309

5

On the Structure of English Partitive NPs and Agreement*

Jong-Bok Kim

(Kyung Hee University)

Syntactic and semantic complexities of English partitive constructions

have given us many troubles in understanding English NP structures. The paper begins with the classification of partitive NPs into two types based on agreement factors and claims that enriched information on lexical entries can provide feasible generalizations for such constructions while capturing their syntactic and semantic properties in a systematic way. Rather than resorting to ad-hoc syntactic operations such as movement, the paper tries to show that the present lexicalist analysis can provide a viable alternative for the constructions in question.

Key words: partitive, agreement, HPSG, measure noun, floating quantifier, preposition, extraposition

1. Introduction

The complexity of the English partitive noun phrases, typically consisting of

* An earlier version of this paper was presented at the 6th Pan-Pacific Association of Applied Linguistics Conference from July 30 to August 1, 2001, Jeju National University. I appreciate the participants of the conference for their constructive criticisms and comments. I am also grateful to Chung Chan, Sae-Youn Cho, Yongjun Jang, Yae-Shiek Lee, William O'grady, among others for comments and suggestions. Two anonymous reviewers also deserve special thanks for the criticisms and suggestions that helped improve the quality of the paper significantly. All misinterpretations and errors are of course mine.

Jong-Bok Kim310

a count noun phrase followed by of-NP, has provided challenges to generative

grammars. The two main research issues that are of concern, in particular, are

agreement and syntactic structure:

(1) Agreement:

a. Most of the people *appreciates/appreciate the complications of the

situation.

b. If no one at this time *answer/answers the door, just move along.

c. At least half of the participants *considers/consider the lectures very

informative.

d. All of the soda *were/was drunk by the little children.

e. Neither of the teams *have/has won any prizes yet.

f. Despite our best efforts, most of the work still *need/needs to be

done.

(2) Structure of the partitive NPs:

a. One of *students/the students came to see me last night.

b. Some of wire/*the wire has been stolen last night.

c. Some of *many problems/the many problems have been solved by the

students, but not all.

d. The teacher could recognize neither of *students/the students.

e. The team examined each of *the suggestion/the suggestions thor-

oughly.

The data in (1) suggest that the seemlingly syntactic heads such as most, no

one, half, all, and most are not the sole determiner in deciding the verb

agreement. It appears that these nouns are ‘number transparent’ in that they

allow the number of the oblique partitive NP (after of) to percolate up to

determine the number of the whole NP. In addition, the examples in (2) tell us

On the Structure of English Partitive NPs and Agreement 311

that the partitive NP seems to require a specific type of noun phrase after the

preposition of, either due to syntactic or semantic reasons. The goal of this

paper is to present a grammar of English partitive NP constructions that can

account for such complex properties in these constructions. The grammatical

framework this paper adopts is a lexicalist, constraint-based one that posits no

movement operations but allows rich lexical information in a systematic way.

2. Two Types of NPs

In English there are two main types of NPs: simple NPs and partitive NPs

given in (3) and (4).

(3) a. some objections

b. most students

c. all students

d. much worry

e. many students

f. neither cars

(4) 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 (4), the partitive phrases have the quantifiers followed by of NP phrase,

designating a set out of which certain individuals are selected. In terms of

Jong-Bok Kim312

semantics, these partitive NPs in (4) are no different from simple NPs in (3).

This semantic reason led traditional grammarians (cf. Jackendoff 1968, Selkirk

1977, Chomsky 1970, Bresnan 1973) to link the two constructions by

transformation operations. For example, within transformation analyses such as

that of Jackendoff (1968), the noun following of is the head of the construction,

and the quantifier occupies the prearticle position as illustrated in (5). The

preposition of is then obligatorily inserted if the article is definite:

(5) NP

Det N

Prearticle Art students ⇒ several of the students

Several the

In a similar spirit, Jackendoff (1977) posits a null PRO to explain why

nothing (e.g. adjectives) can precede quantifiers. His assumed structure is given

in (6):

(6) NP

Q N'

many N PP

PRO of the men

As we can observe here, these syntax-based approaches posit otherwise

On the Structure of English Partitive NPs and Agreement 313

unmotivated abstract elements such as PRO and complex structures. We believe

such an analysis makes the grammar more complicated, missing various

generalizations we discuss in what follows. This paper tries to provide simpler

and more intuitive syntactic structures of English partitive NPs that adopt no

such transformational tools. The paper will show us that a simple analysis of

agreement in partitive NP constructions is possible if we adopt more

fine-grained theory of lexicon.

3. Base-Generation or Transformation?

Let us first discuss issues in deriving partitive phrases from simple NPs

through transformation mechanisms. As noted by Selkirk (1977), such an

approach misses clear contrasts between the two constructions.1

1. An anonymous reviewer points out that it is rather unfair to refute transformational analyses if we adopt a DP anlaysis. The reviewer suggests that within a DP analysis, the partitive pronoun some selects [PP [of [DP the boys]]] whereas the determiner some selects an NP (see (31) and (43) also):

(i) a. [some [NP boy]]b. [some [PP [of [DP the boys]]]]

The present analysis also does not claim that there is a direct relationship between such examples. The only conjecture we make is a correlation between partitive pronouns in (ii)a and pure pronouns in (ii)b:

(ii) a. Some of the students came home.b. Many of the students came home, but some still did not.

Even within such a DP analysis, we need to have a theory that account for how the partitive some in (i)b requires a PP whose FORM value is of and this preposition selects a DP. If this is the generalization we need to capture, we need a system in which some can refer to

Jong-Bok Kim314

First, the lower NP in partitive phrases must be definite and also in the

of-phrase no quantification NP is allowed as in (7):

(7) a. Each student vs. *Each of students

b. Some problems vs. *some of many problems/*all of some men

Second, not all determiners with a quantificational force can appear in

partitive constructions. As illustrated in (8), determiners such as the, every and

no cannot occupy the first position:

(8) a. *the of the students vs. the students

b. *every of his ideas vs. every idea

c. *no of your book vs. no book

Third, the simple NPs and partitive NPs have different restrictions on the

semantic head. Observe the contrast between (9) and (10) (Data from Baker

1989):

(9) 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 frescoes did the flood damage?

d. I read some of the book.

(10) a. *She doesn't believe much story.

b. *We listened to as little speech as possible.

c. *How much frescoes did the flood damage?

its complement's daughter the boys in (i)b without violating the ‘locality condition’. The present analysis can be taken to be a syntactic attempt to solve such an issue within a lexicalist framework, not refuting other possible transformation analyses though there are few to refer to.

On the Structure of English Partitive NPs and Agreement 315

d. *I read some book.

The partitive constructions in (9) 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 shown in (10), the same elements serving as determiners cannot

precede such nouns.

Another issue concerns lexical idiosyncrasies, as noted in Selkirk (1977):

(11) a. One of the people was dying of thirst.

b. Many of the people were dying of thirst.

(12) a. *One people was dying of thirst.

b. Many people were dying of thirst.

The partitives can be headed by quantifiers one and many as shown in (11) and

(12). However, unlike many, the numeral one cannot serve as a determiner

when the head noun is collective as in (12)a.

Extraposition also shows another difference between simple NPs and partitive

NPs. Observe the contrast in (13) and (14):

(13) a. How many of the answers [to this classical mechanical problem] have

been found?

b. *How many of the answers have been found [to this classical

mechanical problem]?

(14) a. How many answers [to this classical mechanical problem] have been

found?

b. How many answers have been found [to this classical mechanical

problem]?

Jong-Bok Kim316

In both partitive and simple constructions, the PP to this classical mechanical

problem serves as the complement of answers. The difference emerges when we

see that the partitive construction does not allow this PP complement to be

extraposed as given in (13).

What the observations we have seen so far tell us is that we cannot simply

derive partitive constructions from simple noun phrases. The two constructions

induce quite different lexical and syntactic properties that no independently

motivated transformation mechanisms can capture.

4. The Structure of Partitive NPs with Quantity words

4.1. Analysis

The starting point of the present analysis is to classify partitive NPs into two

types based on agreement facts: Type I and Type II.

In Type I, the number value of the partitive phrase is always singular (cf.

Celce-Murcia and Larsen-Freeman 2000).

(15) a. Each of the suggestions is acceptable.

b. Neither of the cars has air conditioning.

c. None of these men wants to be president.

In Type II, the number value depends on the head noun in the of-NP phrase.

(16) a. Most of the fruit is rotten.

b. Most of the children are here.

c. Some of the soup needs more salt.

On the Structure of English Partitive NPs and Agreement 317

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 (16), 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. In terms of a semantic perspective indefinite pronouns such as some,

half, most and all may take either singular or plural verbs, depending upon the

meaning of the of-NP phrase. If these phrases tell us how much of something

is meant, the verb is singular; but if they tell us how many of something is

meant, the verb is plural. Expressions in (17) also exhibit similar behavior in

terms of agreement.

(17) half of, part of, the majority of, the rest of, two-thirds of, a number of

(but not the number of),...

We assume that the most effective way of capturing both similarities and

differences between Type I and Type II constructions is to resort to the lexical

properties of partitive nouns. One obvious similarity of the two types is that

they are pronouns serving as the head of the constructions and in addition select

an of-NP[definite] phrase as seen from the contrast in (18):

(18) a. *neither of students, neither of the two linguists

b. *some of water, some of the water

This basic lexical information can be represented in a feature system of

HPSG as given in (19):2

2. See (26)-(30) for what we mean by the fact that a PP is definite.

Jong-Bok Kim318

(19) a. Type I: neither

HEAD pronoun

FORM of

COMP(LEMENT)S PP

DEF +

b. Type II: some

HEAD pronoun

FORM of

COMPS PP

DEF +

As represented in (19), both Type I neither and Type II some are lexically

specified to require a PP whose semantic value is definite.3

3. As an anonymous reviewer points out, there is of course a clear semantic reason for the partitive pronoun to select a definite NP: one cannot select things out of an indeterministic set. However, this does not mean that we can purely rely on semantics in the c(omplement)-selection. Though it is a general tendency that there is an isomophic relation between c-selection and s(emantic)-selection (see Grimshaw 1979), there are cases where such a one-to-one mapping relation does not hold. For example, ‘verbs of becoming’ show us that it is hard to rely on s-selection alone (cf. Pollard and Sag 1994):

(i) a. Kim *grew/*got/turned out/ended up a successb. Kim *grew/*got/*turned out/ended up doing all the workc. Kim grew/got/turned out/*ended up to like enchovies.

The contrasts here seem to be related to the lexical properties or c-selection of the verbs in question. Partitive constructions are no exception: they also need to refer to syntactic as well as semantic aspects of the lexical properties, which will be clear in due course.

What we thus assume is that the grammar needs to access to c-selection as well as s-selection information (syntax as well as semantics), as indicated by various peculiar syntactic properties of the constructions in question. That is, the grammar we assume in the

On the Structure of English Partitive NPs and Agreement 319

There are two things to be noted in order here: the feature DEFINITE and

the lexical representation. The reference to the feature ‘definite’ seems to reflect

a correct generalization in there and partitive constructions, in particular with

determiners such as a, the, and some:

(20) a. *there be [+ definite] N XP (Milsark 1977)

b. *Det of [- definite] N (Milsark 1977)

However, such a pure dichotomy between definiteness and indefiniteness runs

into a problem when checking with more determiners.

(21) a. There are *the/*all/*most/three boys in the garden.

b. Some of the/*most/*twenty boys are red-haired.

c. Half of the/all/*most/*twenty boys are read-haired.

Like numerals but unlike the and all, most is blocked in the crucial position of

partitives. One solution that one could take is to adopt more than just one

dichotomy as suggested in Barwise and Cooper (1981). They distinguish

between a weak-strong among determiners and a definite-indefinite contrast

among NPs. The weak-strong contrast is reflected in there-sentences, whereas

the definite-indefinite contrast is relevant in relation to the partitive construction.

Since the partitive constructions are only sensitive to definite and indefinite

distinction, the feature DEFINITE will do justice to the present analysis.

A second issue is, as noted by an anonymous reviewer, that if we simply list

every lexical item with its own information every time, we then loose

present analysis is the one that can refer to all the grammatical information (e.g., syntax, semantics, phonology, pragmatics) in the same parallel level (cf. Jackendoff 2002). We believe that a better, complete theory would be the one that can capture syntax as well as semantic properties of the constructions in question. This paper focuses on the syntax aspect.

Jong-Bok Kim320

appropriate linguistic generalizations, the generalizations about classes of words

with common behavior. In the present context, both types of partitive nouns

select a definite PP. One effective way of eliminating such a ‘vertical’

redundancy is hierarchical classification of words (cf. Flickinger, Pollard and

Wasow (1985) and Pollard and Sag (1994)).

The concept of hierarchical classification, introduced to avoid the ‘vertical’

redundancy, is essentially assigning words to specific categories (formally

termed), and an assignment of those categories to superordinate categories

(supertypes). For each sort, certain constraints are stated (the constraints are of

course declared in terms of constraints on feature structures). The constraints

each type carries correspond to properties shared by all members of that sort.

The technique of hierarchical inheritance further ensures that a type inherits all

the constraints of its supertypes. Thus, a word assigned to a type obtains all the

features and constraints associated with its supertypes, in addition to its own

constraints. Due to the organization of the lexicon in this hierarchical fashion,

we now can avoid stating redundant information for each lexical entry. That is,

the only information we need to encode in a lexical entry is the information that

is not inherited from the supertypes of that lexical element.

Interpreting this in the present system, we could assume that English noun

system roughly has the following hierarchical classification:

(22) Noun

common-noun pronoun proper-n

relative-pn partitive-pn personal-pn

type1 type2

On the Structure of English Partitive NPs and Agreement 321

Given this inheritance hierarchy, we then can move the general properties of

partitive nouns on the type partitive as in (23):

(23)partitive

FORM ofCOMPS PP

DEF +

type1 type2

Thus the common properties of the pronoun partitives will be inherited onto its

two subtypes type1 and type2 whose membership includes neither and some

respectively.

However, the two types are different in terms of agreement: type1 pronouns

are lexically specified to be singular whereas the number value of

type2pronouns is identical to that of the selected PP. Within our system, this

difference can be teased out in the hierarchy as in (24):

(24) partitive FORM of COMPS PP DEF +

type 1 type 2 HEAD NUM sng HEAD NUM α

COMPS NUM α

The only thing we added is the number value to the head value of the pronoun.

The difference between the two types is then just a matter of the NUM(BER)

value. Given this inheritance system, an instance of type1 and type2 partitive

Jong-Bok Kim322

noun will have the following lexical information, some inherited from its

supertypes and the other as its own constraints:

(25) a. Type I: neither

pronoun HEAD

NUM sing

COMP PP DEF +

b. Type II: some

pronoun HEAD

NUM αDEF +

COMP PP NUM α

If we represent this difference in terms of syntactic structures, it would be

something like those given in (26) and (27):

(26) NP[NUM sng]

N[NUM sng] PP

neither P NP

of the students

(27) NP

N[NUM α] PP[NUM α]

Some P NP[NUM α]

of the students

One question that may rise, as an anonymous reviewer pointed out, is how a

On the Structure of English Partitive NPs and Agreement 323

PP can have number values and be even definite. There are two types of

prepositions: those that function as predicates and those that serve as argument

markers (see Sag and Wasow 1999). As for the PPs headed by these markers,

as in the partitive construction, their semantic features are identical with the

prepositional object NP:

(28) a. John is in the room.

b. John gave a book to Bill.

The preposition in in (28)a is a predicative preposition selecting two arguments,

whereas to in (28)b is simply an argument marker. Put it differently, the

preposition to in (28)b just indicates what role its object NP (Bill) plays in the

situation denoted by the verb gave. Meanwhile the preposition in in (28)b

functions much like a verb, introducing new predicates and having its own

argument structure. Such a distinction can play an important role in binding

facts as noted by Sag and Wasow (1999):

(29) a. The housei had a fence around iti/*itselfi.

b. To make a noose, you wind the rope around itselfi/*iti.

c. Susani wrapped the blanket around heri/herselfi.

Once we accept the view that the preposition around can either an argument-

marking or predicative preposition, we could easily account for the variability of

binding possibilities observed in (29). It seems that around in (29)a functions as

a separate predicate, around in (29)b the preposition around just marks one of

the arguments of the verb wind, and around in (29)c can ambiguously be either

an independent predicate or an argument-marking, as represented in (30):

Jong-Bok Kim324

(30) a. argument-marking-p

PHON around

HEAD preposition

SUBJ

COMPS NP

b. predication-p

PHON around

HEAD preposition

SUBJ NP

COMPS NP

What we need then is to assume that the PP headed by an argument-marking

preposition will have the identical syntactic-semantic information to its object

NP, and to posit a general theory of binding in terms of argument-structure

roughly stating that an anaphor must be bound by a preceding argument in the

ARG-STR (argument-structure) (Sag and Wasow 1999). The argument structure

of the main verbs in (29) will look like the following:

(31) a. verbPHON had ARG-ST NP, NP

b. verbPHON windARG-ST NP, NP, PP [ANAPHOR+]

c. verbPHON wrappedARD-ST NP, NP, (PP[ANAPHOR+])

On the Structure of English Partitive NPs and Agreement 325

Thus, itself in (29), the object of a predicative preposition, is not an argument

of had. This explains why it cannot occur here. Since the verb wind selects the

PP complement around itself in (29)b, the preceding binder rope can binds this

anaphor. But the pronoun cannot occur here since it is bound within the same

argument structure (binding domain). In wrap, when the PP complement is one

of its argument, the prepositional object can be anaphoric. But when it isn't an

argument, the prepositional object can be a pronoun.

In this respect, we assume that of in the partitive construction is just an

argument marker, playing no semantic role. This is identical to the claim that

the complement PP in the partitive construction will have the identical

syntactic-semantic information to its object NP as represented in (31):4

(32) NP

N PP[SYNSEM ꊺ]

... P NP [SYNSEM ꊺ]

of .....

As an anonymous reviewer did, one could wonder how the present analysis

deals with cases where pronouns such as some and neither appear without the

PP complement or when they are used as a determiner:

(33) a. I ate some apples.

b. I bought lots of apples, but some are rotten.

4. The SYNSEM here means syntax-semantics. In addition, this idea could be in a similar line with the assumption of inserting the preposition of in the construction.

Jong-Bok Kim326

As in (33), there are several different usages for words like some, neither,

few, and so forth. The pure pronoun usage in (33)b can be easily predictable by

assuming that all the PP complement is optional. This relation can be easily

captured with a revision of the hierarchical structure in (20) as in (34):5

(34) pronoun

relative-pn indefinite-pn personal-pn

partitive-pn nonpartitive-pn

type1 type2

What this hierarchy implies is that indefinite-pn has two subtypes partitive-pn

and nonpartitive-pn. One main difference between the two types is the

subcategorization information: the former selects a PP[of] whereas the latter

requires nothing. Thus pronouns such as someone and anyone would be

instances of nonpartitive-pn. However, pronouns such as some, neither, most,

each, and so forth will be both partitive-pn and non-partitive. We then see how

the true indefinite pronoun and the partitive pronoun are interconnected.

4.2. Consequences

The present system makes the English grammar much more simple and in

addition bring us several welcoming consequences.

First, the system enables us to provide a straightforward account for the

5. But for the determiner usage, there seems to be no direct syntactic connection with the pronoun usage as we discussed in section 3, though a finer-grained study remains to be developed. Also see (43) for a simple lexical entry for the determiner usage of such pronouns.

On the Structure of English Partitive NPs and Agreement 327

contrast in (35).

(35) a. many of the/those/her apples

b. *many of some/all/no/ apples

(35)b is simply out since many requires an of-PP prhase whose definiteness is

positive.

Second, the lexicalist system also can capture the fact that the quantifier

pronouns affect the number 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 a restriction. This is illustrated

in (36) and (37).

(36) Type I:

a. one of the suggestions/*the suggestion

b. each of the suggestions/*the suggestion

c. neither of the students/*the student

(37) Type II:

a. some of his advice/students

b. most of his advice/students

c. all of his advice/students

The only additional specification we need is that the PP complement of Type I

is plural, as shown in (38). As for Type II, no further specification is required.

Jong-Bok Kim328

(38) a. Type I: neither pronoun

HEADNUM sing

FORM ofCOMPS PP DEF +

NUM plural

Third, this line of approach can thus explain a clear contrast given in (39) and

(40) (cf. Baker 1989):

(39) 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.

(40) 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 tells that not all quantity words are acceptable with singular

count-noun. This contrast falls out naturally within our Type I and Type II

division resorting to enriched lexical information.

Fourth, the system proposed here also predicts the differences between simple

NPs and partitive NPs.6

(41) a. much advice

b. much of the advice

(42) a. *much story

6. The noun agrees with the determiner in the value of NUM as well as COUNTABLE (e.g. much furniture, *many furniture/*much students, many students)

On the Structure of English Partitive NPs and Agreement 329

b. much of the story

(43) a. *many advice

b. *many of the advice

The contrast observed here is easily expected, given the minimal difference

in the lexical information:7

(44) many and much as pronouns:

many mucha. HEAD pronoun b. HEAD pronoun

COMPS PP[def, plural] COMPS PP[def, nonplural]

(45) many and much as determiners:

many mucha. HEAD det b. HEAD det

COUNTABLE + COUNTABLE --

The analysis also can capture extraposition facts, given the general

assumption that the complement of a lexical head can be extraposed but not the

complement of the complement. Let us see the contrast here again (data

repeated here):

(46) a. How many of the answers [to this classical mechanical problem] have

been found?

b. *How many of the answers have been found [to this classical

7. As a reviewer points out, only when the feature COUNTABLE is positive, the NUM is plural. Such a relation could be captured by an underspecification mechanism. Since every feature structure needs to be well-resolved, that is, cannot be left unspecified, the current system needs to resort to a sort of type hierarchy that links the two feature attributes and states the relation.

Jong-Bok Kim330

mechanical problem]?

(47) a. How many answers [to this classical mechanical problem] have been

found?

b. How many answers have been found [to this classical mechanical

problem]?

As an anonymous reviewer points out, one could assume that many of the

answers is a strong QP (DP) whereas many answers is a weak QP (NP), and

that this difference induces the difference in the extraposition. Though such

semantic factors could play an important role in the possibility of extraposition,

the syntactic position of an extraposed phrase is also a key factor (Selkirk

1977):

(48) a. A review came out yesterday [of a new book about French cooking].

b.*A review of a new book came out yesterday [about French cooking].

The contrast here can be hardly relegate to the semantic difference between the

extraposed phrases.

The present analysis does not claim that such the difference in extraposition

is a pure lexical property: It is rather a misunderstanding to take the present

analysis to takes such a claim. We attribute the difference in extraposition to

interfaces among lexical, syntax, and semantic information. Let us see the

structural differences of these two examples within our analysis:

On the Structure of English Partitive NPs and Agreement 331

(49) NP

N PP

many P NP

of D N'

the N PP

answers to this ... problem

(50) NP

N N'

many N PP

answers to this...problem

As seen from the contrast in the structures, the PP to this ... problem in (49)

is much more deeply embedded than the one in (50).8

So far, we have seen that our lexical treatment provides us with an explicit

grammar of two different types of partitive constructions. In particular, it

accounts for their differences and similarities in a systematic way. The structures

projected from the lexical information also leads to an adequate explanation of

the constructions.

8. This is in line with the analysis of Akamajian (1975) suggesting that “no element may be extraposed more than one cycle up from the cycle containing it (where the cyclic domain is S or NP).”

Jong-Bok Kim332

5. Partitive NPs with Measure nouns

5.1. Facts

In addition to the two types (Type I and Type II), there is another type of

partitive constructions. These are partitive NPs with measure nouns. There are

several differences. First, partitive NPs with measure nouns, in addition to

allowing of-phrases with a definite NP, can be followed by an of-NP phrase that

contains a bare NP as illustrated in (52) (Quirk et al. 1985).9

(51) a. one pound of those beans

b. three feet of that wire

c. a quart of Bob's cider

(52) a. one pound of beans

b. three feet of wire

c. a quart of cider

Also measure nouns cannot occur in simple noun phrases (unlike partitives

with quantity words). They require an of-NP phrase obligatorily (cf. Baker

1989).

(53) a. *one pound beans (one pound of beans)

b. *three feet wire (three feet of wire)

c. *a quart cider (a quart of cider)

(54) many beans, some wire, much cider, no yogurt, one strawberry

Further, unlike partitive nouns, quantity words may not be preceded by

9. This is one main point in which measure nouns part company with quantity words: *many of beans, *some of wire, *much of cider, *none of yogurt, *one of strawberries.

On the Structure of English Partitive NPs and Agreement 333

numerals.

(55) a.*one many of the books, *several much of the beer,

b. one pound of beans, three feet of wire

What makes these types more complicated is the existence of defective

measure nouns:

(56) 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

5.2. Analysis

In terms of agreement, these types are similar to Type I: The number of

measure nouns rather than that of the of-NP phrase determines the number

value of the main predicate.

(57) a. A can of tomatoes is/*are ....

b. Two cans of tomatoes are/*is .......

In this respect, I claim the measure noun itself is the head of this

construction and selects an of-NP as its complement. But unlike Type I, there is

no definite restriction on the of-NP complement.

(58) poundHEAD nounSPEC(IFIER) DetCOMPS PP [of]

Jong-Bok Kim334

Notice that our system again resorts to lexical information.

There is a set of words whose behavior leave themselves somewhere between

quantity words and measure nouns. These are words such as dozen, hundred,

and thousand.

(59) a. three hundred of your friends

b.*three hundreds of your friends

c.*three hundreds of friends

d. three hundred friends

(60) a. several thousand of Bill's supporters

b.*several thousands of Bill's supporters

c.*several thousands of supporters

d. several thousand supporters

(61) a. hundreds of friends/*hundreds friends

b. dozens of roses/*dozens roses

c. thousands of supports/*thousands supports

The singular hundred when used as noun obligatorily requires PP[of] as well as

a specifier whereas the plural hundreds requires no specifier though it selects an

of-NP complement. This lexical information is represented in (62):

(62) hundreds hundredHEAD noun HEAD noun

a. SPEC b. SPEC [ ]COMPS PP [of] COMPS PP[of]

It is hard to attribute such peculiar properties to any syntactic process. If the

grammar refers to the enriched lexical information that may be required

independently, we would be expected to encounter such lexical idiosyncrasies.

On the Structure of English Partitive NPs and Agreement 335

6. Floating Quantifiers

The final point we would like to consider is the so called floating quantifier

constructions such as given in (63):

(63) a. all four of your sons

b. all four lions

c. all of the senators

d. all the senators

(64) a. both of the alligators

b. both the alligators

c. half of the money

d. half the money

The free distributional possibilities of quantifiers like all shown in (65) have led

to assume the transformational approaches we have seen earlier.

(65) (all) The students (all) have (all) been (all) being (all) kicked out.

Rather than adopting a movement process, following Kim and Sag (2002),

we treat adverbs like all as a VP modifying adverb rather than a floating

quantifier. The distributional possibilities in (65) can be simply captured by this

as represented in (66):

(66) VP

Adv VP

all ......

Jong-Bok Kim336

Since all is lexically specified to be a VP modifying adverb, it can appear in

any VP position.

(67) [The students vp[have vp[been vp[being vp[kicked out]]]]].

One remaining position that requires an explanation is the sentence initial

position. Considering that only a handful number of quantifiers can appear in

this position, wecan simply assume that they select a definite NP.

This adverbial treatment also renders the grammar simple for adverb

stranding in VP ellipsis constructions. The fact is that an adverb cannot precede

a VP ellipsis position:

(68) a. Kim has often visited grandmother, but Bill never has.

b.*Kim has often visited grandmother, but Bill has never.

(69) a. Only a few of the teachers have check out books, but the students all

have __.

b.*Only a few of the teachers have check out books, but the students

have all __.

Though one could attribute the ungrammaticality to pragmatic factors such as

focus, the present analysis can offer a simpler lexicalist analysis. Given the

assumption that adverbs require elements that they modify as argued in Kim

and Sag (2002), there is nothing for adverbs like never and all to modify in

(68) and (69).

7. Conclusion

We hope to have shown that the complexities of partitive constructions

On the Structure of English Partitive NPs and Agreement 337

cannot be simply captured by syntactic operations that link simple noun phrases

with partitive constructions. The constructions induce various lexical, idiosyncratic

properties that cannot be derived from phrase structure rules or syntactic

operations. The information of their peculiar properties is encoded in our mental

dictionary, lexicon. In midst of these lexicon properties, grammar should provide

generalizations for learners too. The system we have presented here can both

draw generalizations as well as peculiarities.

References

Akmajian, Adrian. 1975. More evidence for the NP cycle. Linguistic Inquiry

6:115129.

Barwise, Jon and Robin Cooper. 1981. Generalized Quantifiers and Natural

Language. Linguistics and Philosophy, 4:159-219.

Baker, C. L. 1989. Syntax, Cambridge, MA: MIT Press.

Bresnan, Joan. 1973. Syntax of the Comparative Cause Construction in English.

LI 4: 275343.

Celce-Murcia, Marianne and Diane Larsen-Freeman. 2000. The Grammar Book.

Heine and Heine.

Chomsky, Noam. 1970. Remarks on Nominalization. In R. Jacobs and P.

Rosenbaum (eds.), Readings in English Transformational Grammar.

Massachusetts: Ginn, Waltham.

Flickinger, Daniel, Carl Pollard and Thomas Wasow. 1985. Structure Sharing in

Lexical Representation. In Proceedings of the 23rd Annual Meeting ofthe

Association for Computational Linguistics. Morristown, N.J.: Association

for Computational Linguistics.

Grimshaw, Jane. 1979. Complement selection and the Lexicon. Linguistic

Jong-Bok Kim338

Inquiry 10:279326.

Jackendoff, Ray. 1968. Quantifiers in English. Foundations of Language 4:

422-442.

Jackendoff, Ray. 1977. X' Syntax: A Study of Phrase Structure. Cambridge: MIT

Press

Jackendoff, Ray. 2002. Foundations of Language. Cambridge: Cambridge Univ.

Press.

Kim, Jong-Bok and Ivan Sag. 2002. Negation Without Movement. Natural

Language and Linguistic Theory, 20.2:339-412.

Milsark, G. 1977. Toward an Explanation of Certain Peculiarities in the

Existential Construction in English. Linguistic Analysis 3:130.

Pollard, Carl and Ivan A. Sag. 1994. Head-driven Phrase Structure Grammar.

Stanford: CSLI publications.

Sag, Ivan and Tom Wasow. 1999. Syntactic Theory: A Formal Approach.

Stanford: CSLI Publications.

Selkirk, Elizabeth. 1977. Some Remarks on Noun Phrase Structure. In A.

Akmajian, P. Culicover, and T. Wasow, (eds), Studies in Formal Syntax.

New York: Academic Press.

Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech and Jan Svartvik. 1985.

Comprehensive Grammar of the English Language. London: Longman.

접수일자: 2002. 3. 31.

게재결정: 2002. 10. 14.