head to its mother is ensured by the following Head Feature Principle: (4) The Head Feature Principle (HFP): A phrase’s head feature (e. H[SPR 1 ] b.   (7) HEAD | POS verb        SPR NP  VAL COMPS NP. *Talked with Bill about the exam. .. VFORM. Head-Modifier Rule: XP → [MOD 1 ]. As seen from the contrast here and as noted in the lexical entry in (7). 74 . The HFP will apply to any features that we declare to be ‘head features’. 1 .) is identical with that of its head. *Kim put in the box. POS. b. n The grammar rules here are well-formedness conditions on possible phrases in English. *Is putting the book in the box. the verb put selects two complements and must combine with all of its complements. Kim put the book in the box. .g. . Consider the following examples: (6) a. indicating what each head combines with and then what happens as the result of the combination. PP We can also see that a finite verb must combine with its subject: (8) a.. The grammar does not allow hypothetical phrases like the following: (5) *VP[POS verb] ƒƒ kkkk ƒƒƒƒƒƒ ƒƒƒ kkkk kkk A[POS adj] PP We have not yet spelled out clearly what ensures a lexical head to combine not just with one of its complements but with all of its COMPS elements. *Kim put the book. The HFP thus ensures that every phrase has its own lexical head with the identical POS value. Head-Specifier Rule: XP SPR → 1 . b. n ].. VFORM being another. Such combinatorial requirements can be formally stated in the revised grammar rules as given in (9): (9) a. Head-Complement Rule: XP COMPS → H[COMPS c. 1 H 1 . . . . etc.

SYN. combines with it. 3 FF  DD    FF DD   FF D     deposits some money in the bank The finite verb deposits selects a subject (a specifier) and two complements. the resulting phrase forms a well-formed head-modifier phrase.1 These three grammar rules. For example. Let us consider one example in a little more detail:2 (10) S   HEAD 4 | POS verb      SPR   COMPS „ xx „„„„„„„ xx „„„H „„„„ xx „ xx x Subj xxxx VP x   xx xx x HEAD 4 | POS verb   1 NP   (( 444 SPR 1    (( 44 COMPS (( 44 ( 4 jjppp pp jjjj (( ( 444 pp jjjj H j pp C (( 44 jjjj j pp C (( 44 pp V pp (( 44 pp   pp 4 (( HEAD 4 | POS verb   2 NP 3 PP Kim   FFF DD SPR 1     FF  DDD FF   DD COMPS 2 . This kind of ‘discharging’ mechanism is further 1 In addition to these three grammar rules. 75 . SEM. suppressing unrelated features. The HFP ensures that the head feature POS values of the verb. When the lexical head combines with its two complements. we have a well-formed head-specifier phrase with the SPR value discharged. its mother VP and S are all identical. interacting with the general principles such as the HFP.For example. the fully specified feature structure in (10) will include VAL as well as PHON. forming a VP in accordance with the Head-Complement Rule. in (9a) when a head. license grammatical sentences in English. etc. as in What did John eat ?. a head combines with all of its COMPS value. when a modifier combines (carrying the MOD feature) with the head it modifies. English employs the Head-Filler Rule that licenses the combination of a head missing one phrasal element with a filler that matches this missing element. in (9b). requiring a SPR. the COMPS value becomes empty. See Chapter 10 for discussion of this rule. in (9c). 2 For convenience reason. we adopt a shorthand system in representing feature structures. it forms a Head-Complement phrase with no further COMPS value. This VP will still need to combine with its SPR in order to form a complete sentence.

He is proud of driving a car. The car was driven by him. This classification means that the plain and base forms are identical with the lexical base (or citation) of the lexeme. He wants to drive a car. and en (past participle). we can take a lexicon to consist of lexemes as headwords. He was driving. driven. He has driven the car. drive. English verbs have 6 grammatical forms. 5.ensured by the following general principle: (11) The Valence Principle (VALP): The mother’s SPR and COMPS value is identical with its head daughters minus the discharged value(s). He has to drive. 3A 76 .2 5. is used for present tense when the subject is non-3rd person. and pln (plain) which are typically realized as -(e)s. and inf (infinitive). Driven by the mentor. drive are forms of the same lexeme DRIVE.3 Even though the two forms are identical in most cases. More generally. For example. -ed form. Using this division. The nonfin forms have the basic forms of bse (base). drove. This principle thus ensures that when the VP in (10) combines with the subject SPR. and null. respectively. driving.2. with all the rest being nonfin(ite) in some way. In this sense. substitution tests by the past form or by the verb be can show us a clear difference: lexeme is thus an abstract unit of morphological analysis. he worked. to drive. pst (past). drive. though identical to the citation form. ing (present participle). it forms a complete S in accordance with the Head-Specifier Rule. He drove a car. in addition to the citation form. driving. They drive a car. drives. Driving a car. The plain form. the VALP ensures that each verb combines in the syntactic structure with exactly all and only the syntactic dependents that its SPR and COMPS values indicate. The fin forms have three subtypes pres (present). The present and past tense forms are usually classified together as fin(ite). he sang a song. drove.1 Feature Specifications on the Complement Values Complements of Verbs Intuitively. the verb drive can take these forms: drives. we might lay out the forms as in (12): (12) Types of English Verb Forms: Finiteness fin Verb forms pres pst pln bse ing nonfin en inf Example He drives a car. For example.

(14) a. satisfying a basic requirement placed on English declarative sentences: (19) English Declarative Sentence Rule: For an English declarative sentence to be well-formed. This has the consequence that there is only one verb in English with an infinitive form: to itself. 77 .(13) a. This in turn shows us that only finite verb forms can be used as the head of the highest VP in a declarative sentence. The student [knows the answers]. They write/wrote to her. Sometimes we want to be able to refer to the type of a value. (18) a. b. The students [know the answers]. b. The student [knew the answers]. The verbs write and are in (13) are plain forms whereas write and be in (14) are base forms with no tense information. *The student [knowing the answers]. b. They want to be/*are kind to her. we propose the following hierarchy for the values of the attribute VFORM: (15) vform i lll iii lll ii ll ii ll lll nonfin fin ‚ i‚‚ yii ‚ yy iii ‚‚‚ yy iii ii ‚‚‚ yy y ii y y ii ‚‚‚ i yy yy y y pres pst pln bse ing en inf The classification of VFORM values here means that the values of VFORM can be typed. The unacceptability of the examples in (18) is due to the fact knowing and known have no expression of tense – they are not finite. [VFORM fin] b. They want to write/*wrote to her. c. and those types have different subtypes. They are/*be kind to her. as in (16b). With these classifications. We also follow the fairly standard generative grammatical analysis of English ‘infinitives’ in which the infinitive part is a head (to) which takes as its complement a verb in the bseform. *The student [known the answers]. Every declarative sentence in English needs to have a finite verb with tense information: (17) a. and sometimes to a particular form. [VFORM ing] The need to distinguish between fin and nonfin is easily determined. as in (16a). b. its verb form value (VFORM) must be finite. (16) a.

there are only two participle forms in English. Seen from this perspective. let us consider a simple example like The student knows the answer.) though strictly speaking. (21) Usages of the present participle (ing) form: a. the project was successful. which each have several functions or constructional usages. showing that the finiteness of the VFORM value is a head feature: (20) S[VFORM fin] n oyy ooo yyyyy yyy ooo o y ooo VP[VFORM fin] NP One thing we need to remember is that the two participle forms (present and past) have many different uses. He is writing another long book about beavers. in different constructions. (1985) or Ginzburg and Sag (2000). The chicken has eaten. Broadly speaking. there is no easy solution. The chicken was eaten. 2 78 . (used as a sentence modifier) Some of these usages have been introduced as VFORM values for the ease of exposition (cf. (part of the the progressive aspect construction) b. (part of the perfect aspect construction) b. He is proud of his son’s passing the bar exam. Here the verb knows will have the following lexical information:   (23) knows       POS verb  HEAD    VFORM pres           1 NP SPR   VAL       2 NP COMPS     ARG-ST 1 . Gazdar et al.The finiteness of a sentence or a VP comes from the head verb. (used in a gerundive construction) (22) Usages of the past participle (en) form: a. (used as a sentence modifier) c. (part of the passive voice construction) c. Every verb will be specified for a value of the head feature VFORM. For example. as partially exemplified in (21) and (22).

This [VFORM pres] value will be projected to the S in accordance with the HFP. the required VFORM value could be different. The monkeys kept [forgetting/*forgot/*forgotten their lines]. the S would have the [VFORM ing] and the result could not be a well-formed declarative sentence. John made Mary [cook/*to cook/*cooking Korean food]. as illustrated by the following lexical entries:   (26) a. There are various constructions in English where we need to refer to VFORM values. keep   HEAD | POS verb      COMPS VP[ing]   b. (ing) b. This is simply because the value ing is a subtype of nonfin. VP[bse] Such lexical specifications on the VP’s VFORM value will make sure that these verbs only combine with a VP complement with the appropriate VFORM value. as represented in the following: (24) S VFORM pres „ o jjjj „„„„„„„ jjjj „ jj jjjj jjjj VP NP II VFORM pres  III t ?  jjjj tttt II  tt jjjj j II tt  tt I  V NP The student ÖVV VFORM pres ÖÖ VVV Ö VV ÖÖ V ÖÖ knows the answer It is easy to verify that if we have knowing instead of knows here. We caught them [eating/*ate/*eat/*eaten the bananas]. make   HEAD | POS  verb     COMPS NP. (bse) Even though each main verb here requires a VP as its complement (the part in brackets). The following structure represents one example: 79 . (ing) c. such as: (25) a.

while intelligent does 80 . The verb forgetting has this VFORM value which is passed up to its mother VP in accordance with the HFP. an adjective like despondent optionally takes a complement.2 Complements of Adjectives There are at least two types of adjectives in English in terms of complement selection: those selecting no complements at all. ensuring that the VFORM value of the S is a subtype of fin. and those taking complements. As shown in the following examples. satisfying the basic English rule for declarative sentences. 5. The Head-Complement Rule allows the combination of the head verb kept with this VP.(27) S   POS verb HEAD 4    VFORM pst       SPR    COMPS † vv †††††††† vv †††† † vv vv vv VP vv   vv vv HEAD 4 v   1 NP   SPR 1 NP    COMPS hhh††††††† †††† hhhh ††† hhhh hhh  V  HEAD 4   SPR 1 NP  COMPS 2 VP 2 VP   5 | VFORM ing HEAD       SPR 2 NP      COMPS r hhhh rrr hhhh rr hhhh rr hh rr rr V rr   rr r HEAD 5   5 NP   SPR 2 NP  ÓYY   ÓÓ YYY Ó YY COMPS 5 ÓÓ YY ÓÓ YY ÓÓ YY ÓÓ Ó forgetting his lines for the play  N John kept The finite verb kept selects as its complement a VP whose VFORM value is ing. In the upper part of the structure. the VFORM value of the verb kept is also passed up to its mother node VP.2.

c. for example:   (30) a. One thing we can note again is that the complements also need to be in a specific VFORM and PFORM value. where PFORM indicates the form of a specific preposition. b. b. possibly of different categories (for example. The contract is subject [to/*for approval by my committee]. e. The monkey seems despondent (that it is in a cage). VP or PP). The chickens seem fond [of/*with the farmer]. fond    HEAD | POS adj      SPR NP   COMPS PP[PFORM of ] Such lexical entries will project sentences like the following: 81 . (29) a. These are similar [to/* with the bottles]. Just like verbs. eager    HEAD | POS adj      SPR NP   COMPS VP[VFORM inf ]   b. adjectives also place restrictions on the VFORM or PFORM value of their complement. Adjectives such as eager.not take any complements: (28) a. He seems intelligent (*to study medicine). The foxes seem compatible [with/*for the chickens]. fond and compatible each select a complement. f. Monkeys are eager [to leave/*leaving the compound]. d. Such restrictions are also specified in the lexical information. as illustrated in examples (29b–f). The teacher is proud [of/*with his students].

though they often may have specifiers. POS and VFORM. faith. In addition. are passed up to the final S. Each structure will satisfy all the relevant constraints and principles. The HFP ensures the HEAD features. Bill’s faith in/*for Fred’s sister c. common nouns like idea. require only a specifier. 5. the adjective eager combines with its VP[inf ] complement in accordance with the Head-Complement Rule.3 Complements of Common Nouns Nouns do not usually select complements. but no complement. the bottom of/*in the barrel 82 .(31) S VFORM pres „„ ppp „„„„„„ „„„„ ppp pp ppp pp VP 3 NP CC VFORM pres  C jjrrr  CCC rr jjjj  rr jjjj CC jj rr  rr CC   rr V C r   2 AP Monkeys VFORM pres   wrr ww rrrr COMPS 2 ww rr w rr ww rr ww rr ww ww 1 VP A are COMPS 1 VFORM inf jjvvvv jjjj vvv jjjj vvv jj vvv V VP eager |ff VFORM inf || fff | ff | ff || f || | to leave the meeting As represented in this simplified tree structure. For example. desire.2. their proximity to their neighbors/*for their neighbors b. such as proximity. the desire to become famous/*for success e. beer etc.3. book. and bottom: (32) a. the king of/*in English d. Yet there also are nouns which do require a specific type of complement. king. this rule also licenses the combination of the infinitival marker to with its VP[bse] complement (see also section 8.4) and the combination of the copula are with its AP complement.

it. but also includes a possessive phrase like John’s. and that. 5. overall we classify nouns as having three different NFORM values: normal. we can have the following lexical entries for the verbs above:   (36) a. the discussion so far has given an idea of what kinds of complement lexical categories select for. and should be represented in the following simplified lexical entries:4   (33) a. Given the NFORM feature. The use of these expletives is restricted to particular contexts or verbs. they are grammatically classified as complements of the nouns. In Chapter 6 we cover NP structures in detail. indicating that they do not have or contribute any meaning. 4 DP 83 . John/Some books/The spy disappeared. The pronouns it and there are often called ‘expletives’. There/*The spy lies a man in the park. One way to specify such lexical specifications for subjects is to make use of a form value specification for nouns: all regular nouns [NFORM norm(al)] as default specification. It/*John/*There rains. though both forms have regular pronoun uses as well. certain English verbs select only it or there as subject: (35) a. most verbs select a regular NP as subject: However.Although these complements can be optional in the right context. rained   SPR NP[NFORM it]    COMPS covers not only simple determiners like a. as noted in the previous chapter. In general. b.3 Feature Specifications for the Subject (34) a. proximity   HEAD | POS noun      SPR  1 DP     2 PP[PFORM to] COMPS   b. and there. b. faith   HEAD | POS noun      SPR  1 DP     2 PP[PFORM in] COMPS Though many more details remain to be covered for the various complement types of lexical categories. The teacher/The monkey/He fooled the students. the.

I think (that) the press has a check-and-balance function. There arose a great storm. consider the complements of think or believe: (41) a. There exists only one truly amphibian mammal. there are verbs selecting not just a phrase but a whole clause as a complement. b. A verb like exist in (37a) takes one argument when used in an example like Only one truly amphibian mammal exists. and the argument will be realized as the SPR.1 Clausal Complement or Subject Verbs Selecting a Clausal Complement We have seen that the COMPS list includes predominantly phrasal elements. There exist few solutions which are cost-effective. we first consider the verb forms which have regular subjects.  fooled   NP[NFORM norm   NP  SPR  COMPS We can also observe that only a limited set of verbs require their subject to be [NFORM there]:5 (37) a. such verbs can introduce there as the subject. c. For the sentences with there subjects. 5 Some verbs such as arise or remain sound a little archaic in these constructions. For example.b. In addition. exists   SPR 1 NP[NFORM there]      COMPS  2 NP   1 NP. b. There is a riot in the park. However. (38) a. exists   SPR  1 NP     COMPS    1 NP ARG-ST   b. b.   (40) a. *There sings a man loudly.4. They believe (that) Charles Darwin’s theory of evolution is just a scientific theory.4 5. There remained just a few problems to be solved. which is the form that occurs in the structure of (37a). 2 NP ARG-ST 5. either finite or nonfinite. 84 . The majority of verbs do not allow there as subject: (39) a. as dictated by the entry in (40a). b. *There runs a man in the park. through the Argument Realization option given in (40b).

suppressing the category information of the sentential complement:6   (43) believe   HEAD | POS verb      COMPS [VFORM fin] We can also find somewhat similar verbs like demand and require: (44) a. While S and CP are saturated categories projected from V.The C (complementizer) that here is optional. Unlike think or believe. the complement can be either S or CP. such words or phrases cannot be complements of verbs like think or believe. This means that we can subsume these two uses into the following single lexical entry. these verbs will have one of the following two COMPS values: (42) a. COMPS COMPS S[VFORM fin] CP[VFORM fin] If the COMPS value only specifies a VFORM value. John demanded [that she stop phoning him]. The rules require [that the executives be polite]. VP and V are not saturated. b. implying that this kind of verb selects for a finite complement clause of some type. This is because complements are typically saturated phrases. which we will notate as a [VFORM fin] clause. That is. these verbs which introduce a subjunctive clause typically only take a CP[VFORM bse] as their complement: the finite verb itself is actually in the bse form. 85 . Observe the structure of (44b): 6 Although the categories V or VP are also potentially specified as [VFORM fin]. b. with no unsatisfied requirements for complements or specifiers.

and this COMPS requirement is discharged at its mother VP: this satisfies the Head-Complement Rule. This encodes the intuition that a complementizer ‘agrees’ with regard to VFORM value with its complement sentence. not from its complement S. NP and CP are abbreviations for feature structure descriptions that include the information [POS noun] and [POS comp]. b. There are also verbs which select a sequence of an NP followed by a CP as complements. c. We told Tom that he should consult an accountant. as represented in this lexical entry:   (46) that         HEAD POS comp     VFORM 1       SPR    COMPS S[VFORM 1 ] This lexical information will then allow us to pass on the VFORM value of S to the head C and then be percolated up to the CP according to the HFP.(45) S € vv €€€€€ v €€€ vv v €€€ vv v vv vv VP NP CC COMPS  C n€€  CCC nnn €€€€€ €€€ nnn CC  nn €€ CC  nnn C  2 CP  V The rules COMPS 2 VFORM bse € nnn €€€€€ €€€ nnn €€€ nnn nnn C require VFORM bse VFORM bse € nnn €€€€€ €€€ nnn €€€ nnn nnn the executives be polite S that The verb require selects a bse CP complement. There is one issue here with respect to the percolation of the VFORM value: the CP must be bse. One way to make sure this is to assume that the VFORM value of C is identical with that of its complement S. Joe warned the class that the exam would be difficult. Mary convinced me that the argument was sound. respectively: (47) a. The COMPS value of such verbs will be as in (48): 86 . but this information comes from the head C.

John would prefer for the children to finish the oatmeal. The structure of (49a) is familiar. headed by the complementizer for. CP[VFORM fin] In addition to the that-type of CP. but now has a nonfinite VFORM value within it: (50) S tƒƒ tt ƒƒƒƒƒƒ ƒƒƒ tt tt tt t tt VP NP 7% 7 COMPS %7 kƒƒƒƒƒ %% % 777 ƒƒƒƒ kkkk %% 77 ƒƒ kkkk kk %% 77 2 CP % V Tom COMPS 2 VFORM inf kƒƒƒƒƒ ƒƒƒƒ kkkk ƒƒ kkkk kk C intends VFORM inf VFORM inf ƒ tt ƒƒƒƒƒƒ tt ƒƒƒƒ t tt tt t tt VP NP VFORM inf t kkkk tttt tt kkkk kkk tt tt t V Sam VP }ee VFORM inf }} eee } ee }} ee }} e }} to review that book S for The structure given here means that the verb intends will have the following lexical information:   (51) intends        POS verb  HEAD      VFORM pres     COMPS CP[VFORM inf ] 87 . there is an infinitive type of CP. b. The data show that verbs like intend and prefer select an infinitival CP clause. Tom intends for Sam to review that book. Some verbs select this nonfinite CP as the complement: (49) a.(48) COMPS NP.

to has a distribution very similar to the finite modal auxiliaries such as will or must. inheriting its VFORM value too.To fill out the analysis. John intends to review the book. In fact. the complementizer for selects an infinitival S as its complement. infinitival VPs. to         POS verb HEAD      VFORM inf     COMPS VP[VFORM bse] Just like the complementizer that. b. By underspecifying the category information of complements. (coordination of inf VPs) c. For [John to tell Bill such a lie] and [Bill to believe it] is outrageous.   (52) a. For John either [to make up such a story] or [to repeat it] is outrageous. we can see that base VPs. 88 . (coordination of inf Ss) Given that only like categories (constituents with the same label) can be coordinated. which we treat as an (infinitive) auxiliary verb. John would prefer to finish the oatmeal. we need explicit lexical entries for the complementizer for and for to. differing only in the VFORM value (see section 8.3. and infinitival Ss are all constituents. For John to either [make up such a story] or [repeat it] is outrageous.4). (coordination of bse VPs) b. we can generalize this subcategorization information:   (55) intend   HEAD | POS  verb     COMPS [VFORM inf ] 7 Tensed VPs can be coordinated regardless of their different tense values as in Kim [alienated cats] and [beats his dog].7 One thing to note here is that the verbs which select a CP[VFORM inf ] complement can also take a VP[VFORM inf ] complement: (54) a. for       POS comp  HEAD     VFORM inf     COMPS S[VFORM inf ]   b. The evidence that the complementizer for requires an infinitival S can be found from coordination data: (53) a.

Such subcategorization differences are hard to predict just from the meaning of verbs: they are simple lexical specifications which language users need to learn. tend. try. Consider the following patterns: (61) a. 8 This 9 For type hierarchy is adopted from Kim and Sag (2006). the function of verbal. vrr vrr vv rrr vv rrr v vv rr rr vv rr vvv rr vv v v noun comp verb According to the hierarchy.. b. There is another generalization that we need to consider with respect to the property of verbs that select a CP: most verbs that select a CP can at first glance select an NP. and others select only a VP[inf ] as attested by the data: (56) a. see Chapter 12. an element specified as [POS nominal] can be realized either as [POS noun] or [POS comp]. respectively. For examples. Tom tends to avoid confrontations. the type nominal is a supertype of both noun and comp. *Tom tends for Mary to avoid confrontations. we can assume that English parts of speech come in families and can profitably be analyzed in terms of a type hierarchy as following:8 (60) part-of-speech ‡‰ v‰‡‰‰‰ gggrr ‡‡‡‰‰‰‰‰ ‰‰‰ ggggg vv rrr ‡‡‡‡‡‡‡‡ ‰‰‰‰ gggg vvv rr ‡‡‡‡‡ ‰‰‰‰‰ ‰ gg v rr ‡‡‡‡‡ ‰‰‰‰‰‰‰ ggggg vv gggg nominal verbal adj prep .. John believed it/that he is honest. 89 . too: (59) a. b. These will correspond to the phrasal types NP and CP.9 The hierarchy implies that the subcategorization pattern of English verbs will refer to (at least) each of these types.Since the specification [VFORM inf ] is quite general. *Joe hoped for Beth to find a solution. b. (58) a. John mentioned the issue to me/mentioned to me that the question is an issue. this does not mean that all verbs behave alike: not all verbs can take variable complement types such as an infinitival VP or S. In accordance with the basic properties of systems of typed feature structures. However. Joe hoped to find a solution. it can be realized either as CP[VFORM inf ] or VP[VFORM inf ]. hope. Tom tried to ask a question. b. Should we have two lexical entries for such verbs or can we have a simple way of representing such a pattern? To reflect such lexical patterns. (57) a. *Tom tried for Bill to ask a question. She pinched [his arm] as hard as she could.

[John] loves Bill b. b. *[That John snores] loves Bill. The contrast here means that verbs like bother can have two realizations of the ARG-ST. . . [That John snores] bothers me. verbal.2 Verbs Selecting a Clausal Subject In addition to CP as a complement. These three patterns illustrate that English transitive verbs come in at least three varieties. We hope [that such a vaccine could be available in ten years].4. the ARG-ST list specifies the argument elements that the verbs select (in the order Subject. b. The part-of-speech type hierarchy in (60) allows us to formulate simple lexical constraints that reflect these subcategorization patterns. . The POS value of a given element is the part-of-speech type that a word passes on to the phrases it projects. b. bothers   SPR 1 [POS nominal]      COMPS  2 NP   1. and nominal: (64) a. but there are other clausal types: 90 . [John] bothers me. In each class. ). Direct Object. whereas those like love allow only one:   (67) a. ARG-ST ARG-ST ARG-ST NP. *We hope [the availability of such a vaccine in ten years]. (62) a. NP. c. [POS nominal]. . we also find some cases where a CP is the subject of a verb: (65) a. . . (66) a. .b. NP[POS noun]. . NP. 5. . CP[POS comp]. . (63) a. . Cohen proved [the independence of the continuum hypothesis]. Cohen proved [that the continuum hypothesis was independent]. b. *She pinched [that he feels pain]. loves   SPR 1 NP      COMPS 2 NP    1. . 2 ARG-ST These different realizations all hinge on the lexical properties of the given verb. A clausal subject is not limited to a finite that-headed CP. making reference to noun. and only some verbs allow the dual realization described by (67a). 2 ARG-ST   b.

For example. *That Fred was unpopular nominated Bill. as in (69): (69) a. *That Tom missed the lecture was enjoyable. whereas eager selects an infinitival CP: (71) a. (wh-question CP subject) Naturally. *How much money Gordon spent is true. Tom seems eager [for her brother to catch a cold]. (infinitival CP or VP subject) c. Tom is insistent [that the witnesses be truthful]. each particular predicate dictates which kinds of subjects are possible. Tom seems eager [to catch a cold]. b. as in (68).(68) a. certain adjectives can also select CPs as their complements. [(For John) to train his horse] would be desirable. the difference between the two verbs nominate and surprise can be seen in these partial lexical entries:   (70) a.4. (72) a. 2   b. We can easily find more adjectives which select a CP complement: 91 .3 Adjectives Selecting a Clausal Complement Like verbs. 5. b. (that-clause CP subject) b. d. 2 Unlike nominate. [Which otter you should adopt first] is unclear. confident and insistent select a finite CP. nominate        1 NP SPR VAL    2 NP COMPS     ARG-ST 1 . (subjunctive that-clause CP subject) d. *For John to remove the mother is undeniable. the first argument of surprised can be a nominal. c. This means that its subject can be either an NP or a CP. b. [That John sold the ostrich] surprised Bill. [That the king or queen be present] is a requirement on all Royal weddings. surprise        1 [POS nominal] SPR VAL    2 NP COMPS     ARG-ST 1 . For example. and which are not. Tom is confident [that the elephants respect him].

e. We were glad it was over. I am ashamed that I neglected you. b. We were thankful that no one had been hurt. We are content for the cleaners to return the drapes next week. and the HFP. for (72b): (75) S mm  mmm mm  mmm VP NP @ mmqqqq mmm "" @@@ qq mm " qq mmm "" @@@ qq " q " V 2 AP Tom q COMPS 2 ww qqq qq ww qq ww qq ww w q ww A seems COMPS 3 3 VP eager VFORM inf  ww  ww  w ww ww ww VP V VFORM bse m mmm   mmm mmm to V NP RR  RRR  catch a cold 92 . I am delighted that Mary finished his thesis.(73) a. c. d. interacting with the Head-Complement Rule. The lexical entries for some adjectives are given in (74):   (74) a. insistent   HEAD | POS adj    COMPS CP[VFORM bse]   c. eager   HEAD | POS adj    COMPS [VFORM inf ] Such lexical entries. the Head-Specifier Rule. confident   HEAD | POS adj    COMPS CP[VFORM fin]   b. can license analyses such as (75).

When the head adjective eager combines with its complement, VP[inf ], it satisfies the HeadComplement Rule. The same rule allows the combination of the verb seem with its AP complement. 5.4.4 Nouns Selecting a Clausal Complement

Nouns can also select a CP complement, for example, eagerness: (76) a. (John’s) eagerness [for Harry to win the election] b. (John’s) eagerness [to win the election] These examples imply that eagerness will have the following lexical information:   (77) eagerness   HEAD | POS noun      COMPS [VFORM inf ] This lexical entry will allow a structure like the following: (78) NP ffff€€€€€ fffff €€€ fff € fffff N DP D n€€ nnn €€€€€  DDD nn  €€ nnn  DDD  2 VP  N John’s COMPS 2 VFORM inf oyy ooo yyyyy yy ooo oo eagerness to win the election

One pattern that we can observe is that when a verb selects a CP complement, if there is a corresponding noun, it also selects a CP: (79) a. Bill alleged that Fred signed the check. b. We believe that the directors were present. c. We convinced him that the operation is safe. (80) a. the allegation that Fred signed the check b. the belief that the directors were present c. his conviction that the operation is safe This shows us that the derivational process which derives a noun from a verb preserves the COMPS value of that verb. A caution here is that not all nouns can of course select a CP complement: (81) a. *his attention that the earth is round b. *his article that the earth is flat
93

c. *the ignorance that James can play the flute d. *the expertise that she knows how to bake croissants These nouns cannot combine with a CP complement, indicating that they do not have CP in the value of COMPS. 5.4.5 Prepositions Selecting a Clausal Complement In general, prepositions in English cannot select a CP complement. (82) a. *Alan is thinking about [that his students are eager to learn English]. b. *Fred is counting on [for Tom to make an announcement]. However, wh-CPs, sometimes known as indirect questions, may serve as prepositional complements. (83) a. The outcome depends on [how many candidates participate in the election]. b. Fred is thinking about [whether he should stay in Seoul]. These facts show us that indirect questions have some feature which distinguishes them from canonical that- or for-CPs, and makes them somehow closer to true nouns (for NP is the typical complement for a preposition).

94

5.5

Exercises
(i) ignore, doubt, deny, prefer, intend, glad, unsure, confident, allegation, ignorance

1. Check if each of the following expressions selects for a clausal complement:

Write out the examples which justify your classifications. 2. Check if each of the following expressions selects a clausal subject or not. (i) annoy, vanish, remain, admire, select, mandatory, enjoyable, apparent Write out the examples which justify your classifications. 3. For each sentence, draw a tree structure and then give the COMPS value (including VFORM and PFORM value) for the italicized word. (i) a. b. c. d. e. f. g. h. i. j. k. The offer made Smith admire the administrators. John tried to make Sam let George ask Bill to keep delivering the mail. The soldiers must enforce Bill to make the baby be quiet. John enjoyed drawing trees for his syntax homework. The picture on the wall reminded him of his country. Free enterprise is compatible with American values and traditions. They were taking a hard look at possible FTA. We need to be in frequent contact with the clients. The contract is subject to approval by my committee. Acknowledge that everyone has limits. We are aware of the existing problems.

4. Identify errors in the following sentences, focusing on the form values of verbs, adjectives, and nouns, and/or their COMPs values. (i) a. b. c. d. e. f. g. h. i. j. k. l. m. n. *Why don’t you leaving me concentrate on my work? *The general commended that all troops was in dress uniform. *My morning routine features swim free styles slowly for one hour. *You should avoid to travel in the rush hour. *You should attempt answering every question. *The authorities blamed Greenpeace with the bombing. *The authorities charged the students of the cheating. *Sharon has been eager finishing the book. *We respect Mary’s desire for becoming famous. *John referred from the building. *John died to heart disease. *John paid me against the book. *We were glad what to do. *She was busy to make lunch.

5. Draw trees for the following examples with detailed NP structures.
95

(i) a. The constant rain forced the abandonment of the next day’s competitions. b. The committee will study the feasibility of setting up a national computer network. c. Aloe may have an analgesic effect on inflammation and minor skin irritations. d. The public never had faith in his ability to handle the job. e. He repeated his claim that the people backed his action. 6. In general the present and base forms of a main verb are identical, but can be distinguished in the case of be, with its present form are: (i) a. We made them take the money. b. *We made them are rude. c. We made them be rude. Do the same substitution test for the following sentences and determine whether the italicized verb is the base or present form. (ii) a. b. c. d. Do not use these words in the beginning of a sentence. We know the witnesses seem eager to testify against the criminal. Jane isn’t sure whether the students keep the books. Why not try to catch the minnows?

7. Read the following text and provide the lexical entries for the underlined words. In doing so, try to specify the VFORM or PFORM value of the complement(s). (i) The study of grammar helps us to communicate more effectively. Quite simply, if we know how English works, then we can make better use of it. For most purposes, we need to be able to construct sentences which are far more complicated than David plays the piano. A knowledge of grammar enables us to evaluate the choices which are available to us during composition. In practice, these choices are never as simple as the choice between David plays the piano and *plays David piano the. If we understand the relationship between the parts of a sentence, we can eliminate many of the ambiguities and misunderstandings which result from poor construction. In the interpretation of writing, too, grammatical knowledge is often crucially important. The understanding of literary texts, for example, often depends on careful grammatical analysis. Other forms of writing can be equally difficult to interpret. Scientific and academic writing, for instance, may be complex not just in the ideas they convey, but also in their syntax. These types of writing can be difficult to understand easily without some familiarity with how the parts relate to each other. The study of grammar enables us to go beyond our instinctive, native-speaker knowledge, and to

96

use English in an intelligent, informed way.10

‘Introducing the Internet Grammar of English’ at http://www.ucl.ac.uk/internet-grammar/ intro/intro.htm.

10 From

97

January. everybody. information. whom. etc. that. difficulty. music. his. gold. Common nouns and proper nouns display clear contrasts in terms of the combinatorial possibilities with determiners as shown in the following chart: 99 . how. laziness. etc. furniture. they. who. but also denote abstract and intangible concepts such as happiness. him. remark. Seoul. places. anywhere. and pronouns.1 Classification of Nouns Nouns not only represent entities like people. Whether a noun is countable or not does not fully depend on its reference. Stanford.6 Noun Phrases and Agreement 6. what. she. butter. which. nouns fall into three major categories: common nouns. or things. when. Kyung Hee. and we have nouns like furniture/*furnitures which are only mass and a chair/chairs which are only count. One important aspect of common nouns is that they are either count or non-count. etc. where. proper nouns. hope. examples like difficulty which is mass (non-count) but difficulties which is count suggest how subtle the distinction can be. proper noun personal relative interrogative indefinite pronoun As shown here. he. etc. Palo Alto. who. why. Such diversity of reference renders it difficult to classify nouns solely according to their meanings. The following chart shows the canonical classification of nouns taking into account semantic differences. but also considering their formal and grammatical properties: (1) Types of Nouns in English: common noun countable uncountable desk. etc. and so forth. Proper nouns denote specific people or places and are typically uncountable. somebody. etc. etc. book. nobody. anybody.

Student discovers planet. Rice is available in most countries. . an. that.(2) Combinatory Possibilities with Determiners: Common N countable uncountable Only N Einstein *book music the + N *the Einstein the book the music a+N *an Einstein a book *a music some + N *some Einstein *some book some music N+s *Einsteins books *musics Proper N neutral cake the cake a cake some cake cakes Proper nouns do not combine with any determiner. (4) a. Army receives high-tech helicopter. with focus on the agreement relationships of nouns with determiners and of subjects with verbs. whereas uncountable nouns combine only with some or the. e. which. 1 The 100 . or plural count nouns.g. cake can be countable when it refers to a specific one as in I made a cake. His friend learned dancing. Students study English for 4 hours a day. as can be seen from the chart..2 Syntactic Structures 6. his. are fully grammatical as bare nouns phrases. As we have seen in Chapters 2 and 4. some common nouns may be either count or non-count..2. b. a book and books). This chapter deals with the structural. nouns are of pivotal importance in English. forming the semantic and structural components of sentences. but can be noncountable when it refers to ‘cake in general’ as in I like cake. no. *Student studies English for 4 hours a day. depending on the kind of reference they have. any. some. there are various kinds of expressions which can serve as determiners including a. We can see here that mass nouns. As noted in Chapter 1. common nouns can have a determiner as a specifier. . 6. count nouns cannot be used without a determiner when they are singular: (3) a. style of English used in headlines does not have this restriction. as well as a possessive phrase: (5) a. how.1 Common Nouns As noted before. For example. some. few. b. *Book is available in most countries.1 This has the consequence for our grammatical analysis that singular countable nouns like student must select a determiner as specifier. In particular. . Together with verbs.g. count nouns have singular and plural forms (e. this. Meanwhile. and functional dimensions of NPs. semantic. much. unlike proper and pronouns.

The president’s bodyguard learned surveillance. The King of Rock and Roll’s records led to dancing.    SPR  SPR NP      COMPS COMPS  brother         friend         HEAD | POS noun  c. c. countable noun brother requires a DP as its specifier. We can also allow an N to directly project into an NP. because they are phrases. These possessive NPs my brother’s or the president’s are not determiners. determiners like my and uncountable nouns like rice select neither specifier nor complement whereas the possessive marker ’s and the countable noun friend requires a NP and DP specifier.  SPR DP  COMPS As given here. respectively.  b. This means that not only a simple determiner but also a possessive NP can function as a specifier.b. we can assume that a countable noun selects not a simple determiner but a DP (determiner phrase) as its specifier: These lexical entries will project NP structures like the following for (5b):2 (7) NP mm‘‘‘‘‘‘‘‘‘‘‘‘‘‘‘ mmm mm mmm N mmm  mmm 1 DP SPR 1 DP   ww   ww COMPS  ww  ww w  w ww ww Det N ww   ww 2 NP SPR 2 NP SPR 1 DP S   ÙÙ SSS COMPS COMPS SS ÙÙ SS ÙÙ ÙÙ my brother ’s friend     2 Keen readers may have noticed that we allow a vacuous projection from N to N and to NP when no specifier or complement is combined. Meanwhile. My bother’s friend learned dancing. Reflecting this generalization. 101 . We take such phrases as DPs headed by the Det ’s (cf. Abney (1987)).  SPR DP  COMPS  HEAD | POS noun  d. d. Let’s consider the lexical entries for the relevant words:     (6) my ’s     HEAD | POS det  HEAD | POS det      a.

I got worried when the neighbors let their/*his dogs out. b. d. b. 6.2. Edward usually remembered to send a copy of his e-mail to himself. John. the noun friend does not select a complement. Bill. It’s nothing like the America I remember. c. *a January. No John Smiths attended the meeting. .2. b. a little something. This John Smith lives in Seoul. gender. Judy threw it/*them into the garbage can. 6. I asked myself why I had faxed invitations to everyone in my office building. we b. Reflexive pronouns are special forms which typically are used to indicate a reflexive activity or action. President Lincoln delivered his/*her Gettysburg Address in 1863. Reflexive pronouns: myself. January. Reciprocal pronoun: each other Personal pronouns refer to specific persons or things and take different forms to indicate person. *a Seoul. itself c. c. e. . and thus projects into an intermediate phrase which in turn combines with the specifier DP my brother’s whose head is the possessive determiner. number. they. After reading the pamphlet. (10) a. b.3 . *a John. he. which can include mental activities. she. They participate in agreement relations with their antecedent. himself.2 Pronouns The core class of pronouns in English includes at least three main subgroups: (8) a. Seoul. the phrase which they are understood to be referring to (indicated by the underlined parts of the examples in (9)). (9) a. and case.3 Proper Nouns Since proper nouns usually refer to something or someone unique.As represented here. My brother is an Einstein at maths. herself. There are three Davids in my class. these personal or reflexive pronouns neither take a determiner nor combine with an adjective except in very restricted constructions.. yourself. . you. However. they do not normally take a plural form and cannot occur with a determiner: (11) a.g. . a certain someone) 102 . Personal pronouns: I. 3 These restricted constructions involve some indefinite pronouns (e. As noted earlier. *a Bill. . proper nouns can be converted into countable nouns when they refer to a particular individual or type of individual: (12) a. it. . After the party.

they are involved in determiner-noun agreement.    SPR  COMPS Common nouns thus impose a specific NUM value on the specifier: 103 .    SPR  COMPS    HEAD  b.3 6. there must be an agreement relationship between the two: (13) a. proper nouns are converted into common nouns.3. When they combine with a determiner. and take other nominal modifiers. H AGR 1 This revised rule guarantees that English head-specifier phrases require their head and specifier to share agreement features.In such cases. *this books/*that books/these books/those books c. All countable nouns are used either as singular or plural. leading us to revise the Head-Specifier Rule as following: (14) Head-Specifier Rule: XP → SPR AGR 1 . First. *few dog/few dogs The data in turn means that the head noun’s number value should be identical to that of its specifier. may select a specifier. 6.1 Agreement Types and Morpho-syntactic Features Noun-Determiner Agreement Common nouns in English participate in three types of agreement. this book/that book b. implying that determiners and nouns have NUM (number) information as their syntactic AGR (agreement) value: (15)  a      POS det    AGR | NUM sing       book      POS noun    AGR | NUM sing     DP[NUM sing]     HEAD  a.

. . . my book. This is in accord with the grammar since the value num is a supertype of sing. *those book. their NUM value is underspecified as num(ber). Formally. with two subtypes sing(ular) and pl(ural): (18) num oyy ooo yyyyy oo yyy oo y ooo sing pl Given this hierarchy. if it is the subject. Notice that the AGR value on the head noun book is passed up to the whole NP. so that it can combine with a singular VP. the grammar of English has the underspecified value num for the feature NUM. nouns like book requiring a singular Det can combine with determiners like the whose AGR value is num. b. marking the whole NP as singular. . 104 . . no book. the book. no and my are not specified for a NUM value. there is nothing preventing a singular noun from combining with a determiner which is not specified at all for a NUM value: (17) a.(16) NP AGR | NUM sing k Óyyyyy yyy ÓÓ yyy ÓÓ ÓÓ Ó ÓÓ N ÓÓ   Ó ÓÓ AGR | NUM sing \   1 DP[NUM sing]   SPR 1 DP    COMPS a N   AGR | NUM sing     SPR 1 DP    COMPS book The singular noun book selects a singular determiner like a. Determiners like the. In addition. . That is. *these book.

b. PER(SON). 105 .3. he/*she has to know the rules of the game. The pronoun he or it here needs to agree with its antecedent not only with respect to the number value but also with respect to person (1st. Let us look at some slightly complex examples: (21) a. This shows us that nouns have also information about person and gender as well as number in the AGR values: (20)  book                   he      POS noun        NUM sing     HEAD     AGR GEN neut  a. a second type of agreement is pronoun-antecedent agreement. The NUM values are shown in (18) above. Throughout it all he seeks to answer the questions about himself. In the book. 2nd or 3rd.3. in particular when the subject is a personal pronoun: (22) a. 3rd) and gender (masculine. You are/*is the only person that I can rely on.       PER 3rd      SPR DP[NUM sing] VAL COMPS      POS noun          NUM sing   HEAD      AGR PER 3rd  b. nouns have NUM. fem(inine) or neut(er). 6. and the person value is also involved in agreement relations.2 Pronoun-Antecedent Agreement As noted earlier. which is one of the most important phenomena in English syntax.3 Subject-Verb Agreement The third type of agreement is subject-verb agreement. b. Students studying English read/*reads Conrad’s Heart of Darkness while at university. and GEN(DER) for their AGR values. the subject and the verb need to have an identical number value. he talks about his ups and downs at McLaren. The characters in Shakespeare’s Twelfth Night *lives/live in a world that has been turned upside-down. 2nd.6. If John wants to succeed in corporate life. It was an in-depth analysis of Plato’s opinions about possible governmental forms.        GEN masc            SPR VAL   COMPS As we have briefly shown. The critique of Plato’s Republic was written from a contemporary point of view. the GEN value can be masc(uline). The PER value can be 1st. or neuter) values too. As we can see here. c. feminine. (19) a. as indicated in (19).

The boys swim/*swims. English verbs will have at least the following selectional information: (24)  swims                  POS verb   HEAD  VFORM pres       PER 3rd   VAL | SPR NP NUM sing The present-tense verb swims thus specifies that its subject (SPR’s value) carries a 3rd singular AGR information. b. In this system. 106 .b. what subject-verb agreement is is structure-sharing between the AGR value of the subject (SPR value of the verb) and that of the NP that the VP combines with. we will use some simpler examples: (23) a. This lexical information will license a structure like the following: (25) S ƒ lll ƒƒƒƒƒƒ lll ƒƒƒƒ ll l ƒƒƒƒ lll lll VP 2 NP         PER 3rd PER 3rd AGR  SPR 2 NPAGR     NUM sing NUM sing D  DDD  DD DD   D  V The boy SPR 2 NP swims Only when the verb combines with a subject satisfying its AGR requirement will we have a well-formed head-subject phrase. To show how the agreement system works. if this verb were to combine with a subject with an incompatible agreement value. In other words. These facts show us that a verb lexically specifies the information about the number as well as person values of the subject that it selects for. we would generate an ungrammatical example like *The boys swims in (23b). He is/*are the only person that I can rely on. The boy swims/*swim.

So. we would not expect the singular verb form to be possible at all in these examples. rather than with a plurality of prawns. closely related to the semantics of a nominal. However. via the notion of ‘index’ that we have just introduced. Both NP and S require agreement between the head and the specifier. If we simply assume that the subject phrase inherits the morphosyntactic agreement features of the head noun (hash) browns in (26b) and (King) prawns in (27). as reflected in the revised Head-Specifier Rule in (14). [The hash browns at table nine] are/*is getting cold. the subject refers to a person who ordered hash browns. When (26b) is spoken by a waiter to another waiter. in addition to the morphological AGR value introduced above. it must be anchored to an individual in the situation described. [The hash browns at table nine] is/*are getting angry. We call this anchoring value the noun phrase’s ‘index’ value. that is. Often what a given nominal refers to in the real world is important for agreement – index agreement. and may be inherited by phrases built out of them. This then requires us to distinguish the morphological AGR value and semantic (SEM(ANTIC)) IND(EX) value. b. The index of hash browns in (26a) must be anchored to the plural entities on the plate.The acute reader may have noticed that there are similarities between noun-determiner agreement and subject-verb agreement. consider now the following examples adopted from Nunberg (1995): (26) a. and somewhat separate from the syntactic agreement feature AGR. and requires that these features match those of the verb.4 Semantic Agreement Features What we have seen so far is that the morphosyntactic AGR values of noun or verb can be specified. In the interpretation of a nominal expression.  boy         POS noun SYN | HEAD    AGR | NUM sing      SEM | IND | NUM sing 107 . Index agreement involves sharing of referential indexes. whereas that of hash browns in (26b) is anchored to a customer who ordered the food. A somewhat similar case is found in (27): (27) King prawns cooked in chili salt and pepper was very much better. (28) a. but context-dependent in various ways. 6. a simple dish succulently executed. Here the verb form was is singular to agree with the dish being referred to. in the way that agreement works inside NP and inside S. each noun will also have a semantic IND value representing what the noun refers to in the actual world. English agreement is not purely morpho-syntactic as described in the sections above.

Different from nouns. etc. s2. s1. If the referent of this subject (its IND value) does not match. etc) whereas that of a verb or predicative adjective will be a situation index such as s0. And the verb will place a restriction on its subject’s IND value rather than its morphological AGR value:4 (29)  swims        HEAD POS verb    SYN  AGR | NUM sing      VAL | SPR NP[IND | NUM sing]   SEM | IND s0             The lexical entry for swims here indicates that it is morphologically marked as singular (the AGR feature) and selects a subject to be linked to a singular entity in the context (by the feature IND). j. k. 4 The 108 .b. the verb’s own IND value is a situation index (s0) in which the individual referred to through the SPR value is performing the action of swimming. we would generate an ungrammatical example like *The boys swims: (30) *S jj„„„„„„ „„„„ jjjj „„„„ jjjj „„ jjjj jj VP 2 NP        PER 3rd PER 3rd    INDj  SPR 2 NPi  NUM sing NUM pl FF  FFF  FF  FF   V The boys SPR 2 NP i swims IND value of a noun will be an individual index (i.  boys         POS noun SYN | HEAD    AGR | NUM pl      SEM | IND | NUM pl The lexical entry for boy indicates that it is syntactically a singular noun (through the feature AGR) and semantically also denotes a singular entity (through the feature IND).

as in examples like (26b).  AGR | NUM pl      1 | NUM pl SEM | IND   hash browns        (when referring to a customer. 5 As indicated here. The feature SYN includes HEAD and SPR and COMPS.. there could be a mismatch between the morphological form of a noun and the index value that the noun refers to. This means that depending on the context. In fact. This is represented in (32). (32) Morpho-syntactic agreement (AGR)  Det  head-noun O verb O . This means that given an appropriate context. or to a dish) POS noun  b. [Five pounds] is/*are a lot of money. SYN | HEAD  AGR | NUM pl      SEM | IND 1 | NUM sing In the lexical entry (31b). but the subject has a different IND value j. but they can be different. we will add more to this part.. As our discussion goes on. In the most usual cases. the required subject has the IND value i. hash browns can have different IND values:5   (31) hash browns        (when referring to the food itself) POS noun SYN | HEAD  a. nothing blocks mismatches between the two (AGR and IND) as long as all the other constraints are satisfied. Given the separation of the morphological AGR value and the semantic IND value. 109 .As we can see here. whereas both subject-verb agreement and pronoun-antecedent agreement are index-based agreement. the AGR and IND value are identical. (33) [Four pounds] was quite a bit of money in 1950 and it was not easy to come by. which is internally plural. Observe further examples in the following: (34) a. Index agreement (IND) Such agreement patterns can be clearly found in examples like the following where the underlined parts have singular agreement with four pounds. the AGR’s NUM value is plural but its IND’s NUM value is singular. The feature SEM is for semantic information. As shown by (26). What this indicates is that subject-verb agreement and noun-specifier agreement are different. English determiner-noun agreement is only a reflection of morpho-syntactic agreement features between determiner and noun. the lexical expression now has two features SYN (SYNTAX) and SEM (SEMANTICS). the reference hash browns can be transferred from cooked potatoes to the customer who ordered it.

b. conceptualized as referring to a single measure. [Two miles] is/*are as far as they can walk. the head noun has to be plural. For proper agreement inside the noun phrase. the index value has to be singular. (35)   pounds         POS noun HEAD        AGR 1 | NUM pl    SYN         SPR DP AGR 1     SEM | IND | NUM sing As indicated in the lexical entry (35). as represented in (35). The nouns pounds and drops here are morphologically plural and thus must select a plural determiner. In the present analysis. this would mean that pounds will combine with a plural determiner but with a singular verb. [Two drops] deodorizes/*deodorize anything in your house. This is possible. but for subject-verb agreement the noun has to be singular. the plural subject combines with a singular verb. The present analysis thus generates the following structure for the sentence (34a): S jj„„„„„„ „„„„ jjjj jjjj „„„„ jjjj „ j  VP AGR 1   3 NP SPR 3 NP i IND i U }ssss }} UU } ss }} UU }} ss } } UU s }} }} UU }} }} } } UU N U   V Det NP AGR 1  QQ SPR 3 NP i AGR 1 | NUM pl   QQQ IND i  QQ  QQ   QQ   Five pounds is a lot of money 110 (36) . Consider the example in (34a). But when these nouns are anchored to the group as a whole – that is. the morpho-syntactic number value of pounds is plural whereas the index value is singular. as argued so far. [Fifteen years] represents/*represent a long period of his life. e. [Fifteen dollars] in a week is/*are not much. c. An apparent conflict arises from the agreement features of the head noun. since the index value is anchored to a singular individual in the context of utterance. In all of these examples with measure nouns. as noted earlier in section 2. d.

The head noun government or team is singular so that it can combine with the singular determiner this. we could represent the relevant information of the expressions participating in these agreement relationships as in (38). (38) a. this and government agree each other in terms of the morphosyntactic agreement number value whereas the index value of government is what matters for subject-verb agreement. 6. team/government         POS noun   SYNHEAD    AGR | NUM sing      SEM | IND | NUM pl As represented in (38a) and (38b).5 Partitive NPs and Agreement 6. b. there are two main types of NP in English: simple NPs and partitive NPs. most students c.5. as in the following authentic examples:6 (37) a.A similar mismatch is also found in cases with terms for social organizations or collections. all students some speakers. the neutral determiner the is much better.  this         POS det   HEAD AGR | NUM sing   b.1 Basic Properties With regard to the NP-internal elements between which we may find instances of agreement. To these people. 6 To 111 . some objections b. (39) a. In preparation for the return fixture this team have/has trained more efficiently than they had in recent months. shown in (39) and (40) respectively. But the conflicting fact is that the singular noun phrase can combine even with a plural verb have as well as with a singular verb has. This in turn means that when government refers to the individuals in this given group. This is possible since the index value of the subject can be anchored either to a singular or to plural kind of entity. the plural verb does not go with the singular subject this team. the whole NP this government carries a plural index value. More precisely. This Government have/has been more transparent in the way they have dealt with public finances than any previous government.

the partitive phrases have a quantifier followed by an of -phrase. neither cars (40) a. determiners such as the. *no of your books vs. the students b. neither of the cars As in (40). little and some to cooccur with a lower of -NP containing a singular count noun. many of the students f. Some problems vs.d. c. and in the of -phrase. First. b. the same elements serving as determiners cannot directly precede such nouns. As shown in (42). no book(s) Third. *each of students b. *the of the students vs. much of her worry e. The partitive constructions in (43) allow a mass (non-count) quantifier such as much. Each student vs. (44) a. most of the students c. We listened to as little of his speech as possible. much worry e. Observe the contrast between (43) and (44): (43) a. *every of his ideas vs. 112 . b. I read some of the book. designating a set with respect to which certain individuals are quantified. simple NPs and partitive NPs have different restrictions relative to the semantic head. as shown in (41): (41) a. (45) a. not all determiners with quantificational force can appear in partitive constructions. She doesn’t believe much of that story. every and no cannot occupy the first position: (42) a. One of the people was dying of thirst. all of the students d. each of the students vs. these partitive NPs are different from simple NPs in several respects. c. the lower NP in partitive phrases must be definite. *We listened to as little speech as possible. many students f. some of the objections b. In terms of semantics. But as we can see in (44). some of the problems vs. How much of the fresco did the flood damage? d. no quantificational NP is allowed. Another difference concerns lexical idiosyncrasies. *I read some book. *How much fresco did the flood damage? d. *some of many problems Second. *She doesn’t believe much story. every idea c.

The partitives can be headed by quantifiers like one and many. as shown in (45) and (46) but unlike many. two-thirds of. Most of the children are here. but if the meaning is about how many of something is meant. b. Many of the people were dying of thirst. the verb is plural. b. (47) Type I: a. All of these cars belong to me. half. c. a number of (but not the number of ) An effective way of capturing the relations between Type I and Type II constructions involves the lexical properties of the quantifiers. (49) half of. part of. neither and none. *some of water 113 . If the meaning of these phrases is about how much of something is meant. (46) a. the number value of the partitive phrase depends on the preceding head noun whereas in Type II. Neither of the cars has air conditioning. Many people were dying of thirst. the verb is singular. most and all may combine either singular or plural verbs.2 Two Types of Partitive NPs We classify partitive NPs into two types based on the agreement facts. f. which select an of -NP inside which the NP is definite: (50) a. the majority of. and call them Type I and Type II. the rest of. From a semantic perspective. Now see Type II: (48) Type II: a. First. depending upon the reference of the of -NP phrase. Type I and Type II involve pronominal forms serving as the head of the constructions. *neither of students. b. Some of the soup needs more salt. d. the number value depends on the head noun inside the of -NP phrase. 6. The expressions in (49) also exhibit similar behavior in terms of agreement. we see that the class of quantificational indefinite pronouns including some. Some of the diners need menus. one cannot serve as a determiner when the head noun is collective as in (46a). the verb is also plural. We can observe here that the verb’s number value is determined by the preceding expression each. when the NP following the preposition of is singular or uncountable. e. All of the land belongs to the government. the main verb is singular. c. Each of the suggestions is acceptable. *One people was dying of thirst. Observe Type I examples. In Type I. When the NP is plural.b. As shown in (48).5. Most of the fruit is rotten. None of these men wants to be president.

we can lexically encode the similarities and differences between Type I and Type II in a simple manner: (52) a. There is no semantic difference (such as definiteness effect represented as the feature DEF in the present system) between the PP of him and the NP him. This will account for the contrast in (50).     POS noun   HEAD  AGR | NUM 1      PFORM of    DEF + COMPS PP  AGR | NUM     1 (52) shows that both Type I neither and Type II some are lexically specified to require a PP complement whose semantic value includes the definite (DEF) feature (with the value +). As for the PPs headed by these markers. b. as in the partitive construction. (51) a. DEF) with its inner NP. I am fond of him. The predicative preposition in here selects two arguments John and the room. However. we have the alternatives in (53): 114 . It is easy to see that there are prepositions whose functions are just grammatical markers. A slight digression is in order. their semantic features are identical with the prepositional object NP. Given this analysis in which the PP in the partitive construction has the identical AGR and semantic features (e. but just functions as a marker to the argument of fond.  neither      POS noun  HEAD  AGR | NUM sing       PFORM of   COMPS PP DEF +  some                           b.b. the preposition of has no predicative meaning. John is in the room. we know that the two types are different in terms of agreement: the pronouns in the Type I construction are lexically specified to be singular whereas the number value for Type II comes from inside the selected PP. neither of the two linguists/some of the water However. the two types are different in terms of their AGR’s NUM value. Showing these differences in the syntactic structures. The NUM value of Type I neither is singular. whereas that of Type II is identified with the PP’s NUM value which is actually coming from its prepositional object NP..g. Meanwhile.

However. We can check a few of the consequences of these different specifications in the two Types. This is illustrated in (55) and (56).(53) a. some of his advice/students b. for Type I. This system also offers a simple way of dealing with the fact that quantifiers like each affect the NUM value as well as the countability of the of -NP phrase. one of the suggestions/*the suggestion/*his advice b. Consider the contrast in (54): (54) a. NP[NUM sing] N oyy ooo yyyyy yyy ooo o y ooo N[NUM sing] PP oyy ooo yyyyy o yyy ooo y ooo NP neither P c  ccc cc   c  of the students NP[NUM 1 ] F oyy ooo yyyyy yyy ooo o y ooo N[NUM 1 ]o PP[NUM 1 ]s ooyyyyy yyy ooo yyy ooo ooo NP[NUM 1 ] some P c  ccc cc   c  of the students b. for Type II. most of his advice/students 115 . it is neither which determines the NUM value of the whole NP phrase. One difference between Type I and Type II is that Type I selects a plural of -NP phrase whereas Type II has no such restriction. *many of some/all/no apples (54b) is ungrammatical since many requires an of -PP phrase whose DEF value is positive. As shown in (53a). neither of the students/*the student/*his advice (56) Type II: a. it is the NP the students which determines the NUM value of the whole NP. (55) Type I: a. each of the suggestions/*the suggestion/*his advice c. many of the/those/her apples b.

The contrast here indicates that Type II pronouns can combine with a PP whose inner NP is singular. *Each of John’s boat has been repainted. few dogs/*few dog/*little dogs/*little dog b. Most of John’s boat has been repainted. b. *One of the story has appeared in your newspaper. c. Consider the following examples: (60) a. Specifically. *Many of the record contained evidence of wrongdoing. a countable noun selects a countable determiner as its specifier. We are also in a position now to understand some differences between simple NPs and partitive NPs. b. To capture this agreement restriction. much furniture/*many furniture/the furniture (61) a. Some of the record contains evidence of wrongdoing. in contrast to those in (59) (cf. This is simply predicted since our analysis allows the inner NP to be either plural or singular (or uncountable). we introduce a new feature COUNT (COUNTABLE): 116 . (59) a. c.c. Much of that theory is unfounded. Type II pronouns do not have such a requirement on the PP complement – note that all the examples in (58) are acceptable. common nouns also place a restriction on the countability value of the selected specifier. Baker 1995): (58) a. all of his advice/students The only additional specification we need for Type I pronouns relates to the NUM value on the PP’s complement as given in (57): (57)  each                   POS noun   HEAD   AGR | NUM sing       PFORM of      COMPS PPDEF +    NUM pl We see that quantifiers like each select a PP complement whose NUM value is plural. many dogs/*much dog/the dogs b. little furniture/*little furnitures/*few furniture/*few furnitures The data here indicate that in addition to the agreement features we have seen so far.

much advice vs. Effectively.  POS det   HEAD COUNT +   little      c. we understand now the contrast between much advice and *many advice or the contrast between *much story and many stories. *much story vs.   POS det  HEAD COUNT +/– Notice here that some determiners such as the are not specified for a value for COUNT.  PFORM of    COMPS PPNUM sing          DEF + The pronoun many requires a PP complement whose inner NP is plural. many of the stories Due to the feature COUNT. 117 .(62)  dogs      furniture      a. much of the advice vs. *many advice b. many stories (65) a. This in turn means that determiners must also carry the feature COUNT: (63)  many   the       a. HEAD | POS noun  SPR DP[COUNT +]  b. licensing combination with either a countable or an uncountable noun (the book or the furniture). HEAD | POS noun  SPR DP[COUNT –] The lexical specification on a countable noun like dogs requires its specifier to be [COUNT +]. Now consider the following contrast: (64) a.  PFORM of  COMPS PPNUM pl        DEF +   HEAD | POS noun           b. whereas much does not.  POS det   HEAD COUNT –      b. but the patterns in the data directly follow from these lexical entries: (66)  many   much             HEAD | POS noun      a. *many of the advice b. much of the story vs. the value can be either + or −. to prevent formations like *much dogs. as (65) shows. The facts in partitive structures are slightly different.

*one many of the books. lot acts more like a noun. 118 . one pound of beans. Consider the following examples: (72) a. Further complications arise due to the existence of defective measure noun phrases. but unlike can. measure noun phrases behave like Type I partitive constructions: (73) a. With respect to few. *one pound beans vs. three feet of wire c. a quart of cider Further. three feet of wire As noted here. one pound of beans b. it appears that a few acts like a complex word. *a can tomatoes/a can of tomatoes/one can of tomatoes b. three feet of that wire c. *several much of the beer b. a quart of cider Notice here that (67) is a kind of partitive construction whereas (68) just measures the amount of the NP after of .6. many or much in the partitive constructions cannot combine with numerals like one or several.3 Measure Noun Phrases There are also so-called ‘measure noun phrase’ constructions. A can of tomatoes is/*are added.5. measure noun phrases require a numeral as their specifier: (71) a. one pound of beans b. measure noun phrases do not require a definite article. one pound of those beans b. there are several more differences between partitive and measure noun phrases. However. *some of wire. measure nouns pound and feet need to combine with a numeral like one or three. They obligatorily require an of -NP phrase: (70) a. which is not an option for the true partitive constructions. *much of cider. unlike partitive constructions. three feet of wire c. For example. a quart of Bob’s cider (68) a. *three feet wire vs. *one of strawberries In addition. a few suggestions/*a few of suggestions/*one few of suggestions c. repeated here: (69) *many of beans. *none of yogurt. it does not allow its specifier to be a numeral. *a quart cider vs. measure nouns cannot occur in simple noun phrases. As the examples show. In terms of agreement. *a lot suggestions/a lot of suggestions/*one lot of suggestions Expressions like few and lot actually behave quite differently. Consider the following contrast: (67) a. which are similar to partitive constructions.

Unlike partitive constructions. *several thousands of supporters d. three hundred of your friends b.  SPR    COMPS PP[PFOR of] 119 . These are words such as dozen. SPR DP  COMPS PP[PFORM of]   HEAD | POS noun     b. and thousand: (75) a. The inner NP in the PP does not affect the NUM value at all. Finally. We can see here that it is the head noun can or cans which determines the NUM value of the whole NP. when used as noun. similar behavior can be observed with thousand and thousands: (76) a. there is no definiteness restriction on the PP complement. hundreds of friends/*hundreds friends Consider the behavior of hundred and hundreds here. The plural hundreds requires no specifier although it also selects a PP complement. hundred. *several thousands of Bill’s supporters c. three hundred friends e. several thousand supporters e. obligatorily requires a PP[of ] complement as well as a numeral specifier. The singular hundred. Not surprisingly. thousands of supporters/*thousands supporters One way to capture these properties is to assign the following lexical specifications to hundred and hundreds: (77)  hundred   hundreds         HEAD | POS noun  a. several thousand of Bill’s supporters b. *three hundreds of friends d. as in (75a). Two cans of tomatoes are/*is added.b. These observations lead us to posit the following lexical entry for a measure noun: (74)  pound     POS noun   HEAD  NUM sing    SPR DP   COMPS PP PFORM of             That is. there is one set of words whose behavior leaves them somewhere between quantity words and measure nouns. a measure noun like pound requires one obligatory SPR and a PP complement. *three hundreds of your friends c.

golden. However. This information is not shown here. *This objection is main. afraid. (79) a. It is the main street. *It is an alive fish. main and mere are only used attributively: (78) a. asleep. can be used only predicatively. aware. a key fact) The predicatively-used adjectives are specified with the feature PRD. awake. (cf. *They are afraid people. drunken. (cf. He is alive. certain adjectives are restricted to their usages.Even though there may be some semantic reasons for all these different kinds of lexical specifications. (80) a. all predicative expressions select one argument. b. addition. a modifying adjective will have the following lexical entry: 7 All 8 In modifiers (adverbial element) will carry the head feature MOD. whereas others such as wooden. He is afraid of foxes. Even though most adjectives can be used either as in a modifying (attributive) function or as a predicate (as in She is tall). and with a MOD value being empty as default. This lexical information will prevent predicative adjectives from also functioning as noun modifiers. for now. and does not have a specification for a MOD value (the value is empty). It is a golden hair. as shown here:7   (82) alive      POS adj       HEADPRD +    MOD This says that alive is used predicatively. (cf.6. *This fact is key.8 In contrast to the predicative adjective. 6. ashamed. b. living fish) b. 120 . Adjectives such alive.1 Modifying an NP Adjectives as Prenominal Modifiers Adjectives are expressions commonly used to modify a noun. stating it all directly in the lexical entries will account at least for the data given here. nervous people) (81) a. not all adjectives can modify nouns. the main objection) b. (cf. their subject (SPR).6 6. c. It is a wooden desk.

[The boy [in the doorway]] waved to his father. e. c. [The money [that you gave me]] disappeared last night. we can say that example (85a) will have the following structure: 121 . [The man [eager to start the meeting]] is John’s sister. [The papers [removed from the safe]] have not been found. b.6.(83)  brave         POS adj  HEAD MOD N This specifies an adjective which modifies any value whose POS is noun. Various phrases can function as such postnominal modifiers: (85) a. Leaving aside detailed discussion of the relative clause(-like) modifiers in b–e until Chapter 12. d. This will license a structure like the following: (84) NP t„„ tt „„„„„„ tt „ t tt tt t N 2 DP SPR 2 DP j„„ jjjj „„„„„„„ jjjj AP the MOD brave 1 1N SPR 2 DP ÙSSS Ù ÙÙ S child 6. [The man [holding the bottle]] disappeared. The only difference is that they come after what they modify. All these postnominal elements bear the feature MOD.2 Postnominal Modifiers Postnominal modifiers are basically the same as prenominal modifiers with respect to what they are modifying.

If it were the case that post-nominal PP could modify any NP. *John in the doorway waved to his father. with no complement or specifier. 122 .(86) NP t„„ tt „„„„„„ t „ tt t tt tt N 2 DP SPR 2 DP jjj„„„„„„„ „„ jjjj jj 1N PP MOD 1 € nnn €€€€€ € nnn n in the doorway the SPR 2 DP boy These modifiers must modify an N . This claim is consistent with the examples above and with the (ungrammatical) examples in (87): (87) a. these examples ought to be acceptable. A proper noun or a pronoun projects directly to the NP. b. but not a complete NP. *He in the doorway waved to his father.

e. Provide the correct VFORM value of the underlined verb and identify the noun that semantically determines this VFORM value.7 Exercises 1. Compare the following examples and assign an appropriate structure to each. 2. Both the workers will wear carnations. Few approve of our remedy. Few doctors approve of our remedy. 123 . e. c. In doing so. Most of the milk your baby ingests during breastfeeding is produced during nursing. What kind of lexical category can you assign to both and few? Can you provide arguments for your decisions? (i) a. All of the plant virus web sites have been conveniently collected in one central location. The effectiveness of teaching and learning depend on several factors. b. (i) a. The committee have/*has all now resigned. provide the correct AGR and IND value of the subject head noun and the main verb. Her family are/*is all avid skiers. None of his customary excuses suffices Edgar now. The crew have/*has both agreed to change sponsor. d. (i) a. The committee were/*was unanimous in their decision. Each of these stones has to be cut and polished. b. b. d. c. focusing on subject-verb agreement. (ii) a. d. f. Neither of these men is worthy to lead Italy. One of major factors affecting the value of diamonds was their weight. Few of the doctors approve of our remedy.6. Provide a detailed analysis for the following examples. One of the problems was the robins. b. (i) a. j. Some of the water from melted snow also goes into the ground for plants. Draw tree structures for the following and mark which expression determines the agreement (AGR) and index values of the subject NP and the main verb. 4. Most of her free time was spent attending concerts and plays or visiting museums and art galleries. g. 3. An example of these substances be tobacco. A variety of styles has/*have been in vogue for the last year. Both will wear carnations. Both of the workers will wear carnations. Both workers will wear carnations. All special rights of voting in the election were abolished. c. h. i. c. b.

c. b. Everyone of us be given a prize. c. j. respectively. That dog is so ferocious. in terms of what the pronoun form must. b. yourself. *I washed me. I washed myself. b. 6. now examine the following data. Some of the fruit be going bad. Fifty pounds seem like a lot of weight to lose in one year. e. *He washed yourself. d. The committeei haven’t yet made up theiri /*itsi mind. it even tried to bite himself. *He washed him. Some of the promoters of ostrich meat compare its taste to beef tenderloin. Ten years be a long time to spend in prison. You washed yourself. *Harry says that Sally dislikes himself. 5. k. Consider the following pairs of examples and explain the subject-verb and pronounantecedent agreement relationships. Once you have your own hypothesis for the above data.) c. him. (iii) a. One of the most serious problems that some students have be lack of motivation. may. (iv) a. *I washed yourself. f. Half of the year be dark and wintry. Consider the distribution of the reflexive pronouns (myself. b. he. (ii) a. g. (ii) a. it even tried to bite itself. revise your hypothesis so that it can extend to these examples: (v) a. He kicked you. (He and him referring to the same person. and how they affect grammaticality: (i) a. b. b. Harry says that Sally dislikes him. You washed me. himself. l. b. you. 124 . Provide rules which can explain their distribution. or must not be coreferential with. (i) a. News of Persephone and Demeter reach the great gods and goddesses of Olympus. Please take the grammaticality judgements as given. He washed me. herself ) and simple pronouns (me. All of his wealth come from real estate investments. and if not. h. The committeei hasn’t yet made up itsi /*theiri mind. and then determine whether your previous hypothesis can account for this extra data. *He washed myself. Do some of your relatives live nearby? i. *That dog is so ferocious. *You washed myself. c. her). c. I washed you.

the first noun is playing the role of an adjective. which is what most people have in mind when we think about modification. 125 . the first noun is said to modify the second. but native speakers of English are quite happy to invent their own combinations of nouns in order to describe things. I train hard. If you have a strong body. play hard.9 (ii) 9 Adopted from Waylink English at http://www. (vii) a. this is particularly true in the workplace where we need constantly to refer to innovations and new concepts.waylink. but nouns can do the job equally well. (i) The power of your mind and the power of your body have a tight connection.uk/. 7. One very important and highly productive feature of nouns in English is that they can be put together to form a new phrase without our having to make any structural changes to the grammar of either noun. It is worth mentioning that not every language has this facility. computer screen. This philosophy gets me through anything and everything. garden fence. chalk board. vacuum cleaner. *Sally believes that herself is brilliant. If you have a strong mind. specify their AGR and IND values. as in tea cup. Sally believes that she is brilliant. too. Read the following passages and provide detailed lexical entries for the underlined expressions. I live hard. events or ideas that they have not come across before. internet facility. b. In a sense. When two or more nouns combine like this. your mind feels pumped and healthy. *Sally wishes that everyone would praise herself. you can craft your body to accomplish amazing things. b. For nouns. and when life snaps at me. Sally wishes that everyone would praise her.co.(vi) a. I focus on constantly developing this double toughness. etc.

.

(2) a. b. John tries to fix the computer. In lingustic terminology. Compare the following pairs of examples: (1) a. Mary expected John to fix the computer. there are several significant differences which motivate two classes. etc. the surface structures of both The cat chased the mouse and The mouse was chased by the cat are derived from the identical deep structure similar to The cat chased the mouse. Control verbs and adjectives: try. etc. As shown here. John seems to fix the computer. Raising verbs and adjectives: seem. At first glance. promise.1 Raising and Control Predicates As noted in Chapter 5. 127 . appear.7 Raising and Control Constructions 7. this deep structure would be proposed and then 1 Deep structure. is a theoretical construct and abstract level of representation that seeks to unify several related observed form and played an important role in the transformational grammar of the 20th century. However. certain. Mary persuaded John to fix the computer. b. Let us consider the ‘deep structure’ of (1a) representing unexpressed subject of the VP complement of tries:1 (4) John tries [(for) John to fix the computer]. For example. linked to surface structure. eager. certain verbs select an infinitival VP as their complement. believe. known as control and raising verbs: (3) a. Verbs like try are called ‘control’ or ‘equi’ verbs. these pairs are structurally isomorphic in terms of complements: both try and tend select an infinitival VP. likely. hope. in this sentence it is John who does the action of fixing the computer. and expect and persuade select an NP and an infinitival VP. where subject is understood to be ‘equivalent’ to the unexpressed subject of the infinitival VP. expect. the subject of the verb is said to ‘control’ the subject of the infinitival complement. b. In the original tranformational grammar approach. persuade. happen.

This chapter discusses the similarities and differences of these two types of verb. b. b. whereas a raising verb seem does not assign any semantic role to its subject (this is what (5) is intended to represent). It seems that John is honest. but is rather about the proposition that John is honest. Since control verbs like try and hope require their subject to have an agent role. Consider the deep structure of (1b): (5) seems [John to fix the computer]. As suggested by the paraphrase. we say that a control verb like try assigns a semantic role to its subject (the ‘agent’ role). and shows how we explain their respective properties in a systematic way. John seems to be honest. b. *It/*There hopes to bother Kim that they resigned. This is why verbs like seem are called ‘raising’ verbs. Such items include the expletives it or there: (8) a. Expletive subjects: Since the raising verb does not assign a semantic role to its subject. which takes no semantic role. an expletive it or there. Due to this difference. It seems to bother Kim that they resigned. The situation is markedly different with control verbs: (9) a. b. These might have paraphrases as follows: (7) a. *It/*There tries to be warm in September. the subject John needs to be raised to the matrix subject position marked by . the one who does the action of trying is John in (6a). John tries to be honest. Meanwhile. John makes efforts for himself to be honest. cannot function as their subject.undergo a rule of ‘Equivalent NP Deletion’ in which the second NP John would be deleted. We can observe the same contrast with respect to raising and control adjectives: 128 . This is why such verbs have the label of ‘equi-verbs’. which we present here. How about (6b)? Is it John who is involved in the situation of ‘seeming’? As represented in its paraphrase (7b). verbs like seem are called ‘raising’ verbs. to produce the output sentence.2 Differences between Raising and Control Verbs There are many differences between the two classes of verb. 7.2. the situation that the verb seem describes is not about the individual John.1 Subject Raising and Control The semantic role of the subject: One clear difference between raising and control verbs is the semantic role assigned to the subject. It tends to be warm in September. 7. certain expressions which do not have a semantic role or any meaning may appear in the subject position. In order to derive the ‘surface structure’ (1b). Let us compare the following examples: (6) a.

Observe the following contrast: (11) a. The VP [to be kept on participants] requires a subject which must be the word tabs in order to induce an idiomatic meaning. we can see that in raising constructions. For example. Subcategorization: If we look into what determines the subject’s properties. This is why John cannot be the subject in (12b). In raising constructions. The VP [to be a letter in the mailbox] allows its subject to be there (cf. *John is likely [to be kept on participants]. On the other hand. whatever category is required as the subject of the infinitival VP. c. d. John/*It is eager to please Maja. c. the VP to be intelligent requires an animate subject. Stephen seemed [to be intelligent]. Sandy tried [to eat oysters]. e. In raising constructions. However. is also required as the subject by the higher VP – hence the intuition of ‘raising’: the requirement for the subject passes up to the higher predicate. *Tabs try [to be kept on Bob by the FBI]. c. the VP to be easy to fool Ben requires the expletive it as its subject. we can have it as its subject. There is likely [to be a letter in the mailbox]. *There tried [to be riots in Seoul]. *It tried [to bother me that Chris lied]. It is the control verb or adjective itself which fully determines the properties of the subject: (13) a. *That he is clever is eager [to be obvious]. Meanwhile. *John is likely [to be a letter in the mailbox]. b. b. The contrast in (c) and (d) is similar. Since the raising adjective easy do not assign any semantic role to its subject. the control adjective eager assigns a role and thus does not allow the expletive it as its subject.(10) a. There is a letter in the mailbox) but not John. Tabs are likely [to be kept on participants]. and this is why (11a) is fine but (12a) is not. *John seems [to be easy to fool Ben]. *There seemed [to be intelligent]. there is no direct selectional relation between the subject of the main verb and that of the infinitival VP. d. It/*John is easy to please Maja. it is not the raising verb or adjective.’ (12) a. d. in the sense of: ‘The participants will be spied on. It seems [to be easy to fool Ben]. b. 129 . but the infinitival complement’s predicate which determines the characteristic of the subject. for control verbs. b. the subject of the raising predicate is selected as the subject of the complement VP.

130 . And consider as well the following examples: (15) a. the verb thank requires a human subject and an object that is at least animate: (14) a. which we have represented in terms of VAL (valence) features. tried is different: just like understand. The color red seems [to be his favorite color]. However. #The cat tries to be out of the bag. and there is no idiomatic meaning. and it has no semantic role at all coming from the raising predicate. This in turn means that the subject of the infinitival VP is the subject of the raising verb. a ‘trier’. The king thanked the man. This implies that an idiom whose meaning is specially composed from its parts will still retain its meaning even if part of it appears as the subject of a raising verb. ( d. c. in the sense of: ‘The secret is out’. #The color red tried [to be his favorite color]. For example. Selectional Restrictions: Closely related to the difference in selection for the type of subject. #The king thanked the throne. we can observe a related similarity with regard to what are known as ‘selectional restrictions’. since the control verb tries assigns a semantic role to its subject the cat. ‘the cat’ must be the one doing the action of trying. b. a control predicate requires its subject to be able to bear the semantic role of agent. The subcategorization frames. the subject of the infinitival VP can be there and it. (17) a. understands requires its subject to be sentient. What we can observe here is that the subject of a raising verb carries the selectional restrictions of the infinitival VP’s subject. the control verb tried requires its subject to be sentient. are themselves syntactic.Regardless of what the infinitival VP would require as its subject. #The color red understands the important issues of the day. The cat seems to be out of the bag. #The castle thanked the deer. in (13b) and (13c). b. For example. Meaning preservation: We have seen that the subject of a raising predicate is that of the infinitival VP complement. the meaning of the idiom The cat is out of the bag is retained. In the raising example (17a). at least. Unlike the verb is. The occurrence of the raising verb seems does not change the selectional restriction on the subject. ?)The king thanked the deer. This selectional restriction then also explains the following contrast: (16) a. However. The color red is his favorite color. but these cannot function as the matrix subject – because the matrix verb tried requires its own subject. b. but verbs also impose semantic selectional restrictions on their subjects or objects. b.

2 Object Raising and Control Similar contrasts are found between what are know as object raising and control predicates: (20) a. but this inference cannot be made in (19). 7. Stephen persuaded Ben to be careful. whereas in (19b). so is the other. it is Pat who is eager to be examined by the dentist. The dentist was believed to have examined Pat. and this forces (19a) and (19b) to differ semantically: in (19a). b. it is lost with the control verb persuaded. However. b. Pat is eager to be examined by the dentist. And in this respect. *Stephen persuaded there to be a fountain in the park.This preservation of meaning also holds for examples like the following: (18) a. b. (22) a. these two verbs look alike in terms of syntax: they both combine with an NP and an infinitival VP complement. and the passive subject in (18b). (18a) and (18b) have more or less identical meanings – the proposition is about the dentist examining Pat. *Stephen persuaded it to be easy to please Maja. The dentist is eager to examine Pat. Stephen believed Ben to be careful. Stephen believed it to be easy to please Maja. (19) a. Pat is likely to be examined by the dentist. One thing we can see here is that unlike believe. b. 131 . in the sense: ‘Stephen believed that the secret was out’. the two are different with respect to the properties of the object NP in relation to the rest of the structure. in active or passive grammatical forms: the active subject is raised in (18a). While the idiomatic reading is retained with the raising verb believed. However. if one of the examples in (18) is true. the verb believe is similar to seem in that it does not assign a semantic role (to its object). Stephen believed there to be a fountain in the park. Observe the differences between believe and persuade in (21): (21) a.2. the control predicate eager assigns a semantic role to its subject. Since the raising predicate likely does not assign a semantic role to its subject. it is the dentist who is eager to examine Pat. The dentist is likely to examine Pat. Active-passive pairs show another contrast: (24) a. b. b. persuade does not license an expletive object (just like try does not license an expletive subject). Stephen believed the cat to be out of the bag. Intuitively. *Stephen persuaded the cat to be out of the bag. The differences show up again in the preservation of idiomatic meaning: (23) a. Once again.

b. b. it is the dentist. for example. This is one more piece of evidence that believe is a raising verb whereas persuade is a control verb. is to derive a surface structure via a derivational process. However. from (26a) to (26b): (26) a. Here the embedded subject Stephen moves not to the matrix subject but to the matrix object position: 132 . To derive (26b). hinted at earlier. there is no strong semantic difference in the examples in (24).b. In (25a). Deep structure: seems [Stephen to be irritating] b.3 A Simple Transformational Approach How then can we account for these differences between raising and control verbs or adjectives? A simple traditional analysis. in (25). with respect to the object. there is a clear difference in who is persuaded. Deep structure: Tom believes [Stephen to be irritating]. Chomsky 1981). This kind of movement to the subject position can be triggered by the requirement that each English declarative have a surface subject (cf. Surface structure: Tom believes Stephen to be irritating. the subject of the infinitival VP in (26a) moves to the matrix subject position. Pat was persuaded to be examined by the dentist. Surface structure: Stephen seems [t] to be irritating. but in (25b). A similar movement process can be applied to the object raising cases: (28) a. Pat was believed to have been examined by the dentist. With the raising verb believe. it is Pat who is persuaded. as represented in the following tree structure: (27) S ƒƒ kkkk ƒƒƒƒƒƒ k ƒƒƒ kkkk kk VP NP ƒƒ kkkk ƒƒƒƒƒƒ k ƒƒƒ kkkk kk V S R ƒƒ kkkk ƒƒƒƒƒƒ k ƒƒƒ kkkk kk VP[inf ] seems NP u ss uuu uu ss s u ss to be irritating Stephen The movement of the subject Stephen to the higher subject position will correctly generate (26b). 7. The dentist was persuaded to examine Pat. (25) a.

and the examples will have the following deep structures: (31) a. The final tree representations of these are as follows: (32) a. John tried [PRO to please Stephen]. S w qqq wwwww qq www qq w qqq VP NP w qqq wwwww www qqq q w qqq V S Johni w qqq wwwww qq www qq w qqq VP[inf ] tried NP uss uu sss ss uu ss uu uu PROi to please Stephen 133 . b. This is traditionally represented as the element called ‘PRO’ (a silent ‘pro’noun). John persuaded Stephen [PRO to be more careful]. John persuaded Stephen to be more careful. and objects.(29) S m mmm   mmm mmm VP NP mˆˆˆˆˆˆˆˆ mmm ˆˆˆˆˆˆ m ˆˆˆˆ mmm mm V S John NP m mmm   mmm mmm believes NP VP[inf ] Q u ss uuu uu ss s u ss to be irritating Mary Control constructions are different: there is no movement operation involved. it is the lower subject position which has special properties. it is posited that there is an unexpressed subject of the infinitival VP to please Stephen and to be more careful. an unfilled position of the kind designated above by cannot be allowed. John tried to please Stephen. Instead. Since try and persuade assign a semantic roles to their subject. Instead. b. Consider the examples in (30): (30) a.

giving the structures in (34): (33) a. it is coindexed with Stephen. believe acts like it has an NP object (separate from the infintival complement). as in (33).4. and hence the structure in (32b) reflects this: the object position is there all along.4 7. For example. and this is reflected in the initial structure. a patient. b. PRO is coindexed with John whereas in (32b).  seems NP VP VFORM inf        SPR    COMPS 134 .1 A Nontransformational Approach Identical Syntactic Structures Instead of the movement approach in which movement operations and various kinds of empty elements or positions play crucial roles. we can observe that both select an infinitival VP. though. and a proposition. In contrast. the verb believe semantically selects an experiencer and a proposition. we simply focus directly on the surface structures of raising and control constructions. These analyses which involve derivations on tree structures are driven by the assumption that the mapping between semantics and syntax is very direct. 7. so to speak. persuade semantically selects an agent. we present a nontransformational account of control and raising.S w qqq wwwww www qqq q w qqq VP NP qq†††††††† †††† qqq †††† qq †† qqq V S John NP w qqq wwwww www qqq q w qqq persuaded Stepheni NP VP[inf ] t tt ttt tt tt t tt tt t tt PROi to be more careful An independent part of the theory of control links PRO in each case to its antecedent. In (32a). The classical transformational approach provides a useful graphical approach to understanding the difference between raising and control. it requires assumptions about the nature of grammar rather different from what we have made throughout this book. in (29). and the raising operation creates this object. In the rest of this chapter. marked by coindexing. Going back to seem and try. In some syntactic respects. However.

VP VFORM inf  b.  tries NP VP VFORM inf       SPR   COMPS (34) a. The object raising verb expect and the control verb persuade also have identical valence (SPR and COMPS) information: (35) a. S kƒƒƒƒƒ ƒƒƒƒ kkkk kk ƒƒ kkkk VP NP kƒƒƒƒƒ ƒƒƒƒ kkkk kkk ƒƒ kkk V VP[inf ] John ƒƒ kkkk ƒƒƒƒƒƒ ƒƒƒ kkkk kkk tries V VP[bse] e }} ee }} eee }} to be honest As shown here.  expects    SPR  NP     COMPS NP. seems and tries actually have identical structures. persuaded  SPR NP   COMPS NP.b. VP VFORM inf      These two lexical entries will license the following structures: 135 . S kkƒƒƒƒƒƒ ƒƒƒƒ kkkk k ƒ kkkk VP NP ƒƒ kkkk ƒƒƒƒƒƒ ƒƒƒ kkkk kkk V John VP[inf ] kkƒƒƒƒƒƒ ƒƒƒƒ kkkk kk ƒ kkk seems V VP[bse] e }} ee }} eee }} to be honest b.

While the subject of a raising predicate is identical to that of the infinitival VP complement. raising and control verbs are not different in terms of their subcategorization or valence requirements. 136 . S i…… iiii ………………… ………… iiii iiii NP VP … GG iiii ………………… ………… iiii  GGG  iii …  ii NP Kim V VP x ppp xxxxx xx ppp pp expects it to rain tomorrow b. the subjct of a control predicate has a different requirement. The subject of a control predicate is coindexed with that of the infinitival VP complement. It seemed to rain. S iii………………… ………… iiii ii …… iiii NP VP G iii…………………  GGG ………… iiii  ii ……  G iiii NP VP Kim V oyy ooo yyyyy oo yy ooo persuaded Mary to leave tomorrow As can be seen here. while the control verb only shares the index (of the subjects). c. Stephen/*It/*There tried to be intelligent. whatever kind of category is required as subject by the infinitival VP is also required as the subject of the predicate. *There tried to be a fountain in the park. (38) a. Stephen/*It/*There seemed to be intelligent. This difference can be represented in the lexical information shown in (39). and so they project similar structures. b. b. There seemed to be a fountain in the park. c. The answer is that their differences follow from the other parts of the lexical information. *It tried to rain. the mapping relations from syntax to semantics.(36) a. Some of the key examples are repeated here: (37) a. The question is then how we can capture the different properties of raising and control verbs.2 Differences in Subcategorization Information We have observed that for raising predicates.. The raising verb involves shared subjects.4. 7. in particular.

the subject of the VP complement is identical with its own subject (notated by 1 ) whereas for tried. the VP complement’s understood subject refers to the same individual as the subject of tried. The example here means that whoever someone might refer to.  seemed 1         VFORM inf    VP SPR 1  NPi   VFORM inf  VP SPR NPi         SPR     COMPS b. that same person left town. That is. This index identity in control constructions is clear when we consider examples like the following: (40) Someonei tried NPi to leave the town. The lexical entries in (39) generate following structures for the intransitive raising and control sentences: 137 .  tried  SPR     COMPS These two lexical entries represent the difference between seem and try: for seemed.(39) a. only the index value of its VP complement is identical to that of its subject.

 HEAD  SPR 4 S  | POS verb   COMPS „ vv „„„„„„„ vv „„„„ v vv vv VP v  vv vv HEAD 4 1 NP  SPR 1    John COMPS jjwwww jjjj www jjjj www jj www V   2 VP HEAD 4     SPR 1  SPR 1   COMPS 2 qqwwwww www qqq qq www qqq ww qqq seems V VP[bse] e }} ee }} eee }} to be honest  HEAD  SPR 3 b.(41) a. S  | POS verb   COMPS „ vv „„„„„„„ vv „„„„ v vv vv VP v  vv vv HEAD 3 1 NPi  SPR 1 NP    John COMPS jjwwww jjjj www jjjj www jj www V   2 VP HEAD 3     SPR 1 NP  SPR NPi   COMPS 2 qqwwwww www qqq qq www qqq ww qqq tries V VP[bse] e }} ee }} eee }} to be honest 138 .

3 || fff ff || ff || ff || || expects it to rain tomorrow 139 .It is easy to verify that these structures conform to all the grammar rules (the Head-Specifier Rule and Head-Complement Rule) and principles such as the HFP and the VALP. The following lexical entries show these properties: (42) a. Raising verbs select a VP complement whose subject is fully identical with the object.  expect 1 NPi         VFORM inf     2 NP . VP 2 NP SPR  NP        VFORM inf    NPi . Object raising and control predicates are no different.  persuade  SPR     COMPS Let us look at the structures these lexical entries eventually project: (43)  HEAD  SPR 4 S  | POS verb   COMPS … t …………… tt ………… tt ……… t tt tt VP t   tt HEAD 4 tt 1 NP   5 SPR 1 NP  '' ' 555 COMPS ' 5 iiyyyy '' ' 555 iiii yyy iii yyy '' 55 iiii yyy '' 55 y V   55 '' 3 VP HEAD 4 '   2 Kim  NP  SPR 1 NP  SPR 2   f COMPS 2 . VP SPR NPi  SPR     COMPS b. Control verbs select a VP complement whose subject’s index value is identical with that of its object.

s0. s1. a predicate represents a semantic property or relation. a noun refers to an individual (e.g. distinguished from syntactic features. hit (j. k. i. As we have seen in Chapter 6. For example.g. for this issue is closely related to the relationship between syntax and semantics. b) This shows that the verb hit takes two arguments with the predicate relation hit. the subject of to rain tomorrow is the NP object of expects. 3 VP }} eee ee }} ee }} ee }} }} persuaded Mary to be more careful As represented here. while the subject of to be more careful is coindexed with the independent object of persuade. We first need to introduce further semantic features.4. 7. John hits a ball.3 Mismatch between Meaning and Structure We have not yet addressed the issue of differences in the assignment of semantic roles. the meaning of the verb hits in (45a) can be represented in canonical first-order predicate logic as in (45b): (45) a.. In addition. etc) whereas a verb denotes a situation (e. The relevant semantic properties can be represented in a feature structure system as follows: 140 . s2. etc).(44)  HEAD  SPR 4 S  | POS verb   COMPS † rr ††††††† †††† rrr †† rr rrr VP rr   r HEAD 4 rrr 1 NP   SPR 1  '' 555 '' 55 COMPS ' 5 hhhyyyy yyy hhhh '' ' 555 hhhh yyy hhh yyy '' 55 yy V '' 55   5 HEAD 4 '' 3 VP  2 NP Kim    i SPR 1  SPR NPi   }ee COMPS 2 NP. That is. j. with the notation to indicate the semantic value.. b. nouns and verbs have IND values.

the lexical entry in (46) is the formal representation of the fact that in X hits Y. This coindexing links the subcategorization information of hit with the arguments in its semantic relation. as a verb. Simply put. hit takes two arguments. The subject of a raising verb like seem is not assigned any semantic role. hit refers to a situation s0 in which an individual i hits an individual j. Now we can use these extra parts of the representation for the semantic differences in raising and control verbs. The AGENT argument in the SEM value is coindexed with the SPR in the SYN value.(46)  hit                        SPR NPi SYN | VAL   COMPS NPj      IND s0         PRED hit  SEM     i  RELS AGENT      PATIENT j In terms of syntax. s1) (‘i tries to (make) s1 (be the case) = s0’) These meaning differences are represented in terms of feature structures as follows: 141 . PRED(ICATE). as shown in the value of the feature SYN(TAX). try (i. hit is a verb selecting a subject and a complement. The semantic information of the verb is represented with the feature SEM(ANTICS). X is the hitter and Y is the one hit. representing what this expression refers to. The semantic relation of hitting is represented using the feature for semantic relations (RELS). seem (s1) (‘s1 seems (to be the case”) = s0’) b. Assuming that ‘s0’ or ‘s1’ stands for situations denoted by an infinitival VP. and PATIENT. while the the PATIENT is coindexed with COMPS. It first has the attribute IND(EX). AGENT. here with three further features. The predicate (PRED) relation is whatever the verb denotes: in this case. while that of a control verb like try is definitely linked to a semantic role. The feature RELS has as its value a list of one feature structure. seem and try will have the following simplified meaning representations: (47) a.

the key difference lies in whether the object (y) receives a semantic role or not: (49) a. try is different. s1) b.(48) a. is a proposition denoted by the VP complement. Just like the contrast between seem and try. these differences are more clearly represented in feature structures: 142 . y. expect (x. Its SPR is coindexed with the AGENT role in the semantics. Once again. s1) What one expects.  try         SPR NPi         VFORM inf    SYN | VAL  COMPS VPSPR NPi        IND s1       IND s0            PRED try   SEM      RELS AGENT i            SIT s1      We can see here that even though the verb seem selects two syntactic arguments. its meaning relation has only one argument: note that the subject (SPR) is not coindexed with any argument in the semantic relation.  seem                                                     SPR 1         VFORM inf    SYN | VAL   COMPS VPSPR 1       IND s1       IND s0        SEM PRED seem     RELS   SIT s1 b. and the SPR is also coindexed with the VP complement’s SPR. x persuades an individual y denoted by the object to perform the proposition denoted by the VP complement. This means that the subject does not receive a semantic role (from seem). Now we look at object-related verbs like expect and persuade. whereas what a person x persuades is not a proposition but rather. persuade (x. as an ‘experiencer’. Meanwhile.

the subject and object respectively is dependent for its semantic properties solely upon the type of VP complement.(50) a. expect has two semantic arguments. and SIT.1 Expletive Subject and Object Recall that for raising verbs like seem and believe. VP SPR 2 NP  COMPS     IND s1       IND s0            PRED expect   SEM        RELS EXPERIENCER i          SIT s1 b. persuade has three semantic arguments: AGENT. EXPERIENCER and SIT: the object is not linked to a semantic argument of expect. while with control predicates. 143 . In contrast. We can thus conclude that raising predicates assign one less semantic role in their argument structures than the number of syntactic dependents.5 Explaining the Differences 7.  expect                                                         SPR NPi         VFORM inf SYN | VAL     2 . There/*It/*John seems [to be a fountain in the park]. 7. VP    IND s1       IND s0             PRED persuade         AGENT i   SEM         RELS  THEME j             SIT s1     As seen in the lexical entries. This fact is borne out by the examples in (51): (51) a. THEME.  persuade        SPR NPi         VFORM inf SYN | VAL    SPR NPj  COMPS NPj .5. there is a one-to-one correlation.

as long as the expressions The dentist examines Pat and Pat is examined by the dentist have roughly the same meaning. but not in a control example like (53b): (53) a. The cat seems to be out of the bag. *There/*It/John tries to leave the country. the idiomatic reading can be preserved. 7. This is once again because the subject of seems does not have any semantic role: its subject is identical with the subject of its VP complement to be out of the bag.2 Meaning Preservation We noted above that in a raising example such as (53a). b. directly assigning a semantic role to the subject or object.5. as given in the following lexical entry: (56)  eager       SPR NPi           SYN | VAL VFORM inf        COMPS VP   IND s1          IND s0            PRED eager    SEM        EXPERIENCER i RELS        SIT s1 144 . Pat is eager to be examined by the dentist.b. control examples are different: (55) a. Exactly the same explanation applies to the following contrast: (54) a. We believed *there/*it/John to try to leave the country. The dentist is eager to examine Pat. b. b. The cat tries to be out of the bag. Control verbs are different. whereas the subject of tries has its own agent role. Hence expletives cannot appear (illustrated here for the subject of try): (52) a. b. The dentist is likely to examine Pat. We believed there/*it/*John [to be a fountain in the park]. the two raising examples will also have roughly the same meaning. However. Since likely is a raising predicate. The control adjective eager assigns a semantic role to its subject independent of the VP complement. Pat is likely to be examined by the dentist.

it is the object me which semantically functions as the subject of the infinitival VP.3 Subject vs. In sum. They persuaded me to leave. whereas in the latter. This difference is straightforwardly represented in their lexical entries: (59)  persuade    SPR NPi          VFORM inf     COMPS NPj . whereas those like persuade are ‘object control’ verbs. Object Control Verbs Consider finally the following two examples: (57) a. *They promised it to rain. we know that when one promises someone to do something. the person who is persuaded will do the action. 7. b. Both persuaded and promised are control verbs since their object is assigned a semantic role (and so is their subject). the two are different with respect to the controller of the infinitival VP.5. Yet. Due to this fact.This then means that (55a) and (55b) must differ in that in the former. when one persuades someone to do something. verbs like promise are known as ‘subject control’ verbs. They promised me to leave. the properties of rasing and control verbs presented here can be summarized as follows: 145 . VPSPR NPi   j       IND s1 Based on world knowledge. b. it is Pat who is eager. this means that the person who makes the promise will do the action. *They persuaded it to rain. In (57a). This in turn means that their object cannot be an expletive: (58) a. Consider who is understood as the unexpressed subject of the infinitival verb here. it is the dentist who is eager to perform the action denoted by the VP complement. Meanwhile. in (57b). VP SPR NPj   promise   SPR NPi          VFORM inf       COMPS NP . However. The lexical entries here reflect this knowledge of the relations in the world. it is the subject they who will do the action of leaving.

Unlike control predicates. With raising predicates. This ensures that whatever category is required by the raising predicate’s VP complement is the raising predicate’s subject (or object). b. and not with control predicates. the VP complement’s unexpressed subject is coindexed with one of the syntactic dependents. *Under the bed wants to be a fun place to hide. (60) a. Under the bed is a fun place to hide. With control predicates. (want is a control verb) role to their subject or object. Under the bed seems to be a fun place to hide. The absence of a semantic role accounts for the possibility of expletives it or there or parts of idioms as subject or object with raising predicates. 146 . raising predicates are unusual in that they do not assign a semantic . the entire syntactic-semantic value of the subject of the infinitival VP is structure-shared with that of one of the dependents of the predicate.. c. Notice that even non-NPs can be subject in certain kinds of example (see (60)).

begin. *Robin is nothing in the box. *Sandy appears that Kim is happy. . there is agreement between the copula be and the postcopular NP in so-called ‘there’ constructions. *John seems to rain. careful. *He believed there to be likely that he won the game. happen. Kim may have admitted to let Mary mow the lawn. raising and control. e. g. *Dana would be unlikely for Pat to be called upon. persuade. decide. . there. consider the it. anxious. There is/*are only one chemical substance involved in nerve transmission. In this chapter. hope. . lucky. b. . b. bound. . f. based on the lexical entries of the predicates: (i) a. sure. refuse 4. reluctant tend. . believe. . as shown again here: (i) a. Frank hopes to persuade Harry to make the cook wash the dishes. John continues to avoid the conflict. h. Gregory appears to have wanted to be loyal to the company. j. we have learned that predicates (verbs and adjectives) can be classified into two main groups. *It is likely to seem to be arrogant. He coaxed his brother to give him the candy. There *is/are more chemical substances involved in nerve transmission. . d. f. *John is likely to appear that he will win the game. c. e. g. *It said that Kim was happy. This kind of agreement relationship can be encoded as a property of the be used in this construction: 147 . apt. . The captain ordered the troops to proceed. *There preferred for Sandy to get the job. as represented in the following simple table: Raising predicates seem. intend. . Draw trees for the following sentences and provide the lexical entries for the italicized verbs: (i) a. John wants it to be clear to Ben that the city plans to honor him. As we have seen in Exercise 4 of Chapter 3. i. *Beth tried for Bill to ask a question. fail. b. Intransitive Transitive Decide in which group each of the following lexical items belongs. 3. liable.7. and idiom tests that this chapter has introduced: (i) (ii) certain. In doing so. Jones would prefer for it to be clear to Barry that the city plans to sue him. Explain why the following examples are ungrammatical. d. manage. . Control predicates try. .6 Exercises 1. 2. h. c.

use the it. There seems/*seem to be no student absent. c. Pat persuaded Leslie to be aggressive. (i) a. Kevin urged Anne to be loyal to herself. *Wei persuaded the dentist to examine ourselvesi . Consider the following data and discuss briefly what can be the antecedent of her and herself . b. Given this. This then will ensure that the verb agrees with the postcopular NP. In this case. c. how might we account for the following contrasts? Provide a structure for each example and explain the rules or principles which are violated in the ungrammatical versions: (ii) a. Now consider the following data and discuss the binding conditions of ourselves and us. 5. *We expect themi to examine themi . There *is/are believed to be sheep in the park. determine the relevance of the ARG-ST list with respect to the possible and impossible binding relations. (ii) a. d. d. (iii) a. there and idiom tests. XP[PRED +] This lexical information specifies that be selects there as its subject and two complements (NP and XP). Also. Pat expected Leslie to be aggressive. (i) a. Discuss the similarities and differences among the following three examples. There is/*are likely to be no student absent. b. the complement NP’s agreement feature AGR is identical with that of the subject there. c. state see the controller is of the infinitival VP in each case. Pat promised Leslie to be aggressive. Wei persuaded the dentist to examine usi . In particular. b. There is/*are believed to be a sheep in the park. b. 6. b. *Wei expect the dentist to examine ourselvesi . We expect them to examine themselves.(ii)  be                      POS verb HEAD    AUX +       NFORM there  SUBJ NP  AGR 1    COMPS NP[AGR 1 ]. Kevin urged Anne to be loyal to her. 148 . Wei expect the dentist to examine usi .

7. and lexical entries for the underlined words. you try to cause that person to do something. We persuaded themi to examine themselvesi . It’s very tough to try to convince other people to buy from you. you also know that this can be one of the most discouraging and difficult things to try to do as a business owner. Read the following passage and provide a tree structure for each sentence. (i) If you’ve ever tried to persuade other people to buy your product or service. especially if it’s against their will. After all. And usually there’s always some kind of pressure involved in this process. this way of trying to get business by trying to persuade other people is one of the factors that causes most business owners to dislike or even hate the process of marketing and selling. In fact.c. d. if you try to persuade someone to buy from you. *We persuaded themi to examine themi . 149 .

.

8 Auxiliary Constructions 8. shall.1 Basic Issues The English auxiliary system involves a relatively small number of elements interacting with each other in complex and intriguing ways. do: has a finite form only with vacuous semantic meaning . have/be: have both finite & nonfinite forms . and Bill Such phenomena provide apparent stumbling blocks to assigning a totally different lexical category to English auxiliary verbs. etc. may. Distinctions between auxiliary and main verbs: One important issue that comes up in the study of the English auxiliary system is that of which words function as auxiliary verbs. Most reliable criteria for auxiliaryhood lie in syntactic phenomena such as negation. though being different in terms of syntactic distribution and semantic contribution. drink beer. John may drink water. inversion. Ontological Issues: One of the main issues in the study of English auxiliary system concerns ontological issues: is it necessary to posit ‘auxiliary’ as an independent part of speech or not? Auxiliary verbs can be generally classified as follows: . b. For example. compared to main verbs. in terms of similarities. both auxiliary and main verbs behave alike in carrying tense information and participating in some identical syntactic constructions such as gapping. as shown in (1): (1) a. This has been one of the main reasons that the system has been one of the most extensively analyzed empirical domains in the literature on generative syntax.: have only finite forms . There have been arguments for treating these auxiliary verbs as simply having the lexical category V. and how we can differentiate them. and ellipsis (usually known as the ‘NICE’ properties): 151 . John drank water and Bill wine. to: has a nonfinite form only with apparently vacuous semantic meaning Such auxiliary verbs behave differently from main verbs in various respects. contraction. modal auxiliary verbs such as will.

Will Tom leave the party now? b. *If anybody keeps spoiling the children. John keeps In addition to these NICE properties. left you? The position of adverbs or so-called floated quantifiers can also be adopted in differentiating auxiliary verbs from main verbs. Ordering Restrictions: The third main issue in the syntax of auxiliaries centers on how to capture the ordering restrictions among them. but not a main verb: (6) a. Tom will not leave. (4) a. She would never believe that story. *John leftn’t the party early.1. John couldn’t leave the party. He must have been being interrogated by the police at that very moment. (10) a. *She believed never his story. Inversion: Only auxiliary verbs undergo subject-auxiliary inversion. c. shouldn’t you? b. 4. and the forms in which they can combine with other auxiliary verbs. They are subject to restrictions which limit the sequences in which they can occur. 2. b. *He has will seeing his children. Negation: Only auxiliary verbs can be followed by not marking sentential negation (including have and be): (2) a. The difference can easily be seen in the following contrasts: (7) a. *Tom kicked not a ball. Observe the following examples: (9) a. John is . 152 . but not a main verb. b. but not of a main verb. You should leave. (8) a. (5) a. b. tag questions are another criterion: an auxiliary verb can appear in the tag part of a tag question. b. b. b. *Our team played all well. The children will have been being entertained. *Left Tom the party already? 3. The boys will all be there. Contraction: Only auxiliary verbs have contracted forms with the suffix n’t. *Margaret has had already left. If anybody is spoiling the children. b. can be elided. . (3) a. Adverbs such as never and floated quantifiers such as all can follow an auxiliary verb. *The house is been remodelling. Ellipsis: The complement of an auxiliary verb. *You didn’t leave.

b. d. For example. introducing the rule in (11). directly stipulates the ordering relations among auxiliary verbs: (11) Aux → Tense (Modal) (have + en) (be + ing) The PS rule in (11) would generate sentences with or without auxiliary verbs as in (12): (12) a. Mary solved the problem. we thus need to address at least the following issues: . Should we posit an auxiliary category? .g. His analysis..1 8. 1 This 153 . c. note that each auxiliary verb requires the immediately following to be in a particular morphological form (e.2 The seminal work on the issues above is that of Chomsky (1957).d. *He has been must being interrogated by the police at that very moment. *has eating). As shown here. the following structure schematizes some examples in (12): (13) S ‡‡ ooo ‡‡‡‡‡‡‡‡ oo ‡‡‡‡‡ ooo ‡‡‡‡‡ ooo NP AUX VP E T oyy ØØ TTT  EEE ooo yyyyy Ø  E oo yyy TT Ø  E oo y ooo  ØØ VP Mary Past will Adv ooyyyyy yyy ooo yyy ooo ooo easily V NP c  ccc cc   c  solve the problem chapter is based on Kim (2000). Mary was solving the problem. In addition. has eaten vs. Mary would solve the problem. How can we account for phenomena (such as the NICE group) which are sensitive to the presence of an auxiliary verb? . In the study of the English auxiliary system. How can we capture the ordering and co-occurrence restrictions among auxiliary verbs? Transformational Analyses This chapter provides answers to these fundamental questions related to the English auxiliary system. Mary would easily solve the problem. How can we distinguish main verbs from auxiliary verbs? . they must come in a certain order. when there are two or more auxiliary verbs.

154 . b.3 A Lexicalist Analysis In the approach we take in this book. *Mary not avoided Bill. typical transformational analyses introduce the English particular rule called ‘ do-support’ for dealing with the NICE properties in clauses that otherwise have no auxiliary verb: (14) a. Kim and Sag 2002). c. identical phrasal constituents can be conjoined. Fred [must have been singing songs] and [probably was drinking beer]. the constituent structure in (13) does not provide constituent properties we find in coordinate structures. the famous ‘Affix Hopping’ rule of Chomsky (1957) ensures that the affixal tense morpheme (Past) in Tense is hopped to M (Modal) (will). (15) a. In addition to the Affix Hopping Rule. but nevertheless it misses several important points. d. As a last-resort option. or over onto the main verb (solve) if the Modal does not appear. ordering restrictions on auxiliay verbs will follow from the correct specification of their lexical properties. because the first auxiliary and the rest of the sentence (the VP) do not form a constituent. Nevertheless. see Kim 2000. interacting with the regular rules of syntactic combination. In this section. shall and must is that they place no semantic restrictions on their subject.To get to the surface structure.g. we discuss several different subtypes of auxiliary. The analysis lays bare the systematicity of the auxiliary system. The coordination examples here indicate that a VP with one auxiliary verb or more behaves just like the one without any. If the modal is there.1 Modals One main property of modal auxiliaries such as will. the position of do diagnoses the position of Tense in the structure.3. Fred must both [have been singing songs] and [have been drinking beer]. Chomsky 1986) use X -theory to provide IP and CP as categories for clausal syntax. giving Mary solved the problem. The analysis requires no movement. or of affixes. Fred must have been both [singing songs] and [drinking beer]. Fred must have both [been singing songs] and [been drinking beer]. there are many problems which transformational analyses cannot easily overcome (for a thorough review. For example. which can deal with the coordination data just given. The presence of not in a position like Adv in the tree (13) has been claimed to prevent the Tense affix from hopping over to the verb (as not intervenes).. the affix Past will hop onto the main verb solve. 8. b. Mary did not avoid Bill. In other words. the grammar introduces the auxiliary verb do onto which the affix Tense is hopped. indicating their status as raising verbs (see Chapter 7). If the modal is not there. More recent analyses in this tradition (e. either of whole words. Past hops onto will and we get Mary would (easily) solve the problem. This would then generate (14b). 8. As we have seen earlier.

(19) a. we notice at least three things: first. In addition. which must have a referential subject.2. the VFORM value fin includes pres. There might be a unicorn in the garden. b. the type of subject in (16) depends on what kind of subject (there or it or a regular NP) is required by the verb right after the modal. John can [kick/*kicked/*kicking/*to kick the ball]. *John musts/musted leave the party early. which differentiates them from main verbs. en. Modal verbs can only occur in finite (plain or past) forms.2 (18) a. a modal auxiliary will have at least the following lexical information:   (21) must       POS verb       HEADVFORM fin           AUX +        SPR 1 NP            VAL VFORM bse     COMPS VP   1 NP SPR In the lexical information given here. and pln whereas nonfin includes ing. b. 2 As 155 . John will [kick/*kicked/*kicking/*to kick the ball]. *The bus hopes to be here at five. c. As seen from the contrast. auxiliary verbs have the head feature AUX. *There hopes to finish the project. b. b. as well as the ungrammatical examples in (20): we have seen in 5. This subcategorization information will rule out examples like the following. It will rain tomorrow. modal verbs select a base VP as their complement: (20) a.(16) a. This is typical of raising verbs. *John wills leave the party early. and different from what we see in examples with a control verb like hope in (17). the rule shows that a modal verb selects a base VP as its complement. John will leave the party earlier. nor a transparent past tense form. and bse. Reflecting these basic lexical properties. Modals do not show 3rd person inflection in the present tense. They cannot occur neither as infinitives nor as participles. b. inf . In terms of their own selectional properties. I hope *to would/*to can/to study in France. (17) a. *John stopped can/canning to sign in tune. pst.1.

As can be easily seen here. *Kim must [VP[fin] baked a cake]. The lexical entry in (21) also specifies that the VP’s subject is identical with the subject of the modal auxiliary (indicated by the box 1 ). b. a crucial property of raising verbs we have discussed in the previous chapter. The VP rain tomorrow requires the expletive subject it. 156 . the modal auxiliary must requires a VP[bse] as its complement: The VP[fin] in (23b) cannot function as the complement of must. This specification. not any other NP. and the VP exist a man in the park requires there and nothing else its subject. *Kim must [VP[fin] bakes a cake]. rules out the ungrammatical versions of examples like the following: (24) a. *Kim must [VP[fin] will bake a cake]. such as John. There/*It may [VP[bse] exist a man in the park].(22) a. In addition. It/*John will [VP[bse] rain tomorrow]. c. b. VP mmtttt mmm tt mmm tt mmm tt m   tt t AUX + 2 VP[bse]  V t tt ttt COMPS 2 VP[bse] tt tt t tt tt tt tt t tt t t tt V[bse] must NP HH  HHH  H  bake a cake *VP mtt mmm ttt mm tt mmm tt mmm tt   tt AUX +   VP[fin] ttt COMPS VP[bse] tt ttt tt tt tt tt tt tt t tt tt t V[fin] must NP HH  HHH  H  baked a cake b. The possible and impossible structures can be more clearly represented in tree format: (23) a. they cannot occur in any environment where finite verbs are prohibited. since modal verbs have the specification [VFORM fin].

and PP. 157 . would be.(25) a. There is no categorical or syntactic reason to distinguish these three. The usage of be actually provides a strong reason why the grammar should allow a verb categorized as ‘V’ to also have the feature specification [AUX +]. On the assumption that every sentence has a main verb. to have/to be). Was the child running to the car? c.2 Be and Have The auxiliary verbs have and be are different from modal verbs. The child (?never) was (never) crazy. is). The child (?never) was (never) deceived. but an inflected nonfinite form. 8. be in (26a) is clearly a verb. He is a fool. (The child (never) became (*never) crazy.3. b. yet it also behaves exactly like an auxiliary. as an aspectual auxiliary with a progressive VP following. c.) b. For example. which is required in almost any analysis. VP. The simple lexical information of modal verbs given in (21). explains the main distributional possibilities of modal verbs. However. they have nonfinite forms (would have. *It is vital that we [VP[fin] will study everyday]. be and have here are main verbs. and as an auxiliary as part of the passive construction: (27) a. In addition. c. though in some cases with different syntax from when they are auxiliaries. and so forth. b. Was the child found? (29) Position of an adverb: a. AP. *We expect there to [VP[fin] will rain]. John was found in the office. John is in the school. they are different from modals in that they also have uses as main verbs. they have a 3rd person inflection form (has. Thus. as we will see below. they select not a base VP as their complement. all three uses have the lexical information given in (30) as their common denominator:3 3 XP here is a variable over phrasal categories such as NP. He has a car. Consider the examples in (26): (26) a. unlike modals. (28) Subject-Aux Inversion: a. their position relative adverbs including floated quantifiers. The verb be has three main uses: as a copula selecting an predicate XP. The child (?never) was (never) running to the car. a striking property of be is that it still shows the properties of an auxiliary: it exhibits all of the NICE properties. Was the child in the school? (*Did the child be in the school?) b. John is running to the car. for they all have NICE properties: they show identical behavior with subject-auxiliary inversion. b.

(30)     HEADPOS verb      AUX +        SPR 1 NP            VAL PRD +     COMPS XP   1 SPR All three bes bear the feature AUX with the + value. John is [AP happy about the outcome]. and select a predicative phrase whose subject is identical with be’s subject. John was [VP[ing] seeing his children]. examples like those in (32) are straightforwardly generated: (32) a. and it selects a past participle VP complement. copula be: COMPS XP b. d. John has sung a song and Mary has . John has not sung a song. b. the copula be needs no further specification: any phrase that can function as a predicate can be its COMPS value. Auxiliary have is rather similar in its properties to auxiliary be. Every use of be thus has the properties of a raising verb. Hence. Has John sung a song? c. (33) a. and the passive be requires its complement to be VP[pass]. The main difference between the three uses lies in the XP’s VFORM value:4 (31) a. Given facts like these. c. passive be: COMPS VP[VFORM pass] As given here. John hasn’t been singing a song. b. part of the perfect aspect construction: 4 See Chapter 9 for the further discussion of passive constructions. progressive be: COMPS VP[VFORM ing] c. 158 . too. we can posit the following information in the lexical entry for auxiliary have. The children are [VP[pass] seen in the yard]. The progressive be requires its complement to be VP[ing].

they have the syntax of auxiliaries.3. This do also exhibits the NICE properties: (40) a. as evidenced by the (b) examples below: (38) a. we can predict the following ordering too: (36) a. b. In no other circumstances does that distinction matter. (37b) is out since the following VP is finite. Are you a student? b. (39) a. For example. which is used as an auxiliary in the absence of other (finite) auxiliaries. c. while the surface syntax of a verb is determined by whether it has the specification [AUX +] or [AUX −]. This fact supports the idea that every sentence has a (main) verb in it. b. In addition. John can [VP[bse] have danced]. the auxiliaries have and be can follow a modal since both have bse as its VFORM value: (35) a. the main verb have also has the specification [AUX +]. b. even though they are main verbs. He will [have [been [seeing his children]]]. 159 . (37) a. John does not like this town. You are a student. *George has [went to America]. *Americans have [paying income tax ever since 1913]. (37a) is ungrammatical since have requires a perfect participle VP. at least. John can [VP[bse] be dancing]. typically in British English. b. 8.3 Periphrastic do Next we discuss the so-called “dummy” do. b. He has [seen his children]. Have you enough money? The main verbs be and have show the NICE properties. In some varieties of English. He must [have [been [being interrogated by the police at that very moment]]]. You have not enough money.(34)  have                     POS verb HEAD   AUX +    SPR 1      VFORM en   COMPS VP 1 SPR The interaction of subcategorization and morphosyntactic information is enough to predict the ordering restrictions among modals.

. John did not come. *John does leave. Except for carrying the grammatical information about tense and agreement (in present tense). John DOES leave. if do is used in a positive statement. unlike other auxiliaries. do appears neither before nor after any other auxiliary: (42) a. no such requirement exists. *He does be leaving. b. *They will do come. *They expected us to do/should leave him. do does not appear in nonfinite clauses.c. c. do is specified to be [AUX +]. Second. it needs to be emphatic (stressed). the verb do has no obvious intrinsic meaning to speak of. First. How long did it last? The most economical way of representing these lexical properties is to give do the lexical entry given in (46). contraction. John DID not come. b. b. just like the other 160 . it has no semantic contribution. b. *He does have been eating. which ensures that do is sensitive to negation. (43) a. d. Did John find the solution? b.) (45) a. (44) a.   (46) do       POS verb        HEADAUX  +         VFORM fin        SPR 1 NP              AUX − VAL      VFORM bse   COMPS VP       1 SPR Like other auxiliaries including modals. inversion. Jane likes these apples even more than Mary does Like the modals. But in negative statements and questions. (more likely in this case: John did NOT come. Third. I found myself needing/*doing need/*should needing sleep. They didn’t leave any food. and ellipsis (NICE properties). There are also some properties which distinguish do from other auxiliaries. (41) a.

auxiliaries. b. b. the VPs following the auxiliary do. *He does [can leave here]. *Jim [DOES [have supported the theory]]. b. for only the first verb in a sequence may be finite. b. The differences stem from the lexical specifications on the feature values for HEAD|POS and its complement VP. (51) a. b. Further. *I [do [not [be happy]]]. John may leave. *John did not rain. This explains their ungrammaticality. do selects a subject NP and a VP complement whose unrealized subject is structure-shared with its subject ( 1 ). This feature specification blocks modals from heading the VP following do. do is specified to be fin. Their properties indicate that they have different lexical information from the verb forms used in non-imperatives.5 are special properties of do in imperatives. c. bear the feature [AUX +] inherited from the auxiliaries have and be. *The proposal [DID [be endorsed by Clinton]]. Do be honest! Don’t be silly! 5 There Do and don’t in imperatives also have one distinct property: only don’t allows the subject you to follow (try inserting you in (ia) and (ib)). John did not leave. as we have seen above: (47) a. (52) a. *John may rain. This property then accounts for why no auxiliary element can precede do. Treating do as a raising verb like other English auxiliaries is based on typical properties of raising verbs. stressed or not. *He does [may leave here]. (i) a. (48) a. c. and different properties with don’t. for modals are specified to be [fin]. predicting the ungrammaticality of the examples in (50): (50) a. b. The lexical entry further specifies that the complement of do is a VP[AUX −]. one of which is that raising verbs allow expletives as their subject. The requirement on the complement VP of the auxiliary do is [VFORM bse]. The [AUX +] specification and raising-verb treatment of do account for the similarities to other auxiliaries and modals. (49) a. Do in imperatives can occur before another auxiliary like be and have. He might [have left]. In (51) and (52). It did not rain. Unlike have and be. 161 . This requirement will correctly predict the ungrammaticality of examples in (51) and (52). It may rain. *He might [do leave]. b. *I [do [not [have sung]]].

John believes Kim not to leave here.4 Explaining the NICE Properties In this section we discuss how we can account for the NICE properties. *John did not leaving here. *John believed Kim to do not leave here.1 Auxiliaries with Negation The English negative adverb not leads a double life: one as a nonfinite VP modifier. *John did not may leave.8. Tom wanted to go home. The verb to is. b. its syntactic behavior puts it in the class of auxiliary verbs (cf. the actual marker of the infinitive in English.4. and negates exactly that constituent: 162 . . which are key diagnostics for presence of auxiliary verbs. 1985): (53) a. and to only in nonfinite contexts. In terms of NICE properties. but Peter didn’t want to b. Even though it has the form of a preposition. of course.4 Infinitival Clause Marker to The auxiliary verbs to and do. Gazdar et al. *John expect to must leave. nonbase forms or modals cannot head the complement VP): (54) a.3. 8. in addition to differing by just one phonological feature. marking constituent negation. differ in an important syntactic property: do appears only in finite contexts. *John believed Kim to leaving here. b. 8. d. to also falls under the VP ellipsis criterion: (55) a. Constituent negation is the name for a construction where negation combines with some constituent to its right. whose complement must be headed by a V in the bse form. voicing. These properties indicate that to should have a lexical entry like this:     (56) POS verb    HEADAUX   +          VFORM inf     SPR 1  NP          VFORM bse     COMPS VP  1 SPR It is an infinitive auxiliary verb. marking sentential negation. These verbs share the property that they obligatorily take bare verbal complements (hence. and the other as a complement of a finite auxiliary verb. c. Lee voted for Bill because his father told him to .

Lee is believed [not VP[inf][to like Kim]]. In the grammatical examples in (60) and (61). Sentential Negation: In contrast to constituent negation. *Lee is believed [to VP[bse][like not Kim]]... not modifies a nonfinite VP. b. (60) a..Constituent Negation: The properties of not as a nonfinite VP modifier can be supported from its similarities with adverbs such as never in nonfinite clauses as given in (57): (57) a. which is the canonical expression of negation. Lee is believed to [not VP[bse][like Kim]. (62) The president could not approve the bill. 163 . c. via the following lexical entry:   (58) not      POS adv      NEG +    HEAD    MOD VP[VFORM nonfin] The adverb not modifies any nonfinite VP: (59) Constituent Negation: VP[nonfin] u lll uuuu lll uuu ll ll uuu lll uuu u Adv 1 VP[nonfin] BB MOD 1 VP[nonfin] BB BB BB not .. b. Taking not to modify a nonfinite VP. [Not [speaking English]] is a disadvantage.. *[Speaking not English] is a disadvantage. One way to distinguish the two types of negation comes from scope possibilities in an example like (62) (cf. we can predict its various positional possibilities in nonfinite clauses. Negation here could have the two different scope readings paraphrased in (63). in the ungrammatical examples the VP[nonfin] lexical constraint is violated. Warner 2000). *Lee likes not Kim. b. Duty made them [never/not [miss the weekly meetings]]. c. We asked him [never/not [to try to call us again]]. Kim regrets [never/not [having seen the movie]]. (61) a. c. there is sentential negation.

but Lee did not . even though they all behave alike in nonfinite contexts. *John [not [left the town]].(63) a. *John [not [could leave the town]]. It would not be possible for the president to approve the bill. We have seen the lexical representation for constituent negation not above. 1 XP        This lexical rule allows a finite auxiliary verb with a complement ( 1 ) to select an extra complement. That is. Observe the following contrast: (67) a. (cf. b. It would be possible for the president not to approve the bill. we can assume that when not is used as a marker of sentential negation. Kim and Sag 1995. This construction shows one clear difference between never and not: not can only modify a nonfinite VP. . Lee never/*not left. b. *Mary sang a song. Mary sang a song. Sentential not may not modify a finite VP: (64) a. and we use contrasts between it and not to reason what the properties of not must be. it is selected by the preceding finite auxiliary verb via a lexical rule: (68) Negative Auxiliary Verb Lexical Rule:       AUX +    +  HEADAUX HEADVFORM fin       VFORM fin  ⇒      NEG +  COMPS 1 XP COMPS Adv[NEG +]. The data here indicate that not behaves differently from adverbs like never in finite contexts. but Lee did never c. The first interpretation is constituent negation. but Lee never did b. Mary sang a song. Another difference between never and not is found in the VP ellipsis construction. John wants [not [to leave the town]]. (66) a. 2002).) b. b. The most economical way to differentiate sentential negation from constituent negation is to assume that sentential negation is actually a syntactic complement of a finite auxiliary verb (cf. . never is a true diagnostic for a VPmodifier. John could [not [leave the town]]. Lee did not leave. a property further illustrated by the following examples: (65) a. Sentential not appears linearly in the same position – following a finite auxiliary verb – but shows different syntactic properties. the second is sentential negation. Lee will never/not leave. marked [NEG +]. This rule gives a lexical entry which licenses the following structure for sentential negation: 164 .

the adverb not and the VP leave the town. the present analysis will assign two different structures for the string (62): (70) a.4). the negative finite auxiliary verb could selects two complements. VP[AUX +] ll‚‚‚‚‚ ‚‚‚ lll ‚‚‚ lll l ‚‚ lll l V[AUX +] VP[VFORM bse] lu lll uuuu uuu lll lll uuu lll uuu u Adv 1 VP[VFORM bse] could ÔXX MOD 1 VP ÔÔ XXX XX ÔÔ XX ÔÔ X ÔÔ Ô not approve the bill 165 . By treating not as an modifier meaning constituent negation and as a complement marking sentential negation. 3 HH   HH  HH   could not leave the town As shown here.(69) VP   VFORM fin     AUX +    COMPS ht hhhh tttt hhhh tt h tt hhhh tt hhhh tt tt V tt tt   tt VFORM fin   2 Adv 3 VP[bse]   HH AUX +     HHH HH  COMPS 2 [NEG +].4. For example. and all together these pro