Citation
Nominals as complements

Material Information

Title:
Nominals as complements
Creator:
Albury, Donald Herbert, 1943-
Publication Date:
Language:
English
Physical Description:
viii, 148 leaves. : 28 cm.

Subjects

Subjects / Keywords:
Adjectives ( jstor )
Denial ( jstor )
Nominalization ( jstor )
Noun phrases ( jstor )
Nouns ( jstor )
Phrases ( jstor )
Predicates ( jstor )
Sentence structure ( jstor )
Sentences ( jstor )
Verbs ( jstor )
Dissertations, Academic -- Linguistics -- UF ( lcsh )
English language -- Grammar, Generative ( lcsh )
English language -- Noun ( lcsh )
English language -- Syntax ( lcsh )
Linguistics thesis Ph. D ( lcsh )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis--University of Florida.
Bibliography:
Bibliography: leaves 145-147.
Additional Physical Form:
Also available online.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
14046201 ( OCLC )
0022768575 ( ALEPH )

Downloads

This item has the following downloads:


Full Text












NOMl-iliALS AS COl-PLBEENTS









By

DONALD HERBERT ALBURY











A DISSERTATION PRESENTED !U THE GRADUATE COUNCIL,
OF THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF TE~E REQUIREiIFNTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY









UNIVERSITY OF FLOR IDA 1974

































Copyright
1974 By
Donald Herbert Albury




















To
Ginny and
Rebecca














AC KNOWLEDGMENTS

I wish to thank first of all my chairman, Jean

Casagrande, who has always encouraged me to approach linguistics in terms of what are the facts of language, and, then, what do the facts say about the theory. He has had the faith to let me try unorthodox approaches until I could either justify them or give them up. He has, above all, helped me "keep the faith" with my orn proposals when my enthusiasm flagged.

I also wish to thank Peter Ivenzel, who has pointed out many errors and inconsistencies in my arguments. Of course, all remaining errors remain my responsibility.

My wife Ginny and daughter Rebecca have had to put up with me for the last year while I worked on this dissertation. am grateful for their patience and support.
T also wish to acknowledge financial support from the University of Florida, in the form of a scholarship and a research assistantship, during the writing of this dissertation.














TABLE OF CONTENTS

Page

ACKNOWLEDGMENTS. . . . . ................ iv

ABSTRACT . . . . . . . . . . . vii

CHAPTER

I INTRODUCTION . . . . . . . . . 1

Statement of Purpose . . . . . . 1

The Lexical Hypothesis . ......... . 3

Derived Nominals as Complements on Nouns . 5 Some Assumptions . . . . . . 9

Outline . . . . .................. 11

NOTES . . . . . . . . . . 13

II DEFINI-NG DERIVED NOMINAL COMPLEMENTS ... . 15 First Definitions . . . . . . . 15

The Surface Structure of Derived Nominal
Complements . . . . . . . . 16

Action Nominals . . . . . . . 25

Agentive Nominals . . . . . . . 42

Summary . . . . . . . . . 44

NOTE S . . . . . . . . . . 45

III THE PRODUCTIVITY OF DERIVED NOMINAL COMPLE.IENTS 48

The Evidence for Lack of Productivity . .. 48 Chomsky's Subcategorization Account . . 51 RAISING Rules . ............... 55

V






vi

CHAPTERS Page

IT-EXTRAPOSITION .... 60

Passive Sentences ......... .......... 66

NP-PREPOSING ..... .. ............... 70

Dummy Subject Insertion ................ 93

EQUI-NP-DELETION ............... 97

DATIVEn-O0VEMENT ..... ............... .102

Psychological Predicates ........... 103

Conclusion ........ ............ .... .108

NOTES .......... ................... 110

IV RULE-ORDERING AND DERIVED NOMINAL COMPLEMENTS 116

Introduction . . ............... 116

Newmeyer's Proposal ............... 117

The Order of the Rules. ........... .121

The Cycle and Complementation .. ........ .127

Complementation as a Process .... .......... 139

NOTES. . .................. 142

V CONCLUSIONS ......... .................. 143

BIBLIOGRAPHY .......... ...................... 145













Abstract of Dissertation Presented to the Graduate Council
of the University of Florida
in Partial Fulfillment of the Requirements for the
Degree of Doctor of Philosophy


NOMINALS AS COMIPLEMVENTS

By

Donald Herbert Albury

March, 1974

Chairman: Jean Casagrande
Major Department: Linguistics
The syntactic construction known as the derived nominal presents problems to a transformational grammar of English.

Noam Chomsy has claimed that derived nominals have restricted productivity, an internal structure 'like that of noun phrases, and idiosyncratic semantic relationships to their associated predicates. Chomsky's claim that the Lexical Hypothesis provides a better explanation of the characteristics of derived nominals than the transformational position is rejected. A predicate-initial analysis of the

underlying structure of English and an analysis of derived nominals as complements of nouns in underlying structure are adopted, it is then shown that the productivity of derived nominals is not as restricted as Chomsky claims, and that the remaining restrictions are due to the failure of certain rules to apply in the derivation of derived nominals.
vii






viii

Frederick Nevmeyer's proposal that such rules fail to apply because DERI/ED NOMINALIZATION applies before they do, and nominalizes the predicate which is a part of the structural description of those rules, is adopted. It is then argued that the same rule-ordering arguments that account for the restrictions on productivity of derived nominals will also account for the noun phrase-like internal structure of derived nominals. It is further argued that the predicate-initial analysis eliminates the need for any rule of extraposition in English, that DERIVED NO MINALIZATION is a cyclic rule, and that the formation of complements from underlying embedded sentences in general is a process of the cycle applying to the embedded sentence, and not of some

higher application of the cycle.













CHAPTER ONE
INTRODUCTION


Statement of Purpose

The question of the extent of regularity in language has been a recurrent theme in linguistics since ancient times, when the Analogists of Alexandria disputed with the Anomalists of Pergamon. This question has recently become once more prominent in the study of transformational grammar. In an important sense, a transformation is an abstraction of a regularity in language. The question of whether a particular structure in a language can be described transformationally is an empirical one, and is equivalent to the question of whether that structure in that language exhibits a regular correspondence to another structure of that language. An assumption of transformational grammar has been that language is basically regular, and, as a consequence, transformational grammars have emphasized the regularity of a language.

Noam Chomsky (1970) has claimed that the English complement construction which he calls the derived nominal constitutes a part of English grammar which is not regular in the sense indicated above, and which cannot be described transformationally. He supports his claim by contrasting derived nominals to gerundive nominals, another type of complement
1






2

which he considers to be derived by transformations from underlying embedded sentences. This is an empirical claim, and may be tested by examining the sentences of English which contain these two types of complements. The forms of these complements, and their relation to each other and to sentences, is illustrated in 1.1, 1.2 and 1.3; the main clauses in 1.1 having the corresponding gerundive nominals in 1.2 and derived nominals in 1.3.

1.1a) John is usually calm under stress, which pleases

b) John completed the assignment early, which
leased Mary.

1.2a). John's usually being calm under stress pleases
LMary.
b) John's completing the assignment early pleased


1.3a) John's usual calmness under stress pleases Mary.
b) John's early completion of the assignment
leased Mary.

The underlined words are intended to supply appropriate contexts for the nominal complements as well as showing how sentences and nominal complements can be alternate realizations of a common underlying structure. Gerundive nominal complements often resemble sentences with progressive aspect. While this may be confusing at times, the intended reading of such forms as gerundive nominal complements may be made clear by putting them into a context such as ...is surprisinz,- or Ann is worried about....

Chomsky's argument that derived nominal complements are not regular, and cannot be derived from underlying embedded sentences, as are gerundive nominal complements, is based on






3
three characteristic properties; A, that productivity is more restricted for derived nominal complements than for gerundive nominal complements, B, that semantic relations between derived nominals and their associated propositions are varied and idiosyncratic, and C, that derived nominal complements have the internal structure of noun phrases, Having concluded that derived nominal complements cannot be derived transformationally, Chomsky claims that the phrase structure rules of English must be complicated to provide the structure of derived nominal complements directly and that derived nominals are lexical entries. This is the Lexical Hypothesis, and amounts to a claim that derived nominal complements show less regularity than other structures in English, and the limits of regularity have been found for this oart of the grammar of English. If Chomsky is krong about the implications of the properties s of derived nominal complements, then his argument for the need for the Lexical Hypothesis is weakened. I propose to show that Chomsky's three properties do not establish the irregularity of the formation of derived nominal complements, that their formation is as regular as other transformational processes, and that a transformational account of derived nominal

complements is at least as well motivated as the Lexical Hypothesi-s.

The Lexical Hypothesis

Chomsky (1970:188) states the lexicalist position as the choice of extending the base rules, with concomitant






4

simplification of the transformational component. By contrast, he gives the transformationalist position as the choice of simplifying the base structure and extending the transformational apparatus. Chomsky states that the choice between the two positions is entirely an empirical one. Chomsky's 1970 study is a presentation of arguments for the validity of the lexicalist position. I will consider those arguments at appropriate points below, but for now I will give a summary of the Lexical Hypothesis as presented by Chomsky.

The theory within which Chomsky is making his claims about derived nominals pictures syntax as being strictly divided between a base component and a transformational component (the "Aspects" model: cf. Katz and Postal, 1964, and Chomsky, 1965). The point of connection between the two components, when all the base (phrase-structure) rules have applied, but before any transformational rules have applied, is called deep structure. A third component is the lexicon, which supplies lexical items to be inserted in, and only in, deep structure.

The Lexical Hypothesis involves the claim that derived nominal complements are produced directly by a subset of the base rules, that derived nominals are supplied directly from the lexicon in deep structure, and that derived nominal complements are not subject to most transformational rules. Chomsky (1970:195) notes that such a claim implies that phrase structure rules must introduce an extensive range of






5

derived nominal complement structures parallel to the structures oft embedded sentences. Indeed, there seem to be few, if an y, sentential deep structures in Chomsky's system which do not have corresponding derived nominal complements.

According to Chomsky (1970:190), in early transformnational theory, "there was no other way to express the fact that the contexts in which refuse appears as a verb and refusal as a noun are closely related" than in terms of transformational rules. Chomsky further states that when contextual features were introduced into the theory (in Chomsky, 1965), it became possible to separate the lexicon from the categorical component of the base, and thus to adopt the Lexical Hypothesis. According to Chomsky (1970t

1.90), refuse is entered in the lexicon with certain features specified, but with no specification of the categorical features [noun] and [verb]. He adds that fairly idiosyncratic morphological rules would be involved in deriving -forms like refusal in derived nominal convolements.

Derived Nominals as Complements on Nouns

The term complement has been used in ways different

enough -to make it worth while indicating here what I mean by gerundive nominal complement and derived nominal complement. On the one hand, anything that completes a structure can be regarded as a complement. Thus, the underlined parts of the examples in 1.4 are all complements of the verbs in their respective sentences.






6

1.4a) John arrived yesterday.
b) I want to leave now.
c) Mike totaled his car.

On the other hand, complement has been used to refer to nominalized (or otherwise reduced) embedded sentences such as those cited in 1.2 and 1.3. By nominal complement I will refer to nominalized embedded sentences which are complements of nouns in noun phrases.

Peter Rosenbaum (1967) argued that the type of complement structure he called Noun Phrase Complementation is a noun phrase in deep structure, consisting of the pronoun it as the head noun, and an embedded sentence as a complement of it. Rosenbaum contrasted Noun Phrase Complementation with Verb Phrase Complementation, in which an embedded sentence is a complement of a verb in deeo structure.

Rosenbaum's claim that the pronoun it is present in

noun phrase complements in deep structure now appears to be wrong. I will present arguments against that claim in the section on TT-EXTRAPOSITION in Chapter Three.

Peter enzel (1969) has argued that gerundive nominal complements (and, less explicitly, derived nominal complements) are complements of one of a certain set of deletable head nouns in underlying structure. He points out (pp. 7781) that verbs which take gerundive nominal complements, with a few exceptions such as verbs of belief and say, declare and claim, also allow the construction Noun + Complement, or just the noun as object. The nouns which can appear in such constructions (i.e., fact, proposition, event,






7

action, etc.) may also be taken as names of the types of complements a verb allows. Thus, John's coming can be both a fact and an event, and we get the sets of possible sentences in 1.5 and 1.6.

1.5a) John's coming occurred at ten o'clock.
b) The event of John's coming occurred at ten
o'clock.
c) The event occurred at ten o'clock.

1.6a) John's coming is surprising.
b) The fact of John's coming is surprising.
c) The fact is surprising.

The noun action can only take subjectless gerundive nominal complements, while other nouns, such as event, require a gerundive nominal complement with a subject (Menzel, 1969t82-83). At the same time, of the nouns which take gerundive nominal complements, only action allows a proposed possessive noun.2 Thus, we find the pattern of acceptability shown in 1.7.

1.7a) the action off*John's eating the meat.
eating the meat
b) the event of fJohnls eating the meaty L*eating.the meat J
c) ons *event of eating the meat


Most of the nouns on Menzel's list also take derived nominal complements, and any derived nominal complement, unless it is a complement of a verb of belief, or of ay, declare, or claim, can be a complement of at least one of those head nouns, as is indicated by the examples in 1.8.

1.8a) The fact of John's departure cannot be contested.
b) The event of the destruction of Jerusalem
occurred in 67 AD.
c) The state of Mike's awareness is unpredictable.





8

The head noun action never takes a derived nominal comnplemnent, no matter where the possessive agent is placed, or even if it is deleted. The examples in 1.9 are all unacceptable, even though the parallel forms in 1.10 are perfectly acceptable.

1.9a) *The action of Humphrey's refusal of the offer
caused trouble.
b) *Humphrey's action of refusal of the offer caused
'trouble.
c) *The action of refusal of the offer caused
trouble.

1.10a) The fact of Humphrey's refusal of the offer
caused trouble.
b) Humphrey's action of refusing the offer caused
trouble.
c) The action of refusing- the offer caused trouble.

Mvenzel (1969251) notes that gerundive nominal complements can also be complements of act. He states that acts are a subclass of actions, with the restriction that "an act is an action which the speaker either (a) disapproves of strongly (in a legal or moral sense); or (b) admires greatly." He also notes that this distinction does not seem to be recognized by some speakers.

While derived nominal complements do not occur as complements of action, a restricted class does occur as c omplements of act. The head noun act does not take derived nominal complements with object prepositional phrases. The examples in 1.11 illustrate the differences between gerundive nominal complements and derived nominal complements in






9

regard to the head nouns action and act. The adjective cowardly is more appropriate for the complements of act.

1.11 (in the conPtext ...was unexpected/cowardly.)

a) John's denying the request
b) John's action of denying the request
c) John's act of denying the request
d) John's denial of the request
e) *John's action of denial of the request
f) *John's act of denial of the request
g) John's act of denial

Gerundive nominal complements and derived nominal complements are alike in that they both can be complements of a limited set of head nouns in surface structure, and seem to always be complements of such nouns in underlying structure. The two types of complement differ in that only gerundive nominal complements can be complements of action.
Some Assumotions

In this study I will be making certain assumptions

bearing on the structures and rules I will discuss. First of all, I will assume that grammatical rules apply in a fixed order and cyclically. The cyclic application of rules means that the full set of rules would first apply in order to the deepest embedded sentence, S2, in 1.12.
1.12 SO


S 1


S 2






10

After all the rules have had a chance to apply to

the rules would apply in the same order to the next higher sentence, S1, which is in turn embedded in So. After all the rules have had a chance to apply to SI, the rules would

apply again in the same order to the highest sentence, S03

I will also adopt here a predicate-initial analysis of the underlying structure of English sentences. McCawley (1970) presents arguments for such an analysis for structural representations comparable to those in Chomsky (1965). A predicate-initial analysis is also at least implicit in the case grammar of Fillmore (1968). The underlying structures in case grammar differ greatly from the underlying structures postulated in Chomsky (1965). In Fillmore's case grammar, the noun phrases in deeply underlying structures are identified by their semantic relationship to the predicate of the sentence, and not by any syntactic relationship. Without adooting case grammar in to+o, I will assume that, before any syntactic rules have applied, there is no ordered relationship between the constituents of the underlying propositions, and that at least some syntactic rules must recognize semantic relationships and transform such semantic rela4
tionshirs into word-order syntactic relationships. For convenience of representation, however, I will assume that underlying sentences have structures like that in 1.13

before any syntactic rule has applied.5

1.13 S

PRED NP NP







Assuming predicate-initial order in structures such as

1.13 is a way of expressing the pivotal role of the predicate in most syntactic rules. An alternative statement could be that noun phrases are assigned linear order in relation to the predicate by syntactic rules.

McCawley (1970) explicitly assumes that underlying sentences have the order Verb-Subject-Object (VSO) before the application of any transformations, but does not present any evidence for preferring that order over one of Verb-Object

-Subject (VOS). I believe that the evidence points to a VOS order, if there is any order at all to the noun phrases to .he right of the predicate in deeply underlying structures.6 In the rest of this work, the relative order of noun phrases in trees representing deeply underlying structures will be arbitrary, and represent no claims as to the actual order.

Outline
This study is concerned primarily with the internal

structure of nominal complements, by which I mean the relationships of the constituents of nominal complements to each other, as opposed to the relationships of nominal complements to items which are not constituents of the said nominal complements. I compare such internal structures to those of sentences and of other clausal complements. I point out that the differences between such structures are regular, and easily accounted for by independently motivated syntactic rules, given certain conditions on the cycle in which the rules apply.






12

I do not discuss the problem of the apparent irregularities in the semantic relationship between related predicates and derived nominals. I will adopt Newmeyer's (1971) claim that, at the worst, the transformationalist position is no less adquate on this point than the Lexical Hypothesis, since, if there are no regularities, the information on restrictions on meanings of derived nominals must be part of the lexical entry& of the underlying predicate with the transformationalist position, of the underlying predicate/nominal with the Lexical Hypothesis.

Chapter Two is a discussion of the surface structure of derived nominal complements, and arguments for including in the class of derived nominal complements certain structures which have not previously been so identified. Chapter Three discusses the rules which do not apply in the derivation of derived nominal complements, or which have different conditions on applicability in derived nominal complements, but which apply without restriction to sentences and gerundive nominal complements. Chapter Four discusses the ordering of rules which will account for the structure of derived nominal complements, and the question of which cycle the appropriate rules apply on. Chapter Five summarizes the thesis, and presents certain questions of theoretical import.














NOTES


A source of possible confusion is the fact that both the nouns formed by adding a suffix to a verb stem, and the complements in which such nouns occur, can be called nominals, To avoid confusion, I will reserve the term nominal for the nouns so formed, and refer to the complements they occur in as nominal complements. I will use nominalization to refer to the process by which nominals are created. I will have more to say about the choice of the term complement below.
2The possessive agent on action is always identical

with the deleted agent of the complement of action. Menzel argues that the agent cannot have been raised from the complement. Among suggestions of possible sources for the possessive agent, he states that "in a grammar based on the transformationalist position...the agent on the head noun action would be derived from an underlying sentence embedded on the noun action, or more probably on the verb act." Ross (1972a) argues that sentences with verbs of action (which are the only sentences that can be embedded as complements of the head noun action) are embedded on the verb do in underlying structures, with the agent of do identical to the agent of the embedded sentence. The substitu13






14

tion of action for do in nominal complements, and the deletion of the lower of two identical agents (cf. EQUI-NP-DELETION in Chapter Three), will account for the above facts within the transformationalist position.
3 There are extensive arguments in the literature for cyclical ordering, I am not concerned here with the arguments over extrinsic vs. intrinsic ordering, but will merely state that assuming ordering permits useful generalizations, as in Chapter Four and Chapter Five.
At this point I am rejecting the Aspects model of grammar. The Aspects model has syntax as the most basic component of the grammar, with both semantics and phonology acting as interpretive components of the output of the syntactic component. I believe that this model is unrealistic, even as a model of competence rather than of performance. Speech is a stream of sound perceived as a linear string of distinct units. Semantic propositions, on the other hand, are unordered in any dimensional sense. There is no dimensional order implied in the statement that someone is the agent t of such-and-such action. Syntax, then, is that part of grammar which relates unordered semantic relationships to linearly ordered phonological strings.
5 James 1cCawley, during a discussion at the 1972 LSA Annual .meeting, summed up the predicate-initial hypothesis in the statement that if there is any linear order so early in the grammar, it is a predicate-initial order.
6 1 will present my arguments for this analysis in Albury (forthcoming).













CHAPTER TWO
DEFINING DERIVED NOMINAL COViPLEMENTS

First Definitions

Chomsky (1970) does not explicitly define derived nominal complements, but a definition may be extracted from the various examples he cites in discussing the characteristics of derived nominals. It is obvious throughout that Chomsky intends the class of derived nominals to include only those nominals which are morphologically derived from a verb by
1
means other than the suffix -ing. He explicitly excludes gerunds (those nomrainals derived by adding -ing to a verb) from the set of derived nominals. Chomsky (19701214) notes certain similarities to derived nominal complements shown by structures called "action nominals" in Lees (1963) and Fraser (1970), which have nominals formed with -ing, but he claims a number of differences which would seem to preclude treating them as derived nominal complements. I will have more to say about these forms below. Chomsky thus limits the term derived nominal to those nominals formed by adding a derivational suffix to a verb, such as refusal (from refuse) and marriage (from marry); nominals identical (except for stress) to verbs, such as search (from search) and

6 ot (from export); and nominals phonologically modified from a verb, such as deed (from did).
15






16

I will temporarily accept Chomsky's implied morphological definition of derived nominals, and define derived nominal complements as those complements in which nominals formed without an -in suffix are found. Starting from this definition of derived nominal complements, I would like to build up a detailed description of the structure of these complements. Chomsky's method of comparing sentences, gerundive nominal complements and derived nominal complements to illustrate the characteristics of derived nominal complements is useful, and I will adopt it here.

The Surface Structure of Derived Nominal Complements

Chomsky (1970:187-88) cites the sentences in 2.1 as having the corresponding gerundive nominal complements in

2.2 and the corresponding derived nominal complements in 2.3. The gerundive nominal complements have gerundive nominals corresponding to the verbs in the sentences, while the derived nominal complements have derived nominals corresponding to the verbs in the sentences.

2.1a) John is eager to please.
b) John refused the offer.
c) John criticized the book.

2.2a) John's being eager to please
b) John's refusing the offer
c) John's criticizing the book
2.3a) John's eagerness to please
b) John's refusal of the offer
c) John's criticism of the book Determiners

One characteristic shared by gerundive nominal complements and derived nominal complements is the presence of






17
possessive nouns corresponding to the subject nouns in the sentences. Chornsky points out that the two types of nominal complement differ in that the possessive nouns in derived nominal complements can be replaced by other determiners, as is shown by the derived nominal complements in 2.4 (of. 2.3), while the possessive nouns in gerundive nominal complements cannot be so replaced, as is indicated by the fact that forms like those in 2.5 (cf. 2.2) do not occur. The derived nominal complements in 2.4 appear to be parallel to the derived nominal complements in 2.3, but the derived nomin al comnplements in 2.6 are closer in meaning to those in 2.3 than the ones in 2.4 are. 2

2.4a) ?the eagerness to please
b) the refusal of the offer
c) the criticism of the book

2.5a) *the being eager to please
b) *the refusing the offer
c) *the criticizing the book

2.6a) ? the eagerness to please by John
b) the refusal of the offer by John
c) the criticism of the book by John Aective-Adverb Correspondences

A second difference between gerundive nominal complements and derived nominal complements which Chomsky discusses is the fact that derived nominal complements can have an adjective preceding the nominal, while gerundive nominal comnplements cannot. These prenominal adjectives correspond to adverbs in sentences and gerundive nominal complements, so that there are sentences like those in 2.7, with the corresponding gerundive nominal complements in 2.8, and the corres-






18

ponding derived nominal complements in 2.9. The fact that strings like those in 2.10 are unacceptable shows that gerundive nominal complements do not take prenominal adjectives.
2.7a) John is overwhelmingly eager to please.
b) John abruptly refused the offer.
c) John criticized the book unmercifully.

2.8a) John's being overwhelmingly eager to please
b) John's abruptly refusing the offer
c) John's criticizing the book unmercifully

2.9a) John's overwhelming eagerness to please
b) John's abrupt refusal of the offer
c) John's unmerciful criticism of the book

2.10a) *John's being overwhelming eager to please
b) *John's abrupt refusing the offer
c) *John's unmerciful criticizing the book
The prenominal adjectives in some derived nominal complements, such as those in 2.11, do not have corresponding adverbs in gerundive nominal complements, as in 2.12, or sentences, as in 2.13.

2.11a) John's troublesome eagerness to please
b) John's untimely refusal of the offer
c) John's unmotivated criticism of the book

2.12a) *John's being troublesomely eager to please
b) *John's untimely refusing the offer
c) *John's criticizing the book unmotivatedly

2.13a) *John is troublesomely eager to please.
b) *John untimely refused the offer.
c) *John criticized the book unmotivatedly.

That the acceptability of the forms in 2.11, and the unacceptability of the forms in 2.12 and 2.13 is due to these adverbs alone can be seen by inspecting the forms in 2.1, 2.2, 2.7 and 2.8. The adjectives which appear in prenominal position in the derived nominal complements in 2.11 can also be predicated of such complements, as in 2.14.






19
2.14a) John's eagerness to please is troublesome.
b) John's refusal of the offer was untimely.
c) John's criticism of the book was unmotivated.
The prenominal adjectives in the derived nominal complements in 2.9 can also be predicated of those complements, as in 2.15.

2.15a) John's eagerness to please is overwhelming.
b) John's refusal of the offer was abrupt.
c) John's criticism of the book was unmerciful.
The parallel I have been drawing between the adverbs in sentences and gerundive nominal complements and the adjectives in derived nominal complements suggests that the adverbs and adjectives are derivationally related. The fact that the adverbs are morphologically derived from the adjectives by adding the suffix -ly reinforces that hypothesis.

The three-way correspondence between prenominal adjectives in derived nominal complements, adjectives predicated of derived nominal complements, and adverbs in sentences and gerundive nominal complements suggests that the adjectives and adverbs have as a common source a higher predicate. If the higher predicate is the highest matrix predicate (ignor,ng abstract higher predicates) then it may be expressed as an adverb of the embedded sentence (raised to surface sentencehood) or as a predicate on a derived nominal complement. If the higher predicate is in turn embedded under a predicate which will appear in surface structure, then it may be expressed as an adverb with a gerundive nominal complement, or as a prenominal adjective with a derived nominal complement. The higher predicate-prenominal adjective relation-






20

ship is also seen in simple noun phrases, so that no new rule need be postulated to derive prenominal adjectives in derived nominal complements from higher predicates.

Chomsky (1970,195) points out that a claim that the prenominal adjectives in derived nominal complements are derivationally related to adverbs leads to the prediction that sentences like those in 2.16 will have corresponding derived nominal complements like the structures in 2.17, which are not acceptable.

2.16a) John refused the offer in a surprising manner.
b) John is sincere to a limited extent.

2.17a) *John's refusal of' the offer in a surprising manner
b) *John's sincerity to a limited extent

However, there are structures like those in 2.18 which include derived nominal complements, and which seem to correspond to the sentences in 2.16'.

2.18a) the surprising manner of John's refusal of
the offer
b) the limited extent of John's sincerity

Adverbs which are morphologically derived from adjectives seem to be included within the scope of structures which have corresponding derived nominal complements, while adverbs which are prepositional phrases seem not to be included within the scope of such structures. It is not necessary to claim that the prenominal adjectives are derived from adverbs to support the claim that the prenominal adjectives and adverbs are derived from the same underlying structures. The restrictions on the inclusion of' adverbial prepositional phrases within the scope of derived nominal complements does not invalidate such a claim.






21

Object Pre positions

Another characteristic which distinguishes derived nominal complements from sentences and' 'gerundive nominal comnplements is the presence of a preposition preceding the noun phrases in derived nominal complements which correspond to the objects of verbs in sentences, if such object noun phrases do not already have a preposition. Among the sentences in 2.19, corresponding gerundive nominal complements in 2.20 and corresponding derived nominal complements in

2.21, only the derived nominal complements have the preposition of preceding the object.3

2.19a) John refused the offer.
b) John criticized the book.
c) John robbed the bank.

2.20a) John's refusing the offer
b) John's criticizing- the book
c) John's robbing the bank

2.21a) John's refusal of the offer
b) John's criticism of the book
c) John's robbery of the bank

When object noun phrases are preceded by prepositions in sentences such as those in 2.22 and in the corresponding Rerundive nominal complements in 2.23, then the same prepositions also appear in the corresponding derived nominal complements, as in 2.24.

2.22a) John was amused at the children's antics.
b) John delighted in teasing Alice.
c) John was doubtful about Dick's honesty.

2.23a) John's being amused at the children s antics
b) John's delighting in teasing Alice
c) John's being doubtful about Dick's honesty

2.24a) John's amusement at the children's antics
b) John's delight in teasing Alice
c) John's doubts about Dick's honesty





22

Pluralization

Another characteristic of derived nominals discussed by Chomsky (1970,189) is that they may be pluralized, while gerundive nominal complements may not. Thus, we find the derived nominal complements in 2.25, which correspond to the sentences in 2.26, while the corresponding gerundive nominal complements in 2.27 are not pluralized.

2.25a) John's three proofs of the theorem
b) John's repeated attempts to scale the wall
c) Agnew's many attacks on the press

2.26a) John proved the theorem three (times, ways, etc.).
b) John repeatedly attempted to scale the wall.
c) Agnew attacked the press many times.

2.27a) John's proving the theorem three (times, ways, etc.)
b) John's repeatedly attempting to scale the wall
c) Agnew's attacking the press many times

The sentences in 2.26 and the gerundive nominal complements in 2.27 express repetitive events. The derived nominal complements in 2.25 also express repetitive events, but with plural nominals rather than adverbs of repetition. Repetition is not always expressed explicitly in sentences such as those in 2.28, which therefore have two sets of corresponding derived nominal complements, those in 2.29, which express explicitly the singularity of the events and those in 2.30, which express explicitly the repetition of the
4
event.

2.28a) John has proved the theorem.
b) John has attempted to scale the wall.
c) Agnew has attacked the press.

2.29a) John's proof of the theorem
b) John's attempt to scale the wall
c) Agnew's attack on the press






23

2.30a) John's proofs of' the theorem
b) John's attempts to scale the wall
c) Agnew's attacks on the press

Not all derived nominals can be pluralized. The sentences in 2.31 have the corresponding derived nominal complements with singular derived norainals in 2.32, but derived nominal complements with plural derived nominals like those in 2.33 corresponding to the sentences in 2.31 do not occur.

2.31a) The enemy has destroyed the city on three
occassions.
b) The crowd laughed repeatedly.

2.32a) the enemy's destruction of' the city on three
occassions
b) the crowd's repeated laughter

2.33a) *the enemy's destructions of' the city on three
occassions
b) *the crowd's repeated laughters Auxiliaries

Another characteristic of the structure of' derived nominal complements discussed by Chomsky (1970:189) is the absence of' any auxiliary verbs. Gerundive nominal compQlements, on the other hand, can have any auxiliary (with the exception exempliied by 2.35c) except modals. Perfective aspect can appear in gerundive nominal complements, so that sentences like those in 2-34a and b have the corresponding

Eerunclive nominal complements in 2.35a and b. Progressive aspect can also appear in gerundive nominal complements, but only in conjunction with perfective aspect, as in 2,35b, so that the sentence in 2.34c does not have a corresponding gerundive nominal complement with ben corresponding to a form of' be which is acceptable (cf. 2.35c and d).-





24

2.34a) John has criticized the book.
b) John had been criticizing the book.
c) John is criticizing the book.

2.35a) John's having criticized the book
b) John's having been criticizing the book
c) *John's being criticizing the book
d) John's criticizing the book
Forms of the verb be which appear as copulas in sentences with predicate adjectives or predicate nouns, or in combination with past/passive participles in passive sentences, such as those in 2.36, also appear in gerundive nominal complements such as those in 2.37. But the derived nominal complements in 2.38 which correspond to the sentences in 2.36 do not have any form of be.

2.36a) John is strong.
b) Alice is beautiful.
c) John is the chairman.
d) The city was destroyed by the enemy.
e) Abby was acquitted by the jury.
2.37a) John's being strong
b) Alice's being beautiful
c) John's being the chairman
d) the city's being destroyed by the enemy
e) Abby's being acquitted by the jury

2.38a) John's strength
b) Alice's beauty 6
c) John's chairmanship
d) the city's destruction by the enemy
e) Abby's acquittal by the jury A Redefinition

I would now like to abandon my first provisional definition of derived nominal complements as those complements with nominals formed without an -1ng suffix, and instead define derived nominal complements as those complements which show one or more of the following features: a wide






25

variety of' determiners; prenominal adjectives instead of' adverbs; pluralization of' the nominal; prepositions preceding object noun phrases when they do not appear in the corresponding sentences; and complete absence of' auxiliaries, including copulas and the passive be. The form of' the nominal in derived nominal complements does not enter into this definition.

Action Nominals

Chomsky's Analysis

There are nominal complements which satisfy the new

definition of' derived nominal complements given above, but which have what appear to be gerundive nominals corresponding to the verb, such as the nominal complements in 2.390 Nominal complements of' this type are called "action nominals" in Lees (1963) and Fraser (1970).

2.39a) John's refusing of the offer
b) john's proving of the theorem
c) the growing of' tomatoes

Chomsky (l970s214.-15) believes that these complements

belong to a third class of' nominal complements distinct from both 'gerundive nominal complements and derived nominal complements. He does not identify them as "action noniinals." Chomsky claims that these nominal complements appear to have the same internal structure of' noun phrases that derived nominal complements have, as evidenced by the possibility of' a determiner- other than a possessive noun appearing (cf'. 2.39c), but that prenominal adjectives seem quite unnatural in such complements. He says that the complement is limited in





26

productivity as well, so that we cannot get structures like those in 2.40. Chomsky (1970:215) finally states that "there is an artificiality to the whole construction that makes it quite resistant to systematic investigation."

2.40a) *the feeling sad
b) *the trying to win
c) *the arguing about money
d) *the leaving

I would like to put aside for now the problems involved in the forms in 2.39a and b and 2.40, and consider in detail the complement in 2.39c, the growing of tomatoes. Part of the data cited by Chomsky (1970:192) to support his claim that the Lexical Hypothesis provides the best explanation for the origin of derived nominals involves transitive verbs derived from intransitive verbs by the rule of CAUSATIVE FORA'ATION (cf. G. Lakoff, 1970), such as grow, as in John grovs tomatoes, derived from Zro as in tomatoes grow. At some point in its derivation, the underlying structure of john grows tomatoes can be represented by the tree in 2.41. The structure underlying tomatoes grow is embedded as S1 in the tree in 2.41,

2,41) 0

V NP NP
CAUSE I


V NP

grow tomatoes John

Chomsky points out that there is a derived nominal complement, the growth of tomatoes, which corresponds to the






27
sentence tomatoes grow, but not to the sentence John grows tomatoes, as would be expected from the apparently parallel example of the derived nominal complement, the rejection of the offer, which corresponds to the verb phrase reject the offer. In other words, the rejection of the offer is an object nominal complement, while the growth of tomatoes is a subject nominal complement, and Chomsky takes the fact that there is no interpretation of the growth of tomatoes as an object nominal complement as proof that there is no derived nominal complement corresponding to John grows tomatoes. In terms of the Lexical Hypothesis, this is to be expected because John grows tomatoes involves a derivation from an underlying structure which includes tomatoes grow as an embedded sentence, as in the tree in 2.41. This is consistent with the claim that derived nominals are associated lexically with the underlying verb. Since grow occurs in the lexicon only in the intransitive, noncausative sense, the transitive sense being derived transformationally by the rule of CAUSATIVE FORMATION, only the intransitive sense can have a corresponding derived nominal, according to the Lexical Hypothesis.

Some Counterexamoles to Chomsky's Analysis

Smith (1972) has pointed out that there are many exceptions to Chomsky's claims concerning the occurrence of derived nominals corresponding to verbs derived by the rule of CAUSATIVE FORMATION. The verb convert occurs as both transitive and intransitive, and both forms have associated






28

derived norninals. Thus, Robert's conversion to hedonism, which at some point in its derivation has an underlying structure like that of the embedded sentence S. in the tree in 2.4I2, and John's conversion of Robert to hedonism, which at some -ooint in its derivation has an underlying structure like that of the whole tree in 2.42, are both acceptable derived nominal complements. The derived nominal complement the conversion of Robert, like many other nominal complements such as the shooting of the hunters, is therefore interpretable as either a subject nominal complement or as an object nominal complement.

2.42 0



CAUSE

VNP NP

convert to Robert hedonism John

So-me verbs listed by Smith which also form derived nominals in both transitive and intransitive forms are exlode, di-vide, accelerate, e2Mand, repeat, neutralize,. conclude and unify. She then points out that all the listed counterexamples share a morphological property their derived nominals are formed with suffixes of Latin origin. She then goes on to claim that there are almost as many verbs of the typoe convert as there are of the type gr and then states what she sees as the conditioning factor in distinguishing the two types; "Whether or not a verb has a transitive






29

nominal depends on how the nominal is formed. If a causative verb takes a nominalizing suffix of Latin origin (-tion, -al, -merit), then it has both transitive and intransitive derived nominals. If a causative verb does not take such a nominalizing suffix, then it occurs only intransitively in derived nominals." She then observes that, in general, verbs of Latin origin form derived nominals with the suffixes of Latin origin, while verbs of what she calls Anglo-Saxon origin form derived nominals with the suffixes

-th, -ness, or Z (null), which can be considered to be native, as opposed to the suffixes borrowed from Latin.

Smith has, based on the data above, reached the conclusion that "the grammar must distinguish at least two classes of 'causative' verbs: those that do and do not have transitive derived nominals." As we have just seen above, that distinction is to be based on something like a native/borrowed-Latin dichotomy. It seems very strange, however, to state that a part of the lexicon associated with borrowed (Latinate) derivational suffixes shows more productivity than another part of the lexicon associated with native derivational suffixes. This raises the possibility that the causative verbs associated with native derivational suffixes also have corresponding derived nominals, but that for some reason SmIt and Chomsky have failed to recognize them. Using the definition of derived nominal complements I have developed above, I will now look for derived nominal complements corresponding to sentences with such causative verbs.





30

Nominals of Causatives

Consider verbs such as ro, raise and move. These

verbs occur in both intransitive and transitive forms, the latter being derived by the rule of CAUSATIVE FORML-ATICN. The verbs in the intransitive form are used in sentences such as those in 2.43, which have the corresponding derived
nominal complements in 2.44 and 2.45.7

2.43a) The tree grew slowly.
b) The temperature rose rapidly.
c) The table moved mysteriously.

2.44a) the tree's slow growth
b) the temperature's rapid rise
c) the table's mysterious movement
2.45a) the slow growth of the tree
b) the rapid rise of the temperature
c) the mysterious movement of the table

The verbs in the transitive form are used in sentences such as those in 2.46, which have the corresponding gerundive nominal complements in 2.47, and the corresponding derived nominal complements in 2.48 and 2.49.

2. 46a) John grows tomatoes skillfully.
b) Tom raised the temperature deliberately.
c) Mike moved the table accidentally.

2.4?a) J ohn's growing tomatoes skillfully
b) Tom's raising the temperature deliberately
c) Mike's moving the table accidentally

2 .48a) John's skillful growing of tomatoes
b) Tom's deliberate raising of the temperature
c) Mike's accidental moving of the table

2.49a) the skillful growing of tomatoes by John
b) the deliberate raising of the temperature by Tom
c) the accidental moving of the table by Miike

The derived nominal complements in 2.48 and 2.49 are

completely consistent with the definition of derived nominal






31

complements which I have adopted. I will therefore assume for now that -g is one of the suffixes forming nominals in derived nominal complements, which happens to be homophonous with the gerundive suffix -_hg. Nominals of Deadjectival Verbs

The rule of CAUSATIVE FORMATION is not the only rule by which a verb may be derived from an underlying form with different properties. Smith (1972) mentions that we find the derived nominal complement the light's dimness, but not
*John's dimness of the light. Dim is homophonous for three senses: adjective, intransitive verb derived by the rule of INCHOATIVE FORMnATION (cf. G. Lakoff, 1970), and transitive verb derived by the rule of CAUSATIVE FORMATION. Other adjective-verb sets related by these rules are low:lower: lower and wide:widenswiden. The adjectives are used in sentences like those in 2.50, which have the corresponding gerundive nominal complements in 2.51, and the corresponding derived nominal complements in 2.52 and 2.53.

2.50a) The light was dim.
b) The bridge was low.
c) The road was wide.

2.51a) the light's being dim
b) the bridge's being low
c) the road's being wide

2.52a) the light's dimness
b) the bridge's lowness
c) the road's width

2.53a) the dimness of the light
b) the lowness of the bridge
c) the width of the road






32

The intrans!ive verbs are used in sentences like those in 2.54, which have the corresponding gerundive nominal comDlements in 2.55, and the corresponding derived nominal complements in 2.56 and 2.57.

2.54a) The light slowly dimmed.
b) The bridge gradually lowered.
c) The road suddenly widened.

2.55a) the light's slowly dirmming
b) the bridge's gradually lowering
c) the road's suddenly widening

2.56a) the light's slow dimming
b) the bridge's gradual lowering
c) the road's sudden widening

2.57a) the slow dinmming of the light
b) the gradual lowering of the bridge
c) the sudden widening of the road

The transitive verbs are used in sentences like those

in 2.58, which have the correspondng gerundive nominal com-,ements- in 2.59, and the corresponding derived nominal complements in 2.60 and 2.61.

2.58a) John suddenly dimmed the light.
b) The tender gradually lowered the bridge.
c) the city widened the road recently,

2.59a) John's suddenly dimming the light
b) the tender's gradually lowering the bridge
c) the city's widening the road recently

2 60a) John's sudden dimming of the light
b) the tender's gradual lowering of the bridge
c) the city's recent widening of the road

2. 61a) the sudden dimming of the light by John
b) the gradual lowering of the bridge by the tender
c) the recent widening of the road by the city
All of the derived nominal complements in 2.56, 2.57,

2.60 and 2.61 have nominals formed with the -ing suffix. Only the nominals in 2.52 and 2.53, which correspond to






33
adjectives, are formed without the -ing suffix. Both this set of adjectives and verbs and the previous set of verbs show the sam e pattern. The form of the verb which is presumably entered in the lexicon (the intransitive form of gro rise and move and the adjective form of di, low and Lj) has a corresponding nominal in derived nominal complements formed by means other than the -ing suffix, while all the derived forms of the verbs have corresponding nominals in derived nominal complements formed with the -ing suffix. This suggests that the form of the associated nominals is specified in the lexicon, but that this specification is not retained when other forms of the verbs or adjectives are derived from the form entered in the lexicon. With verbs like convert, on the other hand, the specification of the form of the associated nominal is retained when another form of the verb is derived from the lexical entry. Other Nominals in -inog

Verbs derived from other underlying verbs are not the

only verbs to correspond to norninals in derived nominal complements formed with the -ing suffix. Sink is another verb which is used both intransitively and transitively. The intransitive use is illustrated in 2.62a, the transitive in

2.62b. These sentences have the corresponding gerunidive no-minal complements in 2.63, and the corresponding derived nominal complements in 2.64 and 2.65.

2.62a) The Bismark sank.
b) The British Navy sank the Bismark.








2.63a) the Bismark's sinking
b) the British Navy's sinking the Bismark

2.64a) the Bismark's sinking
b) the British Navy's sinking of the Bismark

2.65a) the sinking of the Bismark
b) the sinking of the Bismark by the British Navy

Presumably, there is no specification of the form of

the nominal in a derived nominal complement corresponding to sink in the lexical entry, so that the form with -ing is supplied by the grammar for sink intransitive and transitive just as it is supplied for the nominals in derived nominal complements corresponding to sentences with the derived verbs gro raise, move, dim, lower and widen. One consequence of this is that it is possible for the gerundive nominal complement and derived nominal complement corresponding to a particular sentence to be identical in surface structure, as are the forms in 2.63a and 2.64a. It will therefore not always be possible to decide whether a nominal complement is actually a gerundive nominal complement or a derived nominal complement, unless the context supplies such information.
One objection which might be made to the identification of any nominal complements with nominals formed with -ing as derived nominal complements is that they do not readily form plurals. However, it is not true that they never form plurals, as shown by the examples in 2.66.

2.66a) the wanderings of an old mind
b) the leavings of a great feast
c) the makings of a great scholar
d) the cravings of a glutton
e) the comings and goings of the workers





35
As was noted above (cf. 2.33), not all derived nominals form plurals, so the fact that many apparent derived nominals in -in do not is not strong counter-evidence. Paired Nominals

The nominal complements in 2.39a and b (repeated here in 2.67), cited by Chomsky (1970:214), appear to be derived nominal complements with nominals formed with -irZ which are parallel to the derived nominal complements in 2.68, which have nominals formed other than with the suffix -ing.

2.67a) John's refusing of the offer
b) John's proving of the theorem

2.68a) John's refusal of the offer
b) John's proof of the theorem

The derived nominal complements with nominals formed

other than with the -n suffix in 2.68 readily take prenominal adjectives, as was discussed above, so that we have derived nominal complements like those in 2.69t

2.69a) John's abrupt refusal of the offer
b) John's brilliant proof of the theorem
The nominal complements with nominals formed with the

-inz suffix in 2.67 seem to less readily accept prenominal adjectives, as is indicated by the strangeness of the nominal complements in 2.70. That this strangeness is not due to the nominal having an -_g suffix alone is shown by the normal occurrence of prenominal adjectives with derived nominal complements with nominals formed with -ig in 2.48,

2.49, 2.56, 2.57, 2.60 and 2.61 above.

2.7Oa) ?John's abrupt refusing of the offer
b) ?John's brilliant proving of the theorem






36

Chomnsky (1970:215) notes that derived nominal complements with nominals formed with the -incr suffix seem rather clumsy when a derived nominal complement with a nominal formed other than with the -i suffix also exists. It is also these nominal complements which do not readily take a prenominal adjective. Since I have claimed above that the suffix -ing in general appears in derived nominal complements when no other means of forming the nominal is specified, it seems possible that pairs of nominals like refusal: refusing and proof:proving represent a misapplication of the rule supplying the -1;ng suffix, and the full regularity of derived nominal complements with forms like refusing and proving is blocked by the existance of forms like refusal and ro of.

Some Apparent Counterexamples

Chomsky (1970:214) also noted that derives nominal complements with -L seem to be limited in production since we do not get forms like those in 2.40, repeated here as 2.71.

2.7ia) *the feeling sad
b) *the trying to win
c) *the arguing about money
d) *the leaving
The forms in 2.71 would presumably correspond to the verb phrases in 2.72.

2.72a) feel sad
b) try to win
c) argue about money
d) leave
Chomsky (1970:186) argues that sentences like 2.73a

have a structure of Noun Phrase-Verb-Predicate parallel to






37

the Noun Phrase-Verb-Noun Phrase structure of a sentence like 2.73b.

2.73a) John felt sad.
b) John felt sadness.

There is a derived nominal complement which corresponds to 2.73b, John's feeling of sadness. Peter Menzel (personal communication) has pointed out to me that feeling is the nominal of a derived nominal complement, and thus would be followed by a preposition, presumably of, so that the proper question is why we do not get *the feeling of sad. The presence of the preposition means that any complement of feeln within a derived nominal complement must be nominalized. Hence, derived nominalization can apply to an embedded sentence like that in 2.73b, but not to one like that in 2.73a, unless sad is nominalized to sadness as part of the nominalization of feel.

The nominal complement in 2,7!c, *the arguing about money would not be expected to occur because the regular derived nominal complement corresponding to argue about mone is the argument about money.

Chomsky (1970:214) gives as unacceptable the example :in

2.7!d, the leaving. At least some speakers do accept nominal complements like John's hurried leaving (cf. John left hurriedly), and I am told that such forms have appeared in print. My ovm judgment agrees with that of Chomsky, however. The dialects that accept forms like John's hurried leaving present no problem to my analysis, so I will turn to those dialects that do not accept such forms.








In my dialect, leave (=depart), try (attempt) (cf. 2.71b), be (exist, occur, be present), have (=possess) and live (=reside) (and other verbs) are alike in that they have no corresponding nominal forms which appear in derived nominal complememts. (I am refering here to be and-have when they are main surface verbs, not auxiliaries.) The existence of verbs such as these poses a serious challenge to the transformationalist position. It would appear that the rule (or rules) producing derived nominal complements is blocked from applying to underlying structures with certain verbs as their predicates. It is Chomsky's claim that this is one of a number of facts supporting the Lexical Hypothesis over the transformationalist position.

The verbs mentioned above (which are all of Germanic origin) occur in sentences like those in 2.74.

2.74a) John tried vainly to win the race.'
b) John left hurriedly on the bus.
c) There is a God.
d) There was a riot yesterday.
e) There is some wine in the bottle.
f) John has a car.
g) Bill usually lives in a hotel.

The Latinate synonyms of those verbs (for meanings

inherent in the sentences in 2.74) occur in sentences like those in 2.75.

2.75a) John attempted vainly to win the race.
b) John departed hurriedly on the bus.
c) There exists a God.
d) There occurred a riot yesterday.
e) There is some wine present in the bottle.
f) John possesses a car.
g) Bill usually resides in a hotel.






39
I find the sentences in 2.75 to be formal and even awkward in comparison to those in 2.74, a difference presumably

attributable to the choice of predicate.

All of these sentences have acceptable corresponding

gerundive nominal complements, those in 2.76 corresponding

to the sentences in 2.74, and those in 2.77 corresponding to

the sentences in 2.75.

2.76a) John's trying vainly to win the race
b) John's leaving hurriedly on the bus
c) there being a god
d) there being a riot yesterday
e) there being some wine in the bottle
f) John's having a car
g) Bill's usually living in a hotel

2.77a) John's vainly attempting to win--the race
b) John's departing hurriedly on the bus
c) there existing a god
d) there occurring a riot yesterday
e) there being some wine present in the bottle
f) John's possessing a car
g) Bill's usually residing in a hotel

Again, the gerundive nominal complements in 2.76 based

on Germanic verbs seem less formal that the gerundive nominal complements based on Latinate verbs in 2.77. 9

Finally, there are no derived nominal complements corresponding to the sentences in 2.74 and gerundive nominal

complements in 2.76, the forms in 2.78 being unacceptable.

2.78a) *John's vain trying to win the race
b) *John's hurried leaving on the bus
c) *the being of a god
d) *the being of a riot yesterday
e) *the being of some o.,ine in the bottle
f) *John's having of a car
g) *Bill's usual living in a hotel






40

On the other hand, the derived nominal complements in 2.79, which correspond to the sentences in 2.75 and gerundive nominal complements in 2.77, are perfectly acceptable.

2.79a) John's vain attempt to win the race
b) John's hurried departure on the bus
c) the existence of a god
d) the occurrence of a riot yesterday,
e) the presence of some wine in the bottle
f) John's possession of a car
g) Bill's usual residence in a hotel

It seems, therefore, that it is possible for a Germanic (or native) verb to have no corresponding nominal in a derived nominal complement when there is a closely synonymous Latinate verb with such a corresponding nominal. The transformationalist position may be maintained in the face of these examples in at least two ways: verbs like try, leave, be, have and live can be marked in the lexicon to block DE~RIVED NOMINAILIZATION, which ignores the existence of the Latinate synonyms; or, some sort of lexica-t alternation may be -posited, with the Latinate option being mandatory in derived nominal complements. I prefer the second position, and I will discuss some different evidence for such lexical alternation for the synonyms of be in Chapter Three. Action Nominals and the Head Noun Action

in %Chapter One I noted that derived nominal complements did not occur as complements of the head noun action. If action nominal complements are a subclass of derived nominal complements, then they should not occur as complements of action either. The examples in 2.80 show that this is indeed the case.





41

2.80a) *John's action of opening of the door
b) *John's action of sinking of the boat
c) *John's action of dimming of the light
d) *John's action of refusing of the offer
e) *John's action of proving of the theorem

Menzel (1970) argues that the head nouns associated

with nominal complements in specific structures define what the nominal complement is, i.e., a complement of event is an event, a complement of action is an action, etc. It is ironic that those structures called "action nominals" cannot be complements of the head noun action, and thus are not actions.

Summary

I have shown in this section that the so-called "action Inominals" have the surface structure of derived nominal complements. That is, the nominal complements sometimes called "action nominals" share with other derived nominal complements the features of a possible variety of determiners, prenominal adjectives corresponding to adverbs in sentences, prepositions with all objects, the possibility of pluralization (although less for action nominals than is so for other derived nominal complements), the complete absence of auxiliaries, and the restriction from being the complement of the head noun action. The fact that action nominal complements have nominals formed with the suffix -ing, as do gerundive nominal complements, ought to have no bearing on this classification.






~4.

Az-entive Nominais

Agentive nominal complements share the characteristics of derived nominal complements. The examples in 2.81, corresponding to the sentences in 2 82, show that such complements may have determiners other than possessive nouns, prenominal adjectives, object prepositions and pluralization. Agentive nominal complements do not have any reflexes of' auxiliaries.

2.81a) the short-sighted designers of this building
b) the unlucky holders of Imperial Russian bonds
c) the greedy despoilers of the Earth

2.82a) The ones who designed this building were
short -sighted.
b) The ones who held Imperial Russian bonds
were unlucky.
c) The ones who despoil the Earth are greedy.

*Some characteristics of agentive nominal complements

deserve further comment. Agentive nominal complements never have possessive nouns as determiners which correspond to underlying agents. When agentive nominal complements do have possessive nouns as determiners, as in 2.83, the possessive nouns correspond to the underlying objects of the verbs in sentences which correspond to the agentive nominal complements, as in 2.84.

2.83a) Mike's helper
b) General Electric's workers
c) America's educators

2.84a) Someone helps Mike.
b) Some people work for General Electric.
c) Some people educate America.






43

It is usually possible to get an agentless passive sentence closely corresponding to such agentive nominal complements, as in 2.85a and c, although 2.85b seems odd.

2.85a) Mike is helped.
b)??General Electric is worked for.
c) America is educated. (Not the stative reading.)

However, the sentences in 2.84 (and 2.85) are not the only ones corresponding to the agentive nominal complements in 2.83. For instance, on a different reading, 2.83c, America's educators, corresponds to the educators of America.

The absence of possessive nouns corresponding to underlying agents in agentive nominal complements is due to the fact that sentences corresponding to agentive nominal complements always have nonspecified agents. Other derived nominal complements may also correspond to sentences with nonspecific agents, as in the refusal of the offer, which, like the agentless passive sentence, the offer was refused, corresponds to an underlying structure of the form SOMEONE refused the offer.

The prenominal adjectives in agentive nominal complements are not always related to adverbs in corresponding sentences. Vendler (1968) points out that beautiful in the beautiful dancer may correspond in meaning to beautiful in the dancer is beautiful, or to beautiful in SOMEONE dances

beautifully.

The nominals in agentive nominal complements also seem to more readily form compounds with their objects than do the nominals of other derived nominal complements. Thus, we






44

find many compounds like those in 2.86 corresponding to the agentive nominal complements in 2.87.

2.86a) lion tamer
b) bookkeeper
c) shock absorber

2.87a) tamer of lions
b) keeper of books
c) absorber of shocks

But some nominals in other derived nominal complements also form such compounds, as in 2.88, corresponding to the derived nominal complements in 2.89.

2.88a) token payment
b) tax assessment c) art collection

2.89a) payment of a token
b) assessment of a tax
c) collection of art

Agentive nominal complements have the same surface

structure as derived nominal complements. I see no reason to not include agentive nominal complements in the class of derived nominal complements.
Summary :
In this chapter I have defined derived nominal complements in terms of surface structure as those nominal complements which allow a variety of determiners, allow prenominal adjectives which correspond to adverbs in sentences, have prepositions with all objects, and allow the nominal to be pluralized, but which have no auxiliaries. Using this definition, I have then argued that the structures known as action nominals and agentive nominals are really subclasses of the class of derived'nominal complements.













NOTES


1 Chomsky presumably also excludes agentive nominals formed with -er, although he never mentions them except to argue against the positing of abstract verbs underlying nominals which otherwise have no corresponding predicates. Agentive nominal complements share several characteristics with derived nominal complements, and I will argue below that agentive nominal complements are a special class of derived nominal complements.
2 The derived nominal complements in 2.4a and 2.6a seem quite strange to me, but 2.4a, at least, is acceptable in a context such as the eagerness to lease shown by John, which, however, seems to correspond to the sentence, an eagerness to ]lease was shown by John, and not to the sentence in 2.1a.
3 The preposition of is the unmarked form. Some verbs have corresponding derived nominal complements with other prepositions, e.g., attack:attack on.
Simple past tense seems to imply a single occurrence of an event unless otherwise specified. Present perfect tense seems to imply only at least one occurrence of an event.
5 The constraint on the occurrence of adjacent forms with the -in__g suffix, which blocks constructions like that in 2.35c, is discussed in Ross (1972b) and Milsark (1972).
45





46
6 The formation of derived nominal complements corresponding to sentences with predicate nouns does not seem to be completely free. Thus, while John's manhood is possible, it corresponds to a limited meaning of John is a man, and Jane's womanhood seems very strange. Nevertheless, many predicate nouns do have corresponding derived nominals, as can be seen in the sentences in i with the corresponding derived nominal complements in ii.

ia) Joan enjoys being a mother.
b) Ralph was governor recently?
c) Billy is a minor.
d) Stephen was a martyr.
e) John is a member in good standing in the lodge.
f) Sam is legally a pauper.

iia) Joan enjoys motherhood.
b) Ralph's recent governorship
c) Billy's minority
d) Stephen's martyrdom
e) John's membership in good standing in the lodge
f) Sam's legal pauperdom

Chomsky (1970:198-99) argues that the inclusion of verbs and adjectives in a category of predicator, as in G. Lakoff (1970:ll5ff.), is wrong, since nouns share the same distributional properties. The examples above of derived nominals corresponding to predicate nouns supports Chomsky's conclusion. Chomsky further claims, however, that such distributional properties are properties of lexical categories. It appears to be more correct to say that properties such as the stative-active distinction and the possession of a corresponding derived nominal, are properties of predicates. Such properties are exhibited by verbs (which are always predicates), adjectives (which presumably are always predicates in underlying structures) and predicate nouns.






4+7
7The corresponding gerundive nominal complements, as in i, which I find awkward, but still acceptable, are com-. pletely unacceptable to at least some speakers. I have no explanation for this fact.

ia) the tree's growing slowly
b) the temperature's rising rapidly
c) the table's moving mysteriously

More generally, possessive inanimate nouns are often awkward. Arnold Zwicky (personnel communication) pointed out that historically inanimates have -not always been able to form possessives in English. The two facts may be related.
8 Although move is borrowed ultimately from Latin and the derived nominal corresponding to the intransitive form of the verb is formed with -mnt move does not appear to belong to the class of Latinate verbs described by Smith (1972).

I will discuss the fact that there does not take the possessive suffix in gerundive -nominal complements in more detail in Chapter Three.













CHAPTER THREE
THE PRODUCTIVITY OF DERIVED NOMINAL COMPLEMENTS

The Evidence for Lack of Productivity

One of the three reasons Chomsky (1970:187-88) gives for rejecting a transformational account of the origin of derived nominal complements in favor of the Lexical Hypothesis is that gerundive nominal complements "can be formed fairly freely from propositions of subject-predicate form," while "productivity is much more restricted" for derived nominal complements. In support of this statement, Chomsky (1970:188-89) refers to the example of sentences like those in 3.1, which have the corresponding gerundive nominal complements in 3.2, and the corresponding derived nominal complements in 3.3, while the sentences in 3.4 have the corresponding gerundive nominal complements in 3.5, but the expected corresponding derived nominal complements in 3.6 are not

acceptable.

3.1a) John is eager to please.
b) John refused the offer.
c) John criticized the book.

3.2a) John's being eager to please
b) John's refusing the offer
c) John's criticizing the book

3.3a) John's eagerness to please
b) John's refusal of the offer
c) John's criticism of the book
48





49

3.4a) John is easy to pJ.easeb) John is certain to w% in the prize.
c) John amused the children with his stories.

3.5a) John's being easy to please
b) John's being certain to win the prize
c) John's amusing the children with his stories

3.6a) *John's easiness to please
b) *John's certainty to win the prize
c) *John's amusement of the children with his stories

Chomsky (1970:189) points out that there are acceptable derived nominal complements like those in 3.7 which superficially resemble the unacceptable strings in 3.6, and which correspond to the sentences in 3.8 and gerundive nominal compolements in 3.9, and comments, "these discrepancies between gerundive and derived nominal [complements] call -for explanation. Specifical-ly, we must determine why the examples of [3.6] are ruled out although those of [3.7] are permitted."

3.7a) John's eagerness to please
b) John's certainty that Bill will win the prize
c) John's amusement at the children's-antics

3.8a) John is eager to please.
b) John is certain that Bill will win the -Prize.
c) John was amused at the children's antics.

3-9a) John's being eager to please
b) John's being certain that Bill will win the prize
C) John's being amused at the children's antics

in this chapter I will try to answer Chomsky's question by showing t1hat sentences like those in 3.4, which do not nave acceptable corresponding derived nominal complements, have undergone certain transformational rules in their derivations which sentences like those in 3.8, which do have acceptable corresponding derived nominal co mplements, have not undergone. Chomsky (1970:191) claims that this is so






50

because derived nominal complements are noun phrases in deep structure, not embedded propositions. He claims that derived nominal complements correspond only to deep structure phrase markers and a few transforms. I will argue that most rules are blocked from applying to derived nominal complements, or have different conditions on their applicability to derived nominal complements, because they are nominalized embedded propositions. I will also show that many rules can apply in the derivation of gerundive nominal complements that cannot apply in the derivation of derived nominal complements.

I will show in detail how the application of a number

of rules is blocked or modified in the derivation of derived nominal complements. I will discuss the effect of adopting a nredicate-initial analysis on the formulation of rules, and certain consequent simplifications of the rules, particularly with regard to derived nominal complements. i will argue that certain facts of derived nominal complements raise problems for the currently accepted formulation of the rule of IT-EXTRAPOSITION, and that with a predicate-initial analysis, no rule of IT-EXTRAPOSITION is needed in a grammar of English. Throughout this chapter I will show that whenever a sentence does not have an acceptable corresponding derived nominal complement, it is because the sentence has undergone a rule in its derivation which is blocked from applying to derived nominal complements.





51
Chomsky's Subcategorization Account

I will first consider Chomsky's (1970:191) claim that a correct account of the facts of productivity of derived nominal complements should be based on subcategorization features. The pair of sentences in 3.10 and the pair of sentences in 3.11, with their corresponding gerundive and derived nominal complements, present parallel problems.
3.10a) John is easy to please.
b) John is eager to please.
3.11a) John is certain to win the prize.
b) John is certain that Bill will win the prize.
The corresponding gerundive nominal complements in 3.12 and 3.13 and the corresponding derived nominal complements in 3.14b and 3.15b are all acceptable, while the expected corresponding derived nominal complements in 3.14a and 3.15a are not acceptable.
3.12a) John's being easy to please
b) John's being eager to please
3.13a) John's being certain to win the prize
b) John's being certain that Bill will win the prize

3.14a) *JoImh's easiness to please
b) John's eagerness to please
3.15a) *John's certainty to win the prize
b) John's certainty that Bill will win the prize
Chomsky (1970:191) attempts to explain the acceptability of the derived nominal complements in 3.14b and 3.15b, in contrast with the unacceptability of forms like those in
3.14a and 3.15a, in terms of the subcategorization features of e eas and certain. Chomsky states that eager is entered in the lexicon with a strict subcategorization






52
feature indicating, that it can take a sentential complement, as in 3.16, derived from underlying structures something like those in 3.17.

3.16a) John is eager to please.
b) John is eager for us to please.

3.17a) John is eager [sJohn please SOMEONE]
b) John is eager [,we please SOMEONE]s

Chomsky says that no further comment is necessary to

account for the acceptability of the derived nominal complements in 3.18.

3.18a) John's eagerness to please
b) John's eagerness for us to please
According to the Lexical Hypothesis, the lexical entry for ager is also the entry for eagerness, and the strict subcategorization feature applies to adjective and nominal alike. Thus, ee can appear in a construction of the form Noun Phrase-Predicate-Sentential Complement, ad eagerness can appear in a construction of the form Possessive Noun-Nominal-Sentential Complement.

Chomsky claims that, on the other hand, there is no

such suboategorization feature in the lexical entry for easy and that there are no base phrase markers of the form e~sv-Sentential Complement. Chomsky (1970,191) says that easy appears in base phrase markers as an adjective predicated of propositions as subject, as in the sentences in

3.19, with sentences like those in 3.20 derived by I-EXTRAPOSITION, and with sentences like those in 3.21 derived in turn from extraposed sentences by TOUGH-MOVEENT.2





53
3
3.19a) To please John is easy.
b) For us to please John is easy.

3.20a) It is easy to please John.
b) It is easy for us to please John.

3.21a) John is easy to please.
b) John is easy for us to please.

3.22a) John is eager to please.4
b) John is eager for us to please.

Even though we get sentences like those in 3.21, which in surface form exactly parallel those in 3.22, Chomsky argues that just as easy cannot be introduced into structures of the form Subject-Predicate-Sentential Complement, easiness cannot be introduced into structures of the form Possessive Noun-Nominal-Sentential Complement, thus preventing the formation of derived nominal complements like those in 3.23.

3.23a) *John's easiness to please
b) *John's easiness for us to please

Chomsky assumes that gerundive nominal complements are transformationally derived from structures which are in turn derived from base structures. That is, 'ERUNDIVE NOMINALIZATION applies to embedded sentences which have already been subject to all or almost all cyclical rules. It is therefore possible to have gerundive nominal complements like those in 3.24.

3.24la) John's being easy to please
b) John's being easy for us to please

Chomsky is assuming that transformations apply to structures dominated by an S but not to structures dominated by an NP even though the two types of structures may be composed of the same lexical items.






54

Similarly, Chomsky (1970:191) states that certain, with the meaning used in 3.11b, John is certain that Bill will win the prize, also has a subcategorization feature which allows certain in this meaning to take a sentential complement in a structure like FS[NpJohn]NP[VP[ Vbe certain] V[ Bill will win the prize]VP]S, from which 3.11b is derived. Thus, the derived nominal certainty, corresponding to this meaning of certain, has the same subcategorization feature, and can appear in derived nominal complements with a sentential complement, as in 3.15b, John's certainty that Bill will win the prize.

Like the lexical entry for easy, the lexical entry for certain in the meaning used in 3.11a, John is certain to win the prize, has no subcategorization feature allowing certain in this meaning to take a sentential complement. Chomsky says that certain in this meaning appears in base phrase markers as an adjective predicated of propositions as subject, as in for John to win the prize is certain. Applying the same arguments Chomsky used with easy, sentences like it is certain for John to win the prize are derived by IT-EXTRAPOSITION, and sentences like John is certain to win the _rize are derived by RAISING-TO-SUBJECT from extraposed sentences.

According to the Lexical Hypothesis, derived nominals

occur in nominal structures corresponding to the base phrase markers of sentences. Such nominal structures are not subject to most rules which apply to sentences. The derived








nominal certainty corresponding to the second meaning of certain, like easiness, cannot occur in nominal structures corresponding to derived phrase markers.

Chomsky has argued that the lack of productivity of derived nominal complements is explai ned by the Lexical Hypothesis. Under that hypothesis, *John's easiness to please is not acceptable because the lexical entry underlying easy and easiness does not have a subcategorization feature allowing sentential complements. On the other hand, John's eagerness to please is acceptable because the lexical entry underlying eageer and eagerness has a subcategorization feature allowing sentential complements.

RAISING Rules

RAISING-TO-OBJECT

As was indicated in Note 2 of this chapter, RAISING is a general term for three different movement transformations. The first movement I will discuss here is from subject of an embedded sentence to object of the next highest sentence, as has occurred in the derivation of the sentences in 3.25, which are deriLved from the same underlying structures as the sentences in 3.26. 6

3.25a) John believes himself to be heroic.
b) Bill expects them to be here soon.
c) Dick is believed to have been involved by everyone.

3.26a) John believes that he is heroic.
b) Bill expects that they will be here soon.
c) Everyone believes that Dick was involved.

That the nouns in question have been raised to object

position in the higher sentences is indicated by the reflex-





56

ive form himself in 3.25a, the accusative them in 3.2'b and the fact that Dick is the derived subject in a passive sentence in 3.25C. Reflexivization is restricted to clause mates in English, pronouns take the accusative form in object position and an underlying object becomes the surface subject in passive sentences.

The sentences in 3.25, which have undergone RAISING-TO--OBJECT, have the acceptable corresponding gerundive nominal complements in 3.27, but no acceptable corresponding- derived nominal complements, the forms in 3.28 being unacceptable.

3.27a) John's believing himself to be heroic
b) Bill's expecting- them to be here soon
c) Dick's being believed by everyone to have been
involved

3.28a) *John's belief of himself to be heroic
b) *Bill's expectation of them to be here soon
c) *Dick's belief by everyone to have been involved

The sentences in 3.26, which have not undergone RAISING-TO -OBJECT, have the acceptable corresponding gerundive nominal complements in 3.29 and derived nominal complements in 3.30.

3.29a) John's believing that he is heroic
b) Bill's expecting that they will be here soon
c) everyone's believing that Dick was involved

3.'10a) John's belief that he is heroic
b) Bill's expectation that they will be here soon
c) everyone's belief that Dick was involved

?A!ISINGT -TO -SUBJECT

The second movement is from subject of an embedded sentence to subject of the next higher sentence, as has occurred





57

in the derivation of the sentences in 3.31, which are derived from the same underlying structures as the sentences in 3.32.
3.31a) John is certain to win the prize.
b) Bill is likely to be drafted.
c) Jerry appeared to open the door.

3.32a) It is certain that John will win the prize..
b) It is likely that Bill will be drafted.
c) It appeared that Jerry opened the door.

The sentences in 3.31, which have undergone

RAISING-TO-SUBJECT, have the acceptable corresponding gerundive nominal complements in 3.33, but no acceptable corresponding derived nominal complements, the forms in 3.34 being unacceptable.

3.33a) John's being certain to win the prize
b) Bill's being likely to be drafted c) Jerry's appearing to open the door

3.34a) *John's certainty to win the prize
b) *Bill's likelihood to be drafted
c) *Jerry's appearance to open the dock_7

The sentences in 3.32, which have not undergone

RAISING-TO-SUBJECTt have the acceptable corresponding gerundive nominal complements in 3.35.8

3.35a) it(s) being certain that John will win the prize
b) it(s) being likely that Bill will be drafted
c) it(s) appearing that Jerry opened the door

The apparently corresponding derived nominal complements in 3.36 are not acceptable, but the derived nominal complements in 3.37, which have the same word order as the sentences in 3.32, but do not have any reflex of it, are acceptable.






58

3.36a) *its certainty that John will win the prize
b) *its likelihood that Bill will be drafted
c) *its appearance that Jerry opened the door

3.37a) the certainty that John will win the prize
b) the likelihood that Bill will be drafted
c) the appearance that Jerry opened the door

I will argue below (cf. Dummy Subject Insertion) that

it in sentences like those in 3.32 is inserted, and not present in underlying structures. The unacceptability of the forms in 3.36, and the acceptability of the forms in 3.37 is simply explained by blocking IT-INSERTION in derived nominal

complements.

T0U H -M 0V *Ivrf".E

The third RAISING movement is from object of an embedded sentence to subject of the next higher sentence, as has occurred in the derivation of the sentences in 3.38, which are derived from the same underlying structures as the sentences in 3.39.

3.38a) John is easy to please.
b) Algebra is difficult to learn.
c) This test is fun to take.

3.39a) it is easy -to please John.
b) It is difficult to learn algebra.
c) It is fun to take this test.

The sentences in 3.38, which have undergone TOUGH-MOVE.NT, have the acceptable corresponding gerundive nominal complements in 3.40, but no acceptable corresponding derived nominal complements, the forms in 3.41 being unacceptable.

3.40a) John's being easy to please
b) algebra's being difficult to learn
c) this test's being fun to take





59
3.41a) *John's easiness to please
b) *algebra's difficulty to learn
c) *this test's fun to take
The sentences in 3.39, which have not undergone
TOUGH-MOVEMENT, have the acceptable corresponding gerundive nominal complements in 3.42.

3.42a) it(s) being easy to please
b) it(s) being difficult to learn algebra
c) it(s) being fun to take this test
The apparently corresponding derived nominal complements in 3.43 are not acceptable, but those in 3.44 are acceptable. The argument used with the examples in 3.36 and 3.37 applies here as well.

3.43a) *its easiness to please John
b) *its difficulty to learn algebra
c) *its fun to take this test

3.44a) ?the easiness of pleasing John
b) the difficulty of learning algebra
c) the fun of taking this test
RAISING-TO-OBJECT, RAISING-TO-SUBJECT and TOUGH-MOVEmENT do not apply in the derivation of derived nominal complements. The three rules share the property of moving a noun phrase to a higher sentence. They appear to apply at the same point in the cycle.10 G. Lakoff (1968) has argued that there is only one rule of RAISING (IT-R-EPLACEiMENT in his paper). To collapse the three types of RAISING to one, Lakoff has to write a complex rule, which includes simultaneous structural descriptions. I consider such a rule to be unlikely in a natural language. I will, therefore, suspend judgment on whether it is possible to collapse the RAISING rules.






6o

IT-EXTRAPOSiTiO lT

introduction

The rule of __T-EXTRAPOSITION was proposed by Rosenbaum (1967) to account for the relationship of sentences like those in 3.45 to the sentences in 3.46.11

3.45a) That John left so early is surprising.
b) For John to quit now would be a disaster.
c) That Mary will come to her senses is to be
hoped for.
3.46a) It is surprising that John left so early.
b) It would be a disaster for John to quit now.
c) It is to be hoped for that Mary will come to
her senses.

In Rosenbaum's analysis, the subject it in the sentences in 3.46 is present in deep structure as the head noun of noun phrases with the structure of 3.47. The rule of iIT-EX!RAPOSITION is given in 3.48.

3.47) NP
N Sit that John left so early

3.48) 1T-.-zXTR A -TION (Optional)

X N S Y
[+PRO(i.e. it)] 1$ 2l 4+3
1 2 3 4
If iT-EXTRAPOSITION does not apply, the head pronoun it is deleted. if IT-EXTRAPOSITION does apply, moving the embedded sentence to the right end of the next higher sentence, the head pronoun it is isolated in subject position, and no longer subject to deletion.

In this section I will first discuss the derivation of certain examples via the rule of IT-EXTRAPOSITION, showing






61

that the pattern of occurrence of derived nominal complements casts doubt on the validity of IT-EXTRAPOSITION as proposed by Rosenbaum. I will then consider other evidence counter to his analysis, and argue that there is no rule of IT-EXTRAPOSITION in English grammar. In a later section (cf. NP-PREPOSING) I will present an alternate analysis of the relationship between the sentences in 3.45 and 3.46. WNith Sublects of Predicate Adjectives

We have seen above that sentences like those in 3.49

have acceptable corresponding gerundive nominalcomplements, as in 3.50, and derived nominal complements, as in 3.51.

3.49a) It is certain that John will win the prize.
b) It is difficult to learn algebra.
c) It is fun to take this test.

3.50a) it(s) being certain that John will win the prize
b) it(s) being difficult to learn algebra
c) it(s) being fun to take this test

3,51ai) the certainty that John will win-the prize
ii) the certainty of John's winning the prize
b) the difficulty of learning algebra
c) the fun of taking this test

According to Chomsky's (1970) analysis, the sentences in 3.49 are derived from the phrase markers underlying the sentences in 3.52 by the rule of IT-EXTRAPOSITION. The sentences in 3.52 do not have acceptable corresponding gerundive nominal complements or derived nominal complements, as is shown by the unacceptability of the expected forms in

3.53 and 3.54.

3.52ai) That John will win the prize is possible.
ii) John's winning the prize is certain.
bi) To learn algebra is difficult.
ii) Learning algebra is difficult.






62

3.52ci) To take this test is fun.
ii) Taking this test is fun.

3.53ai) *that John will win the prize's being certain
ii) *John's winning the prize's being certain
bi) *to learn algebra's being difficult 1i) *learning algebra's being difficult
ci) *to take this test s being fun ii) *taking this test's being fun
3.54ai) *that John will win the prize's certainty
ii) *John's winning the prize's certainty
bi) *to learn algebra's difficulty
ii) *learning algebra's difficulty
ci) *to take this test's fun ii) *taking this test's fun

The Lexical Hypothesis predicts that there should be derived nominal complements corresponding to the sentences in 3.52, which are supposedly closer to base phrase markers than the extraposed sentences in 3.49, but the unacceptability of the strings in 3.53 and 3.54, in contrast to the acceptability of the nominal complements in 3.50 and 3.51, is counter to that prediction.12 With Passives

By Rosenbaum's analysis, IT-EXTRAPOSITION also applies to passive sentences with complements in the derived subject position. Thus, the extraposed sentences in 3.55 are derived from the phrase markers underlying the passive sentences in 3.56. The sentences in 3.56 do not have any acceptable corresponding derived nominal complements, the expected forms in 3.57 being unacceptable, but the extraposed passive sentences in 3.55 have the corresponding derived nominal complements in 3.58.
3.55a) It was discovered by John that Jerry was a
bigamist.
b) It is doubted by them that you will go.






63
3.56a) That Jerry was a bigamist was discovered by John.
b) That you will go is doubted by them.

3.57a) *that Jerry was a bigamist's discovery by John
b) *that you will go's doubting by them
3.58a) the discovery by John that Jerry was a bigamist
b) the doubting by them that you will go
Chomsky suggests that derived nominal complements such as those in 3.59 are derived by the obligatory application of a rule called AGENT-POSTPOSING (cf. Passive Sentences below) to the underlying noun phrases in 3.60.
3.59a) the necessity for John to leave
b) the likelihood that John will leave
3.60a) [Sfor John to leaves' necessity
b) [Sthat John will leave]s's likelihood
This solution will explain why the examples in 3.51 and
3.59 are acceptable while those in 3.53 and 3.54 are not. AGENT-POSTPOSING seems to be a poor name for a rule which moves complements in such structures, however, and there is a curious problem with forms like those in 3.57 and 3.58. If the fact that the derived nominal complements in 3.58 are acceptable but not those in 3.57 is to be explained by the ad hoc obligatory application of AGENT-POSTPOSING, we are left with the problem of accounting for the post-verbal k_-phrases in 3.58. I must reject Chomsky's implied claim that AGEN-PFOSTPOSING has applied in the derivations of the derived nominal complements in 3.58, but not in those in
3.59, and in so doing, reject his claim that AGENT-POSTPOSING moves any complement within derived nominal complements,






64

With Psvcholozical Predicates

iT-EXTRAPOSITION can also apply to sentences like those in 3.61. These are active sentences with complements as subjects. The verbs which permit this construction share a number of other characteristics, and will be called osycholozical predicates here (cf. Postal, 1971:39-54, and the section Psvcholozical Predicates, below). The sentences in

3.61 are subject to IT-EXTRAPOSITION in Rosenbaum's analysis, yielding the sentences in 3.62.

3.61a) That John is here surprises me.
b) For Mary to be so late worries Bill.
c) For Jim to leave now would disturb Alice,

3.62a) It surprises me that John is here.
b) It worries Bill for Mary to be so late.
c) it would disturb Alice for Bill to leave now.

Neither the sentences in 3.61 nor the sentences in 3.62 have acceptable corresponding derived nominal complements, as shown by the unacceptable forms in 3.63 and 3.64, and there are no acceptable derived nominal complements with determiners which are not possessive nouns corresponding to the extraposed sentences in 3.61, as is shown by the unacceptable examples in 3.65.
3.63a) *that John is here's surprise to me
b) *for Mary to be so late's worry to Bill
c) *for Jim to leave now's disturbance of Alice
3.64a) *its surprise to me that John is here
b) *its worry to bill for Mary to be so late
c) *its disturbance of Alice for Bill to leave now

3.65a) *the surprise to me that John is here
b) *the worry to Bill for Mary to be so late
c) *the disturbance of Alice for Bill to leave now






65

Psychological predicates appear to be the only predic t.es which form extraposed sentences which do not have acceptable corresponding derived nominal complements. Further Problems with the Rule

The acceptable derived nominal complements in 3.51,

3.58 and 3.59 all show the same order of major constituents as the extraposed sentences in 3.46, 3.49 and 3.55. There is no subject pronoun it in these derived nominal complements. This fact can be explained by having it in extraposed sentences supplied by a rule of IT-INSERTION, which does not apply to derived nominal complements (cf. Dummy Subject inserting, below). The arguments for not having it present in deep structure are well summarized in Stockwell et al. (1973:527ff.). Given the absence of it in underlying structures, the derived nominal complements in 3.51, 3.58 and 3.59 may be taken as corresponding directly- to the extraposed sentences cited above. If Chomsky's claim that derived nominal complements correspond only to base phrase markers is at all correct, then extraposed sentences would appear to be more basic than nonextraposed sentences.

Although this conclusion goes against a generally accepted analysis, there is some independent supporting evidence. The intransitive verbs seem, annear and h take subject

complements which must be extraposed.13 The extraposed sentences in 3.66 are acceptable while the nonextraposed ones in 3.67 are not.





66

3.66a) It seems that John is late.
b) It appears that Iildred has fallen down.
c) It happens that Mlike is a brilliant student.

3.67a) *That John is late seems.
b) *That M~ildred has fallen down appears.
c) *That Mike is a brilliant student happens.

It would seem that at least some sentences can be
derived only by obligatory application of IT-EXTRAPOSITION. I have noted above in Chapter One certain arguments for assuming an underlying predicate-initial order in English. All extraposed sentences, including those resulting from the obligatory application of IT-EXTRAPOSITION, have a predicate-initial order, if the presence of the semantically empty pronoun it in subject position is discounted. I propose that there is no rule of IT-EXTRAPOSITION in English, and that nonextraposed sentences are derived from the structures underlying extraposed sentences. I will discuss the details of such derivations in the section on NP-PREPOSING below.
Passive Sentences
I will turn next to a problem which is not raised by
Chomsky's examples; that of derived nominal complements corresponding to passive sentences. The sentences in 3.68 have the corresponding gerundive nominal complements in 3.69 and derived nominal complements in 3.70.

3.68a) John refused the offer.
b) John criticized the book.
c) The enemy destroyed the city.
d) The jury acquitted Abby.

3.69a) John's refusing the offer
b) John's criticizing the book





67

3. 69c) the enemy's destroying- the city
d) the jury's acquitting Abby
3.70a) John's refusal of the offer
b) John's criticism of the book
c) the enemy's destruction of the city
d) the jury's acquittal of Abby

The sentences in 3.68 also have the corresponding passive sentences in 3.71.

3.71a) The offer was refused by John.
b) The book was criticized by John,
c) The city was destroyed by the enemy.
d) Abby was acquitted by the jury.

The passive sentences in 3.71 have the correspDondin'g gerundive nominal complements in 3.72, but only the sentiences in 3.71c and d have the corresponding derived nominal complements in 3.73c and d, while the strings in 3-73a and b, which seem to correspond to the sentences in 3.71a and b, are not acceptable.

3.72a) the offer's being refused by John
b) the book's being criticized by John
c) the city's being destroyed by the enemy
d) Abby's being acquitted by the jury

3.73a) *the offer's refusal by John
b) *'he book's criticism by John
c) he city's destruction by the enemy
d) Abby's acquittal by the jury

The derived nominal complements in 3.74, all of which are acceptable, are related to the derived nominal complements in 3.70 and 3.73, but do not correspond exactly to any sentences. These derived nominal complements would correspond to predicate-initial underlying structures.

3.74a) the refusal of the offer by John
b) the criticism of the book by John
c) the destruction of the city by the enemy
d) the. acquittal of Abby by the jury






68

Chomsky's claim that derived nominal complements correspond only to base phrase markers creates a problem here if a predicate-initial analysis is not adopted. Not only are there derived nominal complements corresponding to active sentences, as in 3.70, there are also derived nominal complements corresponding to some, but not all, passive sentences, as in 3.73, and there are derived nominal complements which do not correspond to any sentences at all (what might be called "half-passives"), as in 3.74.

Chomsky (1970z202ff.) proposes two transformations

which must both apply in the derivation of passive sentences from phrase markers parallel to those underlying active sentences, one of which also can apply to noun phrases (including derived nominal complements). Chomsky proposes a rule of NP-PREPOSING which will transform phrases like those in

3.75 into phrases like those in 3.76. This rL'e thus applies to noun phrases whether they have a simple noun or a derived nominal as their head.

3.75a) the picture of John
b) the bottom of the barrel
c) the destruction of the city
d) the murder of John

3.76a) John's picture
b) the barrel's bottom (the literal meaning)
c) the city's destruction
d) John's murder

There are similar phrases which do not undergo the rule of NP-PREPOSING, and again the phrases can have either a simple noun or a derived nominal as their head. The phrases in

3.77 do not have acceptable corresponding forms like those in 3.78.





69

3.77a) the algebra of revolution
b) the strategy of war
c) the refusal of the offer d) the criticism of the book

3.78a) *revolution's algebra
b) *war's strategy
c) *the offer's refusal
d) *the book's criticism

Chomsky proposed a rule of AGENT-POSTPOSING to move the subject noun phrase to a post-verbal position in phrase markers which are to become passive sentences. NP-PREPOSING would then apply to move the object noun phrase into pre-verbal position. Chomsky goes on to say that since passivizability is a property of verbs (i.e., is a governed process), then derived nominal complements containing nominals corresponding to such verbs can also be passivized. it would seem that the passivizability of a verb is best expressed by having AGENT-POSTPOSING apply optionally to structures with verbs which can appear in passive sentences. That is, verbs which can appear in passive sentences are marked as allowing the optional application of AGENT-POSTPOSING, while verbs which cannot appear in passive sentences are marked as not allowing the application of AGENT-POSTPOSING. On the other hand, NP-PREPOSING applies obligatorily in the derivation of any sentence in which AGENT-POSTPOSING has already applied, and optionally in the derivation of any noun phrase (including derived nominal complements) whose head noun (or nominal) is marked as allowing the rule. A given lexical entry may therefore be marked differently for AGENT-POSTPOSING and for NP-PREPOSING.






70

I will claim here that there is a rule of AGENT-PREPOSING (the reverse of Chomsky's AGENT-POSTPOSING) which applies to predicate-initial underlying structures to yield active sentences and nominal complements such as those in

3.68, 3.69 and 3.70. If AGENT-PREPOSING does not apply, then the resulting obligatory application of NP-PREPOSING gives the passive sentences and gerundive nominal complements in 3.71 and 3.72. If AGENT-PREPOSING does not apply in the derivation of derived nominal complements, then NP-PREPOSING applies optionally with some underlying verbs, giving the pairs of derived nominal complements in 3.73c and d and 3.74c and d, and not at all with other verbs, giving the derived nominal complements in 3.74a and b.
By Chomsky's analysis, passive sentences are derived

from structures underlying active sentences by the application of two rules, both of which must apply. .y my analysis, all sentences are derived from underlying predicate-initial structures, with active sentences derived by one rule and passive sentences by another rule.
NP-PREPOSING

With Simole Noun Phrases
The application of AGENT-PREPOSING to underlying predicate-initial structures with agents yields active sentences like those in 3.79, and their corresponding gerundive nominal complements in 3.80 and derived nominal complements in 3.81.






71

3.79a) The enemy destroyed the city.
b) The jury acquitted Abby.
c) John refused the offer.
d) John criticized the book.

3.80a) the enemy's destroying the city
b) the jury's acquitting Abby c) John's refusing the offer
d) John's criticizing the book

3.81a) the enemy's destruction of the city
b) the jury's acquittal of Abby
c) John's refusal of the offer
d) John's criticism of the book

If AGENT-PREPOSING does not apply to the structures

underlying the active sentences and nominal complements in

3.79, 3.80 and 3.81, then the rule of NP-PR4EPOSING applies

obligatorily in the derivation of passive sentences and gerundive nominal complements like those in 3.82 and 3.83.

3.82a) The city was destroyed by the enemy.
b) Abby was acquitted by the jury.
c) The offer was refused by John.
d) The book was criticized by John.

3.83a) the city's being destroyed by the-enemy
b) Abby's being acquitted by the jury
c) the offer's being refused by John
d) the book's being criticized by John

Either AS-NT-PREPOSING or NP-PREPOSING must apply in

the derivation of sentences and gerundive nominal complements. Forms like those in 3.84 and 3.85, in which neither

rule has applied, are not acceptable.

3.84a) *Was destroyed the city by the enemy.
b) *Was acquitted Abby by the jury.
c) *Was refused the offer by John.
d) *Was criticized the book by John.

3.85a) *being destroyed the city by the enemy
b) *being acquitted Abby by the jury
c) *being refused the offer by John
d) *being criticized the book by John






72

On the other hand, in the derivation of derived nominal complements, if AG NT-PR'POSING does not apply, then the application of NP-PREPOSING is optional for some nominals (or underlying verbs), giving the paired acceptable derived nominal complements in 3.86a and b and 3.87a and b, and blocked for other nominals, so that the derived nominal complements in 3.87c and d are acceptable, but not those in

3.86c and d.

3.86a) the city's destruction by the enemy
b) Abby's acquittal by the jury c) *the offer's refusal by John
d) *the book's criticism by John

3.87a) the destruction of the city by the enemy
b) the acquittal of Abby by the jury c) the refusal of the offer by John
d) the criticism of the book by John

If there is no specified agent noun phrase in an underlying structure, then NP-PREPOSING applies obligatorily in the derivation of sentences, as in 3.88, and gerundive nominal complements, as in 3.89, with the forms in 3.90 and 3.91 being unacceptable.

3.88a) The city was destroyed.
b) Abby was acquitted.
c) 2he offer was refused.
d) The book was criticized.

3.89a) the city's being destroyed
b) Abby's being acquitted
c) the offer's being refused
d) the book's being criticized

3.90a) *Was destroyed the city.
b) *Was acquitted Abby.
c) *Was refused the offer.
d) *Was criticized the book

3.91a) *being destroyed the city
b) *being acquitted Abby
c) *being refused the offer
d) *being criticized the book





73

Again,.when AGENT-PRoPOSING has not applied in the

derivation of' a derived nominal complement, the application of NP-PREPOSING is optional for some norninals and blocked for others, as indicated by the derived nominal complements in 3.92 and 3.93.

3.92a) the city's destruction
b) Abby's acquittal
c) *the offer's refusal
d) *the book's criticism

3.93a) the destruction of the city
b) the acquittal of Abby
c) the refusal of' the offer
d) the criticism of the book

Finally, if the underlying predicate-initial structure contains an intransitive predicate, then NP-PREPOSING applies obligatorily in the derivation of' sentences like those in 3.9)4 and gerundive nominal complements like those in 3.95, with the forms in 3.96 and 3.97 being unacceptable.

3.94a) The boat sank suddenly.
b) John arrived.
c) Nvary is kind.
d) Bill is friendly.

3.95a) the boat' s sinking suddenly
b) John's arriving
c) Mary's being kind
d) Bill's being friendly

3.96a) *Sank the boat suddenly.
b) *Arrived John.
c) *Is kind Mary.
d) *is friendly Bill.

3.97a) *sinking the boat suddenly
b) *arriving John
c) *being kind Mdary
d) *being friendly Bill

Again, the application of NP-PREPOSING in the derivation of derived nominal complements corresponding to






74

intransitive sentences is'optional, so that the derived nominal complements in 3.98 and 3.99 are acceptable.

3.98a) the boat's sudden sinking
b) John's arrival c) Mary's kindness
d) Bill's friendliness

3.99a) the sudden sinking of the boat
b) the arrival of John c) the kindness of Mary
d) the friendliness of Bill

The application of NP-PREPOSING to simple noun phrases shows a simple pattern. If there is no noun phrase in subject position when NP-PREPOSING is applicable (as is the case with all intransitive predicates, transitive verbs with unspecified agents and transitive verbs with specified agents to which AGENT-PREPOSING has not applied), then NP-PREPOSING moves the underlying object noun phrase into subject position. This movement is obligatory in the derivation of sentences and gerundive nominal complements and optional in the derivation of derived nominal complements (with the exception that certain nominals corresponding to transitive verbs block the application of the rule in derived nominal complements). With Nominal Comolements

As I indicated in Chapter One, I have accepted Menzel's (1969) analysis of nominal complements as complements of a restricted set of head nouns in underlying structures. These head nouns are optionally deletable, and in the following examples will be enclosed in parentheses to indicate this optionality.






?5

Again, I will claim that AGENT-PREPOSING has applied to

predicate-initial underlying structures to yield the sentences in 3.100, and their corresponding gerundive nominal

complements in 3.101 and derived nominal complements in 3.102.

3.100a) John complained about Peter's (action of)
insulting Tom.
b) Jerry testified about Bill's (action of)
leaving Peggy.
c) Mike attested to (the fact of) Mary's participation in the crime.
d) David reported on (the event of) Alice's
disappearance.

3.101a) John's complaining about Peter's (action of)
insulting Tom
b) Jerry's testifying about Bill's (action of)
leaving Peggy
c) Mike's attesting to (the fact of) Mary's
participation in the crime
d) David's reporting on (the event of) Alice's
disappearance

3.102a) John's complaint about Peter's (action of)
insulting Tom
b) Jerry's testimony about Bill's (action of)
leaving Peggy
c) 2ike's attestation to (the fact of) Mary's
participation in the crime
d) David's report on (the event of) Alice's
disappearance

The verbs complain about, attest to and report on allow

passivization in my dialect, while testify about does not.

That is, AGENT-PREPOSING applies optionally to the first

three predicates and obligatorily to testify about in the

derivation of sentences. If AGENT-PREPOSING does not apply,

then NP-PREPOSING must apply to yield sentences like those

in 3.103. The expected corresponding gerundive nominal complements in 3.104 and derived nominal complements in 3.105

are not acceptable (cf. Note 12 above).






76

3.103a) Peter's (action of) insulting Tom was complained about by Torn.
b) *Bili's (action of) leaving Peggy was testified
about by Jerry.
c) (The fact of) Mary's participation in the
crime was attested to by Mike..
d) (The event of) Alice's disappearance was
reported on by David.

3.104a) *Peter's (action of) insulting Tomn's being
complained about by John
b) *Bill's (action of) leaving Peggy's being
testified about by Jerry
c) *(the fact of) Mary's participation in the
crime's being attested to by Mike
d) *(the event of) Alice's disappearance's being
reported on by David

3.105a) *Peter's (action of) insulting Tom's complaint
about by John
b) HBill's (action of) leaving Peggy's testimony
about by Jerry
c) *the fact of) Mary's participation in the
crime's attestation to by Mike
d) *(the event of) Alice's disappearance's5 report
on by David

Failure of both AGENT-PREPOSING and NP-PREPOSI:G to

apply in the derivation of sentences and gerundive nominal

complements results in unacceptable forms, as in 3.106 and

3.107, but derived nominal complements to which neither rule

has applied are acceptable, including one with testimony

about (from testify about), as in 3.108.

3.106a) *Wlas complained about by John Peter's (action
of) insulting Tom.
b) *Was testified about by Jerry Bill's (action
of) leaving Peggy.
c) *Wa attested to by Mike (the fact of) Mary's
participation in the crime.
d) *Wlas reported on by David (the event of)
Alice 's disappearance.

3.107a) *being complained about by John Peter's (action
of) insulting Tom
b) *being testified about by Jerry Bill's (action
of) leaving Peggy





77

3.107c) *being attested to by Mike (the fact of) Mary's
participation in the crime
d) *being reported on by David (the event of)
Alice's disappearance

3.108a) the complaint by John about Peter's (action
of) insulting Tom
b) the testimony by Jerry about Bill's (action
of) leaving Peggy
c) the attestation by Mike to (the fact of)
Mary's participation in the crime
d) the report by David on (the event of) Alice's
disappearance

While the head nouns of the embedded nominal complemaents in the examples above may be deleted, they may not be

separated from the complements, i.e., by moving the head

nouns into surface subject position but leaving the complements at the right end of the structure (of. With That-Compolements, below), so that all of the examples in 3.109,

3.110 and 3,111 are unacceptable.

3.109a) *Peter's action was complained about by John of
insulting Tom.
b) *Bill's action was testified about by Mike of
leaving Alice.
c) *The fact was attested to by Mike of Mary's
participation in the crime.
d) *The event was reported on by David of Alice's
disappearance.

3.110a) *?Peterls action's being complained about by
John of insulting Tom
b) *Bill's action's being testified about by Jerry
of leaving Peggy
c) *the fact's being attested to by Mike of Mary's
participation in the crime
d) *the event's being reported on by David of
Alice's disappearance

3.1l1a) *Peterls action's complaint by John about
insulting Torn
b) *Bill's action's testimony by Jerry about
.leaving Peggy
c) *the fact's attestation by Mike to Mary's participation in the crime,
d) *the event's report by David on Alice's disappearance






78

Like destroy, acquit, refuse and criticize, the predicates cmolain about, attest to and report on also form

agentless passives. Again, AGENT-PREPOSING does not apply

to underlying structures with unspecified agents. Thus,

NP-PREPOSING must apply in the derivation of sentences, as

in 3.112, but does not apply in the derivation of gerundive

and derived nominal complements, so that the forms in 3.113

and 3.114 are not acceptable.

3.112a) Peter's (action of) insulting Tom was complained about.
b) (The fact of) Mary's participation in the
crime was attested to.
c) (The event of) Alice's disappearance was
reported on.

3.113a) *Peter's (action of) insulting Tom's being
complained about
b) *(the fact of) Mary's participation in the
crime's being attested to
c) *(the event of) Alice's disappearance's
being reported on

3.114a) *Peter's (action of) insulting Tom's complaint
about
b) *(the fact of) Mary's participation in the
crime's attestation to
c) *(the event of) Alice's disappearance's
report on

Failure of both AGENT-PREPOSING and NP-PREPOSING to

apply in the derivation of sentences and gerundive nominal

complements results in unacceptable forms, as in 3.115 and

3.116, but derived nominal complements to which neither

rule has applied, as in 3.117, are acceptable.

3.115a) *Was complained about Peter's (action of)
insulting Tom.
b) *Was attested to (the fact of) Mary's participation in the crime.
c) *Was reported on (the event of) Alice's disappearance.






79

3.116a) *beinz complained about Peter's (action of)
insulting Tom
b) *being attested to (the fact of) .,iary's participation in the crime
c) *beinz reported on (the event of) Alice's
disappearance
3.117a) the complaint about Peter's (action of)
insulting Tom
b) the attestation to (the fact of) iviary's
participation in the crime
c) the report on (the event of) Alice's disappearance

Again, while the head nouns of the embedded nominal complements in the examples above may be deleted, they may not

be separated from the complements, so that all the examples

in 3.118, 3.119 and 3.120 are unacceptable.

3.118a) *Peter's action was complained about of
insulting Tom.
b) *The fact was attested to of Mary's participation in the crime.
c) *The event was reported on of Alice's disappearance.

3.119a) *Peter's action's being complained about of
insulting Tom
b) *the fact's being attested to of Mary's participation in the crime
c) *the event's being reported on of Alice's
disappearance

3.120a) *Peter's action's complaint about insulting
TOfl
b) *the fact's attestation to Mary's participation in the crime
c) *the event's report on Alice's disappearance

if the underlying predicate-initial structure contains

an intransitive predicate, then NP-PREPOSING applies obligatorily in the derivation of sentences, as in 3.121, but is

blocked from applying in the derivation of nominal complements, so that the forms in 3.122 and 3.123 are unacceptable.






80

3.121a) (The fact of) John's leaving so early was
unexpected.
b) Bill's (action of) refusing the offer is
unfortunate.
c) (The event of) Dick's resignation is likely.
d) (The fact of) Jimmy's testimony was remarkable.

3.122a) *(the fact of) John's leaving so early's being
unexpected
b) *Bill's (action of) refusing the offer's being
unfortunate
c) *(the event of) Dick's resignation's being
likely
d) *(the fact of) Jim-my's testimony's being
remarkable

3.123a) *(the fact of) John's leaving so early's
unexpectedness
b) *Bill's (action of) refusing the offer's
unfortunateness
c) *(the event of) Dick's resignation's likelihood
d) *(the fact of) Jimmy's testimony's remarkableness
Failure of both AG2NT-PREPOSING and NP-PREPOSING to

apply in the derivation of sentences and gerundive nominal
complements results in unacceptable forms, as in 3.124 and
3.125, but derived nominal complements to which neither rule

has applied, as in 3.126, are acceptable.
3.124a) *Was unexpected (the fact of) John's leaving
so early.
b) *Was unfortunate Bill's (action of) refusing
the offer
c) *is likely (the event of) Dick's resignation.
d) *,Tas remarkable (the fact of) Jimmy's testimony.
3.125a) *being unexpected (the fact of) John's leaving
so early
b) *being unfortunate Bill's (action of) refusing
the offer
c) *being likely (the event of) Dick's resignation
d) *being remarkable (the fact of) Jimmy's testimony

3.126a) the unexpectedness of (the fact of) John's
leaving so early
b) the unforunateness of Bill's (action of)
refusing the offer
c) the likelihood of (the event of) Dick's
resignation
d) the remarkableness of (the fact of) Jimmy's
testimony






81

Again, while the head nouns of the embedded nominal complements in the examples above may be deleted, they may not be separated from the complements, so that all the examples in 3.127, 3.128 and 3.129 are unacceptable.

3.127a) *The fact was unexpected of John's leaving so
early.
b) *Bill's action was unfortunate of refusing the
offer.
c) *The event was likely of Dick's resignation.
d) *The fact was remarkable of Jimmy's testimony.

3.128a) *the fact's being unexpected of John's leaving
so early
b) *Bill's action's being unfortunate of refusing
the offer
c) *the event's being likely of Dick's resignation
d) *the fact's being remarkable of Jimmy's
testimony

3.129a) *the fact's unexpectedness of John's leaving
so early
b) *Bill's action's unfortunateness of refusing
the offer
c) *the event's likelihood of Dick's resignation
d) *the fact's remarkableness of Jimmy's
testimony

The examples above show that either AGENT-PREPOSING or NP-PREPOSING must apply in the derivation of sentences with embedded nominal complements, that the only acceptable gerundive nominal complements with embedded nominal complements are those to which AGENT-PREPOSING has applied, and that derived nominal complements with embedded nominal complements to which AGENT-PREPOSING has applied, as well as derived nominal complements with embedded nominal complements to which neither AGENT-PREPOSING nor NP-PREPOSING has applied, are acceptable. No other possibilities are acceptable, not is any case in which the head noun is separated from the embedded nominal complements.





82

44ith That-Complements

That-complements also appear to have head nouns in underlying structure (cf. Menzel, 1969). Like the head nouns of nominal complements, these head nouns are deletable. The optional presence of head nouns in surface structure will again be indicated by the use of parentheses.

As before, I will claim that AGENT-PREPOSING has applied to predicate-initial underlying structures to yield the sentences in 3.130, and their corresponding gerundive nominal complements in 3.131 and derived nominal complements in 3.132.

3.130a) Tom reported (the fact) that the account was
overdrawn.
b) Jerry revealed (the fact) that Alice was smart.
c) INary denied (the claim) that Peggy was
pregnant.

3.131a) Tom's reporting (the fact) that the account
was overdrawn
b) Jerry's revealing (the fact) that Alice was
smart
c) Mary's denying (the claim) that Peggy was
pregnant

3.132a) Tom's report (of the fact) that the account
was overdrawn
b) Jerry's revelation (of the fact) that Alice
was smart
c) ary's denial (of the claim) that Peggy was
pregnant

If AENT-PREPOSING does not apply, then NP-PREPOSING applies obligatorily in the derivation of sentences. However, NP-PRLPOSING applies in two ways. Either the head noun and the that-complement are preposed as a unit, as has happened in the derivation of the sentences in 3.133, or only the head noun is preposed, in which case the sentences in 3.134 result if the head noun is retained in surface






83
.structure, and the sentences in 3.135 result if the head

noun is deleted.

3.133a) (The fact) that, the account was overdrawn was
reported by Tom.
b) (The fact) that Alice was smart was revealed
by Jerry.
c) (The claim) that Peggy was pregnant was denied
by Mary.

3.134a) The fact was reported by Tom that the account
was overdrawn.
b) The fact was revealed by Jerry that Alice was
smart.
c) The claim was denied by M~'ary that Peggy was
pregnant.

3.135a) It was reported by Tom that the account was
overdrawn.
b) It was revealed by Jerry that Alice was smart.
c) It was denied by Mary that Peggy was pregnant.

The sentences in 3.133 are passive. Sentences like

those in 3.134 have been described as being derived by the

rule of EXTRAPOSITICN FROM NP (of which I will have more to say below). The sentences in 3.135 are extraposed (i.e.,

supposedly derived by IT-EXTRAPOSITION).

In no case in which head nouns are retained in surface

structure are both the head noun and the that-complement to

-the right of the predicate in sentences. Thus, there are no

acceptable sentences like those in 3.136.

3.136a) *I't was reported by Tom the fact that the
account was overdrawn.
b' *It was revealed by Jerry the fact that Alice
was smart.
c) *It was denied by Mary -the claim that Peggy
was pregnant.

The sentences in 3.124 do not have any acceptable corresponding gerundive or derived nominal complements, the

forms in 3.137 and 3.138 all being unacceptable.






84

3.137a) *(tUhe fact) that the account was overdrawn's
being reported byA Torn
b) *(the fact) that Alice was smart's being
revealed by Jerry
c) *(the claim) that Peggy was pregnant's being
denied by Mary

3.138a) *(the fact) that the account was overdrawn's
report by Tomn
b) *(the fact) that Alice was smart's revelation
by Jerry
c) *(the claim) that Peggy was pregnant's denial
by Mary

The sentences in 3.134 do not have any acceptable corresponding gerundive or derived nominal complements, the

forms in 3.139 and 3.140 all being unacceptable.

3.139a) *the fact's being reported by Torn that the
account was overdrawn
b) *the fact's being revealed by Jerry that Alice
was smart
c) *the claim's denial by Mary that Peggy was
pregnant

3.140a) *the fact's report by Torn that the account was
overdrawn
b) *the fact's revelation by Jerry that Alice was
smart
c) *the claim's denial by Mary that Peggy was
pregnant

The sentences in 3.135 have the acceptable corresponding

gerundive nominal complements in 3.141 and derived nominal

cornlernen-ts in 3.142.

3.141a) its being reported by Tom that the account
was overdrawn
b) its being revealed by Jerry that Alice was smart
c) it being denied by Mary that Peggy was
pregnant

3-142a). the report by Tomn that the account was overdrawn
b) the revelation by Jerry that Alice was smart
c) the denial by Mary that Peggy was pregnant






85

Finally, although the sentences in 3.136, and their

expected corresponding gerundive nominal complements in 3.143, are all unacceptable, the derived nominal complements in

3.144, with the same linear order as the forms in 3.136 and
3.143, are acceptable.

3.143a) *its being reported by Tom the fact that the
account was overdrawn
b) *its being revealed by Jerry the fact that
Alice was smart
c) *its being denied by Mary the claim that Peggy
was pregnant

3.144a) the report by Tom of the fact that the account
was overdrawn
b) the revelation by Jerry of the fact that
Alice was smart
c) the denial by Mary of the claim that Peggy
was pregnant

By referring to the examples given for simple noun

phrases and nominal complements, the reader should be able to convince himself that the patterns of acceptability exhibited for the examples with transitive pred-icates and specified agents given immediately above also hold for forms with transitive predicates and unspecified agents and for forms with intransitive predicates.

If we understand NP-PREPOSING to apply either to the head noun alone or to the head noun plus its that-complement, we see that the rule has the same conditions on application to that-complements that it has to nominal complements: it applies optionally in the derivation of sentences, and not at all in the derivation of derived nominal complements. Acceptable gerundive nominal complements occur only if the head noun is deleted.






86
EXTA.POSITION FROM NP

Ross (1967) proposed a rule, which he called EXTRAPOSITION FROM NP, to account for the relationship of sentences like those in 3.145 to sentences like those in 3.146.

3.145a) A gun which I had cleaned went off.
b) He let the cats which were meowing out.
c) He expected someone who I was acquainted with
to show up.

3.146a) A gun went off which I had cleaned.
b) He let the cats out which were meowing.
c) He expected someone to show up who I was
acquainted with.
Ross stated the rule in the form given in 3.147.
Although Ross does not state the rule in the context of the cycle, the restrictions he gives on the forms of the variable Y indicate that the rule is cyclic.

3.147 EXTRAPOSITION FROM NP

X N[NP S]I Y
____ OPT
1 2 3 == 1, 3+2
EXTRAPOSITION FROM NP is not needed to account for the relationship of the sentences in 3.145 to the sentences in 3-146 if NP-PREPOSING is stated to apply to such noun-relative clause combinations in the same manner as with that-complements and their head nouns. Thus. NP-PREPOSING would





87

have preposed both the head noun and its relative clause in the sentences in 3.145, while it would have preposed only the head noun in the sentences in 3.146. The examples in

3.145 and 3.146 require a modification of the rule to allow preposing across the separable particle of a verb-particle predicate. This would, however, eliminate any need for a rule to postpose such separable particles across the object of the verb.

NP-PREPOSING and Pied Piping

In light of the behavior of NP-PREPOSING with that-complements and relative clauses, I will modify the rule so that it applies to the head noun of the structures in question, and allow the Pied Piping convention to account for the optional preposing of the that-complements and relative clauses with the head nouns. The Pied Piping convention is proposed by Ross (1967) to handle just such phenomena. The Pied Piping convention is as follows:15

The Pied Pioing Convention
Any transformation which is stated in such a way as to effect the reordering of some specified node NP, where this node is preceded and followed by variables in the struct-ural index of the rule, may apply to this NP or to any non-coordinate NP which dominates it, as long
as there are no occurrences of any coordinate node, nor
the node S, on the branch connecting the higher node
and the specified node.

Ross further mentions that Pied Piping is obligatory in some contexts. The distributional facts discussed above are accounted for if NP-PREPOSING applies to head nouns, with Pied Piping being obligatory for nominal complements and optional for that-complements (and relative clauses).





88

Conditions on Application

The conditions on the application of NP-PREPOSING that i have discussed above fall along two parameters. The contextual parameter opposes sentences and gerundive nominal complements to derived nominal complements. The noun phrase parameter opposes simple noun phrases to head nouns of complements. The interaction of these parameters is indicated in the chart in 3.148.

3.148 Conditions on the Application of NP-PREPOSING
(AGENT-PREPOSING has not applied)

in sentences and in derived nomigerundive nomi- nal complements nal complements

simple noun obligatory optional*
phrases

head nouns of obligatory blocked
complements
*blocked for specific predicates

In Chapter Four I will argue that when NP-PREPOSING

applies in the derivation of gerundive nominal complements, they are still sentences. Thus, NP-PREPOSING is obligatory


when it apDlies within structures dominated by an S. On the other hand, I will argue that when NP-PREPOSING applies in the derivation of derived nominal complements, they are already nominalized. Thus, NP-PREPOSING is optional or blocked when it applies within structures dominated by an ;I. I will have more to say about how this distinction is achieved in Chapter Four.






89

Emonds' POSSESSIVE Transformation

Emonds (1969:78-81) argues that NP-PREPOSING does not

apply in the derivation of "passive" derived nominal complements such as those in 3.149.

3.149a) the city's destruction by the enemy
b) Abby's acquittal by the jury

He argues that the possessive noun phrases the city's and Abby's are preposed by a rule he calls the POSSESSIVE Transformation. First he notes that NP-PREPOSING may prepose a noun phrase over a verb-particle predicate, as in

3.150, but claims that the POSSESSIVE Transformation never does, so that forms like that in 3.151 are blocked.

3.150) The strike was refered to briefly in the report.

3.151) *the strike's brief reference to in the report

Since many derived nominals idiosyncratically block

preposing of object noun phrases, the fact that verbs with lexical prepositions also block preposing is weak evidence for the separate existence of the POSSESSIVE Transformation.
A second reason Emonds gives for requiring as a separate rule the POSSESSIVE Transformation is that NP-PREPOSING is associated with the occurrence of the passive morpheme be-en, which never appears in "passive" derived nominal complements. As was mentioned in Chapter Two, derived nominal complements never have any auxiliaries. All that is needed is a simple rule or principle that insures that no auxiliaries are present in the surface structures of derived nominal







90

complements. If it is assumed that auxiliaries are present in deeply underlying structures, then they are deleted at some point in the derivation of derived nominal complements. If it is assumed that auxiliaries are inserted, then the insertion rules are blocked from applying to derived nominal complements. Thus, there will be no passive morpheme be-en in "passive" derived nominal complements.

A third reason cited by Jmonds is more complex. He

points out that the POSSESSIVE Transformation apparently can apply to noun phrases other than the one immediately following the derived nominal, whereas NP-PREPOSING (in sentences and gerundive nominal complements) applies only to noun phrases which immediately follow the predicate (or gerundive nominal). That is, he claims that the POSSESSIVE Transformation can apply to noun phrases which are not objects of the predicate. He cites the occurrence of prepositionless time adverbials preceding the derived nominal, as in 3.152, which, according to Emonds, represent derivations from the structures underlying the derived nominal complements in 3.153.

3.152a) last week's discussion of novels by the
librarian
b) this morning's speech to the nation by the
president

3.153a) the discussion of novels by the librarian
last week
b) the speech to the nation by the president
this morning






91
The application of NP-PkREPOSING in this way is not possible in sentences, so that we find the sentences in 3.154, but the sentences in 3.155 are not acceptable.

3.154a) The librarian discussed novels last week.
b) The president spoke to the nation this morning.
3.155a) *Last week was discussed novels by the
librarian,
b) *This morning was spoken to the nation by the
president.

But it cannot be the POSSESSIVE Transformation which

has moved the prepositionless time adverbials to a position preceding the derived nominals in 3.152, since the forms in

3.156 are not acceptable, indicating that the POSSESSIVE Transformation (or NP-PREPOSING) is blocked for discussion and speech in derived nominal complements.

3.156a) *novel's discussion by the librarian last week
b) *the nation's speech by the president this morning

All time adverbials are subject to fronting in sentences, as in 3.157.

3.15?a) Last week the librarian discussed novels.
b) This morning the president spoke to the nation.
The application of a rule of ADYVERB-FRONTING to the structures underlying the derived nominal complements in

3.153 (with subsequent addition of the suffix 's) will produce the derived nominal complements in 3.152.

Other time adverbials are also subject to ADVERB-FRONTING, but do not form determiners of derived nominals. The time adverbials with prepositions, however, are subject to deletion of the preposition, and then behave just like other




Full Text
122
included in its structural description. If Newmeyer's pro
posal is to stand, then AGENT-POSTPOSING (or AGENT-PREPOSING)
must precede DERIVED NOMINALIZATIQN. Newmeyer states that
there is no evidence that AGENT-POSTPOSING is cyclic, so
that it may be assumed to precede DERIVED NOMINALIZATIQN.
The question of whether or not AGENT-PREPOSING is cyc
lic is worth considering in detail, as it bears on the ques
tion of whether or not DERIVED NOKINALIZATION is cyclic.
The best evidence for the cyclic nature of a rule is the
existence of some RULE X which would apply after AGENT-PRE
POSING at some point in a derivation and before AGENT-PREPOS
ING somewhere else. In fact, there Is no such rule known in
the grammar of English, leading to Newmeyer's claim that
there is no evidence for the cyclic nature of AGENT-POSTPOS
ING (or, here, AGENT-PREPOSING). However, there is reason
to believe that there could be no such evidence, even though
AGENT-PREPOSING were a cyclic rule. If AGENT-PREPOSING were
the first rule of the cyclce, or if it were preceded in the
cycle only by rules with which it could never interact, then
it would be impossible to find evidence of the type discussed
above for the cyclic nature of AGENT-PREPOSING.
AGENT-PREPOSING makes an Agent noun phrase into the
subject of a sentence. No matter how the rule Is formulated
(even if it is formulated as AGENT-POSTPOSING), the Agent
noun phrase is dominated by the same S, with no intervening
S, both before and after the application of the rule. No
RULE X can be shown to apply before AGENT-PREPOSING, since


78
Like destroy, acquit, refuse and criticize, the predi
cates complain about, attest to and report on also form
agentless passives. Again, AGENT-PREPOSING does not apply
to underlying structures with unspecified agents. Thus,
NP-PREPOSING must apply in the derivation of sentences, as
in 3.112, but does not apply in the derivation of gerundive
and derived nominal complements, so that the forms in 3.113
and 3.114 are not acceptable.
3.112a) Peters (action of) insulting Tom was com
plained about.
b) (The fact of) Marys participation in the
crime was attested to.
c) (The event of) Alice's disappearance was
reported on.
3.113a) '"'Peters (action of) insulting Tom's being
complained about
b) *(the fact of) Mary's participation in the
crime's being attested to
c) *(the event of) Alice's disappearance's
being reported on
3.114a) "Peter's (action of) insulting Tom's complaint
about
b) *(the fact of) Mary's participation in the
crime's attestation to
c) *(the event of) Alice's disappearance's
report on
Failure of both AGENT-PREPOSING and NP-PREPOSING to
apply in the derivation of sentences and gerundive nominal
complements results in unacceptable forms, as in 3.115 and
3.116, but derived nominal complements to which neither
rule has applied, as in 3.117 are acceptable.
3.115a) *Was complained about Peter's (action of)
insulting Tom.
b) *Was attested to (the fact of) Mary's partici
pation in the crime.
c) *Was reported on (the event of) Alice's dis
appearance.


NOTES
^ Chomsky presumably also excludes agentive nominis
formed with -er, although he never mentions them except to
argue against the positing of abstract verbs underlying
nominis which otherwise have no corresponding predicates.
Agentive nominal complements share several characteristics
with derived nominal complements, and I will argue below
that agentive nominal complements are a special class of
derived nominal complements.
O
The derived nominal complements in 2.4a and 2.6a seem
quite strange to me, but 2.4a, at least, is acceptable in a
context such as the eagerness to -please shown by John, which,
however, seems to correspond to the sentence, an eagerness
to olease was shown by John, and not to the sentence in 2.1a.
3
'The preposition of is the unmarked form. Some verbs
have corresponding derived nominal complements with other
prepositions, e.g., attack;attack on.
4
Simple past tense seems to imply a single occurrence
of an event unless otherwise specified. Present perfect
tense seems to imply only at least one occurrence of an event.
^ The constraint on the occurrence of adjacent forms
with the -ing suffix, which blocks constructions like that
in 2.35c, is discussed in Ross (1972b) and Milsark (1972) .
45


11?
will also argue that the rule known as COMPLEMENTIZER PLACE
MENT is not compatible v/ith the theory of the cycle, and
that no such rule is necessary in any case for any comple
ments other than nominal complements. Although it will
appear that COMPLEMENTIZER PLACEMENT will still be needed to
properly account for nominal complements, I will argue that
it is better to have the rules of DERIVED NOMINALIZATION and
GERUNDIVE NOMINALIZATION apply on the cycle of the comple
ment, and to allow these rules to 'look up' into the next
higher sentence and include the predicate of the next higher
sentence in their environments than it is to retain COMPLE
MENTIZER PLACEMENT. Finally, I will argue that the specific
characteristics of derived nominal complements are not the
direct result of the application of DERIVED NOMINALIZATION,
but rather the result of a process that begins with the
application of DERIVED NOMINALIZATION.
Newmeyer's Proposal
The argument that appropriate ordering of the rules
will account for the facts of productivity was presented
independently in Albury (1971) and Newmeyer (1971).
Newmeyer argues that DERIVED NOMINALIZATION precedes all the
cyclic rules, and by changing the predicate of the embedded
sentence to a nominal, prevents the application of any
later rule which mentions the predicate in its structural
description.
Newmeyer's claim that DERIVED NOMINALIZATION precedes
all cyclic rules seems to imply that the rule is precyclic.


22
Pluralizaticn
Another characteristic of derived, nominis discussed by
Chomsky (I97O1I89) is that they may be pluralizad* while
gerundive nominal complements may not. Thus, we find the
derived nominal complements in 2.25* which correspond to the
sentences in 2.26* while the corresponding gerundive nominal
complements in 2.2? are not pluralizad.
2.25a) Johns three proofs of the theorem
b) John's repeated attempts to scale the wall
c) Agnews many attacks on the press
2.2oa) John proved the theorem three (times, ways, etc.).
b) John repeatedly attempted to scale the wall.
c) A.gnew attacked the press many times.
2.27a) John's proving the theorem three (times, ways, etc.)
b) John's repeatedly attempting to scale the wall
c) A.gnew's attacking the press many times
The sentences in 2,26 and the gerundive nominal comple
ments in 2.27 express repetitive events. The derived nomi
nal complements in 2.25 also express repetitive events, but
with plural nominis rather than adverbs of repetition,
Repetition is not always expressed, explicitly in sentences
such as those in 2.28, which therefore have two sets of cor
responding derived nominal complements, those in 2.29, which
express explicitly the singularity of the event, and those
in 2.30, which express explicitly the repetition of the
4
event.
2.28a) John has proved the theorem.
b) John has attempted to scale the wall.
c) A.gnew has attacked the press.
2.29a) John's proof of the theorem
b) John's attempt to scale the wall
c) Agnew's attack on the press


138
nominal complements are sometimes deleted by EQUI-NP-DELE
TION without creating infinitival complements. Deletion on
Agent-subject identity has occurred in 4.22, and on Dat
ive-subject identity in 4.23*^
4.22a) John anticipated winning the race.
b) Bill delighted in telling tall stories.
c) Mary enjoyed reading Russian novels.
4.23a) Mark imitated Jane putting on a girdle.
b) Alice watched Jerry running the race.
c) Mike worried about Mary walking through town.
If the gerundive nominal complements are formed before
EQUI-NP-DELETION deletes their subjects, there is no possibi
lity of the deletion leading to the creation of an infiniti
val complement. In any case, while infinitival complements
are always the result of such deletions, gerundive nominal
complements are not, as many retain their subjects- If ger
undive nominal complements are formed by the application of
GERUNDIVE NOMINALIZATION on the cycle of the embedded sen-
tence (the cycle before the application of EQUI-NP-DELETION),
the desired results are obtained.
Norainaliz atio n
I have presented arguments that the rule of COMPLEMENT
IZER PLACEMENT not only violates the principle of the cycle,
but is not necessary to account for that-complements and
infinitival complements. All of the rules needed to derive
that-complements and infinitival complements either apply on
the cycle above the complement, or are post-cyclic rules.
On the other hand, I have argued that the rules of
DERIVED NOMINALIZATION and GERUNDIVE NOMINALIZATION apply on


23
any previous application of RULE X would be to an S lower
than the S of the Agent noun phrase in question, and would
thus have no effect on the higher 3. Given the nature of
AGENT-PREPOSING and the assumption that it is one of the
earliest rules in the cycle, it is impossible to show that
it follows any other cyclic rule.
Since it would appear that it is impossible to discover
direct evidence of the cyclic nature of AGENT-PREPOSING, I
would like to consider some indirect evidence. AGENT-PRE
POSING can apply in more than one S in a single derivation,
as can be seen in the sentences in 4.4.
4.4a) John discovered that Mary hated him.
b)Jerry hoped that Mary would find out that Mack
really loved Suzie.
This in itself does not show that AGENT-PREPOSING is
cyclic, since a precyclic rule could conceivably prepose all
Agents within their respective sentences at one time. How
ever, AGENT-PREPOSING does not necessarily apply to all
Agents in a single derivation, so that passive sentences may
be embedded in active sentences and vice versa, as in 4.5*
4.5a) John hoped that the package would be handled
carefully by the Post Office.
b) Mary claimed that she was being followed by a
strange man.
c) That Bob had betrayed Tom was not known to the
other members of the cell.
d) That the Earth is not a perfect sphere was demon
strated by the first artificial satellite.
For a precyclic rule to produce such structures, it
would need to not only recognize Agents within individual
embedded sentences, but it would also have to recognize the


NOTES
A source of possible confusion is the fact that both
the nouns formed by adding a suffix to a verb stem, and the
complements in which such nouns occur, can be called nomi
nis. To avoid confusion, I will reserve the term nominal
for the nouns so formed, and refer to the complements they
occur in as nominal complements. I will use nominalization
to refer to the process by which nominis are created. I
will have more to say about the choice of the term comple
ment below.
2
The possessive agent on action is always identical
with the deleted agent of the complement of action. Menzel
argues that the agent cannot have been raised from the com
plement. Among suggestions of possible sources for the
possessive agent, he states that "in a grammar based on
the transformationalist position...the agent on the head
noun action would be derived from an underlying sentence
embedded on the noun action, or more probably on the verb
act." Ross (1972a) argues that sentences with verbs of act
ion (which are the only sentences that can be embedded as
complements of the head noun action) are embedded on the
verb do in underlying structures, with the agent of do iden
tical to the agent of the embedded sentence. The substitu-
13


6?
3*69c)
d)
the enemy's destroying the city
the jury's acquitting Abby
3-70a)
b)
John's refusal of the offer
John's criticism of the book
c)
d)
the enemy's destruction of the city
the jury's acquittal of Abby
The sentences in 3-68 also have the corresponding pas
sive sentences in 3-71*
3*71a)
b)
c)
d)
The offer was refused by John.
The book was criticized by John.
The city was destroyed by the enemy.
Abby was acquitted by the jury.
The passive sentences in 3-71 have the corresponding
gerundive nominal complements in 372, but only the sen
tences in 3-71c and d have the corresponding derived nominal
complements in 3-73c and d, while the strings in 3* 73a and b,
which seem to correspond to the sentences in 3-71a and b,
are not acceptable.
3 72a)
b)
c)
d)
the offer's being refused by John
the book's being criticized by John
the city's being destroyed by the enemy
Abby's being acquitted by the jury
3.73a)
b)
c)
d)
*the offer's refusal by John
*the book's criticism by John
the city's destruction by the enemy
Abby's acquittal by the jury
The derived nominal complements in 3*74, all of which
are acceptable, are related to the derived nominal comple
ments in 3*70 and 3*73* but do not correspond exactly to any
sentences. These derived nominal complements would corres
pond to predicate-initial underlying structures.
3*7^a)
b)
c)
d)
the refusal of the offer by John
the criticism of the book by John
the destruction of the city by the enemy
the acquittal of Abby by the jury


61
that the pattern of occurrence of derived nominal comple
ments casts doubt on the validity of IT-EXTRAPOSITION as pro
posed by Rosenbaum. I will then consider other evidence
counter to his analysis* and argue that there is no rule of
IT-EXTRAPOSITION in English grammar. In a later section
(cf. NP-PREPQ5ING) I will present an alternate analysis of
the relationship between the sentences in 3-45 and 3-46.
With Subjects of Predicate Ad.iectives
We have seen above that sentences like those in 3*49
have acceptable corresponding gerundive nominal.complements,
as in 3*50, and derived nominal complements, as in 3ol
3.49a) It is certain that John will win the prize.
b) It is difficult to learn algebra.
c) It is fun to take this test.
3*50s.) it(s) being certain that John will win the prize
b) it(s) being difficult to learn algebra
c) it(s) being fun to take this test
3*51ai) the certainty that John will win'-the prize
ii) the certainty of John's winning the prize
b) the difficulty of learning algebra
c) the fun of taking this test
According to Chomsky's (1970) analysis, the sentences
in 3.49 are derived from the phrase markers underlying the
sentences in 3*52 by the rule of IT-EXTRAPOSITION. The sen
tences in 3.52 do not have acceptable corresponding gerund
ive nominal complements or derived nominal complements, as
is shown by the unacceptability of the expected forms in
3.53 and 3-54.
3-52ai) That John will win the prize is possible,
ii) John's winning the prize is certain.
bi) To learn algebra is difficult,
ii) Learning algebra is difficult.


ive form himself in 3*25a, the accusative them in 3.25b and
the fact that Dick is the derived subject in a passive sen
tence in 3*25c. Reflexivization is restricted to clause
mates in English, pronouns take the accusative form in
object position and an underlying object becomes the surface
subject in passive sentences.
The sentences in 3-25* which have undergone RAIS-
ING-TO-OBJECT, have the acceptable corresponding gerundive
nominal complements in 3-2?, but no acceptable correspond
ing derived nominal complements, the forms in 3-28 being
unacceptable.
3.27a) John's believing himself to be heroic
b) Bill's expecting them to be here soon
c) Dick's being believed by everyone to have been
involved
3.28a) -'John's belief of himself to be heroic
b) *3ill's expectation of them to be here soon
c) *Dicks belief by everyone to have been involved
-
The sentences in 3-26, which have not undergone RAIS-
ING-TG-OBJECT, have the acceptable corresponding gerundive
nominal complements in 3*29 and derived nominal complements
in 330.
3-29a) John's believing that he is heroic
b) Bill's expecting that they will be here soon
c) everyone's believing that Dick was involved
3.30a) John's belief that he is heroic
b) Bill's expectation that they will be here soon
c) everyone's belief that Dick was involved
RAISING-TO-SUBJECT
The second movement is from subject of an embedded sen
tence to subject of the next higher sentence, as has occurred


70
I will claim here that there is a rule of AGSNT-PR3PQ3-
ING (the reverse of Chomsky's AGENT-POSTPOSING) which
applies to predicate-initial underlying structures to yield
active sentences and nominal complements such as those in
3.68, 3-69 and 3*70. If AGENT-PREPOSING does not apply,
then the resulting obligatory application of NP-PREPOSING
gives the passive sentences and gerundive nominal comple
ments in 3*71 and 3-72. If AGENT-PREPOSING does not apply
in the derivation of derived nominal complements, then
NP-PREPOSING applies optionally with some underlying verbs,
giving the pairs of derived nominal complements in 3-73c
and d and 3-7^c and d, and not at all with other verbs,
giving the derived nominal complements in 3*7^a and b.
By Chomsky's analysis, passive sentences are derived
from structures underlying active sentences by the applica
tion of two rules, both of which must apply. Ey my analysis,
all sentences are derived from underlying predicate-initial
structures, with active sentences derived by one rule and
passive sentences by another rule.
NP-PREPOSING
With Simle Noun Phrases
The application of AGSNT-PRBPOSING to underlying pre
dicate-initial structures with agents yields active sen
tences like those in 3*79 and their corresponding gerundive
nominal complements in 3-80 and derived nominal complements
in 3.81.


11
Assuming predicate-initial order in structures such as
1.13 is a way of expressing the pivotal role of the predi
cate in most syntactic rules. An alternative statement
could he that noun phrases are assigned linear order in rela
tion to the predicate by syntactic rules.
McCawley (1970) explicitly assumes that underlying sen
tences have the order Verb-Subject-Object (VSO) before the
application of any transformations, but does not present any
evidence for preferring that order over one of Verb-Object
-Subject (VOS), I believe that the evidence points to a VOS
order, if there is any order at all to the noun phrases to
£
the right of the predicate in deeply underlying structures.
In the rest of this work, the relative order of noun phrases
in trees representing deeply underlying structures will be
arbitrary, and represent no claims as to the actual order.
Outline
This study is concerned primarily with the internal
structure of nominal complements, by which I mean the rela
tionships of the constituents of nominal complements to each
other, as opposed to the relationships of nominal comple
ments to items which are not constituents of the said nomi
nal complements, I compare such internal structures to
those of sentences and of other clausal complements, I
point out that the differences between such structures are
regular, and easily accounted for by independently motivated
syntactic rules, given certain conditions on the cycle in
which the rules apply.


53
3.36a) *its certainty that John will win the prize
b) "its likelihood that Bill will be drafted
c) "its appearance that Jerry opened the door
3.37a) the certainty that John will win the prize
b) the likelihood that Bill will be drafted
c) the appearance that Jerry opened the door
I will argue below (cf. Dummy Subject Insertion) that
it in sentences like those in 3*32 is inserted, and not pre
sent in underlying structures. The unacceptability of the
forms in 3*36, and the acceptability of the forms in 3*37 is
simply explained by blocking IT-INSERTION in derived nominal
complements.
TOUGH-MOVEMENT
The third RAISING movement is from object of an embed
ded sentence to subject of the next higher sentence, as has
occurred in the derivation of the sentences in 3-33, which
are derived from the same underlying structures as the sen
tences in 3*39.
3.33a) John is easy to please.
b) Algebra is difficult to learn,
c) This test is fun to take.
3.39a) It is easy to please John.
b) It is difficult to learn algebra.
c) It is fun to take this test.
The sentences in 3'3Sj which have undergone TOUGH-MOVE
MENT, have the acceptable corresponding gerundive nominal
complements in 3-^0, but no acceptable corresponding derived
nominal complements, the forms in 3-^1 being unacceptable. .
3>40a) John's being easy to please
b) algebras being difficult to learn
c) this test's being fun to take


100
3.1?a) John condescended to speak to Mary,
b) Mike quickly learned to analyze sentences.
c) Bill expected to leave early.
3.177a) John's condscending to speak to Mary
b) Mike's quickly learning to analyze sentences
c) Bill's expecting to leave early
3.178a) *John's condescension to speak to Mary
b) *Mike's quick learning to analyze sentences
c) ^Bill's expectation to leave early
That the unacceptability of the forms in 3.178 is con
nected with the application of EQUI-NP-DELETION is indicated
by the fact that the sentences in 3.179 with the same basic
predicates as in 3.176, have the corresponding gerundive
nominal complements in 3.180 and derived nominal complements
in 3.181.
3.179a) John was condescending towards Mary.
b) Mike quickly learned the multiplication tables.
c) Bill expected that Alice would leave,
3.180a) John's being condescending towards Alice
b) Mike's quickly learning the multiplication
tables
c) Bill's expecting that Alice would leave
3.181a) John's condescension towards Mary
b) Mike's quick learning of the multiplication
tables
c) Bills expectation that Alice would leave
Other predicates allow the application of EQUI-NP-DELE
TION in the derivation of derived nominal complements. The
sentences in 3.182, which have undergone EQUI-NP-DELETION in
their derivations, have the corresponding gerundive nominal
complements in 3.183 and derived nominal complements in 3.184.
3.182a)
b)
c)
d)
John intended to leave early.
Mike desired to leave early.
Bill refused to leave early.
Jerry attempted to leave early.


7
action, etc.) may also be taken as names of the types of
complements a verb allows. Thus, Johns coming can be both
a fact and an event, and we get the sets of possible sen
tences in 1.5 and 1.6.
1.5a) John's coming occurred at ten o'clock.
b) The event of John's coming occurred at ten
o'clock.
c) The event occurred at ten o'clock.
1.6a) John's coming is surprising.
b) The fact of John's coming is surprising.
c) The fact is surprising.
The noun action can only take subjectless gerundive
nominal complements, while other nouns, such as event,
require a gerundive nominal complement with a subject
(Menzel, 1969582-83). At the same time, of the nouns which
take gerundive nominal complements, only action allows a
2
preposed possessive noun. Thus, we find the pattern of
acceptability shown in 1.7.
1.7a)
the
action of -j
b)
the
event of -f!
c)
, faction
J ohn
s I* event
Most of the nouns on Menzels list also take derived
nominal complements, and any derived nominal complement,
unless it is a complement of a verb of belief, or of say,
declare, or claim, can be a complement of at least one of
those head nouns, as is indicated by the examples in 1.8.
1.8a) The fact of Johns departure cannot be contested,
b) The event of the destruction of Jerusalem
occurred in 67 AD.
c) The state of Mike's awareness is unpredictable.


54
Similarly, Chomsky (1970:191) states that certain, with
the meaning used in 3.lib, John is certain that Bill will
win the prize, also has a subcategorization feature which
allows certain in this meaning to take a sentential comple
ment in a structure like ohn]^p[~ypfybe certain^j^Bill
will win the prize~)^,~|^p~j^, from which 3.11b is derived.
Thus, the derived nominal certainty, corresponding to this
meaning of certain, has the same subcategorization feature,
and can appear in derived nominal complements with a senten
tial complement, as in 3.15b, John's certainty that Bill
will win the -prize.
Like the lexical entry for easy, the lexical entry for
certain in the meaning used in 3.11a John is certain to win
the prize, has no subcategorization feature allowing certain
in this meaning to take a sentential complement. Chomsky
says that certain in this meaning appears in bse phrase
markers as an adjective predicated of propositions as sub
ject, as in for John to win the prize is certain. Applying
the same arguments Chomsky used with easy, sentences like
it is certain for John to win the prize are derived by
PT-EXTKAPQSITIQN, and sentences like John is certain to win
the prize are derived by RAISING-TO-SUBJECT from extraposed
5
sentences,
According to the Lexical Hypothesis, derived nominis
occur in nominal structures corresponding to the base phrase
markers of sentences. Such nominal structures are not sub
ject to most rules which apply to sentences. The derived


129
4.12a) That John won surprises me.
b) For John to win would surprise me.
c) John's winning surprises me.
4.13)
S
0
Pred
NP
NP
(would) surprise
S
1
I/me
Pred
NP
win J ohn
There is good reason for having COMPLEMENTIZER PLACE
MENT apply on the cycle of SQ in 4.13. This allows COMPLE
MENTIZER PLACEMENT to apply to the higher sentence which
includes the predicate which governs the type of complement
the embedded sentence may become, and it prevents COMPLE
MENTIZER PLACEMENT from applying to underlying sentences
which are not embedded.
Bresnan (1970:299-300) argues that this formulation of
Xr
COMPLEMENTIZER PLACEMENT is peculiar in that "it violates
an otherwise well-motivated universal stated by Chomsky
(1965l46), namely, that while transformations may remove
material from embedded sentences, no transformation can
insert morphological material into 'lower* sentences." I
will now examine this peculiarity of COMPLEMENTIZER PLACE
MENT in more detail.
To facilitate this discussion I would like to introduce
the terms proper constituent and embedded constituent. I
will define a proper constituent of a sentence as a con
stituent dominated by S but not dominated by an S which
J n J m


37
the Noun Phrase-Verb-Noun Phrase structure of a sentence
like 2.73b.
2.73^) John felt sad.
b) John felt sadness.
There is a derived nominal complement which corresponds
to 2.73b, John's feeling of sadness. Peter Menzel (personal
communication) has pointed out to me that feeling is the nom
inal of a derived nominal complement, and thus would be fol
lowed by a preposition, presumably of, so that the proper
question is why we do not get *the feeling of sad. The pre
sence of the preposition means that any complement of feel
ing within a derived nominal complement must be nominalized.
Hence, derived nominalization can apply to an embedded sen
tence like that in 2.73b, but not to one like that in 2.73a,
unless sad is nominalized to sadness as part of the nominal-
isation of feel.
The nominal complement in 2.71c, ^the arguing about
money, would not be expected to occur because the regular
derived nominal complement corresponding to argue about
money is the argument about money.
Chomsky (19?0s2l4) gives as unacceptable the example.:in
2.71d, the leaving. At least some speakers do accept nomi
nal complements like John's hurried leaving (cf, John left
hurriedly), and I am told that such forms have appeared in
print. My own judgment agrees with that of Chomsky, however.
The dialects that accept forms like John's hurried leaving
present no problem to my analysis, so I will turn to those
dialects that do not accept such forms.


92
prepositionless time adverbials. Time adverbials with the
-ly suffix form pre-ncminal adjectives just like other -ly
adverbs in derived nominal complements, as was discussed in
Chapter Two. Thus, the sentences in 3.158 have the corres
ponding derived nominal complements in 3.159.
3.158a) Recently the librarian discussed novels,
b) Daily the president speaks to the nation.
3.159a) the librarian's recent discussion of novels
b) the president's daily speech to the nation
Emonds choice of the name POSSESSIVE for the rule which
preposes objects in derived nominal complements is unfortun
ate. Last week and this morning in 3.152 were preposed by
ADVERB-FRONTING, and the librarian and the -president in
3.159 were preposed by AGENT-PRSPOSING, yet they also form
possessive determiners. The formation of a possessive is
obviously distinct from any preposing rule, and operates on
whatever noun phrase is in pre-nominal position^ at some late
point in the cycle.
The last reason Emonds cites for distinguishing the POS
SESSIVE Transformation from NP-PREPOSING is that the condi
tions for preposing are not the same in derived nominal com
plements as they are in sentences and gerundive nominal com
plements, as was indicated in 3.1^8. Since those conditions
do differ, there seems to be some merit in the proposal for
two rules. The two rules would have so much in common, how
ever, that to say that they are unrelated would miss signifi
cant generalizations. I therefore reject Emonds claim, and
retain my earlier analysis of one rule of NP-PREPOSING with


141
predicate applies to some derived nominis and all gerundive
nominis.



52
feature indicating.that it can take a sentential complement,
as in 3>l6, derived from underlying structures something
like those in 3-17*
3.16a) John is eager to please.
b) John is eager for us to please.
3.17a) John is eager f^John please SOMEONE"];,
b) John is eager f,-,we please SOMEONE ~L
Chomsky says that no further comment is necessary to
account for the acceptability of the derived nominal comple
ments in 3*18.
3.18a) John's eagerness to please
b) John's eagerness for us to please
According to the Lexical Hypothesis, the lexical entry
for eager is also the entry for eagerness and the strict
subcategorization feature applies to adjective and nominal
alike. Thus, eager can appear in a construction of the form
Noun Phrase-Predicate-Sentential Complement, and eagerness
can appear in a construction of the form Possessive
Noun-Nominal-3 entential Complement.
Chomsky claims that, on the other hand, there is no
such subcategorization feature in the lexical entry for easy
and that there are no base phrase markers of the form
easy-5entential Complement. Chomsky (1970tl91) says that
easv appears in base phrase markers as an adjective predi
cated of propositions as subject, as in the sentences in
3.19, with sentences like those in 3-20 derived by IT-EXTRA-
POSITION, and with sentences like those in 3-21 derived in
- 2
turn from extraposed sentences by TO LJ OH -MO v .MEN T.


131
which move constituents down into embedded sentences. These
rules move constituents of abstract higher sentences down
into the surface structure sentence, such as NEGATIVE-M0V2-
MENT (cf. R. Lakoff, 1968). Such rules appear to violate
Chomskys (1965) universal, but are outside the scope of
this discussion.
Another subdivision includes rules which delete embed
ded constituents, such as EQUI-NP-DELETION, which deletes an
embedded constituent when it is identical to a specified
proper constituent.
The last subdivision consists of COMPLEMENTIZER PLACE
MENT. This rule inserts complementizers into embedded sen
tences, but has no effect on any proper constituent. Its
only connection with any proper constituent is that the
predicate of the higher sentence governs the type of comple-
4
ment formed.
The RAISING rules, EQUI-NP-DELETION and COMPLEMENTIZER
PLACEMENT are all governed by the predicate of the higher
sentence. RAISING converts an embedded constituent into a
proper constituent, EQUI-NP-DELETION requires identity
between an embedded constituent and a proper constituent,
but Rosenbaums (1967) COMPLEMENTIZER PLACEMENT has no con
nection with the higher sentence that it applies on other
than that it is governed by the higher predicate.
There is only one justification for having a rule like
COMPLEMENTIZER PLACEMENT apply in the cycle of the higher
sentence in which the complement is embedded, and that is to


53
3.19a) To please John is easy.
b) For us to please John is easy.
3.20a) It is easy to please John.
b) It is easy for us to please John.
3-21a) John is easy to please.
b) John is easy for us to please.
. 4
3.22a) John is eager to please.
b) John is eager for us to please.
Even though we get sentences like those in 3*21, which
in surface form exactly parallel those in 3*22, Chomsky
argues that just as easy cannot be introduced into struc
tures of the form Subject-Predicate-Sentential Complement,
easiness cannot be introduced into structures of the form
Possessive Noun-Nominal-Sentential Complement, thus prevent
ing the formation of derived nominal complements like those
in 3-23-
3.23a) ''John's easiness to please
b) ^John's easiness for us to please
Chomsky assumes that gerundive nominal complements are
transformationally derived from structures which are in turn
derived from base structures. That is, GERUNDIVE NOMINALI-
ZATION applies to embedded sentences which have already been
subject to all or almost all cyclical rules. It is there
fore possible to have gerundive nominal complements like
those in 3-24.
3*24a) John's being easy to please
b) John's being easy for us to please
Chomsky is assuming that transformations apply to struc
tures dominated by an S but not to structures dominated by
an NP even though the two types of structures may be com
posed of the same lexical items.


104
The first reading results from the derivation of 3-193a
from the underlying structure illustrated in 3-197*
CAUSE John
stories V NP NP
tell stories John
The other reading results from the derivation of 3-193a
20
from the underlying structure illustrated in 3-198.
stories V NP NP
tell stories John
The second semantic reading is also associated with the
sentence in 3-199-
3*199) John's stories amused the children.
The structure of 3^ in 3-198 is the same as that of
in 3-197* This means that the structure underlying the
second semantic reading of 3-193a is embedded in the struc
ture underlying the first semantic reading of that sentence.
The derivation of the sentence in 3-193a from the underlying
structure in 3-19? involves the reduction of stories
[pJohn told stories^p^.Tp to John's stories, the pronominal-


55
nominal certainty corresponding to the second meaning of
certain^ like easiness, cannot occur in nominal structures
corresponding to derived phrase markers,
Chomsky has argued that the lack of productivity of
derived nominal complements is explained by the Lexical
Hypothesis, Under that hypothesis, *John's easiness to
please is not acceptable because the lexical entry underly
ing easy and easiness does not have a subcategorization
feature allowing sentential complements. On the other hand,
John's eagerness to please is acceptable because the lexical
entry underlying eager and eagerness has a subcategorization
feature allowing sentential complements.
RAISING Rules
RAI5ING-T0-0BJECT
As was indicated in Note 2 of this chapter, RAISING is
a general term for three different movement transformations.
The first movement I will discuss here is from subject of an
embedded sentence to object of the next highest sentence, as
has occurred in the derivation of the sentences in 3.25
which are derived from the same underlying structures as the
sentences in 3.26.^
3,25a) John believes himself to be heroic.
b) Bill expects them to be here soon.
c) Dick is believed to have been involved by everyone.
3.26a) John believes that he is heroic.
b) Bill expects that they will be here soon.
c) Everyone believes that Dick was involved,
That the nouns in question have been raised to object
position in the higher sentences is indicated by the reflex-


33
In my dialect* leave (=depart), try (=attempt) (cf.
2.71b) _be (-exist, occur, be present) have (=possess)
and live (-reside) (and other verbs) are alike In that they
have no corresponding nominal forms which appear in derived
nominal complememts., (I am refering here to be and have
when they are main surface verbs, not auxiliaries.) The
existence of verbs such as these poses a serious challenge
to the transformationalist position. It would appear that
the rule (or rules) producing derived nominal complements is
blocked from applying to underlying structures with certain
verbs as their predicates. It is Chomskys claim that this
is one of a number of facts supporting the Lexical Hypothe
sis over the transformationalist position.
The verbs mentioned above (which are all of Germanic
origin) occur in sentences like those in 2,7^.
2.7^a) John tried vainly to win the race,"
b) John left hurriedly on the bus.
c) There Is a God.
d) There was a riot yesterday.
e) There is some wine in the bottle.
f) John has a car.
g) Bill usually lives in a hotel.
The Latinate synonyms of those verbs (for meanings
inherent in the sentences in 2,70 occur in sentences like
those in 2.75.
2.75a) John attempted vainly to win the race.
b) John departed hurriedly on the bus,
c) There exists a God.
d) There occurred a riot yesterday.
e) There is some wine present in the bottle.
f) John possesses a car.
g) Bill usually resides in a hotel.


26
productivity as well, so that we cannot get structures like
those in 2.40. Chomsky (1970:215) finally states that
"there is an artificiality to the whole construction that
makes it quite resistant to systematic investigation."
2.40a) *the feeling sad
b) *the trying to win
c) *the arguing about money
d) *the leaving
I would like to put aside for now the problems involved
in the forms in 2.39a and b and 2.40, and consider in detail
the complement in 2.39c, the growing of tomatoes. Part of
the data cited by Chomsky (1970:192) to support his claim
that the Lexical Hypothesis provides the best explanation
for the origin of derived nominis involves transitive verbs
derived from intransitive verbs by the rule of CAUSATIVE
FORMATION (cf. G. Lakoff, 1970), such as grow, as in John
grows tomatoes, derived from grow, as in tomatoes grow. At
some point in its derivation, the underlying structure of
John grows tomatoes can be represented by the tree in 2.41,
The structure underlying tomatoes grow is embedded as 3-^ in
the tree in 2,41
2.41)
V
NP
NP
CADS I
V
NP
grow tomatoes John
Chomsky points out that there is a derived nominal com
plement, the growth of tomatoes, which corresponds to the


CHAPTER TWO
DEFINING DERIVED NOMINAL COMPLEMENTS
First Definitions
Chomsky (1970) does not explicitly define derived nomi
nal complements, but a definition may be extracted from the
various examples he cites in discussing the characteristics
of derived nominis. It Is obvious throughout that Chomsky
intends the class of derived nominis to include only those
nominis which are morphologically derived from a verb by
means other than the suffix -ing.~ He explicitly excludes
gerunds (those nominis derived by adding -ing to a verb)
from the set of derived nominis. Chomsky (1970j214) notes
certain similarities to derived nominal complements shown by
structures called "action nominis" in Lees (1963) and
Fraser (1970), which have nominis formed with -ing, but he
claims a number of differences which would seem to preclude
treating them as derived nominal complements, I will have
more to say about these forms below, Chomsky thus limits
the term derived nominal to those nominis formed by adding
a derivational suffix to a verb, such as refusal (from
x~efuse) and marriage (from marry) ; nominis identical
(except for stress) to verbs, such as search (from search) and
export (from exp5rt); and nominis phonologically modified
from a verb, such as deed (from did).
15


90
complements. If it is assumed that auxiliaries are present
in deeply underlying structures, then they are deleted at
some point in the derivation of derived nominal complements.
If it is assumed that auxiliaries are inserted, then the
insertion rules are blocked from applying to derived nominal
complements. Thus, there will be no passive morpheme be-en
in "passive" derived nominal complements.
A third reason cited by Emonds is more complex. He
points out that the POSSESSIVE Transformation apparently can
apply to noun phrases other than the one immediately follow
ing the derived nominal, whereas NP-PREPOSING (in sentences
and gerundive nominal complements) applies only to noun
phrases which immediately follow the predicate (or gerundive
nominal). That is, he claims that the POSSESSIVE Transforma
tion can apply to noun phrases which are not objects of the
predicate. He cites the occurrence of prepositionless time
adverbials preceding the derived nominal, as in 3*152, which
according to Emonds, represent derivations from the struc
tures underlying the derived nominal complements in 3*153*
3152a) last week's discussion of novels by the
librarian
b) this morning's speech to the nation by the
president
3*153a) the discussion of novels by the librarian
last week
b) the speech to the nation by the president
this morning


NOMINAIS AS COMPLEMENTS
By
DONALD HERBERT ALBURY
DISSERT
PARTIAL
AT ION PRESENTED T-J THE GRADUATE COUNCIL
OF THE UNIVERSITY OF FLORIDA
, FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA


18
ponding derived nominal complements in 2.9* The fact that
strings like those in 2.10 are unacceptable shows that ger
undive nominal complements do not take pronominal adjectives.
2.7a) John is overwhelmingly eager to please.
b) John abruptly refused the offer.
c) John criticised the book unmercifully.
2.8a) John's being overwhelmingly eager to please
b) John's abruptly refusing the offer
c) Johns criticizing the book unmercifully
2.9a) John's overwhelming eagerness to please
b) John's abrupt refusal of the offer
c) John's unmerciful criticism of the book
2.10a) "John's being overwhelming eager to please
b) ^John's abrupt refusing the offer
c) ^John's unmerciful criticizing the book
The prenominal adjectives in some derived nominal com
plements, such as those in 2.11, do not have corresponding
adverbs in gerundive nominal complements, as in 2.12, or
sentences, as in 2.13.
2.11a) John's troublesome eagerness to please
b) John's untimely refusal of the offer
c) John's unmotivated criticism of the book
2.12a) ^Johns being troublesomely eager to please
b) *John,s untimely refusing the offer
c) ^John's criticizing the book unmotivatedly
2.13a) *John is troublesomely eager to please.
b) -John untimely refused the offer.
c) *John criticized the book unmotivatedly.
That the acceptability of the forms in 2.11, and the
unacceptability of the forms in 2.12 and 2.13 is due to
these adverbs alone can be seen by inspecting the forms in
2-1, 2.2, 2.7 and 2.8. The adjectives which appear in pre
nominal position in the derived nominal complements in 2*11
can also be predicated of such complements, as in 2.14.


97
the predicate because NP-PREPOSING has failed to apply, a
surrogate subject must be supplied. This is done by
THEfLi;-INSERTION if the predicate belongs to the class of
existential predicates, and by If-INSERTION if the predicate
has a sentential complement, or is a weather predicate. In
a derived nominal complement, no surrogate subject is sup
plied by the grammar.
eQ.U I-NP -DELETION
In this section I will discuss another problem which is
not raised by Chomsky (1970): the lack of derived nominal
complements in which EQUI-NP-DELETION has applied.
Stockwell et al. (1973*553ff) present an analysis
which has EQUI-NP-DELETION apply when there is an Agent (of
the higher sentence)-subject (of the embedded sentence) iden
tity or a Dative (of the higher sentence)-subject (of the
embedded sentence) identity, with the Dative-subject condi-
19
tion taking precedence over the Agent-subject condition. 7
Stockwell et al. state that require takes an optional Dative
noun phrase with a sentential complement, with the applica
tion of EQUI-NP-DELETION being optional if a Dative noun
phrase is present and identical to the subject of the embed
ded sentencej that command takes an optional Dative noun
phrase with the application of EQUI-NP-DELETION being obli
gatory if a Dative noun phrase is present and identical to
the subject of the embedded sentence? and that force takes
an obligatory Dative noun phrase with obligatory application
of EQUI-NP-DELETION if the Dative noun phrase is identical


87
have preposed both the head noun and its relative clause in
the sentences in 3.1^5* while it would have preposed only
the head noun in the sentences in 3.1^6. The examples in
3.1^5 and 3.1^6 require a modification of the rule to
allow preposing across the separable particle of a verb-par
ticle predicate. This would, however, eliminate any need
for a rule to postpose such separable particles across the
object of the verb.
NP-PREPOSING and Pied Piping
In light of the behavior of NP-PREPOSING with that-com-
plements and relative clauses, I will modify the rule so that
it applies to the head noun of the structures in question,
and allow the Pied Piping convention to account for the
optional preposing of the that-complaments and relative
clauses with the head nouns. The Pied Piping convention is
proposed by Ross (196?) to handle just such phenomena. The
Pied Piping convention is as follows;^
The Pied Piping Convention
Any transformation which is stated in such a way as to
effect the reordering of some specified node NP, where
this node is preceded and followed by variables in the
structural index of the rule, may apply to this NP or
to any non-coordinate NP which dominates it, as long
as there are no occurrences of any coordinate node, nor
the node S, on the branch connecting the higher node
and the specified node.
Ross further mentions that Pied Piping is obligatory in
some contexts. The distributional facts discussed above are
accounted for if NP-PREPOSING applies to head nouns, with
Pied Piping being obligatory for nominal complements and
optional for that-complaments (and relative clauses).


27
sentence tomatoes grow, but not to the sentence John grows
tomatoes. as would be expected from the apparently parallel
example of the derived nominal complement, the rejection of
the offer, which corresponds to the verb phrase reject the
offer- In other words, the rejection of the offer is an
object nominal complement, while the growth of tomatoes is a
subject nominal complement, and Chomsky takes the fact that
there is no interpretation of the growth of tomatoes as an
object nominal complement as proof that there is no derived
nominal complement corresponding to John grows tomatoes. In
terms of the Lexical Hypothesis, this is to be expected
because John grows tomatoes involves a derivation from an
underlying structure which includes tomatoes grow as an
embedded sentence, as in the tree in 2.41. This is consis
tent with the claim that derived nominis are associated
lexically with the underlying verb. Since grow occurs in
the lexicon only in the intransitive, noncausative sense,
the transitive sense being derived transformationally by the
rule of CAUSATIVE FORMATION, only the intransitive sense can
have a corresponding derived nominal, according to the
Lexical Hypothesis.
Some Counterexamples to Chomsky *s .Analysis
Smith (1972) has pointed out that there are many excep
tions to Chomskys claims concerning the occurrence of
derived nominis corresponding to verbs derived by the rule
of CAUSATIVE FORMATION. The verb convert occurs as both
transitive and intransitive, and both forms have associated


50
because derived nominal complements are noun phrases in deep
structure, not embedded propositions. He claims that
derived nominal complements correspond only to deep struc
ture phrase markers and a few transforms. I will argue that
most rules are blocked from applying to derived nominal com
plements, or have different conditions on their applicabili
ty to derived nominal complements, because they are nominal-
ized embedded propositions. I will also show that many
rules can apply in the derivation of gerundive nominal com
plements that cannot apply in the derivation of derived nom
inal complements.
I will show in detail how the application of a number
of rules is blocked or modified in the derivation of derived
nominal complements. I will discuss the effect of adopting
a predicate-initial analysis on the formulation of rules,
and certain consequent simplifications of the rules, parti
cularly with regard to derived nominal complements. I will
argue that certain facts of derived nominal complements
raise problems for the currently accepted formulation of the
rule of IT-EXTRAPOSITION, and that with a predicate-initial
analysis, no rule of IT-EXTRAPOSITION is needed in a grammar
of English. Throughout this chapter I will show that when
ever a sentence does not have an acceptable corresponding
derived nominal complement, it is because the sentence has
undergone a rule in its derivation which is blocked from
applying to derived nominal complements.^


I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
Assistant Professor of Spanish
and Portuguese and Lin
guistics
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
Associate Professor of Speech
and Linguistics
This dissertation was submitted to the Graduate Faculty
of the Program in Linguistics in the College of Arts and
Sciences and to the Graduate Council, and was accepted as
partial fulfillment of the requirements for the degree of
Doctor of Philosophy.
March, 1974
Dean, Graduate School


128
4.9a) John's disruption of the meeting created panic,
b) John's disrupting the meeting created panic.
4.10) SQ
disrupt John meeting
Pred NP NP
b) NP
disruption John meeting
disrupt John meeting
The description of the operation of nominalization
given in the above paragraph is not in accord with the usual
concept of the nature of rules of complementation found in
the literature. Rosenbaum (1967) presents a rule he calls
o-
COMPLEMENTIZER PLACEMENT which is intended to account for
the placement of that in that-complements, for and to in
infinitival complements, and the -'s. and -Ing suffixes in
gerundive nominal complements. With that-complements, COM
PLEMENTIZER PLACEMENT simply adjoins that to an embedded
sentence. With infinitival or gerundive nominal complements,
COMPLEMENTIZER PLACEMENT inserts for and to or -s and -ing
into the embedded sentence. The type of complement formed
is governed by the predicate In the higher sentence. In the
derivation of the sentences in 4.12 from the underlying
structure in 4.13, COMPLEMENTIZER PLACEMENT would apply on
the cycle of In each case.


BIOGRAPHICAL SKETCH
Donald Herbert Albury was born July 22, 1943, at Miami,
Florida, In June, 1961, he was graduated from North Miami
Senior High School. In December, 1964, he received the
degree of Bachelor of Arts with a major in English from the
University of Florida. From 1965 to 1967 he worked for the
Trailways Bus System and the University of Florida. From
1967 until 1969 he served in the Adjutant General Corps of
the United States .Army, including service in Vietnam. Fol
lowing his discharge from the Army in 1969, he enrolled In
the Graduate School of the University of Florida, where he
received the degree of Master of Arts with a major in Speech
in December, 1970, and has pursued his work toward the degree
of Doctor of Philosophy.
Montes.
Donald Herbert Albury is married to Virginia Elenor
They have one daughter, Rebecca Lynne. He is a
member of the Southeastern Conference on Linguistics, the
Linguistic Society of America, the Modern Language Associ
ation, and. the American Dialect Society.
143


3.121a)
b)
c)
d)
3.122a)
b)
c)
d)
3.123a)
b)
c)
d)
Failure
apply in the
complements
80
(The fact of) Johns leaving so early was
unexpected.
Bill's (action of) refusing the offer is
unfortunate.
(The event of) Dicks resignation is likely.
(The fact of) Jimmys testimony was remarkable.
*(the fact of) John's leaving so early's being
unexnected
^Bill's faction of) refusing the offer's being
unfortunate
*(the event of) Dick's resignation's being
likely
*(the fact of) Jimmy's testimony's being
remarkable
*(the fact of) John's leaving so early's
une xp e c te dne s s
^Bill's (action of) refusing the offer's
unfortunateness
*(the event of) Dick's resignation's likelihood
*(the fact of) Jimmy's testimony's remarkableness
of both AGENT-PREP03ING and NP-PREP03ING to
derivation of sentences and gerundive nominal
results in unacceptable forms, as in 3-124 and
3.125, but derived nominal complements to which neither rule
has applied, as in 3*126, are acceptable.
3.124a) *Was unexpected (the fact of) John's leaving
so early.
b) *Was unfortunate Bill's (action of) refusing
the offer
c) *Is likely (the event of) Dick's resignation.
d) *Vas remarkable (the fact of) Jimmy's testimony.
3.125a) *being unexpected (the fact of) John's leaving
so early
b) *being unfortunate Bill's (action of) refusing
the offer
c) *being likely (the event of) Dick's resignation
d) *being remarkable (the fact of) Jimmy's testimony
3.12oa) the unexpectedness of (the fact of) Johns
leaving so early
b) the unforunateness of Bill's (action of)
refusing the offer
c) the likelihood of (the event of) Dick's
resignation
d) the remarkableness of (the fact of) Jimmy's
testimony


33
adjectives, are formed without the -ing suffix. Both this
set of adjectives and verbs and the previous set of verbs
show the same pattern. The form of the verb which is pre
sumably entered in the lexicon (the intransitive form of
grow, rise and move and the adjective form of dim, low and
wide) has a corresponding nominal in derived nominal comple
ments formed by means other than the -ing suffix, while all
the derived forms of the verbs have corresponding nominis
in derived nominal complements formed with the -ing suffix.
This suggests that the form of the associated nominis is
specified in the lexicon, but that this specification is not
retained when other forms of the verbs or adjectives are
derived from the form entered in the lexicon. With verbs
like convert, on the other hand, the specification of the
form of the associated nominal is retained when another form
of the verb is derived from the lexical entry. *
Other Nominis in -ing
Verbs derived from other underlying verbs are not the
only verbs to correspond to nominis in derived nominal com
plements formed with the -ing suffix. Sink is another verb
which is used both intransitively and transitively. The
intransitive use is illustrated in 2.o2a, the transitive in
2.62b. These sentences have the corresponding gerundive
nominal complements in 2.63, and the corresponding derived
nominal complements in 2.64 and 2.65*
2.62a) The Bismark sank.
b) The British Navy sank the Bismark.


VI
CHAPTERS Page
IT-EXTRAPOSITION 60
Passive Sentences 66
NP -PREPOSING 70
Dummy Subject Insertion 93
EQUI-NP-DELETION 97
DATIVE-MOVEMENT 102
Psychological Predicates 103
Conclusion 108
NOTES 110
IV RULE-ORDERING AND DERIVED NOMINAL COMPLEMENTS . 11.6
Introduction 116
Newmeyers Proposal ... 117
The Order of the Rules 121
The Cycle and Complementation 127
Complementation as a Process 139
NOTES 142
V CONCLUSIONS 143
BIBLIOGRAPHY 145


93
conditions on application in derived nominal complements
which differ from those in sentences and gerundive nominal
complements.
Summary
I have argued in this section that the subjects of sen
tences with intransitive verbs and the derived subjects of
passive sentences and their corresponding nominal comple
ments are derived by the rule of NP-PREPOSING. I have fur
ther argued that extraposed sentences and derived nominal
complements with nominal-initial order have resulted from
the failure of either AGENT-PREPOSING or NP-PREPOSING to
apply in their derivations. I have also argued that NP-PRE
POSING may apply to structures consisting of a noun (phrase)
plus an embedded sentences with the embedded sentence becom
ing a nominal complement, a that-complement, or a relative
clause. This thus eliminates any need for the rules of
IT-EXTRAPOSITION and EXTRAPOSITION FROM NP. Finally, I
have examined, and rejected, arguments presented by Emonds
in support of his claim that NP-PREPOSING does not apply in
the derivation of derived nominal complements.
Dummy Subject Insertion
Newmeyer (1971) notes that there are no derived nominal
complements with there as possessive noun determiner. The
sentences in 3.160 have the corresponding gerundive nominal
complements in 3.161, but there are no acceptable derived
nominal complements corresponding to the sentences in 3.160,
the forms in 3*162 being unacceptable.


NOMINAIS AS COMPLEMENTS
By
DONALD HERBERT ALBURY
DISSERT
PARTIAL
AT ION PRESENTED T-J THE GRADUATE COUNCIL
OF THE UNIVERSITY OF FLORIDA
, FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA


Copvright
197^
By
Donald Herbert Albury

To
Ginny
and
Rebecca

ACKNOWLEDGMENTS
I wish to thank first of all my chairman, Jean
Casagrande, who has always encouraged me to approach lin
guistics in terms of what are the facts of language, and,
then, what do the facts say about the theory. He has had
the faith to let me try unorthodox approaches until I could
either justify them or give them up. He has, above all,
helped me "keep the faith" with my own proposals when my
enthusiasm flagged.
I also wish to thank Peter Menzel, who has pointed out
many errors and inconsistencies in my arguments. Of course,
all remaining errors remain my responsibility.
My wife Ginny and daughter Rebecca have had to put up
with me for the last year while I worked on this disserta
tion. I am grateful for their patience and support,
I also wish to acknowledge financial support from the
University of Florida, in the form of a scholarship and a
research assistantship, during the writing of this disser
tation.

TABLE OF CONTENTS
Page
ACKNOWLEDGMENTS iv
ABSTRACT vii
CHAPTER
IINTRODUCTION 1
Statement of Purpose ..... 1
The Lexical Hypothesis 3
Derived Nominis as Complements on Nouns ... 5
Some Assumptions 9
Outline 11
NOTES 13
IIDEFINING DERIVED NOMINAL COMPLEMENTS . 15
First Definitions 15
The Surface Structure of Derived Nominal
Complements 16
Action Nominis 25
Agentive Nominis 42
Summary 44
NOTES 45
IIITHE PRODUCTIVITY OF DERIVED NOMINAL COMPLEMENTS 48
The Evidence for Lack of Productivity .... 48
Chomsky's Subcategorization Account 51
RAISING Rules 55
v

VI
CHAPTERS Page
IT-EXTRAPOSITION 60
Passive Sentences 66
NP -PREPOSING 70
Dummy Subject Insertion 93
EQUI-NP-DELETION 97
DATIVE-MOVEMENT 102
Psychological Predicates 103
Conclusion 108
NOTES 110
IV RULE-ORDERING AND DERIVED NOMINAL COMPLEMENTS . 11.6
Introduction 116
Newmeyers Proposal ... 117
The Order of the Rules 121
The Cycle and Complementation 127
Complementation as a Process 139
NOTES 142
V CONCLUSIONS 143
BIBLIOGRAPHY 145

Abstract of Dissertation Presented to the Graduate Council
of the University of Florida
in Partial Fulfillment of the Requirements for the
Degree of Doctor of Philosophy
NOMINAIS AS COMPLEMENTS
3y
Donald Herbert Albury
March, 197^
Chairmans Jean Gasagrande
Major Departments Linguistics
The syntactic construction Anown as the derived nominal
presents problems to a transformational grammar of English.
Noam Chomsky has claimed that derived nominis have
restricted productivity, an internal structure "like that of
noun phrases, and idiosyncratic semantic relationships to
their associated predicates. Chomsky's claim that the Lexi
cal Hypothesis provides a better explanation of the charac
teristics of derived nominis than the transformational pos
ition is rejected. A predicate-initial analysis of the
underlying structure of English and an analysis of derived
nominis as complements of nouns in underlying structure are
adopted. It is then shown that the productivity of derived
nominis is not as restricted as Chomsky claims, and that
the remaining restrictions are due to the failure of cer
tain rules to apply in the derivation of derived nominis.
vi x

VI11
Frederick Newmeyer's proposal that such rules fail to apply
because DERIVED NOMINALIZATION applies before they do, and
noninalizes the predicate which is a part of the structural
description of those rules, is adopted. It is 'then argued
that the sane rule-ordering arguments that account for the
restrictions on productivity of derived nominis will also
account for the noun phrase-like internal structure of
derived nominis. It Is further argued that the predi
cate-initial analysis eliminates the need for any rule of
extraposition in English, that DERIVED NOMINALIZATION is a
cyclic rule, and that the formation of complements from
underlying embedded sentences in general is a process of the
cycle applying to the embedded sentence, and not of some
higher application of the cycle.

CHAPTER ONE
INTRODUCTION
Statement of Purpose
The question of the extent of regularity in language
has been a recurrent theme in linguistics since ancient
times, when the Analogists of Alexandria disputed with the
Anomalists of Pergamon, This question has recently become
once more prominent in the study of transformational gram
mar. In an important sense, a transormation is an abstrac
tion of a regularity in language. The question of whether a
particular structure in a language can be described trans
formationally is an empirical one, and is equivalent to the
question of whether that structure in that language exhibits
a regular correspondence to another structure of that lan
guage. An assumption of transformational grammar has been
that language is basically regular, and, as a consequence,
transformational grammars have emphasized the regularity of
a language.
Noam Chomsky (1970) has claimed that the English comple
ment construction which he calls the derived nominal consti
tutes a part of English grammar which is not regular in the
sense indicated above, and which cannot be described trans
formationally. He supports his claim by contrasting derived
nominis to gerundive nominis, another type of complement
1

2
which he considers to be derived by transformations from
underlying embedded sentences. This is an empirical claim,
and may be tested by examining the sentences of English
which contain these two types of complements. The forms of
these complements, and their relation to each other and to
sentences, is illustrated in 1.1, 1.2 and 1.3; the main
clauses in 1.1 having the corresponding gerundive nominis
1
in 1.2 and derived nominis in 1.3*
1.1a) John is usually calm under stress, which pleases
Mary.
b) John completed the assignment early, which
pleased Mary.
1.2a) John's usually being calm under stress pleases
Mary.
b) John's completing the assignment early pleased
Mary.
1.3a) Johns usual calmness under stress pleases Mary.
b) John's early completion of the assignment
pleased Mary.
The underlined words are intended to supply appropriate
contexts for the nominal complements as well as showing how
sentences and nominal complements can be alternate realiza
tions of a common underlying structure. Gerundive nominal
complements often resemble sentences with progressive aspect.
While this may be confusing at times, the intended reading
of such forms as gerundive nominal complements may be made
clear by putting them into a context such as ...is surpris
ing.- or Ann is worried about....
Chomsky's argument that derived nominal complements are
not regular, and cannot be derived from underlying embedded
sentences, as are gerundive nominal complements, Is based on

3
three characteristic properties; A, that productivity is
more restricted for derived nominal complements than for
gerundive nominal complements, B, that semantic relations
between derived nominis and their associated propositions
are varied and idiosyncratic, and C, that derived nominal
complements have the Internal structure of noun phrases.
Having concluded that derived nominal complements cannot be
derived transformationally, Chomsky claims that the phrase
structure rules of English must be complicated to provide
the structure of derived nominal complements directly and
that derived nominis are lexical entries. This is the
Lexical Hypothesis, and amounts to a claim that derived
nominal complements show less regularity than other struc
tures in English, and the limits of regularity have been
found for this part of the grammar of English, If Chomsky
is wrong about the implications of the properties of derived
nominal complements, then his argument for the need for the
Lexical Hypothesis is weakened. I propose to show/ that
Chomskys three properties do not establish the irregularity
of the formation of derived nominal complements, that their
formation Is as regular as other transformational processes,
and that a transformational account of derived nominal
complements is at least as well motivated as the Lexical
Hypothesis.
The Lexical Hypothesis
Chomsky (1970:188) states the lexicalist position as
the choice of extending the base rules, with concomitant

4
simplification of the transformational component. By con
trast, he gives the transformationalist position as the
choice of simplifying the base structure and extending the
transformational apparatus. Chomsky states that the choice
between the two positions is entirely an empirical one.
Chomsky's 1970 study is a presentation of arguments for the
validity of the lexicalist position. I will consider those
arguments at appropriate points below, but for now I will
give a summary of the Lexical Hypothesis as presented by
Chomsky,
The theory within which Chomsky is making his claims
about derived nominis pictures syntax as being strictly
divided between a base component and a transformational
component (the "Aspects81 models cf. Katz and Postal, 1964,
and Chomsky, 1965). The point of connection between the
two components, when all the base (phrase-structure) rules
have applied, but before any transformational rules have
applied, is called deep structure. A third component is
the lexicon, which supplies lexical items to be inserted
in, and only in, deep structure.
The Lexical Hypothesis involves the claim that derived
nominal complements are produced directly by a subset of the
base rules, that derived nominis are supplied directly from
the lexicon in deep structure, and that derived nominal com
plements are not subject to most transformational rules.
Chomsky (1970:195) notes that such a claim implies that
phrase structure rules must introduce an extensive range of

5
derived nominal complement structures parallel to the struc
tures of embedded sentences. Indeed, there seem to be few,
if any, sentential deep structures in Chomsky's system which
do not have corresponding derived nominal complements.
According to Chomsky (1970:190), In early transforma
tional theory, "there was no other way to express the fact
that the contexts in which refuse appears as a verb and
refusal as a noun are closely related" than in terms of
transformational rules, Chomsky further states that when
contextual features were introduced into the theory (in
Chomsky, 1965) it became possible to separate the lexicon
from the categorical component of the base, and thus to
adopt the Lexical Hypothesis. According to Chomsky (1970:
190), refuse is entered in the lexicon with certain features
specified, but with no specification of the categorical fea
tures £nounj and [verbj. He adds that fairly idiosyncratic
morphological rules would, be involved In deriving forms like
refusal in derived nominal complements.
Derived Nominis as Complements on Nouns
The term complement has been used in ways different
enough to make it worth while Indicating here what I mean by
gerundive nominal complement and derived nominal complement.
On the one hand, anything that completes a structure can be
regarded as a complement. Thus, the underlined parts of the
examples In 1,4 are all complements of the verbs in their
respective sentences.

6
1.4a) John arrived yesterday.
b) I want to leave now.
c) Mike totaled his car.
On the other hand, complement has been used to refer to
nominalized (or otherwise reduced) embedded sentences such
as those cited in 1.2 and 1.3* By nominal complement I will
refer to nominalized embedded sentences which are comple
ments of nouns in noun phrases.
Peter Rosenbaum (1967) argued that the type of comple
ment structure he called Noun Phrase Complementation is a
noun phrase in deep structure, consisting of the pronoun it
as the head noun, and an embedded sentence as a complement
of it. Rosenbaum contrasted Noun Phrase Complementation
with Verb Phrase Complementation, in which an embedded sen
tence is a complement of a verb in deep structure.
Rosenbaum's claim that the pronoun it is present in
noun phrase complements in deep structure now appears to be
wrong. I will present arguments against that claim in the
section on IT-EXTRAPOSITION in Chapter Three.
Peter Menzel (1969) has argued that gerundive nominal
complements (and, less explicitly, derived nominal comple
ments) are complements of one of a certain set of deletable
head nouns in underlying structure. He points out (pp. 77-
3l) that verbs which take gerundive nominal complements,
with a few exceptions such as verbs of belief and say,
declare and claim, also allow the construction Noun + Com
plement, or just the noun as object. The nouns which can
appear in such constructions (i.e., fact, -pro-position, event.

7
action, etc.) may also be taken as names of the types of
complements a verb allows. Thus, Johns coming can be both
a fact and an event, and we get the sets of possible sen
tences in 1.5 and 1.6.
1.5a) John's coming occurred at ten o'clock.
b) The event of John's coming occurred at ten
o'clock.
c) The event occurred at ten o'clock.
1.6a) John's coming is surprising.
b) The fact of John's coming is surprising.
c) The fact is surprising.
The noun action can only take subjectless gerundive
nominal complements, while other nouns, such as event,
require a gerundive nominal complement with a subject
(Menzel, 1969582-83). At the same time, of the nouns which
take gerundive nominal complements, only action allows a
2
preposed possessive noun. Thus, we find the pattern of
acceptability shown in 1.7.
1.7a)
the
action of -j
b)
the
event of -f!
c)
, faction
J ohn
s I* event
Most of the nouns on Menzels list also take derived
nominal complements, and any derived nominal complement,
unless it is a complement of a verb of belief, or of say,
declare, or claim, can be a complement of at least one of
those head nouns, as is indicated by the examples in 1.8.
1.8a) The fact of Johns departure cannot be contested,
b) The event of the destruction of Jerusalem
occurred in 67 AD.
c) The state of Mike's awareness is unpredictable.

8
The head noun action never takes a derived nominal com
plement, no matter where the possessive agent is placed, or
even if it is deleted. The examples in 1.9 are all unaccep
table, even though the parallel forms in 1.10 are perfectly
acceptable.
l-9a) *The action of Humphrey's refusal of the offer
caused trouble.
b) ^Humphrey's action of refusal of the offer caused
' trouble.
c) *The action of refusal of the offer caused
trouble.
1.10a) The fact of Humphrey's refusal of the offer
caused trouble.
b) Humphrey's action of refusing the offer caused
trouble.
c) The action of refusing the offer caused trouble.
Menzel (1969851) notes that gerundive nominal comple
ments can also be complements of act. He states that acts
are a subclass of actions, with the restriction that "an act
is an action which the speaker either (a) disapproves of
strongly (in a legal or moral sense); or (b) admires greatly."
He also notes that this distinction does not seem to be
recognised by some speakers.
While derived nominal complements do not occur as com
plements of action, a restricted class does occur as comple
ments of act. The head noun act does not take derived nomi
nal complements with object prepositional phrases. The
examples in 1.11 illustrate the differences between gerund
ive nominal complements and derived nominal complements in

9
regard to the head nouns action and act. The adjective
cowardly is more appropriate for the complements of act.
1.11 (in the context ...was unexpected/cowardly.)
a) Johns denying the request
b) Johns action of denying the request
c) John's act of denying the request
d) John's denial of the request
e) ^Johns action of denial of the request
f) *John's act of denial of the request
g) John's act of denial
Gerundive nominal complements and derived nominal com
plements are alike in that they both can be complements of
a limited set of head nouns in surface structure, and seem
to always be complements of such nouns in underlying struc
ture, The two types of complement differ in that only ger
undive nominal complements can be complements of action.
Some Assumptions
In this study I will be making certain assumptions
bearing on the structures and rules I will discuss. First
of ail, I will assume that grammatical rules apply in a
fixed order and cyclically. The cyclic application of rules
means that the full set of rules would first apply in order
to the deepest embedded sentence, S?, in 1.12,
1.12
S
0
S
1
S
2

10
After all the rules have had a chance to apply to 3
the rules would apply in the same order to the next higher
sentence, S-, which is in turn embedded in SQ, After all
the rules have had a chance to apply to 3-, the rules would
3
apply again in the same order to the highest sentence, Sq.
I will also adopt here a predicate-initial analysis of
the underlying structure of English sentences. McCawley
(19?0) presents arguments for such an analysis for struc
tural representations comparable to those in Chomsky (1965).
A predicate-initial analysis is also at least implicit in
the case grammar of Fillmore (1968). The underlying struc
tures in case grammar differ greatly from the underlying
structures postulated in Chomsky (1965). In Fillmore's case
grammar, the noun phrases in deeply underlying structures
are identified by their semantic relationship to the predi
cate of the sentence, and not by any syntactic relationship.
Without adopting case grammar in toto, I will assume that,
before any syntactic rules have applied, there is no ordered
relationship between the constituents of the underlying pro
positions, and that at least some syntactic rules must recog
nize semantic relationships and transform such semantic rela-
4
tionships into word-order syntactic relationships. For con
venience of representation, however, I will assume that
underlying sentences have structures like that in 1.13
before any syntactic rule has applied.^
1.13
S
PRED
NP
NP

11
Assuming predicate-initial order in structures such as
1.13 is a way of expressing the pivotal role of the predi
cate in most syntactic rules. An alternative statement
could he that noun phrases are assigned linear order in rela
tion to the predicate by syntactic rules.
McCawley (1970) explicitly assumes that underlying sen
tences have the order Verb-Subject-Object (VSO) before the
application of any transformations, but does not present any
evidence for preferring that order over one of Verb-Object
-Subject (VOS), I believe that the evidence points to a VOS
order, if there is any order at all to the noun phrases to
£
the right of the predicate in deeply underlying structures.
In the rest of this work, the relative order of noun phrases
in trees representing deeply underlying structures will be
arbitrary, and represent no claims as to the actual order.
Outline
This study is concerned primarily with the internal
structure of nominal complements, by which I mean the rela
tionships of the constituents of nominal complements to each
other, as opposed to the relationships of nominal comple
ments to items which are not constituents of the said nomi
nal complements, I compare such internal structures to
those of sentences and of other clausal complements, I
point out that the differences between such structures are
regular, and easily accounted for by independently motivated
syntactic rules, given certain conditions on the cycle in
which the rules apply.

12
I do not discuss the problem of the apparent irregular
ities in the semantic relationship between related predi
cates and derived nominis, I will adopt Newmeyer's (1971)
claim that, at the worst, the transformationalist position
is no less adquate on this point than the Lexical Hypothe
sis, since, if there are no regularities, the information
on restrictions on meanings of derived nominis must be
part of the lexical entrys of the underlying predicate with
the transformationalist position, of the underlying predi
cate/nominal with the Lexical Hypothesis.
Chapter Two is a discussion of the surface structure of
derived nominal complements, and arguments for including in
the class of derived nominal complements certain structures
which have not previously been so identified. Chapter Three
discusses the rules which do not apply in the derivation of
derived nominal complements, or which have different condi
tions on applicability in derived nominal complements, but
which apply without restriction to sentences and gerundive
nominal complements. Chapter Four discusses the ordering of
rules which will account for the structure of derived nominal
complements, and the question of which cycle the appropriate
rules apply on. Chapter Five summarizes the thesis, and
presents certain questions of theoretical import.

NOTES
A source of possible confusion is the fact that both
the nouns formed by adding a suffix to a verb stem, and the
complements in which such nouns occur, can be called nomi
nis. To avoid confusion, I will reserve the term nominal
for the nouns so formed, and refer to the complements they
occur in as nominal complements. I will use nominalization
to refer to the process by which nominis are created. I
will have more to say about the choice of the term comple
ment below.
2
The possessive agent on action is always identical
with the deleted agent of the complement of action. Menzel
argues that the agent cannot have been raised from the com
plement. Among suggestions of possible sources for the
possessive agent, he states that "in a grammar based on
the transformationalist position...the agent on the head
noun action would be derived from an underlying sentence
embedded on the noun action, or more probably on the verb
act." Ross (1972a) argues that sentences with verbs of act
ion (which are the only sentences that can be embedded as
complements of the head noun action) are embedded on the
verb do in underlying structures, with the agent of do iden
tical to the agent of the embedded sentence. The substitu-
13

tion of action for do in nominal complements, and the dele
tion of the lower of two identical agents (cf. SQUI-NP-DELE
TION in Chapter Three), will account for the above facts
within the transformationalist position.
3
There are extensive arguments in the literature for
cyclical ordering. I am not concerned here with the argu
ments over extrinsic vs. intrinsic ordering, but will merely
state that assuming ordering permits useful generalizations,
as in Chapter Four and Chapter Five.
4
At this point I am rejecting the Aspects model of
grammar. The Aspects model has syntax as the most basic
component of the grammar, with both semantics and phonology
acting as interpretive components of the output of the syn
tactic component. I believe that this model is unrealistic,
even as a model of competence rather than of performance.
Speech is a stream of sound perceived as a linear string of
distinct units. Semantic propositions, on the other hand,
are unordered in any dimensional sense. There is no dimen
sional order implied in the statement that someone is the
agent of such-and-such action. Syntax, then, is that part
of grammar which relates unordered semantic relationships
to linearly ordered phonological strings.
James McCawley, during a discussion at the 1972 LSA
Annual meeting, summed up the predicate-initial hypothesis
in the statement that if there is any linear order so early
in the grammar, it is a predicate-initial order.
^ I will present my arguments for this analysis in
Albury (forthcoming).

CHAPTER TWO
DEFINING DERIVED NOMINAL COMPLEMENTS
First Definitions
Chomsky (1970) does not explicitly define derived nomi
nal complements, but a definition may be extracted from the
various examples he cites in discussing the characteristics
of derived nominis. It Is obvious throughout that Chomsky
intends the class of derived nominis to include only those
nominis which are morphologically derived from a verb by
means other than the suffix -ing.~ He explicitly excludes
gerunds (those nominis derived by adding -ing to a verb)
from the set of derived nominis. Chomsky (1970j214) notes
certain similarities to derived nominal complements shown by
structures called "action nominis" in Lees (1963) and
Fraser (1970), which have nominis formed with -ing, but he
claims a number of differences which would seem to preclude
treating them as derived nominal complements, I will have
more to say about these forms below, Chomsky thus limits
the term derived nominal to those nominis formed by adding
a derivational suffix to a verb, such as refusal (from
x~efuse) and marriage (from marry) ; nominis identical
(except for stress) to verbs, such as search (from search) and
export (from exp5rt); and nominis phonologically modified
from a verb, such as deed (from did).
15

16
I will temporarily accept Chomsky's implied morphologi
cal definition of derived nominis, and define derived nomi
nal complements as those complements in which nominis
formed without an -ing suffix are found. Starting from this
definition of derived nominal complements, I would like to
build up a detailed description of the structure of these
complements. Chomskys method of comparing sentences, ger
undive nominal complements and derived nominal complements
to illustrate the characteristics of derived nominal comple
ments is useful, and I will adopt it here.
The Surface Structure.of Derived Nominal Complements
Chomsky (1970s187-88) cites the sentences in 2.1 as
having the corresponding gerundive nominal complements in
2,2 and the corresponding derived nominal complements in 2.3.
The gerundive nominal complements have gerundive nominis
corresponding to the verbs in the sentences, while the
derived nominal complements have derived nominis corres
ponding to the verbs in the sentences.
2.1a) John is eager to please,
b) John refused the offer,
c) John criticized the book.
2.2a) John's being eager to please
b) John's refusing the offer
c) John's criticizing the book
2.3a) John's eagerness to please
b) John's refusal of the offer
c) John's criticism of the book
Determiners
One characteristic shared by gerundive nominal comple
ments and derived nominal complements is the presence of

17
possessive nouns corresponding to the subject nouns in the
sentences. Chomsky points out that the two types of nominal
complement differ in that the possessive nouns in derived
nominal complements can be replaced by other determiners, as
is shown by the derived nominal complements in 2.4 (cf. 2-3).
while the possessive nouns in gerundive nominal complements
cannot be so replaced, as is indicated by the fact that forms
like those in 2-5 (cf. 2.2) do not occur. The derived nomi
nal complements in 2.4 appear to be parallel to the derived
nominal complements in 2-3? but the derived nominal comple
ments in 2.6 are closer in meaning to those in 2.3 than the
2
ones in 2.4 are.
2.4a) ?the eagerness to please
b) the refusal of the offer
c) the criticism of the book
2.5a) *the being eager to please
b) *tha refusing the offer
c) *the criticizing the book
2.6a) ?the eagerness to please by John
b) the refusal of the offer by John
c) the criticism of the book by John
Adjective-Adverb Correspondences
A second difference between gerundive nominal comple
ments and derived nominal complements which Chomsky discus
ses is the fact that derived nominal complements can have an
adjective preceding the nominal, while gerundive nominal com
plements cannot. These pronominal adjectives correspond to
adverbs in sentences and gerundive nominal complements, so
that there are sentences like those in 2.7, with the corres
ponding gerundive nominal complements in 2.8, and the corres-

18
ponding derived nominal complements in 2.9* The fact that
strings like those in 2.10 are unacceptable shows that ger
undive nominal complements do not take pronominal adjectives.
2.7a) John is overwhelmingly eager to please.
b) John abruptly refused the offer.
c) John criticised the book unmercifully.
2.8a) John's being overwhelmingly eager to please
b) John's abruptly refusing the offer
c) Johns criticizing the book unmercifully
2.9a) John's overwhelming eagerness to please
b) John's abrupt refusal of the offer
c) John's unmerciful criticism of the book
2.10a) "John's being overwhelming eager to please
b) ^John's abrupt refusing the offer
c) ^John's unmerciful criticizing the book
The prenominal adjectives in some derived nominal com
plements, such as those in 2.11, do not have corresponding
adverbs in gerundive nominal complements, as in 2.12, or
sentences, as in 2.13.
2.11a) John's troublesome eagerness to please
b) John's untimely refusal of the offer
c) John's unmotivated criticism of the book
2.12a) ^Johns being troublesomely eager to please
b) *John,s untimely refusing the offer
c) ^John's criticizing the book unmotivatedly
2.13a) *John is troublesomely eager to please.
b) -John untimely refused the offer.
c) *John criticized the book unmotivatedly.
That the acceptability of the forms in 2.11, and the
unacceptability of the forms in 2.12 and 2.13 is due to
these adverbs alone can be seen by inspecting the forms in
2-1, 2.2, 2.7 and 2.8. The adjectives which appear in pre
nominal position in the derived nominal complements in 2*11
can also be predicated of such complements, as in 2.14.

19
2.l4a) John's eagerness to please is troublesome,
b) John's refusal of the offer was untimely.
c) John's criticism of the book was unmotivated.
The prenominal adjectives in the derived nominal com
plements in 2.9 can also be predicated of those complements,
as in 2,15.
2.15a) John's eagerness to please is overwhelming.
b) John's refusal of the offer was abrupt.
c) John's criticism of the book was unmerciful.
The parallel I have been drawing between the adverbs in
sentences and gerundive nominal complements and the adjec
tives in derived nominal complements suggests that the ad
verbs and adjectives are derivationally related. The fact
that the adverbs are morphologically derived from the adjec
tives by adding the suffix -ly reinforces that hypothesis.
The three-way correspondence between prenominal adjec
tives in derived nominal complements, adjectives predicated
of derived nominal complements, and adverbs in sentences and
gerundive nominal complements suggests that the adjectives
and adverbs have as a common source a higher predicate. If
the higher predicate is the highest matrix predicate (ignor
ing abstract higher predicates) then it may be expressed as
an adverb of the embedded sentence (raised to surface sen-
tencehood) or as a predicate on a derived nominal complement.
If the higher predicate is in turn embedded under a predi
cate which will appear in surface structure, then it may be
expressed as an adverb with a gerundive nominal complement,
or as a prenominal adjective with a derived nominal comple
ment. The higher predicate-prenominal adjective relation-

20
ship is also seen in simple noun phrases, so that no new
rule need be postulated to derive prenominal adjectives in
derived nominal complements from higher predicates.
Chomsky (I970sl95) points out that a claim that the
prenominal adjectives in derived nominal complements are
derivationally related to adverbs leads to the prediction
that sentences like those in 2.1 will have corresponding
derived nominal complements like the structures in 2.1?,
which are not acceptable.
2.16a) John refused the offer in a surprising manner,
b) John is sincere to a limited extent.
2.17a) ^John's refusal of the offer in a surprising manner
b) ^Johns sincerity to a limited extent
However, there are structures like those in 2.18 which
include derived nominal complements, and which seem to cor
respond to the sentences in 2.1.
2.18a) the surprising manner of John's refusal of
the offer
b) the limited extent of John's sincerity
Adverbs which are morphologically derived from adjec
tives seem to be included within the scope of structures
which have corresponding derived nominal complements, while
adverbs which are prepositional phrases seem not to be
included within the scope of such structures. It is not
necessary to claim that the prenominal adjectives are
derived from adverbs to support the claim that the prenomi
nal adjectives and adverbs are derived from the same under
lying structures. The restrictions on the inclusion of ad
verbial prepositional phrases within the scope of derived
nominal complements does not invalidate such a claim.

21
Object Prepositions
Another characteristic which.distinguishes derived nom
inal complements from sentences and gerundive nominal com
plements is the presence of a preposition preceding the noun
phrases in derived nominal complements which correspond to
the objects of verbs in sentences, if such object noun
phrases do not already have a preposition. Among the sen
tences in 2.19, corresponding gerundive nominal complements
in 2.20 and corresponding derived nominal complements in
2.21, only the derived nominal complements have the prepos-
3
ition of preceding the object.
2.19a) John refused the offer.
b) John criticized the book.
c) John robbed the bank.
2.20a) Johns refusing the offer
h) John's criticizing the book
c) John's robbing the bank
2.21a) Johns refusal of the offer
b) John's criticism of the book
c) John's robbery of the bank
When object noun phrases are preceded by prepositions
in sentences such as those in 2.22 and in the corresponding
gerundive nominal complements in 2.23 then the same pre
positions also appear in the corresponding derived nominal
complements, as in 2.24.
2.22a) John was amused at the children's antics.
b) John delighted in teasing Alice.
c) John was doubtful about Dick's honesty.
2.23a) John's being amused at the childrens antics
b) John's delighting in teasing Alice
c) John's being doubtful about Dick's honesty
2.24a) John's amusement at the children's antics
b) John's delight in teasing Alice
c) John's doubts about Dick's honesty

22
Pluralizaticn
Another characteristic of derived, nominis discussed by
Chomsky (I97O1I89) is that they may be pluralizad* while
gerundive nominal complements may not. Thus, we find the
derived nominal complements in 2.25* which correspond to the
sentences in 2.26* while the corresponding gerundive nominal
complements in 2.2? are not pluralizad.
2.25a) Johns three proofs of the theorem
b) John's repeated attempts to scale the wall
c) Agnews many attacks on the press
2.2oa) John proved the theorem three (times, ways, etc.).
b) John repeatedly attempted to scale the wall.
c) A.gnew attacked the press many times.
2.27a) John's proving the theorem three (times, ways, etc.)
b) John's repeatedly attempting to scale the wall
c) A.gnew's attacking the press many times
The sentences in 2,26 and the gerundive nominal comple
ments in 2.27 express repetitive events. The derived nomi
nal complements in 2.25 also express repetitive events, but
with plural nominis rather than adverbs of repetition,
Repetition is not always expressed, explicitly in sentences
such as those in 2.28, which therefore have two sets of cor
responding derived nominal complements, those in 2.29, which
express explicitly the singularity of the event, and those
in 2.30, which express explicitly the repetition of the
4
event.
2.28a) John has proved the theorem.
b) John has attempted to scale the wall.
c) A.gnew has attacked the press.
2.29a) John's proof of the theorem
b) John's attempt to scale the wall
c) Agnew's attack on the press

23
2,30a) John's proofs of the theorem
b) John's attempts to scale the wall
c) Agnew's attacks on the press
Not all derived nominis can be pluralizad. The sen
tences in 2,31 have the corresponding derived nominal comple
ments with singular derived nominis in 2,32, but derived
nominal complements with plural derived nominis like those
in 2.33 corresponding to the sentences in 2.31 do not occur.
2,31a) 'The enemy has destroyed the city on three
occassions,
b) The crowd laughed repeatedly.
2.32a) the enemy's destruction of the city on three
occassions
b) the crowd's repeated laughter
2.33a) *the enemy's destructions of the city on three
occassions
b) *the crowd's repeated laughters
Auxiliaries
Another characteristic of the structure of derived nom
inal .complements discussed by Chomsky (1970:189) is the
absence of any auxiliary verbs, Gerundive nominal comple
ments, on the other hand, can have any auxiliary (with the
exception exemplified by 2.35c) except modals. Perfective
aspect can appear in gerundive nominal complements, so that
sentences like those In 2.3^a and b have the corresponding
gerundive nominal complements in 2.35a and b. Progressive
aspect can also appear in gerundive nominal complements, but
only in conjunction with perfective aspect, as in 2.35b, so
that the sentence in 23^'C does not have a corresponding
gerundive nominal complement with being corresponding to a
form of be which is acceptable (cf. 2,35c and d),-^

24
2.34a) John has criticized the book.
b) John had been criticizing the book.
c) John is criticizing the book.
2.35a) Johns having criticized the book
b) John's having been criticizing the book
c) ^John's being criticizing the book
d) John's criticizing the book
Forms of the verb be which appear as copulas in sen
tences with predicate adjectives or predicate nouns, or in
combination with past/passive participles in passive sen
tences, such as those in 2.36, also appear in gerundive nom
inal complements such as those in 2=37* But the derived
nominal complements in 2.38 which correspond to the sen
tences in 2.36 do not have any form of be.
2.36a) John is strong.
b) Alice is beautiful.
c) John is the chairman.
d) The city was destroyed by the enemy.
e) Abby was acquitted by the jury.
2.37a) John's being strong
b) Alice's being beautiful
c) John's being the chairman
d) the citys being destroyed by the enemy
e) Abby's being acquitted by the jury
2.38a) Johns strength
b) Alice's beauty ^
c) John's chairmanship0
d) the city's destruction by the enemy
e) Abby's acquittal by the jury
A Redefinition
I would now like to abandon my first provisional defi
nition of derived nominal complements as those complements
with nominis formed without an -inn suffix, and instead
define derived nominal complements as those complements
which show one or more of the following features: a wide

25
variety of determiners; prenominal adjectives instead of
adverbs; pluralization of the nominal; prepositions preced
ing object noun phrases when they do not appear in the cor
responding sentences; and complete absence of auxiliaries,
including copulas and the passive be. The form of the nom
inal in derived nominal complements does not enter into
this definition.
Action Nominis
Chomsky's Analysis
There are nominal complements which satisfy the new
definition of derived nominal complements given above, but
which have what appear to be gerundive nominis correspond
ing to the verb, such as the nominal complements in 2.39-
Nominal complements of this type are called "action nomi
nis in Lees (1963) and Fraser (1970).
2.39a) John's refusing of the offer
b) John's proving of the theorem
c) the growing of tomatoes
Chomsky (1970s214-15) believes that these complements
belong to a third class of nominal complements distinct from
both gerundive nominal complements and derived nominal com
plements. He does not identify them as "action nominis."
Chomsky claims that these nominal complements appear to have
the same internal structure of noun phrases that derived nom
inal complements have, as evidenced by the possibility of a
determiner other than a possessive noun appearing (cf. 2.39c),
but that prenominal adjectives seem quite unnatural in such
complements. He says that the complement is limited in

26
productivity as well, so that we cannot get structures like
those in 2.40. Chomsky (1970:215) finally states that
"there is an artificiality to the whole construction that
makes it quite resistant to systematic investigation."
2.40a) *the feeling sad
b) *the trying to win
c) *the arguing about money
d) *the leaving
I would like to put aside for now the problems involved
in the forms in 2.39a and b and 2.40, and consider in detail
the complement in 2.39c, the growing of tomatoes. Part of
the data cited by Chomsky (1970:192) to support his claim
that the Lexical Hypothesis provides the best explanation
for the origin of derived nominis involves transitive verbs
derived from intransitive verbs by the rule of CAUSATIVE
FORMATION (cf. G. Lakoff, 1970), such as grow, as in John
grows tomatoes, derived from grow, as in tomatoes grow. At
some point in its derivation, the underlying structure of
John grows tomatoes can be represented by the tree in 2.41,
The structure underlying tomatoes grow is embedded as 3-^ in
the tree in 2,41
2.41)
V
NP
NP
CADS I
V
NP
grow tomatoes John
Chomsky points out that there is a derived nominal com
plement, the growth of tomatoes, which corresponds to the

27
sentence tomatoes grow, but not to the sentence John grows
tomatoes. as would be expected from the apparently parallel
example of the derived nominal complement, the rejection of
the offer, which corresponds to the verb phrase reject the
offer- In other words, the rejection of the offer is an
object nominal complement, while the growth of tomatoes is a
subject nominal complement, and Chomsky takes the fact that
there is no interpretation of the growth of tomatoes as an
object nominal complement as proof that there is no derived
nominal complement corresponding to John grows tomatoes. In
terms of the Lexical Hypothesis, this is to be expected
because John grows tomatoes involves a derivation from an
underlying structure which includes tomatoes grow as an
embedded sentence, as in the tree in 2.41. This is consis
tent with the claim that derived nominis are associated
lexically with the underlying verb. Since grow occurs in
the lexicon only in the intransitive, noncausative sense,
the transitive sense being derived transformationally by the
rule of CAUSATIVE FORMATION, only the intransitive sense can
have a corresponding derived nominal, according to the
Lexical Hypothesis.
Some Counterexamples to Chomsky *s .Analysis
Smith (1972) has pointed out that there are many excep
tions to Chomskys claims concerning the occurrence of
derived nominis corresponding to verbs derived by the rule
of CAUSATIVE FORMATION. The verb convert occurs as both
transitive and intransitive, and both forms have associated

28
derived nominis. Thus* Roberts conversion to hedonism,
which at some point in its derivation has an underlying
structure like that of the embedded sentence S^ in the tree
in 2.42, and Johns conversion of Robert to hedonism, which
at some point in its derivation has an underlying structure
like that of the whole tree in 2.42, are both acceptable
derived nominal complements. The derived nominal complement
the conversion of Robert, like many other nominal complements
such as the shooting: of the hunters, is therefore interpreta
ble as either a subject nominal complement or as an object
nominal complement.
2.42 SQ
V NP NP
convert to Robert hedonism John
Some verbs listed by Smith which also form derived nom
inis in both transitive and intransitive forms are explode.
divide, accelerate. expand. repeat, neutralize. conclude
and unify. She then points out that all the listed counter
examples share a morphological propertys their derived nomi
nis are formed with suffixes of Latin origin. She then
goes on to claim that there are almost as many verbs of the
type convert as there are of the type grow, and then states
what she sees as the conditioning factor in distinguishing
the two types; "Whether or not a verb has a transitive

29
nominal depends on how the nominal is formed. If a causa
tive verb takes a nominalizing suffix of Latin origin
(-tion, -al, -ment), then it has both transitive and intrans
itive derived nominis. If a causative verb does not take
such a nominalizing suffix, then it occurs only intransi
tively in derived nominis," She then observes that, in
general, verbs of Latin origin form derived nominis with
the suffixes of Latin origin, while verbs of what she calls
Anglo-Saxon origin form derived nominis with the suffixes
-th, -ness, or & (null), which can be considered to be
native, as opposed to the suffixes borrowed from Latin.
Smith has, based on the data above, reached the conclu
sion that "the grammar must distinguish at least two classes
of 'causative' verbs: those that do and do not have transi
tive derived nominis." As we have just seen above, that
distinction is to be based on something like a native/bor
rowed- Latin dichotomy. It seems very strange, however, to
state that a part of the lexicon associated with borrowed
(Latinate) derivational suffixes shows more productivity
than another part of the lexicon associated with native
derivational suffixes. This raises the possibility that the
causative verbs associated with native derivational suffixes
also have corresponding derived nominis, but that for some
reason Smith and Chomsky have failed to recognize them.
Using the definition of derived nominal complements I have
developed above, I will now look for derived nominal comple
ments corresponding to sentences with such causative verbs.

30
Nominis of Causatives
Consider verbs such as grow, raise and move. These
verbs occur in both intransitive and transitive forms, the
latter being derived by the rule of CAUSATIVE FORMATION.
The verbs in the intransitive form are used in sentences
such as those in 2.43, which have the corresponding derived
7
nominal complements in 2.44 and 2.45.
2.43a)
b)
c)
The tree grew slowly.
The temperature rose rapidly.
The table moved mysteriously.
the tree's slow growth
the temperature's rapid rise
the table's mysterious movement
2.45a)
b)
c)
the slow growth of the tree
the rapid rise of the temperature
the mysterious movement of the table
The verbs in the transitive form are used in sentences
such as those in 2.46, which have the corresponding gerund
ive nominal complements in
derived nominal complements
2.4?, and the corresponding
in 2.43 and 2.49.8
2.46a) John grows tomatoes skillfully.
b) Tom raised the temperature deliberately.
c) Hike moved the table accidentally.
2.47a) John's growing tomatoes skillfully
b) Toms raising the temperature deliberately
c) Mike's moving the table accidentally
2.48a) John's skillful growing of tomatoes
b) Toms deliberate raising of the temperature
c) Mikes accidental moving of the table
;/ ci j
b)
c)
the skillful growing of tomatoes by John
the deliberate raising of the temperature by Tom
the accidental moving of the table by Mike
The derived nominal complements in 2.48 and 2.49 are
completely consistent with the
definition of derived nominal

31
complements which I have adopted. I will therefore assume
for now that -ing is one of the suffixes forming nominis
in derived nominal complements, which happens to be homo-
phonous with the gerundive suffix -ing.
Nominis of Dead.iectival Verbs
The rule of CAUSATIVE FORMATION is not the only rule by
which a verb may be derived from an underlying form with dif'
ferent properties. Smith (1972) mentions that we find the
derived nominal complement the lights dimness, but not
^John's dimness of the light. Dim is homophonous for three
sensess adjective, intransitive verb derived by the rule of
INCHOATIVE FORMATION (cf. G. Lakoff, 1970), and transitive
verb derived by the rule of CAUSATIVE FORMATION. Other
adjective-verb sets related by these rules are low:lower;
lower and wide¡widens widen. The adjectives are used in sen
tences like those in 2*50, which have the corresponding ger
undive nominal complements In 2.51 and the corresponding
derived nominal complements in 2.52 and 2.53
2.50a) The light was dim.
b) The bridge was low.
c) The road was wide.
2.51a) the light's being dim
b) the bridge's being low
c) the road's being wide
2.52a) the light's dimness
b) the bridge's lowness
c) the road's width
2.53a) the dimness of the light
b) the lowness of the bridge
c) the width of the road

32
The Intransitive verbs are used in sentences like those
in 2.54, which have the corresponding gerundive nominal com-
plenient s in
2.55 and the corresponding derived nominal com-
plements in
2.56 and 2.57.
2.54a)
b)
c)
The light slowly dimmed.
The bridge gradually lowered.
The road suddenly widened.
2,55a)
b)
c)
the light's slowly dimming
the bridges gradually lowering
the road's suddenly widening
2.56a)
b)
c)
the light's slow dimming
the bridge's gradual lowering
the road's sudden widening
2.57a)
b)
c)
the slow dimming of the light
the gradual lowering of the bridge
the sudden widening of the road
The transitive verbs are used in sentences like those
in 2.53 which have the corresponding gerundive nominal com
plements in 2,59 and the corresponding derived nominal com
plements in 2,60 and 2,61,
2.58a)
b)
c)
^ J
John suddenly dimmed the light.
The tender gradually lowered the bridge,
the city widened the road recently.
2.59a)
b)
c)
John's suddenly dimming the light
the tender's gradually lowering the bridge
the city's widening the road recently
2.60a)
0)
c)
John's sudden dimming of the light
the tender's gradual lowering of the bridge
the city's recent widening of the road
2,61a)
b)
c)
the sudden dimming of the light by John
the gradual lowering of the bridge by the tende
the recent widening of the road by the city
All of the derived nominal complements in 2.56, 2,57,
.60 and 2.61 have nominis formed with the -ing suffix,
niy the nominis in 2.52 and 2.53, which correspond to

33
adjectives, are formed without the -ing suffix. Both this
set of adjectives and verbs and the previous set of verbs
show the same pattern. The form of the verb which is pre
sumably entered in the lexicon (the intransitive form of
grow, rise and move and the adjective form of dim, low and
wide) has a corresponding nominal in derived nominal comple
ments formed by means other than the -ing suffix, while all
the derived forms of the verbs have corresponding nominis
in derived nominal complements formed with the -ing suffix.
This suggests that the form of the associated nominis is
specified in the lexicon, but that this specification is not
retained when other forms of the verbs or adjectives are
derived from the form entered in the lexicon. With verbs
like convert, on the other hand, the specification of the
form of the associated nominal is retained when another form
of the verb is derived from the lexical entry. *
Other Nominis in -ing
Verbs derived from other underlying verbs are not the
only verbs to correspond to nominis in derived nominal com
plements formed with the -ing suffix. Sink is another verb
which is used both intransitively and transitively. The
intransitive use is illustrated in 2.o2a, the transitive in
2.62b. These sentences have the corresponding gerundive
nominal complements in 2.63, and the corresponding derived
nominal complements in 2.64 and 2.65*
2.62a) The Bismark sank.
b) The British Navy sank the Bismark.

2.63a)
b)
34
the Bismark's sinking
the British Navy's sinking the Bismark
2.64a) the Bismark's sinking
b)the British Navy's sinking of the Bismark
2.65a) the sinking of the Bismark
b) the sinking of the Bismark by the British Navy
Presumably, there is no specification of the form of
the nominal in a derived nominal complement corresponding to
sink in the lexical entry, so that the form with -ing is sup
plied by the gra.mmar for sink intransitive and transitive
just as it is supplied for the nominis in derived nominal
complements corresponding to sentences with the derived
verbs grow, raise, move, dim, lower and widen. One conse
quence of this is that it is possible for the gerundive nomi
nal complement and derived nominal complement corresponding
to a particular sentence to be identical in surface struc
ture, as are the forms in 2.63a and 2,64a. It jvill there
fore not always be possible to decide whether a nominal com
plement is actually a gerundive nominal complement or a
derived nominal complement, unless the context supplies such
information.
One objection which might be made to the identification
of any nominal complements with nominis formed with -ing as
derived nominal complements is that they do not readily form
plurals, However, it is not true that they never form plur
als, as shown by the examples in 2,66.
2,66a) the wanderings of an old mind
b) the leavings of a great feast
c) the makings of a great scholar
d) the cravings of a glutton
e) the comings and goings of the workers

35
As was noted, above (cf. 2-33) not all derived nominis
form plurals, so the fact that many apparent derived nomi
nis in -ins do not is not strong counter-evidence.
Paired Nominis
The nominal complements in 2.39a and b (repeated here
in 2.67), cited by Chomsky (1970:214), appear to be derived
nominal complements with nominis formed with -ing which are
parallel to the derived nominal complements in 2.68, which
have nominis formed other than v/ith the suffix -ing.
2.67a) John's refusing of the offer
b) John's proving of the theorem
2.68a) John's refusal of the offer
b) John's proof of the theorem
The derived nominal complements with nominis formed
other than v/ith the -ing suffix in 2.68 readily take prenora-
inal adjectives, as was discussed above, so that we have
derived nominal complements like those in 2.69..
2.69a) John's abrupt refusal of the offer
b) John's brilliant proof of the theorem
The nominal complements with nominis formed with the
-ing suffix in 2.67 seem to less readily accept prenominal
adjectives, as is indicated by the strangeness of the nom
inal complements in 2.70. That this strangeness is not due
to the nominal having an -ing suffix alone is shown by the
normal occurrence of prenominal adjectives v/ith derived nom
inal complements with nominis formed with -ing in 2.48,
2.49, 2.56, 2.57 2.6o and 2.61 above.
2.70a) ?John's abrupt refusing of the offer
b) ?John's brilliant proving of the theorem

36
Chomsky (1970s215) notes that derived nominal comple
ments with nominis formed with the -ing suffix seem rather
clumsy when a derived nominal complement with a nominal
formed other than with the -ing suffix also exists. It is
also these nominal complements which do not readily take a
prenominal adjective. Since I have claimed above that the
suffix -ing in general appears in derived nominal comple
ments when no other means of forming the nominal is speci
fied j it seems possible that pairs of nominis like refusal
refusing and proof:proving represent a misapplication of the
rule supplying the -ing suffix, and the full regularity of
derived nominal complements with forms like refusing and
proving is blocked by the existance of forms like refusal
and proof.
Some Apparent Counterexamples
Chomsky (1970:214) also noted that derived" nominal com
plements with -ing seem to be limited in production since we
do not get forms like those in 2.40, repeated here as 2.71=
2.71a) *the feeling sad
b) *the trying to win
c) *the arguing about money
d) *the leaving
The forms in 2,71 would presumably correspond to the
verb phrases in 2.72.
2.72a) feel sad
b) try to win
c) argue about money
d) leave
Chomsky (1970:186) argues that sentences like 2.73a
have a structure of Noun Phrase-Verb-Predicate parallel to

37
the Noun Phrase-Verb-Noun Phrase structure of a sentence
like 2.73b.
2.73^) John felt sad.
b) John felt sadness.
There is a derived nominal complement which corresponds
to 2.73b, John's feeling of sadness. Peter Menzel (personal
communication) has pointed out to me that feeling is the nom
inal of a derived nominal complement, and thus would be fol
lowed by a preposition, presumably of, so that the proper
question is why we do not get *the feeling of sad. The pre
sence of the preposition means that any complement of feel
ing within a derived nominal complement must be nominalized.
Hence, derived nominalization can apply to an embedded sen
tence like that in 2.73b, but not to one like that in 2.73a,
unless sad is nominalized to sadness as part of the nominal-
isation of feel.
The nominal complement in 2.71c, ^the arguing about
money, would not be expected to occur because the regular
derived nominal complement corresponding to argue about
money is the argument about money.
Chomsky (19?0s2l4) gives as unacceptable the example.:in
2.71d, the leaving. At least some speakers do accept nomi
nal complements like John's hurried leaving (cf, John left
hurriedly), and I am told that such forms have appeared in
print. My own judgment agrees with that of Chomsky, however.
The dialects that accept forms like John's hurried leaving
present no problem to my analysis, so I will turn to those
dialects that do not accept such forms.

33
In my dialect* leave (=depart), try (=attempt) (cf.
2.71b) _be (-exist, occur, be present) have (=possess)
and live (-reside) (and other verbs) are alike In that they
have no corresponding nominal forms which appear in derived
nominal complememts., (I am refering here to be and have
when they are main surface verbs, not auxiliaries.) The
existence of verbs such as these poses a serious challenge
to the transformationalist position. It would appear that
the rule (or rules) producing derived nominal complements is
blocked from applying to underlying structures with certain
verbs as their predicates. It is Chomskys claim that this
is one of a number of facts supporting the Lexical Hypothe
sis over the transformationalist position.
The verbs mentioned above (which are all of Germanic
origin) occur in sentences like those in 2,7^.
2.7^a) John tried vainly to win the race,"
b) John left hurriedly on the bus.
c) There Is a God.
d) There was a riot yesterday.
e) There is some wine in the bottle.
f) John has a car.
g) Bill usually lives in a hotel.
The Latinate synonyms of those verbs (for meanings
inherent in the sentences in 2,70 occur in sentences like
those in 2.75.
2.75a) John attempted vainly to win the race.
b) John departed hurriedly on the bus,
c) There exists a God.
d) There occurred a riot yesterday.
e) There is some wine present in the bottle.
f) John possesses a car.
g) Bill usually resides in a hotel.

39
I find the sentences in 2.75 to be formal and even awk
ward in comparison to those in 2,7^ a difference presumably
attributable to the choice of predicate.
All of these sentences have acceptable corresponding
gerundive nominal complements, those in 2.76 corresponding
to the sentences in 2.7^ and those in 2.77 corresponding to
the sentences in 2.75.
2,76a) John's trying vainly to win the race
b) John's leaving hurriedly on the bus
c) there being a god
d) there being a riot yesterday
e) there being some wine in the bottle
f) John's having a car
g) Bill's usually living in a hotel
2.77a) John's vainly attempting to win the race
b) John's departing hurriedly on the bus
c) there existing a god
d) there occurring a riot yesterday
e) there being some wine present in the bottle
f) John's possessing a car
g) Bill's usually residing in a hotel
Again, the gerundive nominal complements in 2.76 based
on Germanic verbs seem less formal than the gerundive nomi-
Q
nal complements based on Latinate verbs in 2.77.
Finally, there are no derived nominal complements cor
responding to the sentences in 2.7^ and gerundive nominal
complements in 2.76, the forms in 2.78 being unacceptable.
2.78a) '"John's vain trying to win the race
b) *John's hurried leaving on the bus
c) *the being of a god
d) *the being of a riot yesterday
e) *the being of some wine in the bottle
f) ^John's having of a car
g) ^Bill's usual living in a hotel

4q
On the other hand, the derived nominal complements in
2.79 which correspond to the sentences in 2*75 and gerund
ive nominal complements in 2.77 > are perfectly acceptable.
2.79a) John's vain attempt to win the race
b) John's hurried departure on the bus
c) the existence of a god
d) the occurrence of a riot yesterday
e) the presence of some wine in the bottle
f) John's possession of a car
g) Bill's usual residence in a hotel
It seemsj therefore, that it is possible for a Germanic
(or native) verb to have no corresponding nominal in a
derived nominal complement when there is a closely synony
mous Latinate verb with such a corresponding nominal. The
transformationalist position may be maintained in the face
of these examples in at least two ways verbs like try,
leave. be, have and live can be marked in the lexicon to
block DERIVED NOMINALIZATION, which ignores the existence of
the Latinate synonyms} or, some sort of lexicaX alternation
may be posited, with the Latinate option being mandatory in
derived nominal complements. I prefer the second position,
end I will discuss some different evidence for such lexical
alternation for the synonyms of be, in Chapter Three.
Action Nominis and the Head Noun Action
in Chapter One I noted that derived nominal complements
did not occur as complements of the head noun action. If
action nominal complements are a subclass of derived nominal
complements, then they should not occur as complements of
action either. The examples in 2.80 show that this is
indeed the case.

2.80a) ^John's action of opening of the door
b) *John's action of sinking of the boat
c) *John's action of dimming of the light
d) ^John's action of refusing of the offer
e) ^John's action of proving of the theorem
Menzel (1970) argues that the head nouns associated
with nominal complements in specific structures define what
the nominal complement is, i.e., a complement of event is an
event, a complement of action is an action, etc. It is
ironic that those structures called "action nominis" cannot
be complements of the head noun action, and thus are not
actions.
Summary
I have shown in this section that the so-called "action
nominis have the surface structure of derived nominal com
plements. That is, the nominal complements sometimes called
"action nominis" share with other derived nominal comple
ments the features of a possible variety of determiners,
prenominai adjectives corresponding to adverbs in sentences,
prepositions with all objects, the possibility of penaliza
tion (although less for action nominis than is so for other
derived nominal complements), the complete absence of auxi
liaries, and the restriction from being the complement of
the head noun action. The fact that action nominal comple
ments have nominis formed with the suffix -ing, as do ger
undive nominal complements, ought to have no bearing on this
classification.

42
Agentive Nominis
Agentive nominal complements share the characteristics
of derived nominal complements. The examples in 2.81, cor
responding to the sentences in 2 82, show that such comple
ments may have determiners other than possessive nouns, pre-
nominal adjectives, object prepositions and pluralization.
Agentive nominal complements do not have any reflexes of
auxiliaries.
2.8la) the short-sighted designers of this building
b) the unlucky holders of Imperial Russian bonds
c) the greedy despoilers of the .Earth
2,82a) The ones who designed this building were
short-sighted.
b) The ones who held Imperial Russian bonds
were unlucky.
c) The ones who despoil the Earth are greedy.
Some characteristics of agentive nominal complements
deserve further comment. Agentive nominal complements never
have possessive nouns as determiners which correspond to
underlying agents. When agentive nominal complements do
have possessive nouns as determiners, as in 2.83, the posses
sive nouns correspond to the underlying objects of the verbs
in sentences which correspond to the agentive nominal comple
ments, as in 2.84.
2.83a) Mike's helper
b) General Electric's workers
c) America's educators
2.84a) Someone helps Mike.
b) Some people work for General Electric.
c) Some people educate America.

It is usually possible to get an agentless passive sen
tence closely corresponding to such agentive nominal comple
ments, as in 2,85a and c, although 2.85b seems odd,
2,85a) Mike is helped,
b)??General Electric is worked for.
c) America is educated. (Not the stative reading.)
However, the sentences in 2,84 (and 2.85) are not the
only ones corresponding to the agentive nominal complements
in 2.83, For instance, on a different reading, 2.83c, Amer
icas educators, corresponds to the educators of America.
The absence of possessive nouns corresponding to under
lying agents in agentive nominal complements is due to the
fact that sentences corresponding to agentive nominal comple
ments always have nonspecified agents. Other derived nomi
nal complements'may also correspond to sentences with nonspe
cific agents, as in the refusal of the offer, which, like
the agentless passive sentence, the offer was refused, cor
responds to an underlying structure of the form SOMEONE
refused the offer.
The prenominal adjectives in agentive nominal comple
ments are not always related to adverbs in corresponding
sentences, Vendler (1968) points out that beautiful in the
beantifu1 dancer may correspond in meaning to beautiful in
the dancer is beautiful, or to beautifully in SOMEONE dances
be autifully,
The nominis in agentive nominal complements also seem
to more readily form compounds with their objects than do
the nominis of other derived nominal complements, Thus, we

44
find many compounds like those in 2.86 corresponding to the
agentive nominal complements in 2.87.
2.86a) lion tamer
b) bookkeeper
c) shock absorber
2.87a) tamer of lions
b) keeper of books
c) absorber of shocks
But some nominis in other derived nominal complements
also form such compounds, as in 2.88, corresponding to the
derived nominal complements in 2.89.
2.88a) token payment
b) tax assessment
c) art collection
2.89a) payment of a token
b) assessment of a tax
c) collection of art
Agentive nominal complements have the same surface
structure as derived nominal complements. I see no reason
1* _
to not include agentive nominal complements in the class of
derived nominal complements.
Summary
In this chapter I have defined derived nominal comple
ments in terms of surface structure as those nominal comple
ments which allow a variety of determiners, allow prenominal
adjectives which correspond to adverbs in sentences, have
prepositions with all objects, and allow the nominal to be
pluralized, but which have no auxiliaries. Using this defi
nition, I have then argued that the structures known as
action nominis and agentive nominis are really subclasses
of the class of derived nominal complements.

NOTES
^ Chomsky presumably also excludes agentive nominis
formed with -er, although he never mentions them except to
argue against the positing of abstract verbs underlying
nominis which otherwise have no corresponding predicates.
Agentive nominal complements share several characteristics
with derived nominal complements, and I will argue below
that agentive nominal complements are a special class of
derived nominal complements.
O
The derived nominal complements in 2.4a and 2.6a seem
quite strange to me, but 2.4a, at least, is acceptable in a
context such as the eagerness to -please shown by John, which,
however, seems to correspond to the sentence, an eagerness
to olease was shown by John, and not to the sentence in 2.1a.
3
'The preposition of is the unmarked form. Some verbs
have corresponding derived nominal complements with other
prepositions, e.g., attack;attack on.
4
Simple past tense seems to imply a single occurrence
of an event unless otherwise specified. Present perfect
tense seems to imply only at least one occurrence of an event.
^ The constraint on the occurrence of adjacent forms
with the -ing suffix, which blocks constructions like that
in 2.35c, is discussed in Ross (1972b) and Milsark (1972) .
45

46
The formation of derived nominal complements corres
ponding to sentences with predicate nouns does not seem to
be completely free. Thus, while Johns manhood is possible,
it corresponds to a limited meaning of John is a man, and
Jane's womanhood seems very strange. Nevertheless, many
predicate nouns do have corresponding derived nominis, as
can be seen in the sentences in i with the corresponding
derived nominal complements in ii.
i.a) Joan enjoys being a mother.
b) Ralph was governor recently.
c) Billy is a minor.
d) Stephen was a martyr.
e) John is a member in good standing in the lodge.
f) Sam is legally a pauper.
iia) Joan enjoys motherhood.
b) Ralph's recent governorship
c) Billy's minority
d) Stephen's martyrdom
e) John's membership in good standing in the lodge
f) Sam's legal pauperdom
Chomsky (1970:198-99) argues that the inclusion of verbs
and adjectives in a category of nredicator, as in G. Lakoff
(1970:115ff)j is wrong, since nouns share the same distri
butional properties. The examples above of derived nominis
corresponding to predicate nouns supports Chomsky's conclu
sion. Chomsky further claims, however, that such distribu
tional properties are properties of lexical categories. It
appears to be more correct to say that properties such as the
stative-active distinction and the possession of a corres
ponding derived nominal, are properties of predicates. Such
properties are exhibited by verbs (which are always predi
cates), adjectives (which presumably are always predicates
in underlying structures) and predicate nouns.

47
7
The corresponding gerundive nominal complements, as
in i, which I find awkward, but still acceptable, are com
pletely unacceptable to at least some speakers. I have no
explanation for this fact.
ia) the tree's growing slowly
b) the temperature's rising rapidly
c) the table's moving mysteriously
More generally, possessive inanimate nouns are often
awkward. Arnold Zwicky (personnel communication) pointed
out that historically inanimates have not always been able
to form possessives in English. The two facts may be
related.
O
Although move is borrowed ultimately from Latin and
the derived nominal corresponding to the intransitive form
of the verb is formed with -ment. move does not appear to
belong to the class of Latinate verbs described by
Smith (1972).
Q
' I v/ill discuss the fact that there does not take the
possessive suffix in gerundive nominal complements in more
detail in Chanter Three.

CHAPTER THREE
THE PRODUCTIVITY OF DERIVED NOMINAL COMPLEMENTS
The Evidence for Lack of Productivity
One of the three reasons Chomsky (19?0:187-88) gives
for rejecting a transformational account of the origin of
derived nominal complements in favor of the Lexical Hypothe
sis is that gerundive nominal complements "can be formed
fairly freely from propositions of subject-predicate form,"
while "productivity is much more restricted" for derived
nominal complements. In support of this statement, Chomsky
(1970:188-89) refers to the example of sentences like those
in 3-1 which have the corresponding gerundive nominal com
plements in 3.2, and the corresponding derived'nominal com
plements in 3-3 while the sentences in 3* 4 have the corres
ponding gerundive nominal complements in 35 but the expec
ted corresponding derived nominal complements in 3*6 are not
acceptable.
3.1a) John is eager to please.
b) John refused the offer.
c) John criticized the book.
3.2a) John's being eager to please
b) John's refusing the offer
c) John's criticizing the book
33a) John's eagerness to please
b) Johns refusal of the offer
c) John's criticism of the book
48

49
3.4a) John is easy to please.
b) John is certain to win the prize.
c) John amused the children with his stories.
3.5a) John's being easy to please
b) Johns being certain to win the prize
c) John's amusing the children with his stories
3.6a) ^John's easiness to please
b) *John's certainty to win the prize
c) *John's amusement of the children with his stories
Chomsky (1970:189) points out that there are acceptable
derived nominal complements like those in 3-7 which superfi
cially resemble the unacceptable strings in 3-6, and which
correspond to the sentences in 38 and gerundive nominal com
plements in 3-9, and comments, "these discrepancies between
gerundive and derived nominal [complements] call for expla
nation. Specifically, we must determine why the examples of
[3.6] are ruled out although those of [3*7] are permitted."
3.7a) John's eagerness to please
b) John's certainty that Bill will win the prize
c) John's amusement at the children's^antics
3-8a) John is eager to please.
b) John is certain that Bill will win the prize.
c) John was amused at the children's antics.
3.9a) John's being eager to please
b) John's being certain that Bill will win the prize
c) John's being amused at the children's antics
In this chapter I will try to answer Chomsky's question
by showing that sentences like those in 3-^ which do not
have acceptable corresponding derived nominal complements,
have undergone certain transformational rules in their deri
vations which sentences like those in 3*8, which do have
acceptable corresponding derived nominal complements, have
not undergone. Chomsky (1970:191) claims that this is so

50
because derived nominal complements are noun phrases in deep
structure, not embedded propositions. He claims that
derived nominal complements correspond only to deep struc
ture phrase markers and a few transforms. I will argue that
most rules are blocked from applying to derived nominal com
plements, or have different conditions on their applicabili
ty to derived nominal complements, because they are nominal-
ized embedded propositions. I will also show that many
rules can apply in the derivation of gerundive nominal com
plements that cannot apply in the derivation of derived nom
inal complements.
I will show in detail how the application of a number
of rules is blocked or modified in the derivation of derived
nominal complements. I will discuss the effect of adopting
a predicate-initial analysis on the formulation of rules,
and certain consequent simplifications of the rules, parti
cularly with regard to derived nominal complements. I will
argue that certain facts of derived nominal complements
raise problems for the currently accepted formulation of the
rule of IT-EXTRAPOSITION, and that with a predicate-initial
analysis, no rule of IT-EXTRAPOSITION is needed in a grammar
of English. Throughout this chapter I will show that when
ever a sentence does not have an acceptable corresponding
derived nominal complement, it is because the sentence has
undergone a rule in its derivation which is blocked from
applying to derived nominal complements.^

51
Chomsky's Subcategorization Account
I will first consider Chomsky's (1970s191) claim that a
correct account of the facts of productivity of derived nom
inal complements should be based on subcategorization fea
tures. The pair of sentences in 3-10 and the pair of sen
tences in 3-11 with their corresponding gerundive and
derived nominal complements, present parallel problems.
3*10a) John is easy to please,
b) John is eager to please.
3.11a) John is certain to win the prize.
b) John is certain that Bill will win the prize.
The corresponding gerundive nominal complements in 3-12
and 3*13 and the corresponding derived nominal complements
in 3.14b and 3*154 are all acceptable, while the expected
corresponding derived nominal complements in 3l4a and 315a
are not acceptable,
3.12a) John's being easy to please
b) John's being eager to please
3,13a) John's being certain to win the prize
b) John's being certain that Bill will win the prize
3.14a) *John's easiness to please
b) John's eagerness to please
3.15a) ^John's certainty to win the prize
b) John's certainty that Bill will win the prize
Chomsky (1970:191) attempts to explain the acceptabili
ty of the derived nominal complements in 3>l4b and 3-154., in
contrast with the unacceptability of forms like those in
3.14a and 3-15a, in terms of the subcategorization features
of eager, easy and certain. Chomsky states that eager is
entered in the lexicon with a strict subcategorization

52
feature indicating.that it can take a sentential complement,
as in 3>l6, derived from underlying structures something
like those in 3-17*
3.16a) John is eager to please.
b) John is eager for us to please.
3.17a) John is eager f^John please SOMEONE"];,
b) John is eager f,-,we please SOMEONE ~L
Chomsky says that no further comment is necessary to
account for the acceptability of the derived nominal comple
ments in 3*18.
3.18a) John's eagerness to please
b) John's eagerness for us to please
According to the Lexical Hypothesis, the lexical entry
for eager is also the entry for eagerness and the strict
subcategorization feature applies to adjective and nominal
alike. Thus, eager can appear in a construction of the form
Noun Phrase-Predicate-Sentential Complement, and eagerness
can appear in a construction of the form Possessive
Noun-Nominal-3 entential Complement.
Chomsky claims that, on the other hand, there is no
such subcategorization feature in the lexical entry for easy
and that there are no base phrase markers of the form
easy-5entential Complement. Chomsky (1970tl91) says that
easv appears in base phrase markers as an adjective predi
cated of propositions as subject, as in the sentences in
3.19, with sentences like those in 3-20 derived by IT-EXTRA-
POSITION, and with sentences like those in 3-21 derived in
- 2
turn from extraposed sentences by TO LJ OH -MO v .MEN T.

53
3.19a) To please John is easy.
b) For us to please John is easy.
3.20a) It is easy to please John.
b) It is easy for us to please John.
3-21a) John is easy to please.
b) John is easy for us to please.
. 4
3.22a) John is eager to please.
b) John is eager for us to please.
Even though we get sentences like those in 3*21, which
in surface form exactly parallel those in 3*22, Chomsky
argues that just as easy cannot be introduced into struc
tures of the form Subject-Predicate-Sentential Complement,
easiness cannot be introduced into structures of the form
Possessive Noun-Nominal-Sentential Complement, thus prevent
ing the formation of derived nominal complements like those
in 3-23-
3.23a) ''John's easiness to please
b) ^John's easiness for us to please
Chomsky assumes that gerundive nominal complements are
transformationally derived from structures which are in turn
derived from base structures. That is, GERUNDIVE NOMINALI-
ZATION applies to embedded sentences which have already been
subject to all or almost all cyclical rules. It is there
fore possible to have gerundive nominal complements like
those in 3-24.
3*24a) John's being easy to please
b) John's being easy for us to please
Chomsky is assuming that transformations apply to struc
tures dominated by an S but not to structures dominated by
an NP even though the two types of structures may be com
posed of the same lexical items.

54
Similarly, Chomsky (1970:191) states that certain, with
the meaning used in 3.lib, John is certain that Bill will
win the prize, also has a subcategorization feature which
allows certain in this meaning to take a sentential comple
ment in a structure like ohn]^p[~ypfybe certain^j^Bill
will win the prize~)^,~|^p~j^, from which 3.11b is derived.
Thus, the derived nominal certainty, corresponding to this
meaning of certain, has the same subcategorization feature,
and can appear in derived nominal complements with a senten
tial complement, as in 3.15b, John's certainty that Bill
will win the -prize.
Like the lexical entry for easy, the lexical entry for
certain in the meaning used in 3.11a John is certain to win
the prize, has no subcategorization feature allowing certain
in this meaning to take a sentential complement. Chomsky
says that certain in this meaning appears in bse phrase
markers as an adjective predicated of propositions as sub
ject, as in for John to win the prize is certain. Applying
the same arguments Chomsky used with easy, sentences like
it is certain for John to win the prize are derived by
PT-EXTKAPQSITIQN, and sentences like John is certain to win
the prize are derived by RAISING-TO-SUBJECT from extraposed
5
sentences,
According to the Lexical Hypothesis, derived nominis
occur in nominal structures corresponding to the base phrase
markers of sentences. Such nominal structures are not sub
ject to most rules which apply to sentences. The derived

55
nominal certainty corresponding to the second meaning of
certain^ like easiness, cannot occur in nominal structures
corresponding to derived phrase markers,
Chomsky has argued that the lack of productivity of
derived nominal complements is explained by the Lexical
Hypothesis, Under that hypothesis, *John's easiness to
please is not acceptable because the lexical entry underly
ing easy and easiness does not have a subcategorization
feature allowing sentential complements. On the other hand,
John's eagerness to please is acceptable because the lexical
entry underlying eager and eagerness has a subcategorization
feature allowing sentential complements.
RAISING Rules
RAI5ING-T0-0BJECT
As was indicated in Note 2 of this chapter, RAISING is
a general term for three different movement transformations.
The first movement I will discuss here is from subject of an
embedded sentence to object of the next highest sentence, as
has occurred in the derivation of the sentences in 3.25
which are derived from the same underlying structures as the
sentences in 3.26.^
3,25a) John believes himself to be heroic.
b) Bill expects them to be here soon.
c) Dick is believed to have been involved by everyone.
3.26a) John believes that he is heroic.
b) Bill expects that they will be here soon.
c) Everyone believes that Dick was involved,
That the nouns in question have been raised to object
position in the higher sentences is indicated by the reflex-

ive form himself in 3*25a, the accusative them in 3.25b and
the fact that Dick is the derived subject in a passive sen
tence in 3*25c. Reflexivization is restricted to clause
mates in English, pronouns take the accusative form in
object position and an underlying object becomes the surface
subject in passive sentences.
The sentences in 3-25* which have undergone RAIS-
ING-TO-OBJECT, have the acceptable corresponding gerundive
nominal complements in 3-2?, but no acceptable correspond
ing derived nominal complements, the forms in 3-28 being
unacceptable.
3.27a) John's believing himself to be heroic
b) Bill's expecting them to be here soon
c) Dick's being believed by everyone to have been
involved
3.28a) -'John's belief of himself to be heroic
b) *3ill's expectation of them to be here soon
c) *Dicks belief by everyone to have been involved
-
The sentences in 3-26, which have not undergone RAIS-
ING-TG-OBJECT, have the acceptable corresponding gerundive
nominal complements in 3*29 and derived nominal complements
in 330.
3-29a) John's believing that he is heroic
b) Bill's expecting that they will be here soon
c) everyone's believing that Dick was involved
3.30a) John's belief that he is heroic
b) Bill's expectation that they will be here soon
c) everyone's belief that Dick was involved
RAISING-TO-SUBJECT
The second movement is from subject of an embedded sen
tence to subject of the next higher sentence, as has occurred

57
in the derivation of the sentences in 3.31? which are
derived from the same underlying structures as the sentences
in 3.32.
3.3.1a) John is certain to win the prize.
b) Bill is likely to be drafted.
c) Jerry appeared to open the door.
3.32a) It is certain that John will win the prize.
b) It is likely that Bill will be drafted.
c) It appeared that Jerry opened the door.
The sentences in 331s which have undergone
RAISING-TO-SUBJECT, have the acceptable corresponding ger
undive nominal complements in 3.33* tut no acceptable cor
responding derived nominal complements the forms in 3.34
being unacceptable,
3.33a) Johns being certain to win the prize
b) Bills being likely to be drafted
c) Jerry's appearing to open the door
3.34a) ^John's certainty to win the prize
b) ^Bills likelihood to be drafted
c) *Jerrys appearance to open the door'
The sentences in 332, which have not undergone
RAISING-TO-SUBJECT, have the acceptable corresponding ger-
8
undive nominal complements in 3.35.
3.35a) it(s) being certain that John will win the prize
b) it(s) being likely that Bill will be drafted
c) it(s) appearing that Jerry opened the door
The apparently corresponding derived nominal comple
ments in 3.36 are not acceptable, but the derived nominal
complements in 3.37* which have the same word order as the
sentences in 3.32, but do not have any reflex of it, are
acceptable.

53
3.36a) *its certainty that John will win the prize
b) "its likelihood that Bill will be drafted
c) "its appearance that Jerry opened the door
3.37a) the certainty that John will win the prize
b) the likelihood that Bill will be drafted
c) the appearance that Jerry opened the door
I will argue below (cf. Dummy Subject Insertion) that
it in sentences like those in 3*32 is inserted, and not pre
sent in underlying structures. The unacceptability of the
forms in 3*36, and the acceptability of the forms in 3*37 is
simply explained by blocking IT-INSERTION in derived nominal
complements.
TOUGH-MOVEMENT
The third RAISING movement is from object of an embed
ded sentence to subject of the next higher sentence, as has
occurred in the derivation of the sentences in 3-33, which
are derived from the same underlying structures as the sen
tences in 3*39.
3.33a) John is easy to please.
b) Algebra is difficult to learn,
c) This test is fun to take.
3.39a) It is easy to please John.
b) It is difficult to learn algebra.
c) It is fun to take this test.
The sentences in 3'3Sj which have undergone TOUGH-MOVE
MENT, have the acceptable corresponding gerundive nominal
complements in 3-^0, but no acceptable corresponding derived
nominal complements, the forms in 3-^1 being unacceptable. .
3>40a) John's being easy to please
b) algebras being difficult to learn
c) this test's being fun to take

59
3.4la) *John's easiness to please
b) '"'algebra's difficulty to learn
c) "'this test's fun to take
The sentences in 3-39i which have not undergone
TOUGH-MOVEMENT. have the acceptable corresponding gerundive
nominal complements in 3-42.
3.42a) it(s) being easy to please
b) it(s) being difficult to learn algebra
c) it(s) being fun to take this test
The apparently corresponding derived nominal complements
in 3-43 are not acceptable, but those in 3*44 are acceptable.^
The argument used with the examples in 3-36 and 3*37 applies
here as well.
3.43a) *its easiness to please John
b) *its difficulty to learn algebra
c) "its fun to take this test
3-44a) ?the easiness of pleasing John
b) the difficulty of learning algebra
c) the fun of taking this test
RAISING-TO-OBJECT. RAISING-TO-SUBJECT and.TOUGH-MOVE
MENT do not apply in the derivation of derived nominal com
plements. The three rules share the property of moving a
noun phrase to a higher sentence. They appear to apply at
the same point in the cycle.G. Lakoff (1968) has argued
that there is only one rule of RAISING (IT-REPLACEMENT in
his paper). To collapse the three types of RAISING to one,
Lakoff has to write a complex rule, which includes simul
taneous structural descriptions. I consider such a rule to
be unlikely in a natural language. I will, therefore, sus
pend judgment on whether it is possible to collapse the
RAISING rules.

IT-EXTRAPOSITION
Introduction
The rule of IT-EXTRAPOSITION was proposed by Rosenbaum
(1967) to account for the relationship of sentences like
those in 3*45 To the sentences in 3*46.'*'"*
3-45a) That John left so early is surprising.
b) For John to quit now would be a disaster.
c) That Mary will come to her senses is to be
hoped for.
346a) It is surprising that John left so early.
b) It would be a disaster for John to quit now.
c) It is to be hoped for that Mary will come to
her senses.
In Rosenbaum's analysis, the subject t in the sentences
in 3-46 is present in deep structure as the head noun of
noun phrases with the structure of 3-47. 'The rule of IT-EX-
TRAPOSITION is given in 3.48.
it that John left so early
3.48) IT-EXTRAPOSITION (Optional)
X N S Y
[+PRO(i.e. it)] ==> 1, 2, 4+3
1 2 3 4
If IT-EXTRAPOSITION does not apply, the head pronoun it
is deleted. If IT-EXTRAPOSITION does apply, moving the
embedded sentence to the right end of the next higher sen
tence, the head pronoun it is isolated in subject position,
and no longer subject to deletion.
In this section I will first discuss the derivation of
certain examples via the rule of IT-EXTRAPOSITION, showing

61
that the pattern of occurrence of derived nominal comple
ments casts doubt on the validity of IT-EXTRAPOSITION as pro
posed by Rosenbaum. I will then consider other evidence
counter to his analysis* and argue that there is no rule of
IT-EXTRAPOSITION in English grammar. In a later section
(cf. NP-PREPQ5ING) I will present an alternate analysis of
the relationship between the sentences in 3-45 and 3-46.
With Subjects of Predicate Ad.iectives
We have seen above that sentences like those in 3*49
have acceptable corresponding gerundive nominal.complements,
as in 3*50, and derived nominal complements, as in 3ol
3.49a) It is certain that John will win the prize.
b) It is difficult to learn algebra.
c) It is fun to take this test.
3*50s.) it(s) being certain that John will win the prize
b) it(s) being difficult to learn algebra
c) it(s) being fun to take this test
3*51ai) the certainty that John will win'-the prize
ii) the certainty of John's winning the prize
b) the difficulty of learning algebra
c) the fun of taking this test
According to Chomsky's (1970) analysis, the sentences
in 3.49 are derived from the phrase markers underlying the
sentences in 3*52 by the rule of IT-EXTRAPOSITION. The sen
tences in 3.52 do not have acceptable corresponding gerund
ive nominal complements or derived nominal complements, as
is shown by the unacceptability of the expected forms in
3.53 and 3-54.
3-52ai) That John will win the prize is possible,
ii) John's winning the prize is certain.
bi) To learn algebra is difficult,
ii) Learning algebra is difficult.

62
352ci) To take this test is fun.
ii) Taking this test is fun.
3-53ai) *that John will win the prize's being certain
ii) ^John's winning the prizes being certain
bi) *to learn algebra's being difficult
ii) *learning algebra's being difficult
ci) *to take this test's being fun
ii) *taking this test's being fun
3o4ai) *that John will win the prize's certainty
ii) ^John's winning the prize's certainty
bi) *to learn algebras difficulty
ii) ^learning algebra's difficulty
ci) *to take this test's fun
ii) ^taking this test's fun
The Lexical Hypothesis predicts that there should be
derived nominal complements corresponding to the sentences
in 3*52f which are supposedly closer to base phrase markers
than the extraposed sentences in 3*49, but the unacceptabi-
litv of the strings in 3*53 and 3*5^ in contrast to the
acceptability of the nominal complements in 3-50 and 3*51
12
is counter to that prediction.
With Passives
By Rosenbaum's analysisf IT-EXTRAPOSITION also applies
to passive sentences with complements in the derived subject
position. Thus, the extraposed sentences in 3*55 are
derived from the phrase markers underlying the passive sen
tences in 3-56. The sentences in 3-56 do not have any accep
table corresponding derived nominal complements, the expec
ted forms in 3-57 being unacceptable, but the extraposed pas
sive sentences in 3-55 have the corresponding derived nomi
nal complements in 3*58.
3-55a) It was discovered by John that Jerry was a
bigamist.
b) It is doubted by them that you will go.

3-5oa) That Jerry was a bigamist was discovered by John,
b) That you will go is doubted by them.
3-5?a) *that Jerry was a bigamist's discovery by John
b) '-'that you will go's doubting by them
3.53a) the discovery by John that Jerry was a bigamist
b) the doubting by them that you will go
Chomsky suggests that derived nominal complements such
as those in 3-59 are derived by the obligatory application
of a rule called AGENT-POSTPOSING (cf. Passive Sentences
below) to the underlying noun phrases in 3*60.
3*59a) the necessity for John to leave
b) the likelihood that John will leave
3-Oa) Tofor John to leaver's necessity
** w
b) that John will leaver's likelihood
This solution will explain why the examples in 3*51 and
3.59 are acceptable while those in 3-53 and 35^ are not.
AGENT-POSTPOSING seems to be a poor name for a rule which
moves complements in such structures, however, "and there is
a curious problem with forms like those in 3-57 and 3-53.
If the fact that the derived nominal complements in 3-53 are
acceptable but not those in 3*57 is to be explained by the
ad hoc obligatory application of AGENT-POSTPOSING, we are
left with the problem of accounting for the post-verbal
by-phrases in 3-58. I must reject Chomsky's implied claim
that AGENT-POSTPOSING has applied in the derivations of the
derived nominal complements in 3*58, but not in those in
3.591 and in so doing, reject his claim that AGENT-POSTPOS
ING moves any complement within derived nominal complements.

64
With Psychological Predicates
IT-EXTRAPOSITION can also apply to sentences like those
)
in 3*6l. These are active sentences with complements as
subjects. The verbs which permit this construction share a
number of other characteristics, and will be called psycho
logical predicates here (cf. Postal, 197109-54, and the
section Psychological Predicates, below). The sentences in
3.61 are subject to IT-EXTRAPOSITION in Rosenbaum's analysis,
yielding the sentences in 3*62.
3-6la) That John is here surprises me.
b) For Mary to be so late worries Bill.
c) For Jim to leave now would disturb Alice.
3.62a) It surprises me that John is here.
b) It worries Bill for Mary to be so late.
c) It would disturb Alice for Bill to leave now.
Neither the sentences in 3-61 nor the sentences in 3*62
have acceptable corresponding derived nominal complements,
as shown by the unacceptable forms in 3*63 and ^3*64, and
there are no acceptable derived nominal complements with
determiners which are not possessive nouns corresponding to
the extraposed sentences in 36l, as is shown by the unac
ceptable examples in 3-65-
3.63a) that John is here's surprise to me
b) *for Mary to be so late's worry to Bill
c) "'for Jim to leave now's disturbance of Alice
3.64a) *its surprise to me that John is here
b) "its worry to bill for Mary to be so late
c) -its disturbance of Alice for Bill to leave now
3.65a) *the surprise to me that John is here
b) *the worry to Bill for Mary to be so late
c) "the disturbance of Alice for Bill to leave now

65
Psychological predicates appear to be the only predi
cates which form extraposed sentences which do not have
acceptable corresponding derived nominal complements.
Further Problems with the Rule
The acceptable derived nominal complements in 3-51*
3.58 and 3-59 all show the same order of major constituents
as the extraposed sentences in 3*^6, 3-^9 and 3-55- There
is no subject pronoun it in these derived nominal comple
ments. This fact can be explained by having rt in extra
posed sentences supplied by a rule of IT-INSERTION. which
does not apply to derived nominal complements (cf. Dummy
Sub.iect Insertion, below) The arguments for not having it
present in deep structure are well summarized in Stockwell
et al. (19?3s52?ff.). Given the absence of it in underlying
structures, the derived nominal complements in 3-51 3-58
and 3-59 may be taken as corresponding directly- to the extra
posed sentences cited above. If Chomsky's claim that derived
nominal complements correspond only to base phrase markers
is at all correct, then extraposed sentences would appear to
be more basic than nonextraposed sentences.
Although this conclusion goes against a generally accep
ted analysis, there is some independent supporting evidence.
The intransitive verbs seem, appear and happen take subject
13
complements which must be extraposed.'"^ The extraposed sen
tences in 366 are acceptable while the nonextraposed ones
in 3*67 are not.

66
3.66a) It seems that John is late.
b) It appears that Mildred has fallen down.
c) It happens that Mike is a brilliant student.
3.o7a) -That John is late seems.
b) *That Mildred has fallen down appears.
c) *That Mike is a brilliant student happens.
It would seem that at least some sentences can be
derived only by obligatory application of IT-EXTRAPOSITION.
I have noted above in Chapter One certain arguments for
assuming an underlying predicate-initial order in English.
All extraposed sentences, including those resulting from the
obligatory application of IT-EXTRAPOSITION. have a predi
cate-initial order, if the presence of the semantically
empty pronoun it in subject position is discounted. I pro
pose that there is no rule of IT-EXTRAPOSITION in English,
and that nonextraposed sentences are derived from the struc
tures underlying extraposed sentences. I will discuss the
details of such derivations in the section on NP-PREPOSINC-
below.
Passive Sentences
I will turn next to a problem, which is not raised by
Chomsky's examplesj that of derived nominal complements cor
responding to passive sentences. The sentences in 3*68 have
the corresponding gerundive nominal complements in 3-69 and
derived nominal complements in 3.70.
3.68a) John refused the offer.
b) John criticized the book.
c) The enemy destroyed the city.
d) The jury acquitted Abby.
3.69a) John's refusing the offer
b) John's criticizing the book

6?
3*69c)
d)
the enemy's destroying the city
the jury's acquitting Abby
3-70a)
b)
John's refusal of the offer
John's criticism of the book
c)
d)
the enemy's destruction of the city
the jury's acquittal of Abby
The sentences in 3-68 also have the corresponding pas
sive sentences in 3-71*
3*71a)
b)
c)
d)
The offer was refused by John.
The book was criticized by John.
The city was destroyed by the enemy.
Abby was acquitted by the jury.
The passive sentences in 3-71 have the corresponding
gerundive nominal complements in 372, but only the sen
tences in 3-71c and d have the corresponding derived nominal
complements in 3-73c and d, while the strings in 3* 73a and b,
which seem to correspond to the sentences in 3-71a and b,
are not acceptable.
3 72a)
b)
c)
d)
the offer's being refused by John
the book's being criticized by John
the city's being destroyed by the enemy
Abby's being acquitted by the jury
3.73a)
b)
c)
d)
*the offer's refusal by John
*the book's criticism by John
the city's destruction by the enemy
Abby's acquittal by the jury
The derived nominal complements in 3*74, all of which
are acceptable, are related to the derived nominal comple
ments in 3*70 and 3*73* but do not correspond exactly to any
sentences. These derived nominal complements would corres
pond to predicate-initial underlying structures.
3*7^a)
b)
c)
d)
the refusal of the offer by John
the criticism of the book by John
the destruction of the city by the enemy
the acquittal of Abby by the jury

63
Chomsky's claim that derived nominal complements corres
pond only to base phrase markers creates a problem here if a
predicate-initial analysis is not adopted. Not only are
there derived nominal complements corresponding to active
sentences, as in 3-70, there are also derived nominal comple
ments corresponding to some, but not all, passive sentences,
as in 3-73t and there are derived nominal complements which
do not correspond to any sentences at all (what might be
called "half-passives"), as in 3*7^-
Chomsky (1970202ff.) proposes two transformations
which must both apply in the derivation of passive sentences
from phrase markers parallel to those underlying active sen
tences, one of which also can apply to noun phrases (includ
ing derived nominal complements). Chomsky proposes a rule
of NP-PREPOSING which will transform phrases like those in
3-75 into phrases like those in 3-76. This rul'e thus applies
to noun phrases whether they have a simple noun or a derived
nominal as their head.
3<75a) the picture of John
b) the bottom of the barrel
c) the destruction of the city
d) the murder of John
3.76a) John's picture
b) the barrel's bottom (the literal meaning)
c) the city's destruction
d) John's murder
There are similar phrases which do not undergo the rule
of NP-PREPOSING, and again the phrases can have either a sim
ple noun or a derived nominal as their head. The phrases in
3-77 do not have acceptable corresponding forms like those
in 3-78.

69
377a) the algebra of revolution
b) the strategy of war
c) the refusal of the offer
d) the criticism of the book
3-78a) ^revolution's algebra
b) *war's strategy
c) *the offer's refusal
d) *the book's criticism
Chomsky proposed a rule of AGENT-PQSTPOSING to move the
subject noun phrase to a post-verbal position in phrase mark
ers which are to become passive sentences. NP-PREPOSING
would then apply to move the object noun phrase into
pre-verbal position. Chomsky goes on to say that since pas-
sivizability is a property of verbs (i.e., is a governed
process), then derived nominal complements containing nomi
nis corresponding to such verbs can also be passivized. It
would seem that the passivizability of a verb is best expressed
by having AGENT-P0STP03ING apply optionally to structures
with verbs which can appear in passive sentences. That is,
verbs v/hich can appear in passive sentences are marked as
allowing the optional application of AGENT-P0STP0SING, while
verbs which cannot appear in passive sentences are marked as
not allowing the application of AGENT-POSTPOSING. On the
other hand, NP-PREPQSING applies obligatorily in the deriva
tion of any sentence in which AGENT-PQ3TP0SIKG has already
applied, and optionally in the derivation of any noun phrase
(including derived nominal complements) whose head noun (or
nominal) is marked as allowing the rule. A given lexical
entry may therefore be marked differently for AGENT-POST-
POSING and for NP-PREPOSING.

70
I will claim here that there is a rule of AGSNT-PR3PQ3-
ING (the reverse of Chomsky's AGENT-POSTPOSING) which
applies to predicate-initial underlying structures to yield
active sentences and nominal complements such as those in
3.68, 3-69 and 3*70. If AGENT-PREPOSING does not apply,
then the resulting obligatory application of NP-PREPOSING
gives the passive sentences and gerundive nominal comple
ments in 3*71 and 3-72. If AGENT-PREPOSING does not apply
in the derivation of derived nominal complements, then
NP-PREPOSING applies optionally with some underlying verbs,
giving the pairs of derived nominal complements in 3-73c
and d and 3-7^c and d, and not at all with other verbs,
giving the derived nominal complements in 3*7^a and b.
By Chomsky's analysis, passive sentences are derived
from structures underlying active sentences by the applica
tion of two rules, both of which must apply. Ey my analysis,
all sentences are derived from underlying predicate-initial
structures, with active sentences derived by one rule and
passive sentences by another rule.
NP-PREPOSING
With Simle Noun Phrases
The application of AGSNT-PRBPOSING to underlying pre
dicate-initial structures with agents yields active sen
tences like those in 3*79 and their corresponding gerundive
nominal complements in 3-80 and derived nominal complements
in 3.81.

71
3-79a) The enemy destroyed the city.
b) The jury acquitted Abby.
c) John refused the offer.
d) John criticized the book.
3.80a) the enemy's destroying the city
b) the jury's acquitting Abby
c) John's refusing the offer
d) John's criticizing the book
38la) the enemy's destruction of the city
b) the jury's acquittal of Abby
c) John's refusal of the offer
d) John's criticism of the book
If AGENT-PROPOSING does not apply to the structures
underlying the active sentences and nominal complements in
3-79 3-80 and 3*81, then the rule of NP-PREPOSING applies
obligatorily in the derivation of passive sentences and ger
undive nominal complements like those in 3-82 and 3*83-
3.82a) The city vas destroyed by the enemy.
b) Abby was acquitted by the jury.
c) The offer was refused by John.
d) The book was criticized by John.
3.83a) the city's being destroyed by the*enemy
b) Abby's being acquitted by the jury
c) the offers being refused by John
d) the book's being criticized by John
Either AGENT-PREPOSING or NP-PREPOSING must apply in
the derivation of sentences and gerundive nominal comple
ments. Forms like those in 3-84 and 3-85 in which neither
rule has applied, are not acceptable.
3.84a) *Was destroyed the city by the enemy.
b) *Was acquitted Abby by the jury.
c) *Was refused the offer by John.
d) *Was criticized the book by John.
3.85a) being destroyed the city by the ene;ny
b) "'being acquitted Abby by the jury
c) *being refused the offer by John
d) *being criticized the book by John

72
On the other hand, in the derivation of derived nominal
complements, if AGENT-PROPOSING does not apply, then the
application of NP-PROPOSING is optional for some nominis
(or underlying verbs), giving the paired acceptable derived
nominal complements in 3-86a and b and 3* 87a and b, and
blocked for other nominis, so that the derived nominal com
plements in 3-37c and d are acceptable, but not those in
3.86c and d.
3.86a) the citys destruction by the enemy
b) Abby's acquittal by the jury
c) *the offer's refusal by John
d) *the book's criticism by John
3.87a) the destruction of the city by the enemy
b) the acquittal of Abby by the jury
c) the refusal of the offer by John
d) the criticism of the book by John
If there is no specified agent noun phrase in an under
lying structure,
then NP-PREPOSING applies obligatorily in
the derivation of sentences, as in 3*83 and gerundive nominal
complements, as in 3-89, with the forms in 3-90 and 3-91
being unacceptable.
3.88a) The city was destroyed.
b) Abby was acquitted.
c) The offer was refused.
d) The book was criticized.
3.89a) the city's being destroyed
b) Abby's being acquitted
c) the offer's being refused
d) the book's being criticized
3.90a) *Was destroyed the city.
b) *Was acquitted Abby.
c) *Was refused the offer.
d) *Was criticized the book
3-91a) *being destroyed the city
b) *being acquitted Abby
c) *being refused the offer
d) *being criticized the book

73
Again, when AGENT-PROPOSING has not applied in the
derivation of a derived nominal complement, the application
of NP-PREPOSING is optional for some nominis and blocked
for others, as indicated by the derived nominal complements
in 3-92 and 3*93*
3-92a) the city's destruction
b) Abby's acquittal
c) *the offer's refusal
d) *the book's criticism
3-93a) the destruction of the city
b) the acquittal of Abby
c) the refusal of the offer
d) the criticism of the book
Finally, if the underlying predicate-initial structure
contains an intransitive predicate,
then NP-PREPOSING
applies obligatorily in the derivation of sentences like
those in 3*9^ and gerundive nominal complements like those
in 3*95 with the forms in 3*96 and 3*97 being unacceptable.
3-9^a) The boat sank suddenly.
b) John arrived.
c) Mary is kind.
d) Bill is friendly.
3.95a) the boat's sinking suddenly
b) John's arriving
c) Mary's being kind
d) Bill's being friendly
3.96a) *3ank the boat suddenly.
b) ^Arrived John.
c) *Is kind Mary.
d) *Is friendly Bill.
3.97a) *sinking the boat suddenly
b) ^arriving John
c) *being kind Mary
d) *being friendly Bill
Again, the application of NP-PREPOSING in the deri
vation of derived nominal complements corresponding to

7^
intransitive sentences is'optional, so that the derived nom
inal complements in 3.98 and 3-99 are acceptable.
3.98a) the boat's sudden sinking
b) John's arrival
c) Mary's kindness
d) Bill's friendliness
3.99a) the sudden sinking of the boat
b) the arrival of John
c) the kindness of Mary
d) the friendliness of Bill
The application of NP-PRHPQSING to simple noun phrases
shows a simple pattern. If there is no noun phrase in sub
ject position when NP-PRBPOSING is applicable (as is the
case with all intransitive predicates, transitive verbs with
unspecified agents and transitive verbs with specified
agents to which AGENT-PREPOSING has not applied), then
NP-PREPOSING moves the underlying object noun phrase into
subject position. This movement is obligatory in the deri-
vation of sentences and gerundive nominal complements and
optional in the derivation of derived nominal complements
(with the exception that certain nominis corresponding to
transitive verbs block the application of the rule in
derived nominal complements).
With Nominal Complements
As I indicated in Chapter One, I have accepted Menzel's
(19o9) analysis of nominal complements as complements of a
restricted set of head nouns in underlying structures.
These head nouns ai'e optionally deletable, and in the follow
ing examples will be enclosed in parentheses to Indicate
this optionality.

75
Again, I will claim that AGENT-PREP0SING has applied to
predicate-initial underlying structures to yield the sen
tences in 3.100, and their corresponding gerundive nominal
complements in 3.101 and derived nominal complements in 3.102
3.100a)
John complained about Peter's (action of)
b)
insulting Tom.
Jerry testified about Bill's (action of)
c)
leaving Peggy.
Mike attested to (the fact of) Mary's parti
dj
cipation in the crime.
David reported on (the event of) Alices
disappearance.
3.101a)
John's complaining about Peter's (action of)
b)
insulting Tom
Jerry's testifying about Bill's (action of)
c)
leaving Peggy
Mikes attesting to (the fact of) Mary's
d)
participation in the crime
Davids reporting on (the event of) Alice's
disappearance
3.102a)
John's complaint about Peter's (action of)
b)
insulting Tom
Jerry's testimony about Bill's (action of)
c)
leaving Peggy
Mike's attestation to (the fact of) Mary's
d)
participation in the crime
David's report on (the event of) Alice's
disappearance
The verbs complain about, attest to and report on allow
passivization in my dialect, while testify about does not.
That is, AGENT-PREPOSING applies optionally to the first
three predicates and obligatorily to testify about in the
derivation of sentences. If AGENT-PREPOSING does not apply,
then NP-PREPOSING must apply to yield sentences like those
in 3.103, The expected corresponding gerundive nominal com
plements in 3.104 and derived nominal complements in 3.105
are not acceptable (cf. Note 12 above).

76
3.103a)
b)
c)
d)
3 104a)
b)
c)
d)
3.105a)
b)
c)
d)
Failure
Peter's (action of) insulting Tom was com
plained about by Tom.
^Bill's (action of) leaving Peggy v/as testified
about by Jerry.
(The fact of) Mary's participation in the
crime was attested to by Mike.
(The event of) Alice's disappearance was
reported on by David.
^Peter's (action of) insulting Tom's being
complained about by John
^Bill's (action of) leaving Peggy's being
testified about by Jerry
*(the fact of) Mary's participation in the
crime's being attested to by Mike
*(the event of) Alice's disappearance's being
reported on by David
^Peter's (action of) insulting Tom's complaint
about by John
^Bill's (action of) leaving Peggy's testimony
about by Jerry
*(the fact of) Mary's participation in the
crime's attestation to by Mike
*(the event of) Alice's disappearance's report
on by David
of both AGENT-PROPOSING and NP-PREPOSING to
apply in the derivation of sentences and gerundive nominal
complements results in unacceptable forms, as in 3*106 and
3.10?, but derived nominal complements to which neither rule
has applied are acceptable, including one with testimony
about (from testify about) as in 3-108.
3.106a) *Was complained about by John Peter's (action
of) insulting Tom.
b) *Was testified about by Jerry Bill's (action
of) leaving Peggy.
c) -"Was attested to by Mike (the fact of) Mary's
participation in the crime.
d) *Was reported on by David (the event of)
Alice's disappearance.
3.107a) -being complained about by John Peter's (action
of) insulting Tom
b) -being testified about by Jerry Bill's (action
of) leaving Peggy

77
3.107c) *being attested to by Mike (the fact of) Mary's
participation in the crime
d) *being reported on by David (the event of)
Alice's disappearance
3.108a) the complaint by John about Peter's (action
of) insulting Tom
b) the testimony by Jerry about Bill's (action
of) leaving Peggy
c) the attestation by Mike to (the fact of)
Mary's participation in the crime
d) the report by David on (the event of) Alice's
disappearance
While the head nouns of the embedded nominal comple
ments in the examples above may be deleted, they may not be
separated from the complements, i.e., by moving the head
nouns into surface subject position but leaving the comple
ments at the right end of the structure (cf. With That-Com
plements, below), so that all of the examples in 3.109
3.110 and 3.111 are unacceptable.
3.109a)
b)
c)
d)
3.110a)
b)
c)
d)
3.1Ha)
b)
c)
d)
^Peter's action was complained about by John of
insulting Tom,
^Bill's action was testified about, by Mike of
leaving Alice.
-The fact was attested to by Mike of Mary's
participation in the crime.
*The event was reported on by David of Alice's
disappearance
^Peter's action's being complained about by
John of insulting Tom
*Bill's action's being testified about by Jerry
of leaving Peggy
*the fact's being attested to by Mike of Mary's
participation in the crime
*the event's being reported on by David of
Alice's disappearance
*Peter5s action's complaint by John about
insulting Tom
^Bill's action's testimony by Jerry about
leaving Peggy
*the fact's attestation by Mike to Mary's par
ticipation in the crime
"the event's report by David on Alice's dis
appearance

78
Like destroy, acquit, refuse and criticize, the predi
cates complain about, attest to and report on also form
agentless passives. Again, AGENT-PREPOSING does not apply
to underlying structures with unspecified agents. Thus,
NP-PREPOSING must apply in the derivation of sentences, as
in 3.112, but does not apply in the derivation of gerundive
and derived nominal complements, so that the forms in 3.113
and 3.114 are not acceptable.
3.112a) Peters (action of) insulting Tom was com
plained about.
b) (The fact of) Marys participation in the
crime was attested to.
c) (The event of) Alice's disappearance was
reported on.
3.113a) '"'Peters (action of) insulting Tom's being
complained about
b) *(the fact of) Mary's participation in the
crime's being attested to
c) *(the event of) Alice's disappearance's
being reported on
3.114a) "Peter's (action of) insulting Tom's complaint
about
b) *(the fact of) Mary's participation in the
crime's attestation to
c) *(the event of) Alice's disappearance's
report on
Failure of both AGENT-PREPOSING and NP-PREPOSING to
apply in the derivation of sentences and gerundive nominal
complements results in unacceptable forms, as in 3.115 and
3.116, but derived nominal complements to which neither
rule has applied, as in 3.117 are acceptable.
3.115a) *Was complained about Peter's (action of)
insulting Tom.
b) *Was attested to (the fact of) Mary's partici
pation in the crime.
c) *Was reported on (the event of) Alice's dis
appearance.

79
3-lla) -'being complained, about Peter's (action of)
insulting Tom
b) *being attested to (the fact of) Mary's parti
cipation in the crime
c) *being reported on (the event of) Alice's
disappearance
3-117a) the complaint about Peter's (action of)
insulting Tom
b) the attestation to (the fact of) Mary's
participation in the crime
c) the report on (the event of) Alice's dis
appearance
Again while the head nouns of the embedded nominal com
plements in the examples above may be deleted, they may not
be separated from the complements, so that all the examples
in 3*118, 3*119 and 3*120 are unacceptable.
3*ll8a) ^Peter's action was complained about of
insulting Tom.
b) *fhe fact was attested to of Mary's partici
pation in the crime.
c) *The event was reported on of Alice's dis
appearance .
3* 119a) '^Peter's action's being complained about of
insulting Tom
b) "the fact's being attested to of Mary's parti
cipation in the crime
c) *the event's being reported on of Alices
disappearance
3.120a) "Peter's action's complaint about insulting
Tom
b) *the fact's attestation to Mary's participa
tion in the crime
c) '"the event's report on Alice's disappearance
If the underlying predicate-initial structure contains
an intransitive predicate, then NP-PROPOSING applies obliga
torily in the derivation of sentences, as in 3*121, but is
blocked from applying in the derivation of nominal comple
ments, so that the forms in 3*122 and 3*123 are unacceptable.

3.121a)
b)
c)
d)
3.122a)
b)
c)
d)
3.123a)
b)
c)
d)
Failure
apply in the
complements
80
(The fact of) Johns leaving so early was
unexpected.
Bill's (action of) refusing the offer is
unfortunate.
(The event of) Dicks resignation is likely.
(The fact of) Jimmys testimony was remarkable.
*(the fact of) John's leaving so early's being
unexnected
^Bill's faction of) refusing the offer's being
unfortunate
*(the event of) Dick's resignation's being
likely
*(the fact of) Jimmy's testimony's being
remarkable
*(the fact of) John's leaving so early's
une xp e c te dne s s
^Bill's (action of) refusing the offer's
unfortunateness
*(the event of) Dick's resignation's likelihood
*(the fact of) Jimmy's testimony's remarkableness
of both AGENT-PREP03ING and NP-PREP03ING to
derivation of sentences and gerundive nominal
results in unacceptable forms, as in 3-124 and
3.125, but derived nominal complements to which neither rule
has applied, as in 3*126, are acceptable.
3.124a) *Was unexpected (the fact of) John's leaving
so early.
b) *Was unfortunate Bill's (action of) refusing
the offer
c) *Is likely (the event of) Dick's resignation.
d) *Vas remarkable (the fact of) Jimmy's testimony.
3.125a) *being unexpected (the fact of) John's leaving
so early
b) *being unfortunate Bill's (action of) refusing
the offer
c) *being likely (the event of) Dick's resignation
d) *being remarkable (the fact of) Jimmy's testimony
3.12oa) the unexpectedness of (the fact of) Johns
leaving so early
b) the unforunateness of Bill's (action of)
refusing the offer
c) the likelihood of (the event of) Dick's
resignation
d) the remarkableness of (the fact of) Jimmy's
testimony

81
Again, while the head nouns of the embedded nominal com
plements in the examples above may be deleted, they may not
be separated from the complements, so that all the examples
in 3.12?, 3.128 and 3.129 are unacceptable.
3,127a) *The fact was unexpected of John's leaving so
early.
b) ^Bill's action was unfortunate of refusing the
offer,
c) *The event was likely of Dick's resignation,
d) *The fact was remarkable of Jimmy's testimony.
3.128a) *the fact's being unexpected of John's leaving
so early
b) ^Bill's action's being unfortunate of refusing
the offer
c) *the event's being likely of Dick's resignation
d) *the fact's being remarkable of Jimmys
testimony
3.129a) *the fact's unexpectedness of John's leaving
so early
b) ^Bill's action's unfortunateness of refusing
the offer
c) *the event's likelihood of Dick's resignation
d) *the fact's remarkableness of Jimmy's
testimony
i*-
The examples above show that either AGENT-PREPOSING or
NP-PREPCSING must apply in the derivation of sentences with
embedded nominal complements, that the only acceptable ger
undive nominal complements with embedded nominal complements
are those to which AGENT-PREPOSING has applied, and that
derived nominal complements with embedded nominal comple
ments to which AGENT-PREPOSING has applied, as well as
derived nominal complements with embedded nominal comple
ments to which neither AGENT-PREPOSING nor NP-PREPOSING has
applied, are acceptable. No other possibilities are accep
table, not is any case in which the head noun is separated
from the embedded nominal complements.

82
<¡ith That-Complements
That-complements also appear to have head nouns in
underlying structure (cf. Menzel, 1969) Like the head
nouns of nominal complements, these head nouns are deletable.
The optional presence of head nouns in surface structure
will again be indicated by the use of parentheses.
As before, I will claim that AGENT-PREPOSING has applied
to predicate-initial underlying structures to yield the sen
tences in 3*130 and their corresponding gerundive nominal
complements in 3*131 and derived nominal complements in 3*132.
3.130a) Tom reported (the fact) that the account was
overdrawn.
b) Jerry revealed (the fact) that Alice was smart.
c) Mary denied (the claim) that Peggy was
pregnant.
3.131a) Tom's reporting (the fact) that the account
was overdrawn
b) Jerry's revealing (the fact) that Alice was
smart
c) Mary's denying (the claim) that Peggy was
pregnant
3.132a) Tom's report (of the fact) that the account
was overdrawn
b) Jerry's revelation (of the fact) that Alice
v/as smart
c) Mary's denial (of the claim) that Peggy was
pregnant
If AGENT-PR3P0SING does not apply, then NP-PREPOSING
applies obligatorily in the derivation of sentences. How
ever, NP-PREPOSING applies in two ways. Either the head
noun and the that-complement are preposed as a unit, as has
happened in the derivation of the sentences in 3-133 or
only the head noun is preposed, in which case the sentences
in 3*13^ result if the head noun is retained in surface

83
structure, and the sentences in 3.135 result if the head
noun is deleted.
3.133a)
(The fact) that the account was overdrawn was
b)
reported by Tom.
(The fact) that Alice was smart was revealed
c)
by Jerry.
(The claim) that Peggy was pregnant was denied
by Mary.
3.134a)
The fact was reported by Tom that the account
was overdrawn.
b)
The fact was revealed by Jerry that Alice was
smart.
c)
The claim was denied by Mary that Peggy was
pregnant.
3.135a)
It was reported by Tom that the account was
overdrawn.
b)
c)
It v/as revealed by Jerry that Alice was smart.
It was denied by Mary that Peggy was pregnant.
The sentences in 3.133 are passive. Sentences like
those in 3.134 have been described as being derived by the
rule of EXTRAPOSITION PROP/I NP (of which I will have more to
say below). The sentences in 3.135 are extraposed (i.e.,
supposedly derived by IT-EXTRAPOSITION).
In no case in which head nouns are retained in surface
structure are both the head noun and the that-complement to
the right of the predicate in sentences. Thus, there are no
acceptable sentences like those in 3.136.
3.136a)
*It v/as reported by Tom the fact that the
account was overdrawn.
b)
*It was revealed by Jerry the fact that Alice
was smart.
c)
*'It was denied by Mary the claim that Peggy
was pregnant.
The sentences in 3.124 do not have any acceptable cor
responding gerundive or derived nominal complements, the
forms in 3.137 and 3.138 all being unacceptable.

3.137a)
*(the fact) that the account was overdrawn's
being reported by Tom
b)
*(the fact) that Alice was smart's being
c)
revealed by Jerry
*(the claim) that Peggy was pregnant's being
denied by Mary
3.138a)
*(the fact) that the account was overdrawn's
b)
report by Tom
*(the fact) that Alice was smart's revelation
c)
by Jerry
*(the claim) that Peggy was pregnant's denial
by Mary
The sentences in 3.134 ho not have any acceptable cor
responding gerundive or derived nominal complements, the
forms in 3.139 and 3.140 all being unacceptable.
3.139a)
*the fact's being reported by Tom that the
account was overdrawn
b)
*the fact's being revealed by Jerry that Alice
was smart
c)
*the claim's denial by Mary that Peggy was
pregnant
3.l40a)
*the fact's report by Tom that the account was
overdrawn
b)
*the fact's revelation by Jerry that Alice was
smart
c)
*the claim's denial by Mary that Peggy was
pregnant
The sentences in 3,135 have the acceptable corresponding
gerundive nominal complements in 3.141 and derived nominal
complements :
in 3.142.
3.141a)
its being reported by Tom that the account
was overdrawn
b)
its being revealed by Jerry that Alice was
c)
smart
it being denied by Mary that Peggy was
pregnant
3.142a)
the report by Tom that the account was over
drawn
b)
c)
the revelation by Jerry that Alice was smart
the denial by Mary that Peggy was pregnant

85
Finally, although the sentences in 3*136, and their
expected corresponding gerundive nominal complements in 3.1*0
are all unacceptable, the derived nominal complements in
3,144, with the same linear order as the forms in 3.136 and
3.143, are acceptable.
3,143a) *its being reported by Tom the fact that the
account was overdrawn
b) *its being revealed by Jerry the fact that
Alice was smart
c) *its being denied by Mary the claim that Peggy
was pregnant
3.144a) the report by Tom of the fact that the account
was overdrawn
b) the revelation by Jerry of the fact that
Alice was smart
c) the denial by Mary of the claim that Peggy
was pregnant
By referring to the examples given for simple noun
phrases and nominal complements, the reader should be able
to convince himself that the patterns of acceptability
exhibited for the examples with transitive predicates and
specified agents given immediately above also hold for forms
with transitive predicates and unspecified agents and for
forms with intransitive predicates.
If we understand NP-PREPOSING to apply either to the
head noun alone or to the head noun plus its that-comple
ment, we see that the rule has the same conditions on appli
cation to that-complements that it has to nominal comple
ments j it applies optionally in the derivation of sentences,
and not at all in the derivation of derived nominal comple
ments. Acceptable gerundive nominal complements occur only
if the head noun is deleted.

86
EXTRAPOSITION FROM NP
Ross (1967) proposed a rule, which he called EXTRAPOSI
TION FROM NP, to account for the relationship of sentences
like those in 3*145 to sentences like those in 3*146.
3-l45a) A gun which I had cleaned went off.
b) He let the cats which were meowing out.
c) He expected someone who I was acquainted with
to show up.
3*l46a) A gun went off which I had cleaned.
b) He let the cats out which were meowing.
c) He expected someone to show up who I was
acquainted with.
Ross stated the rule in the form given in 3*147-
Although Ross does not state the rule in the context of the
cycle, the restrictions he gives on the forms of the varia
ble Y indicate that the rule is cyclic.
3.147 EXTRAPOSITION FROM NP
NpO^ S]Np
1 2
Y
OPT -
3 ==> i, 0, 3+2
EXTRAPOSITION FROM NP is not needed to account for the
relationship of the sentences in 3*145 to the sentences in
3-146 if NP-PRSPOSING is stated to apply to such noun-rela
tive clause combinations in the same manner as with that-com-
plements and their head nouns. Thus, NP-PREPOSING would

87
have preposed both the head noun and its relative clause in
the sentences in 3.1^5* while it would have preposed only
the head noun in the sentences in 3.1^6. The examples in
3.1^5 and 3.1^6 require a modification of the rule to
allow preposing across the separable particle of a verb-par
ticle predicate. This would, however, eliminate any need
for a rule to postpose such separable particles across the
object of the verb.
NP-PREPOSING and Pied Piping
In light of the behavior of NP-PREPOSING with that-com-
plements and relative clauses, I will modify the rule so that
it applies to the head noun of the structures in question,
and allow the Pied Piping convention to account for the
optional preposing of the that-complaments and relative
clauses with the head nouns. The Pied Piping convention is
proposed by Ross (196?) to handle just such phenomena. The
Pied Piping convention is as follows;^
The Pied Piping Convention
Any transformation which is stated in such a way as to
effect the reordering of some specified node NP, where
this node is preceded and followed by variables in the
structural index of the rule, may apply to this NP or
to any non-coordinate NP which dominates it, as long
as there are no occurrences of any coordinate node, nor
the node S, on the branch connecting the higher node
and the specified node.
Ross further mentions that Pied Piping is obligatory in
some contexts. The distributional facts discussed above are
accounted for if NP-PREPOSING applies to head nouns, with
Pied Piping being obligatory for nominal complements and
optional for that-complaments (and relative clauses).

88
Conditions on Application
The conditions on the application of NP-PREPOSING that
I have discussed above fall along two parameters, 'The con
textual parameter opposes sentences and gerundive nominal
complements to derived nominal complements. The noun phrase
parameter opposes simple noun phrases to head nouns of com
plements. The interaction of these parameters is indicated
in the chart in 3.148.
3.148 Conditions on the Application of NP-PREPOSING
(AGENT-PREPOSING has not applied)
simple noun
phrases
head nouns of
complements
in sentences and
gerundive nomi
nal complements
obligatory
obligatory
in derived nomi
nal complements
optional*
blocked
^blocked for specific predicates
In Chapter Four I will argue that when NP-PREPOSING
applies in the derivation of gerundive nominal complements,
they are still sentences. Thus, NP-PREPOSING is obligatory
when it applies within structures dominated by an S, On the
other hand, I will argue that when NP-PREPOSING applies in
the derivation of derived nominal complements, they are
already nominaiized. Thus, NP-PREPOSING is optional or
blocked when it applies within structures dominated by an
NP. I will have more to say about how this distinction is
achieved in Chapter Four.

89
Emonds* POSSESSIVE Transformation
Emonds (1969:78-81) argues that NP-PROPOSING does not
apply in the derivation of "passive" derived nominal comple
ments such as those in 3.149,
3-149a) the citys destruction by the enemy
b) Abbys acquittal by the jury
Ke argues that the possessive noun phrases the city's
and Abbys are preposed by a rule he calls the POSSESSIVE
Transformation. First he notes that NP-PREPOSING may pre
pose a noun phrase over a verb-particle predicate, as in
3.150, but claims that the POSSESSIVE Transformation never
does, so that forms like that in 3*151 are blocked.
3.150) The strike was refered to briefly in the report.
3.151) *the strike's brief reference to in the report
Since many derived nominis idiosyncratically block
preposing of object noun phrases, the fact that verbs with
lexical prepositions also block preposing is weak evidence
for the separate existence of the POSSESSIVE Transformation.
A second reason Emonds gives for requiring as a sepa
rate rule the POSSESSIVE Transformation is that NP-PREPOSING
is associated with the occurrence of the passive morpheme
be-en, which never appears in "passive" derived nominal com
plements. As was mentioned in Chapter Two, derived nominal
complements never have any auxiliaries. All that Is needed
is a simple rule or principle that insures that no auxiliar
ies are present in the surface structures of derived nominal

90
complements. If it is assumed that auxiliaries are present
in deeply underlying structures, then they are deleted at
some point in the derivation of derived nominal complements.
If it is assumed that auxiliaries are inserted, then the
insertion rules are blocked from applying to derived nominal
complements. Thus, there will be no passive morpheme be-en
in "passive" derived nominal complements.
A third reason cited by Emonds is more complex. He
points out that the POSSESSIVE Transformation apparently can
apply to noun phrases other than the one immediately follow
ing the derived nominal, whereas NP-PREPOSING (in sentences
and gerundive nominal complements) applies only to noun
phrases which immediately follow the predicate (or gerundive
nominal). That is, he claims that the POSSESSIVE Transforma
tion can apply to noun phrases which are not objects of the
predicate. He cites the occurrence of prepositionless time
adverbials preceding the derived nominal, as in 3*152, which
according to Emonds, represent derivations from the struc
tures underlying the derived nominal complements in 3*153*
3152a) last week's discussion of novels by the
librarian
b) this morning's speech to the nation by the
president
3*153a) the discussion of novels by the librarian
last week
b) the speech to the nation by the president
this morning

91
The application of NP-PREPOSING in this way is not pos
sible in sentences, so that we find the sentences in 3-15^
but the sentences in 3*1.55 are not acceptable.
3*15^a) The librarian discussed novels last week.
b) The president spoke to the nation this morning.
3*155a) *Last week was discussed novels by the
librarian.
b) *This morning was spoken to the nation by the
president
But it cannot be the POSSESSIVE Transformation which
has moved the prepositionless time adverbials to a position
preceding the derived nominis in 3*152, since the forms in
3.155 are not acceptable, indicating that the POSSESSIVE
Transformation (or NP-PREPOSING) is blocked for discussion
and speech in derived nominal complements.
3*155a) ^novel's discussion by the librarian last week
b) *the nations speech by the president this
morning
All time adverbials are subject to fronting in sen
tences, as in 3*157*
3*15?a) Last week the librarian discussed novels.
b) This morning the president spoke to the nation.
The application of a rule of ADVERB-FRONTING to the
structures underlying the derived nominal complements in
3*153 (with subsequent addition of the suffix *.s) will pro
duce the derived nominal complements in 3*152.
Other time adverbials are also subject to ADVERB-FRONT
ING, but do not form determiners of derived nominis. The
time adverbials with prepositions, however, are subject to
deletion of the preposition, and then behave just like other

92
prepositionless time adverbials. Time adverbials with the
-ly suffix form pre-ncminal adjectives just like other -ly
adverbs in derived nominal complements, as was discussed in
Chapter Two. Thus, the sentences in 3.158 have the corres
ponding derived nominal complements in 3.159.
3.158a) Recently the librarian discussed novels,
b) Daily the president speaks to the nation.
3.159a) the librarian's recent discussion of novels
b) the president's daily speech to the nation
Emonds choice of the name POSSESSIVE for the rule which
preposes objects in derived nominal complements is unfortun
ate. Last week and this morning in 3.152 were preposed by
ADVERB-FRONTING, and the librarian and the -president in
3.159 were preposed by AGENT-PRSPOSING, yet they also form
possessive determiners. The formation of a possessive is
obviously distinct from any preposing rule, and operates on
whatever noun phrase is in pre-nominal position^ at some late
point in the cycle.
The last reason Emonds cites for distinguishing the POS
SESSIVE Transformation from NP-PREPOSING is that the condi
tions for preposing are not the same in derived nominal com
plements as they are in sentences and gerundive nominal com
plements, as was indicated in 3.1^8. Since those conditions
do differ, there seems to be some merit in the proposal for
two rules. The two rules would have so much in common, how
ever, that to say that they are unrelated would miss signifi
cant generalizations. I therefore reject Emonds claim, and
retain my earlier analysis of one rule of NP-PREPOSING with

93
conditions on application in derived nominal complements
which differ from those in sentences and gerundive nominal
complements.
Summary
I have argued in this section that the subjects of sen
tences with intransitive verbs and the derived subjects of
passive sentences and their corresponding nominal comple
ments are derived by the rule of NP-PREPOSING. I have fur
ther argued that extraposed sentences and derived nominal
complements with nominal-initial order have resulted from
the failure of either AGENT-PREPOSING or NP-PREPOSING to
apply in their derivations. I have also argued that NP-PRE
POSING may apply to structures consisting of a noun (phrase)
plus an embedded sentences with the embedded sentence becom
ing a nominal complement, a that-complement, or a relative
clause. This thus eliminates any need for the rules of
IT-EXTRAPOSITION and EXTRAPOSITION FROM NP. Finally, I
have examined, and rejected, arguments presented by Emonds
in support of his claim that NP-PREPOSING does not apply in
the derivation of derived nominal complements.
Dummy Subject Insertion
Newmeyer (1971) notes that there are no derived nominal
complements with there as possessive noun determiner. The
sentences in 3.160 have the corresponding gerundive nominal
complements in 3.161, but there are no acceptable derived
nominal complements corresponding to the sentences in 3.160,
the forms in 3*162 being unacceptable.

3.l60a)
b)
c)
3.161a)
b)
c)
3.l62ai)
ii)
bi)
ii)
ci)
ii)
There appeared, to be no hope.
There seemed to he a disturbance.
There happened to be some wine in the bottle.
there appearing to be no hope^
there seeming to be a disturbance
there happening to be some wine in the bottle
17
^there's appearance to be no hope
*there*s appearance of being no hope
^there's semblance to be a disturbance
*there's semblance of being a disturbance
*there's happening to be some wine in the
bottle
*there's happening of being some wine in the
bottle
The forms in 3.162 do not provide any evidence of the
relation between THERE-INSERTION and DERIVED NOMINALIZATION,
however, as the sentences in 3.160 have been derived by RAIS-
ING-TO-SUBJECT from the structures underlying the sentences
in 3.163, which have the corresponding gerundive nominal com-
18
plements in 3.164- and derived nominal complements in 3.165*
3.163a) It appeared that there was no hope.
b) It seemed that there was a disturbance.
c) It happened that there was some wine in the
bottle.
3.164a) it(s) appearing that there v/as no hope
b) it(s) seeming that there was a disturbance
c) it(s) happening that there was some wine in
the bottle
3.1652-) the appearance that there was no hope
b) ?the semblance that there was a disturbance
c) the happenstance that there was some wine in
the bottle
If we take the embedded that-complements of the sen
tences in 3.163 we have sentences in which there has not
been raised from an embedded sentence, as in 3.166. These
sentences have the corresponding gerundive nominal comple
ments in 3,16?. It would seem that the derived nominal

95
complements corresponding to the sentences in 3.166 are the
simple noun phrases in 3.163.
3.166a) There was no hope.
b) There was a disturbance,
c) There was some wine in the bottle.
3.167a) there being no hope
b) there being a disturbance
c) there being some wine in the bottle
3.168a) no hope
b) a disturbance
c) some wine in the bottle
The sentences in 3.166 all have the surface structure
there-Copula-Predicate Nominal. THERE-INSERTION is presumed
to have applied in each sentence because no noun phrase was
left in subject position. In an analysis assuming underly
ing SVO order, this requires the postposing of the subject
with certain intransitive verbs. In an analysis assuming
predicate-initial order, THERE-INSERTION applies in case no
other rule has moved a noun phrase into subject" position.
In the section on IT-EXTRAPOSITION above I indicated
that ft is inserted in subject position if no noun phrase or
complement is moved into subject position. I will claim
here that ft is the surrogate subject if the predicate has a
sentential complement, or if the predicate is one which
takes no arguments (i.e., weather verbs such as rain, etc.),
and that there is the surrogate subject if the predicate
has at least one argument, but no sentential complement.
The derived nominal complements I have given in 3.168
are really just the noun phrases v/hich occur as predicate
nominis in the sentences in 3*166. There are variants of

96
those sentences, as in 3.169 "to which THERE-INSERT ION has
not applied. These sentences have the corresponding gerund
ive nominal complements in 3.170 and derived nominal comple
ments in 3.171. The forms in 3.172, with the same word
order as the sentences in 3*169, are not acceptable.
3.169a) No hope existed.
b) A disturbance occurred.
c) Some wine was present in the bottle.
3.170a) no hope('s) existing
b) a disturbance('s) occurring
c) some wine('s) being present in the bottle
3.1?la) the existance of no hope
b) the occurrence of a disturbance
c) the presence of some wine in the bottle
3,172a) *no hope's existance
b) *a disturbance's occurrence
ci) *some wine in the bottle's presence
ii) *some wine's presence in the bottle
Certain points emerge with these examples. The occur
rences of be in the sentences in 3.166 are not really the
v
copula, but rather an alternate representation of the exis
tential verbs in 3.169. The only acceptable derived nominal
complements corresponding to the sentences in 3.169 show pre
dicate-initial order. For this set of predicates, then,
NP-PREP03ING seems to apply optionally in sentences, and not
at all in derived nominal complements. This pattern is paral
lel to the application of the rule with an embedded that-com-
plement.
THERE-INSBRTION and PI-INSERTION are both blocked from
applying to derived nominal complements, and are both gov
erned by very similar conditions. If no noun phrase precedes

97
the predicate because NP-PREPOSING has failed to apply, a
surrogate subject must be supplied. This is done by
THEfLi;-INSERTION if the predicate belongs to the class of
existential predicates, and by If-INSERTION if the predicate
has a sentential complement, or is a weather predicate. In
a derived nominal complement, no surrogate subject is sup
plied by the grammar.
eQ.U I-NP -DELETION
In this section I will discuss another problem which is
not raised by Chomsky (1970): the lack of derived nominal
complements in which EQUI-NP-DELETION has applied.
Stockwell et al. (1973*553ff) present an analysis
which has EQUI-NP-DELETION apply when there is an Agent (of
the higher sentence)-subject (of the embedded sentence) iden
tity or a Dative (of the higher sentence)-subject (of the
embedded sentence) identity, with the Dative-subject condi-
19
tion taking precedence over the Agent-subject condition. 7
Stockwell et al. state that require takes an optional Dative
noun phrase with a sentential complement, with the applica
tion of EQUI-NP-DELETION being optional if a Dative noun
phrase is present and identical to the subject of the embed
ded sentencej that command takes an optional Dative noun
phrase with the application of EQUI-NP-DELETION being obli
gatory if a Dative noun phrase is present and identical to
the subject of the embedded sentence? and that force takes
an obligatory Dative noun phrase with obligatory application
of EQUI-NP-DELETION if the Dative noun phrase is identical

93
to the subject of the embedded sentence. The case of
require without a Dative noun phrase is represented by
3.173a require with a Dative noun phrase but without appli
cation of EQUI-NP-DELETION by 3.173, require with a Dative
noun phrase and application of EQUI-NP-DELETION by 3.173c
comraand without a Dative noun phrase by 3.173d, command with
a Dative noun phrase by 3.173e, and force by 3.1731. The
failure of EQUI-NP-DELETION to apply for whatever reason pro
duces that-complements from the embedded sentences, as in
3.173a b and d. The sentences in 3.173 have corresponding
gerundive nominal complements, as in 3.174-, but only 3.173a,
d and e have acceptable corresponding derived nominal comple
ments, as indicated by the forms in 3.175.
3.173a) I require that you solve the problem.
b) I require of you that you solve the problem.
c) I require you to solve the problem.
d) I commanded that he solve the problem.
e) I commanded him to solve the problem.
f) I forced him to solve the problem.
3.174-a) my requiring that you solve the problem
b) my requiring of you that you solve the problem
c) my requiring you to solve the problem
d) my commanding that he solve the problem
e) my commanding him to solve the problem
f) my forcing him to solve the problem
3.175a) my requirement that you solve the problem
b) *my requirement of you that you solve the problem
c) *my requirement of you to solve the problem
d) my command that he solve the problem
e) my command to him to solve the problem
f) *my forcing of him to solve the problem
There are no derived nominal complements corresponding
to sentences with require which have a Dative noun phrase
whether or not EQUI-NP-DELETION has applied to the sentence.

99
I have no explanation for this fact other than the sugges
tion that some unknown rule has applied in the derivation of
the sentences in 3-173b and c which has not applied in the
derivation of the sentence in 3-173a, and that this unknown
rules blocks DERIVED NOMINALIZATION. On the other hand,
there are derived nominal complements corresponding to sen
tences with command without regard to whether the sentences
have undergone EQUI-NP-DELETION. I will show below that
DERIVED NOMINALIZATION blocks the application of EQUI-NP-DE-
LETION for only some underlying predicates, so that there is
nothing troublesome about its failure to do so for command.
That the unacceptability of 3*1751 is linked to the applica
tion of EQUI-NP-DELETION is shown by the acceptability of
my forcing: of the issue corresponding to I forced the issue,
which has not undergone EQUI-NP-DELETION. 'Thus, it is not
the case that DERIVED NOMINALIZATION is blocked idiosyncrati-
cally for force. The examples in 3*175 do not provide clear
evidence on the interaction of EQUI-NP-DELETION and DERIVED
NOMINALIZATION. To find such evidence we must turn to the
Agent-subject condition on EQUI-NP-DELETION.
The sentences in 3*176 result from the deletion of the
subject of the embedded sentence when it is identical to the
Agent noun phrase of the next higher sentence. The corres
ponding gerundive nominal complements are given in 3*177*
There are no acceptable corresponding derived nominal com
plements, the expected forms in 3*178 being unacceptable.

100
3.1?a) John condescended to speak to Mary,
b) Mike quickly learned to analyze sentences.
c) Bill expected to leave early.
3.177a) John's condscending to speak to Mary
b) Mike's quickly learning to analyze sentences
c) Bill's expecting to leave early
3.178a) *John's condescension to speak to Mary
b) *Mike's quick learning to analyze sentences
c) ^Bill's expectation to leave early
That the unacceptability of the forms in 3.178 is con
nected with the application of EQUI-NP-DELETION is indicated
by the fact that the sentences in 3.179 with the same basic
predicates as in 3.176, have the corresponding gerundive
nominal complements in 3.180 and derived nominal complements
in 3.181.
3.179a) John was condescending towards Mary.
b) Mike quickly learned the multiplication tables.
c) Bill expected that Alice would leave,
3.180a) John's being condescending towards Alice
b) Mike's quickly learning the multiplication
tables
c) Bill's expecting that Alice would leave
3.181a) John's condescension towards Mary
b) Mike's quick learning of the multiplication
tables
c) Bills expectation that Alice would leave
Other predicates allow the application of EQUI-NP-DELE
TION in the derivation of derived nominal complements. The
sentences in 3.182, which have undergone EQUI-NP-DELETION in
their derivations, have the corresponding gerundive nominal
complements in 3.183 and derived nominal complements in 3.184.
3.182a)
b)
c)
d)
John intended to leave early.
Mike desired to leave early.
Bill refused to leave early.
Jerry attempted to leave early.

101
3.133a) John's intending to leave early
b) Mike's desiring to leave early
c) Bill's refusing to leave early
d) Jerry's attempting to leave early
3.184a) John's intention to leave early
b) Mike's desire to leave early
c) Bill's refusal to leave early
d) Jerry's attempt to leave early
It appears from these examples that EQUI-NP-DELETION,
like NP-PREPOSING, can apply to some derived nominal comple
ments s but not to others. The applicability of EQUI-NP-DELE-
TION seems to be governed by a feature on the predicate.
That the conditions on applicability may be more compli
cated is indicated by the fact that while 3.178c, *Bill's
expectation to leave early, seems hopelessly bad, the
related form, Bill's expectation of an early departure, is
much better, A further point is that in the derived nominal
complements in 3.184, all of the derived nominis except
refusal will also take embedded gerundive or deprived nominal
complements to which EQUI-NP-DELETICN has applied, as in
3.185.
3.185a) John's intention of leaving early
b) Mike's desire for leaving early/an early
departure
c) *Bill's refusal of leaving early/an early
departure
a) Jerry's attempt at leaving early/an early
departure
These facts indicate a difference in the application of
EQUI-NP-DELETION to derived nominal complements with embedded
infinitival complements and to those with embedded gerundive
or derived nominal complements. I will have more to say
about that difference in Chapter Four,

102
DAT IVE -*M0 VEM EN T
Newmeyer (1971) points out that sentences which have
undergone DATIVE-MOVEMENT in their derivations do not have
corresponding derived nominal complements. The sentences
in 3*186, which have undergone DATIVE-MOVEMENT, have the cor
responding gerundive nominal complements in 3*18?, but the
expected corresponding derived nominal complements in 3*188
are not acceptable,
3.186a) John gave Mary the book.
b) Bill told Alice the story.
c) Jane bought Jim a soda.
3*187a) John's giving Mary the book
b) Bill's telling Alice the story
c) Jane's buying Jim a soda
3.188a) ^Johns gift to Mary of the book
b) ^Bill's telling to Alice of the story
c) *Jane's buying for Jim of a soda
The sentences in 3*189 which have not undergone DAT
IVE-MOVEMENT, have the corresponding gerundive-nominal comple
ments in 3*190 and derived nominal complements in 3*191*
The nominal-initial derived nominal complements correspond
ing to this set of sentences are given in 3*192.
3*189a) John gave the book to Mary.
b) Bill told the story to Alice.
c) Jane bought a soda for Jim,
3*190a) John's giving the book to Mary
b) Bill's telling the story to Alice
c) Jane's buying a soda for Jim
3-191a) John's gift of the book to Mary
b) Bill's telling of the story to Alice
c) Jane's buying of a soda for Jim
3.192a) the gift of the book to Mary by John
b) the telling of the story to Alice by Bill
c) the buying of a soda for Jim by Jane

103
Psychological Predicates
Chomsky presents a set of examples which show a unique
pattern of acceptability for derived nominal complements.
The sentences in 3.193 have the corresponding gerundive nom
inal complements in 3.194, hut only 3.193b has an acceptable
corresponding derived nominal complement in 3.195b, with the
form in 3.195a being unacceptable.
3.193a) John amused the children with his stories,
b) John was amused at the children's antics.
3.194a) John's amusing the children with his stories
b) John's being amused at the children's antics
3.195a) *John's amuement of the children with his
stories
b) John's amusement at the children's antics
The unacceptability of 3.195a raises a couple of pro
blems. The first problem is that of instrumental phrases
such as with his stories. The examples from Chomsky which
a*
I have just repeated above involve sentences with the verb
amuse, which belongs to the set of psychological predicates.
The sentence in 3,193a has at least two semantic readings.
One reading involves the presupposition that John intended
to amuse the children, while the other reading does not
include that presupposition. The two readings may be illus
trated by the use of the adverbs cleverly and inadvertently,
as in the sentences in 3.187.
3.187a) John cleverly amused the children with his
stories.
b) John inadvertently amused the children with
his stories.

104
The first reading results from the derivation of 3-193a
from the underlying structure illustrated in 3-197*
CAUSE John
stories V NP NP
tell stories John
The other reading results from the derivation of 3-193a
20
from the underlying structure illustrated in 3-198.
stories V NP NP
tell stories John
The second semantic reading is also associated with the
sentence in 3-199-
3*199) John's stories amused the children.
The structure of 3^ in 3-198 is the same as that of
in 3-197* This means that the structure underlying the
second semantic reading of 3-193a is embedded in the struc
ture underlying the first semantic reading of that sentence.
The derivation of the sentence in 3-193a from the underlying
structure in 3-19? involves the reduction of stories
[pJohn told stories^p^.Tp to John's stories, the pronominal-

105
isation of the second occurrence of John, and something like
CAUSATIVE FORMATION or PREDICATE RAISING to incorporate
21
CAUSE into amuse.
The derivation of either the sentence in 3-193a or the
one in 3>190 from the structure in 3-198 can be explained by
allowing John to be optionally moved out of the embedded sen
tence to subject position of the next higher sentence in the
case of 3-193a* with the alternative being to move the
reduced form of the NP dominating 3^, John's stories, into
subject position.
The sentence in 3*199 is repeated in the set of sen
tences in 3*200s which have the corresponding passive sen
tences in 3*201. The sentences in 3*202 which resemble the
passive sentences in 3*201, but have some preposition other
than by preceding the Agent phrase, are found only with verbs
belonging to the set of psychological predicates.
3.200a) John's stories amused the children.
b) The children's antics amused John.
c) The storm frightened Mary.
d) The ceremony pleased 3ob.
3>201a) The children were amused by John's stories.
b) John was amused by the children's antics.
c) Mary v/as frightened by the storm.
d) Bob was pleased by the ceremony.
3*202a) The children were amused at John's stories.
b) John was amused at the children's antics.
c) Mary v/as frightened of the storm.
d) Bob v/as pleased with the ceremony.
The sentences in 3*200, 3*201 and 3*202 have the respec
tive corresponding gerundive nominal complements in 3*203,
3*204 and 3-205*

io6
3203a) John's stories amusing the children
the children's antics amusing John
o)
c)
d)
20 4a)
b)
c)
3205a)
b)
c)
d)
the storms frightening Mary
the ceremony's pleasing 3ob
the children's being amused by John's stories
John's being amused by the children's antics
Mary's being frightened by the storm
d) Bob's being pleased by the ceremony
the children's being amused at John's stories
John's being amused at the children's antics
Mary's being frightened of the storm
Bob's being pleased with the ceremony
While there are the derived nominal complements in
208 corresoonding to the sentences in 3-202 which are
acceptable, the forms in 3-206 and 3-20?, which would appear
to be the expected derived nominal complements corresponding
to the sentences in 3*200 and 3-201 respectively, are not
acceptable,
3-206a)
b)
c)
d)
3.207a)
b)
c)
d)
3208a)
b)
c)
d)
*John's stories' amusement of the children
"'"the childrens antics' amusement of John
*the storm's fright of Mary
*the ceremony's pleasure of Bob
Hu-
*the children's amusement by John's stories
*John's amusement by the childrens antics
*Ivlary's fright by the storm
*3ob's pleasure by the storm
the children's amusement at John's stories
John's amusement at the childrens antics
Mary's fright of the storm
Bob's pleasure with the ceremony
Only the pseudopassive sentences in 3-202 have accepta
ble corresponding derived nominal complements. There are no
derived nominal complements based on psychological predi
cates which show nominal-initial order such as we have found
associated with other types of predicates. The forms in
3-209 are not acceptable.

10?
3.209a) *the amusement of the children at Johns stories
b) '*the amusement of John at the children's antics
c) *the fright of Mary of the storm
d) *tne pleasure of Bob with the ceremony
In Chapter Two I noted that eagerness does not occur
in derived nominal complements with nominal-initial order,
so that *the eagerness of John to please is unacceptable.
The adjective eager describes a mental state, and belongs
to the class of emotive predicates, of which psychological
predicates are a part. The lack of nominal-initial derived
nominal complements seems to be characteristic of emotive
predicates.
Following Chomsky's claim that derived nominal comple
ments correspond to base phrase markers, the sentences in
3.193a, 3*200 and 3.201 do not have acceptable correspond
ing derived nominal complements because certain rules have
applied in their derivations which are not allowed to apply
in the derivation of derived nominal complements. I have
indicated that one rule has applied in the derivation of
3.193a which has not applied in the derivations of 3.202,
and which appears to be related to the RAISING rules,
It is not clear v/hat rules have applied in the deriva
tions of the sentences in 3.200 and 3.201 which have not
applied in the derivations of the sentences in 3,202. I
will assume an ad hoc rule of PSYCH-CHANGE as a way of refer
ring in the next chapter to the pattern of acceptability
of derived nominal complements associated with psycholo
gical predicates.

108
Conclusion
In this chapter I first argued that sentences which
have undergone any of the RAISING rules do not have corres
ponding derived nominal complements. I then showed that
sentences which have undergone IT-EXTRAPOSITION in their
derivations have corresponding derived nominal complements,
while sentences which met the structural description of
IT-EXTRAPOSITION, but to which the rule did not apply in
their derivation, do not have corresponding derived nominal
complements. I then showed that while all simple active
sentences have corresponding derived nominal complements,
only some passive sentences have corresponding derived
nominal complements, and that for every pair of active and
passive sentences there is a derived nominal complement to
which only the AGENT-POSTPOSING part of Passivization has
applied. I then argued that with a predicate-initial ana
lysis both extraposed and nonextraposed and active and pas
sive sentences can be related, at least in part, by the
same rule, NP-PREPOSING. I have also discussed certain
interesting aspects of NP-PREP03ING, and concluded that the
application of the rule in the derivation of derived nominal
complements is governed by a set of conditions dependent on
different aspects of the environment. Finally, I have
argued that ^-INSERTION, THERE-INSERTION and PSYCH-CHANGE
do not apply in the derivation of derived nominal comple
ments, and that the application of EQUI-NP-DELETION is
blocked in derived nominal complements for certain underly
ing verbs.

109
In this chapter I have shown that the adoption of a
predicate-initial analysis of underlying structure allows
the statement of restrictions on the occurrence of derived
nominal complements in terms of the application of certain
rules being blocked in the derivation of derived nominal
complements from underlying embedded sentences. I will
discuss how such rules are blocked in Chapter Four.
The analysis in this chapter shows that, contrary to
Chomsky's claim, the restrictions on the productivity of
derived nominal complements can be easily stated within a
transformational framework. In Chapter Four I will discuss
how the noun phrase-like internal structure of derived nom
inal complements can also be accounted for in a transforma
tional framework, based on the analysis given in this
chapter.

NOTES
One exception to this claim is that verbs like be and
have do not function as main verbs in sentences with corres
ponding derived nominal complements, as was discussed in
Chapter Two.
^ TOUGH-MOVEMENT is a RAISING rule. It raises the
object of an embedded sentence to subject of the next higher
sentence. The other RAISING rules are RAISING-TO-OBJECT,
which raises the subject of an embedded sentence to object
of the next higher sentence, and RAISING-*TO-SUBJECT, which
raises the subject of an embedded sentence to subject of the
next higher sentence. These three RAISING rules are dis
cussed in detail in Stoekwell et al. (1973)- TOUGH-MOVEMENT
is also discussed in Postal (1971)- The relation of IT-EX
TRAPOSITION and the RAISING rules to derived nominal comple
ments is discussed in the next two sections.
3
An indefinite object of please has been deleted.
Jj.
John as the subject of olease has been deleted by
EQUI-NP-DELETION (discussed in the section under that name
below). The unspecified object of olease has also been
deleted.
^ This derivation is questionable. See the section on
IT-EXTRAPOSITION below.
110

Ill
Removal of the subject of a that-complement (here, by
RAISING-TO-SUBJECT) results in the formation of an infini
tival complement, as in 3*25- This will be discussed in more
detail in Chapter Four.
^ This is unacceptable on the reading of *Jerrv's
appearance of opening the door. A reading based on some
thing like Jerry appeared in order to open the door is not
intended.
8
The possessive suffix does not seem to be necessary
with it in gerundive nominal complements, and its absence
may improve such complements for some speakers. Some speak
ers also apparently accept nouns without the possessive suf
fix in gerundive nominal complements. Ross (1972b) calls
such forms accusative-ins: complements. Fraser (1970) dis
cusses apparent accusative-ing complements, and analyzes
them as Object+Complement constructions. The forms in 3-35
seem to be gerundive nominal complements, rather than some
such construction.
o
Fasv. and several other adjectives and predicate
nouns which govern TOUGH-MOVEMENT. present a special problem,
The only satisfying use of a derived nominal corresponding
to easy seems to be in a construction like the ease with
which John is pleased, which corresponds to something like
John is pleased with ease/easily. I have no explanation for
this exception to DERIVED N0MINALIZATION.
10
In a grammar which includes the rules of PASSI7IZA-
TIGN and IT-EXTRAPOSITION. those two rules must be ordered

112
after RAISING-TO-OBJECT but before TOUGH-MOVEMENT and RAIS
ING -TO-SUBJECT. A sentence like John was believed by Mary
to be a fool is derived by PASSIVTZATION from the structure
underlying Mary believes John to be a fool, which is derived
in turn by RAI3ING-T0-OBJECT from the structure underlying
Mary believes that John is a fool. This indicates that RAIS
ING -TO-OBJECT must precede PAS3IVIZATIQN. The sentence it
is believed by Mary that John is a fool is derived by IT-EX
TRAPOSITION from the structure underlying that John is a
fool is believed by Mary, which is derived in turn by PASSI-
VTZAT'IQN from the structure underlying Mary believes that
John is a fool. This indicates that PASSIVTZATION must pre
cede IT-EXTRAPOSITION. On the other hand, the sentence John
is easy to olease is derived by TOUGH-MOVEMENT from the
structure underlying it is easy to olease John, which is in
turn derived by IT-EXTRAPOSITION from the structure underly
ing to olease John is easy. Similarly, the sentence John is
likely to win the prize is derived by RAISING-TO-SUBJECT
from the structure underlying it is likely that John will
win the arize, which is in turn derived by IT-EXTRAPOSITION
from the structure underlying that John will win the prize
is likely. This indicates that IT-EXTRAPOSITTON must pre
cede TOUGH-MOVEMENT and RAISING-T0-SU3JECT. Assuming trans
itivity for the ordering relationships exemplified above,
the order of the five rules within a single cycle must be?
RAISING-TO-OBJECT, PASSIVTZATION, IT-EXTRAPOSITION and
TOUGH-MOVEMENT and RAISING-TO-SUBJECT. However, I will

113
argue below ( cf. IT-EXTRA?OSI PION and NP-PkwPOSIKG) that:
PAS SIVIZATION and IT-EXTRAPOSITION are not rules of English,
and that the relationships to be accounted for by those
rules are accounted for by the rule of NP-PREPOSING. Under
this analysis, RAI3ING-T0-03JECT precedes NP-PROPOSING, and
TOUGH-MOVEMENT and RAISING-TO-SUBJECT are not ordered with
respect to NP-PREPOSING.
^ The rule is called EXTRAPOSITION in Rosenbaum (1967).
The name IT-EXTRAPOSITION has come into use to distinguish
it from other rules of EXTRAPOSITION.
"2 The unacceptability of the forms in 3-53 and 3-5^
can be "explained" by the statement that sentential comple
ments are not allowed to form possessives in English. This
is an observation, however, not an explanation. Indeed, this
restriction also applies to derived nominal complements, as
in ^.Dannys departure's imminence, while ordinary noun
phrases of greater length freely form possessives, as in the
King of England's crown and the man next door's son-in-law.
In this case, derived nominal complements behave like embed
ded sentences, not like noun phrases.
13
J It would appear that sentences like those in i are
counter-evidence to this statement.
a) That John is late seems likely.
b) That Mildred has fallen down appears certain.
c) That Mike is a brilliant student happens to be true
However, such sentences are derived by RAISING-T0-SU3JECT
from the structures underlying the sentences in ii.

114
iia) It seems that it is likely that John is late.
b) It appears that it is certain that Mildred has
fallen down.
c) It happens that it is true that Mike is a
brilliant student.
Ik .
Ross5 example (1.10).
^ Ross' example (4.ISO).
1 £
x There does not allow the possessive suffix at ail in
gerundive nominal complements (cf. Note 8 above).
I have indicated the alternate forms in 3162 since
Newmeyer paired 3*l62aii with 3-lOa in his discussion of
this point.
-j 0
The form in 3-l65o seems pretty bad to me. The
choice of happenstance as the derived nominal corresponding
to happen seems more appropriate here than the other possi
bility, happening.
^ SQUI-NP-DELETION applies to the end-cyclic subject
of an embedded sentence, as shown by its application to the
derived subjects of embedded passive sentences. Both i and
ii are possible results of applying the rule.
i) Someone intended to leave Mary behind.
ii) Mary did not intend to be left behind by anyone.
20
Lee (I9?0:?0ff.) suggests that the surface subjects
of sentences like John amused the children by telling
stories (non-intentional reading) are derived by a RAISING
rule from the embedded sentence John told stories. The para
phrase John's telling stories amused the children would be
derived by moving the whole embedded sentence into subject
position. In the sentence John amused the children with his

115
stories, his stories (=John's stories) is a reduced form of
stories which John told,. Hence, it would appear that a RAIS
ING rule could move John out of with John's stories (or,
rather, copy it out, leaving the pronoun) and into subject
position. Lee's suggestion would involve extending the list
of predicates governing RAISING- rules to psychological predi
cates. My suggestion would involve postulating a new RAIS
ING rule which leaves a pronoun behind, and would require
independent motivation for the substitution of with for the
predicate to derive phrases like with his stories-

CHAPTER FOUR
RULE-ORDERING AND DERIVED NOMINAL COMPLEMENTS
Introduction
In Chapter Three I showed for a number of examples that
whether or not a particular sentence has a corresponding
derived nominal complement depends on whether or not that
sentence has undergone one or more of a certain set of trans
formations in its derivation. Those sentences which have
undergone one of the rules in question do not have corres
ponding derived nominal complements. In this chapter I will
offer an explanation for the above fact.
In Chapter One I stated my assumption that derived nomi
nal complements are derived transformationally'-from underly
ing embedded sentences. In Chapter Two I showed that, with
the exception of the Germanic member of Germanic-Latinate
synonym pairs, all predicates have corresponding derived
nominis. Finally, in Chapter Three I showed that the only
other restrictions on the productivity of derived nominal
complements are due to the failure of certain transforma
tional rules to apply in the derivation of derived nominal
complements. In this chapter I will argue that certain
rules are blocked from applying in derived nominal comple
ments because DERIVED NOMINALIZATION precedes those rules,
and the rules cannot apply to nominalized structures. I
116

11?
will also argue that the rule known as COMPLEMENTIZER PLACE
MENT is not compatible v/ith the theory of the cycle, and
that no such rule is necessary in any case for any comple
ments other than nominal complements. Although it will
appear that COMPLEMENTIZER PLACEMENT will still be needed to
properly account for nominal complements, I will argue that
it is better to have the rules of DERIVED NOMINALIZATION and
GERUNDIVE NOMINALIZATION apply on the cycle of the comple
ment, and to allow these rules to 'look up' into the next
higher sentence and include the predicate of the next higher
sentence in their environments than it is to retain COMPLE
MENTIZER PLACEMENT. Finally, I will argue that the specific
characteristics of derived nominal complements are not the
direct result of the application of DERIVED NOMINALIZATION,
but rather the result of a process that begins with the
application of DERIVED NOMINALIZATION.
Newmeyer's Proposal
The argument that appropriate ordering of the rules
will account for the facts of productivity was presented
independently in Albury (1971) and Newmeyer (1971).
Newmeyer argues that DERIVED NOMINALIZATION precedes all the
cyclic rules, and by changing the predicate of the embedded
sentence to a nominal, prevents the application of any
later rule which mentions the predicate in its structural
description.
Newmeyer's claim that DERIVED NOMINALIZATION precedes
all cyclic rules seems to imply that the rule is precyclic.

118
I will not accept Newmeyer's claim as it stands, but will
present arguments below that DERIVED NGMINALIZATION is
cyclic, and ordered very early in the cycle, although not
necessarily first in the cycle.
Newmeyer's proposal that the nominalization of the
underlying predicate blocks the later application of rules
mentioning predicate in their structural descriptions
appears to be well motivated. In Chapter 'Three I showed
that there are no acceptable derived nominal complements
corresponding to sentences which have undergone one of the
RAISING rules, IT-INSERTION, THERE-INSERTION. DATIVE-MOVE
MENT, and/or PSYCH-CHANGE in their derivations.
Each of the RAISING rules raises a noun phrase into a
position defined by its relationship to the predicate of the
higher sentence. In RAISING-TO-QBJECT, the noun phrase is
raised into a position immediately to the right of the pre
dicate of the higher sentence. In RAISING-TO-SUBJECT (and
the similar rules governed by psychological predicates) and
TOUGH-MOVEMENT, the noun phrase is raised into a position
immediately to the left of the predicate of the higher
sentence.
IT-INSERTION and THERE-INSERTION introduce, in appro
priate circumstances, a semantically empty word as the sub
ject of the predicate. That is, these rules insert it or
there Immediately to the left of the predicate.
DATIVE-MOVEMENT involves a switch in position between
direct and indirect objects, and thus has the effect of

119
moving an indirect object to a position immediately to the
right of the predicate.
Although I have not said anything about the form of
PSYCH-CHANGE, this rule (or set of rules) involves the
choice of the noun phrase to be subject of the sentence in
surface structure, as well as differences in the surface
form of the predicate.
Each of these rules must refer explicitly to the struc
tural relationships between a noun phrase and a predicate.1
The structural description of each rule must include the
mention of predicate within the sentence to which the rule
is directly applying.
The rule of NP-PREPOSING presents some difficulty at
this point. This rule also moves a noun phrase to a posi
tion immediately to the left of a predicate, thus making it
the subject of the predicate in surface structure. By
Newmeyer's proposal, this rule should not apply in the deri
vation of derived nominal complements. In fact, it does so
apply, although its application is very limited. The appli
cation of NP-PREPOSING in the derivation of derived nominal
complements is limited to simple noun phrases and to a sub
set of predicates, and moreover, is optional even when it
does apply, while the application of NP-PREPOSING to head
nouns is always obligatory in the derivation of sentences,
if the structural description of the rule is met. There
would be no conflict with Newmeyer's proposal if it were
assumed that a different rule applied in the derivation of

120
derived nominal complements than did in the derivation of
sentences, but as I showed in Chapter Three, there is no
other reason to postulate two rules of NP-PREPOSING. I see
no non-ad hoc way around this problem. .Either NP-PREPOSING
violates Newmeyer's proposed constraint, or two rules of
NP-PREP03ING are admitted on the sole basis of different
conditions on application in the derivation of sentences
and of derived nominal complements.
Newmeyer bolsters the argument for his proposal by
pointing to the rules of EQUI-NP-DELETION and REFLEXIVE.
SQUI-NP-DELETION applies to many, but not all, derived nom
inal complements which meet the structural description of the
rule. Applicability seems to be governed by many interact
ing factors, two of which are the previous application of
DERIVED NOMINALIZATION and the predicate nominal!zed, as
mentioned in the section on that rule in Chapter* Three.
There appears to be no restriction on the application of
REFLEXIVE to derived nominal complements, as indicated by
the sentences in 4.1, with their corresponding gerundive
nominal complements in 4.2 and derived nominal complements
in 4.3
4.1a) Bill hates himself.
b) The president promoted himself to generalsimo.
c) John described himself as intelligent.
4.2a) Bill's hating himself.
b) the president's promoting himself to generalsimo
c) John's describing himself as intelligent
4.3a) Bill's hatred of himself
b) the president's promotion of himself to
generalsimo
c) Johns description of himself as intelligent

121
REFLEXIVE applies between two identical noun phrases
dominated by a single 3 node with no intervening 3 node
(clause mates). There is no need to mention predicate in
the structural description of the rule. EQUI-NP-DELETION
applies between two identical noun phrases also, but the
noun phrases require careful definition. One noun phrase
(the one deleted) is the subject of an embedded sentence
(or complement). While subject is defined in terms of the
relation of a noun phrase to a predicate, this particular
occurrence of subject is embedded within the sentence the
rule is applying to, and does not bear on Newmeyer's claim,
which refers to the higher sentence which is being nominal-
ized. In the analysis of this rule presented in Stoekwell
et al. (1973) the appropriate noun of the higher sentence
is defined by case role, which is independent of any mention
of predicate. As long as EQUI-NP-DELETION canbe stated
without mention of the predicate of the higher sentence,
there is no contradiction of Newmeyer's proposal.
The Order of the Rules
The Cyclic Nature of AGENT-PREPOSING
Newmeyer states that DERIVED NOMINALIZATION precedes
all the cyclic rules. As was mentioned in Chapter Three,
the first rule in Chomsky's (1970) analysis of passivization
is AGENT-P03TP0SING (equivalent to the nonapplication of
AGENT-PREP03ING in a predicate-initial analysis).
AGENT-P0STP0SING (or AGENT-PREPOSING) applies freely in
derived nominal complements, even though predicate is

122
included in its structural description. If Newmeyer's pro
posal is to stand, then AGENT-POSTPOSING (or AGENT-PREPOSING)
must precede DERIVED NOMINALIZATIQN. Newmeyer states that
there is no evidence that AGENT-POSTPOSING is cyclic, so
that it may be assumed to precede DERIVED NOMINALIZATIQN.
The question of whether or not AGENT-PREPOSING is cyc
lic is worth considering in detail, as it bears on the ques
tion of whether or not DERIVED NOKINALIZATION is cyclic.
The best evidence for the cyclic nature of a rule is the
existence of some RULE X which would apply after AGENT-PRE
POSING at some point in a derivation and before AGENT-PREPOS
ING somewhere else. In fact, there Is no such rule known in
the grammar of English, leading to Newmeyer's claim that
there is no evidence for the cyclic nature of AGENT-POSTPOS
ING (or, here, AGENT-PREPOSING). However, there is reason
to believe that there could be no such evidence, even though
AGENT-PREPOSING were a cyclic rule. If AGENT-PREPOSING were
the first rule of the cyclce, or if it were preceded in the
cycle only by rules with which it could never interact, then
it would be impossible to find evidence of the type discussed
above for the cyclic nature of AGENT-PREPOSING.
AGENT-PREPOSING makes an Agent noun phrase into the
subject of a sentence. No matter how the rule Is formulated
(even if it is formulated as AGENT-POSTPOSING), the Agent
noun phrase is dominated by the same S, with no intervening
S, both before and after the application of the rule. No
RULE X can be shown to apply before AGENT-PREPOSING, since

23
any previous application of RULE X would be to an S lower
than the S of the Agent noun phrase in question, and would
thus have no effect on the higher 3. Given the nature of
AGENT-PREPOSING and the assumption that it is one of the
earliest rules in the cycle, it is impossible to show that
it follows any other cyclic rule.
Since it would appear that it is impossible to discover
direct evidence of the cyclic nature of AGENT-PREPOSING, I
would like to consider some indirect evidence. AGENT-PRE
POSING can apply in more than one S in a single derivation,
as can be seen in the sentences in 4.4.
4.4a) John discovered that Mary hated him.
b)Jerry hoped that Mary would find out that Mack
really loved Suzie.
This in itself does not show that AGENT-PREPOSING is
cyclic, since a precyclic rule could conceivably prepose all
Agents within their respective sentences at one time. How
ever, AGENT-PREPOSING does not necessarily apply to all
Agents in a single derivation, so that passive sentences may
be embedded in active sentences and vice versa, as in 4.5*
4.5a) John hoped that the package would be handled
carefully by the Post Office.
b) Mary claimed that she was being followed by a
strange man.
c) That Bob had betrayed Tom was not known to the
other members of the cell.
d) That the Earth is not a perfect sphere was demon
strated by the first artificial satellite.
For a precyclic rule to produce such structures, it
would need to not only recognize Agents within individual
embedded sentences, but it would also have to recognize the

124
appropriate contextual information governing the application
of the rule within each embedded sentence. The rule would
also have to be constrained to prevent it from moving an
Agent out of its own sentence. All of this means that if
AGENT-PROPOSING were precyclic, it would have to be formu
lated to apply to possibly many different sentences within
a single tree, but at the same time confine each application
to a single sentence, and also be governed as to whether or
not to apply in any given sentence by contextual information
specific to that sentence. All-in-all, this does not seem
to be a very desirable type of rule. All of the above com
plications may be avoided by assuming that AGENT-PREPOSING
is a cyclic rule.
The Place of Rules of Nominalization in the Cycle
Since, as was mentioned before, AGENT-PREPOSING
involves the mention of predicate in its structural descrip
tion, and can apply freely in the derivations of derived
nominal complements, it must precede DERIVED NOMINALIZATION.
This means that DERIVED NOMINALIZATION is also cyclic. That
this Is so is also shown Indirectly by the existence of the
derived nominal complements in 4.6, which in turn have embed
ded derived nominal complements.
4.6a) Harvey's discovery of Mary's disappearance
b) Bill's search for evidence of John's betrayal
of Alice
c) Jane's reluctance to admit her hatred of Jerry
The same arguments given above to show the undesirabi
lity of making AGENT-PREPOSING a precyclic rule also apply
to DERIVED NOMINALIZATION.

125
It can be shown that all of the other rules discussed
above may be ordered after DERIVED NOMINALIZATION The
rules which mention predicate in their structural descrip
tions are blocked if the predicate has been nominalized;
hence, any rule of this sort which does not apply to any
derived nominal complement may be presumed to follow' DERIVED
NOMINALIZATION. EQUI-NP-DELETION is blocked for some
derived nominis, but not for their corresponding predicates.
As was mentioned in Chapter Three, there seem to be several
factors at work blocking EQUI-NP-DELETION in the derivations
of derived nominal complements, but whatever they are, the
simplest explanation for the fact that it is blocked in some
derived nominal complements is that the rule follows DERIVED
NOMINALIZATION.
REFLEXIVE, which does not interact with DERIVED NOMINAL
IZATION, can nevertheless be shown to follow a.\ least one
rule which follows DERIVED NOMINALIZATION. The sentences in
4." have reflexive nouns as objects. The reflexive nouns
have been raised to object position from the embedded sen
tences (in 4.8) underlying the infinitive and gerundive nom
inal complements by RAISING-TO-OBJECT.3
4.7a) John believes himself to be handsome.
b) Mary considers herself to be intelligent.
c) Bill likes to v/atch himself acting in the movies.
4.8a) John believes [V,John is handsome^
b) Mary considers [gMary is intelligent^
c) Bill likes ["q. Bill watch Fc Bill act in the
S1 S2
movies'^ Iq
"1

126
Since REFLEXIVE cannot cross sentence boundaries in its
application, the subjects of the embedded sentences cannot
become reflexive until they have been raised to object of
the predicate in the higher sentence. Hence, REFLEXIVE must
follow RAISING, which follows DERIVED NOMINALIZATION. 'The
fact that REFLEXIVE is not blocked by DERIVED NOMINALIZATION
supports Newmeyer's claim that only rules that mention predi
cate are regularly blocked by DERIVED NOMINALIZATION.
All of the rules discussed above which are blocked by
DERIVED NOMINALIZATION apply freely in gerundive nominal com
plements. GERUNDIVE NOMINALIZATION also changes the predi
cate so that it no longer functions as such. No modals are
present in gerundive nominal complements, and -ing replaces
the tense marker. I see no reason why this cannot be called
nominalization of the predicate. The presence of be and
have as auxiliaries and the absence of the preposition of
following gerundive nominis can be explained by ordering
the rules responsible after DERIVED NOMINALIZATION and
before GERUNDIVE NOMINALIZATION. If be and have are
inserted, nominalization blocks the rule. If they are pre
sent in deeply underlying structures, they are deleted by a
rule that applies to noun phrases, not to sentences. If of
is inserted, the rule applies to noun phrases, not to sen
tences. If of is present in deeply underlying structures,
it is deleted by a rule that is blocked by nominalization.
I have just argued, in effect, that most cyclic rules
are ordered between DERIVED NOMINALIZATION and GERUNDIVE

127
NOMINALIZATION. This would mean that derived nominal comple
ments would function as noun phrases for most of the cycle,
while gerundive nominal complements would function as sen
tences for most of the cycle. One of Chomskys (1970) argu
ments for denying an origin as underlying embedded sentences
to derived nominal complements is that they have a noun
phrase-like internal structure, while gerundive nominal com
plements have a sentence-like internal structure. Having
derived nominal complements function as noun phrases through
most of the cycle while gerundive nominal complements do not
would account for the observed differences. Before pursuing
this proposal further, I must turn to a discussion of the
nature of the cycle.
The Cycle g.nd Complementation
COMFLEMENTIZ£R PLACEMENT
An unstated assumption underlying the discussion in the
previous section is that DERIVED NQMINALIZATION (and GERUND
IVE NGMINALIZATION) apply in the same cycle as the other
rules discussed in connection with the examples presented in
Chapter Three. This means that in the derivation of the sen
tences in 4.9 from the underlying structure in 4.10, DERIVED
NOhIMALIZATION (in the case of 4.9a) and GERUNDIVE NOMINALI-
2ATION (in the case of 4.9b) apply to the sentence dominated
by the node S^. Rules which refer to -predicate in their
structural description are blocked if the nominalization
operation in 4.11 has been performed preceding the rule
within the same cycle.

128
4.9a) John's disruption of the meeting created panic,
b) John's disrupting the meeting created panic.
4.10) SQ
disrupt John meeting
Pred NP NP
b) NP
disruption John meeting
disrupt John meeting
The description of the operation of nominalization
given in the above paragraph is not in accord with the usual
concept of the nature of rules of complementation found in
the literature. Rosenbaum (1967) presents a rule he calls
o-
COMPLEMENTIZER PLACEMENT which is intended to account for
the placement of that in that-complements, for and to in
infinitival complements, and the -'s. and -Ing suffixes in
gerundive nominal complements. With that-complements, COM
PLEMENTIZER PLACEMENT simply adjoins that to an embedded
sentence. With infinitival or gerundive nominal complements,
COMPLEMENTIZER PLACEMENT inserts for and to or -s and -ing
into the embedded sentence. The type of complement formed
is governed by the predicate In the higher sentence. In the
derivation of the sentences in 4.12 from the underlying
structure in 4.13, COMPLEMENTIZER PLACEMENT would apply on
the cycle of In each case.

129
4.12a) That John won surprises me.
b) For John to win would surprise me.
c) John's winning surprises me.
4.13)
S
0
Pred
NP
NP
(would) surprise
S
1
I/me
Pred
NP
win J ohn
There is good reason for having COMPLEMENTIZER PLACE
MENT apply on the cycle of SQ in 4.13. This allows COMPLE
MENTIZER PLACEMENT to apply to the higher sentence which
includes the predicate which governs the type of complement
the embedded sentence may become, and it prevents COMPLE
MENTIZER PLACEMENT from applying to underlying sentences
which are not embedded.
Bresnan (1970:299-300) argues that this formulation of
Xr
COMPLEMENTIZER PLACEMENT is peculiar in that "it violates
an otherwise well-motivated universal stated by Chomsky
(1965l46), namely, that while transformations may remove
material from embedded sentences, no transformation can
insert morphological material into 'lower* sentences." I
will now examine this peculiarity of COMPLEMENTIZER PLACE
MENT in more detail.
To facilitate this discussion I would like to introduce
the terms proper constituent and embedded constituent. I
will define a proper constituent of a sentence as a con
stituent dominated by S but not dominated by an S which
J n J m

130
is also dominated by S^. In other words, a proper consti
tuent is any constituent of a sentence which is not also a
constituent of a sentence embedded in the first sentence.
An embedded constituent of a sentence S is a constituent
which is dominated by a sentence which is dominated by
the sentence S In other words, an embedded constituent of
n
a sentence is a proper constituent of a sentence embedded in
the first sentence.
The transformations discussed here may be divided into
three classes on the basis of their effect on the proper and
embedded constituents of the sentences they apply to. Some
rules affect only proper constituents. These rules may move
proper constituents, as with AGENT-PREPOSING, NP-PREPOSING
and DATIVE-MOVEMENT, create proper constituents, as with
IT-INSERTION and THERE-INSERTION. or delete proper consti
tuents, as with the rule or rules which deleter-indeterminate
subjects or objects to derived sentences like those in 4.14
from the underlying structures in 4.15*
4.l4a) It is easy to please John,
b) John is eager to please.
4.15a) easy [ypEssoraeone please John^g^yp
b) John eager [[^pj^John please someone Jg^jp
Some rules affect both proper constituents and embedded
constituents of the sentences they apply to. The RAISING
rules convert embedded constituents into proper constituents.
Finally, some rules affect only embedded constituents.
There are three subdivisions of this group. There are rules

131
which move constituents down into embedded sentences. These
rules move constituents of abstract higher sentences down
into the surface structure sentence, such as NEGATIVE-M0V2-
MENT (cf. R. Lakoff, 1968). Such rules appear to violate
Chomskys (1965) universal, but are outside the scope of
this discussion.
Another subdivision includes rules which delete embed
ded constituents, such as EQUI-NP-DELETION, which deletes an
embedded constituent when it is identical to a specified
proper constituent.
The last subdivision consists of COMPLEMENTIZER PLACE
MENT. This rule inserts complementizers into embedded sen
tences, but has no effect on any proper constituent. Its
only connection with any proper constituent is that the
predicate of the higher sentence governs the type of comple-
4
ment formed.
The RAISING rules, EQUI-NP-DELETION and COMPLEMENTIZER
PLACEMENT are all governed by the predicate of the higher
sentence. RAISING converts an embedded constituent into a
proper constituent, EQUI-NP-DELETION requires identity
between an embedded constituent and a proper constituent,
but Rosenbaums (1967) COMPLEMENTIZER PLACEMENT has no con
nection with the higher sentence that it applies on other
than that it is governed by the higher predicate.
There is only one justification for having a rule like
COMPLEMENTIZER PLACEMENT apply in the cycle of the higher
sentence in which the complement is embedded, and that is to

132
allow the rule to be governed by the predicate of the higher
sentence. This allows the rule to properly account for the
restrictions on the type of complement a predicate may take,
and for the fact that only embedded underlying sentences
become complements in surface structure. The principle of
the cycle requires that all cyclic rules apply to an embed
ded sentence before any cyclic rule applies to the higher
sentence in which the first sentence is embedded. COMPLE
MENTIZER PLACEMENT reaches down into an embedded sentence to
which the cycle has previously applied, and thus stretches
the principle of the cycle to the breaking point. Chomsky's
(1965s146) claim that no transformation can insert morpholo
gical material into an embedded sentence supports the above
analysis of COMPLEMENTIZER PLACEMENT as violating the prin
ciple of the cycle.
Bresnan (1970) offers, as a solution to the problems
with COMPLEMENTIZER PLACEMENT discussed above, the
Phrase-structure Hypothesis, which is a proposal that com
plementizers are specified in deep structure by a
phrase-structure rule that rewrites 3 as COMP+S, giving deep
structure trees like that in 4.16, and that predicates are
subcategorized for the form of the complementizer which is
lexically inserted under the node COMP.
4, l6
S
0
NP
VP
s
1
Pred NP
COMP 3
2

133
This Phrase-structure Hypothesis raises its own pro
blems- It requires the ad hoc extension of the principle of
subcategorization features to allow such features to apply
across sentence boundaries. While it is easy enough to
state that a particular predicate takes such-and-such comple
ments in surface structure, the expression of such restric
tions in underlying structure is not simple. The
Phrase-structure Hypothesis requires that a subcategoriza
tion feature on the predicate in Sq in 4.16 govern the selec
tion of a lexical item (the complementizer) within the embed
ded sentence S^. At the same time, the complementizer under
COMP must eventually be inserted into the embedded sentence
S?. Any possible modification of the Phrase-structure Hypo
thesis would still require either that subcategorization fea
tures reach across sentence boundaries, or that complement
izers be inserted into lower sentences. Bresnan's
Phrase-structure Hypothesis hides the problem of COMPLEMENT-
IZER PLACEMENT without resolving it.
I will show below that COMPLEMENTIZER PLACEMENT can be
eliminated from the grammar without resort to anything like
the Phrase-structure Hypothesis. I will first discuss the
formation of infinitival complements, and then return to
nominal complements. As I have already mentioned, the forma
tion of that-complements does not require the insertion of
any complementizer into the embedded sentence, and thus will
not enter into this discussion.

134
Infinitival Complementation
In Rosenbaums analysis, infinitival complements are
created by the insertion into the embedded sentence of for
preposed to the subject and to preposed to the part of the
predicate which would other wise carry the tense marker. A
later rule would delete the preposes for s of most infiniti
val complements. Thus, the underlying structures in 4.17
are transformed into the intermediate structures in 4.18 by
COMPLEMENTIZER PLACEMENT, and eventually become the sen
tences in 4.19.
4.17a) Tg John want Bill join John's fraternityjg Jg
b) [g Jerry expect [g Jerry win the prizejg Jg
c) [g Mary be anxious [g Mike arrive^ Jg
4.18a) [g John want [g for Bill to join Johns
fraternity]g ]s
b) Jerry expect fo for Jerry to win the
*1 ^2
Prize]S ]S
2 1
c) fo Mary be anxious fo for Mike to arrive"L 1Q
i>l b2 bl
4.19a) John wants Bill to join his fraternity.
b) Jerry expects to win the prize.
c) Mary is anxious for Mike to arrive.
Kiparsky and Kiparsky (1972) offer an alternative to
Rosenbaum's analysis. They propose that infinitival comple
ments result from the failure of the verb to agree with the
subject because the subject is no longer present as such.
Stockwell et al. (1973 s 546ff.) list three ways in v/hich the

135
subject is made no longer available for verb-agreement in
embedded sentences: when it takes the preposition for with
an emotive predicate in the higher sentence;^ when
EQUI-NP-DELETION has applied; and when RAISING-TO-SUBJECT or
RAISING-TO-OBJECT has applied. If the subject is not removed
by one of these means, a that-complement results. Pizzini
(1972) reduces this list by having the for associated with
emotive predicates adjoined to the embedded sentence, and
the subject of the embedded sentence then raised to become
the object of for. By this analysis, for remains a proper
constituent of the higher sentence.
Starting with the underlying structures in 4.20, under
this analysis, the application of RAISING-TO-OBJECT to 4.20a,
EQUI-NP-DELETION to 4.20b and the adjunctiontion of for to
the embedded sentence in 4.20c, with subsequent raising of
the embedded subject to become object of for, gives the
intermediate structures in 4.21. In each case the embedded
sentence has lost Its subject, and the predicate cannot
agree with a subject it does not have. This causes the for
mation of an infinitive, and results eventually in the sur
face sentences in 4,19.
4.20a) !"- want John fq join Bill John's fraternity!q !q
-^1 >-^2 s2 1
b) T- expect Jerry j~q win Jerry the prizel-, !q
al 2 a2 1
c) fq anxious for Mary To arrive Mike!-, !-,
bl b2 2 1
4,21a) Tq John want Bill [q join John's fraternity]]-. ]jq
1 2 2 1
b) [e Jerry expect [q win the prize]]q ]q
1 2 &2 1
c) Mary be anxious for Mike arrive!-, !,,
S1 S2 a2 *1

136
It would appear at this point that jfco is inserted into
the embedded sentence in each case, in violation of Chomsky's
claim that such insertion by a transformation is unnatural.
This would be so if TO-INSERTION were a cyclical rule.
Chomsky (1965s 146) specifically states that no morphologi
cal material can be introduced into a configuration domi
nated by S once the cycle of transformational rules has
already completed its application to this configuration."
I will argue, however, first, that TO-INSERTION is a
post-cyclic rule, and second, that post-cyclic rules are not
necessarily subject to the restriction stated by Chomsky.
TO-INSERTION (or INFINITIVE FORMATION) applies if
verb-agreement is not possible. The easiest way to handle
this fact in the grammar is to order TO-INSERTION after the
rule (or rules) responsible for verb-agreement, with TO-IN-
SERIION blocked by the application of the rule JfERB-AGREE
MENT. Ordering TO-INSERTION before VERB-AGREEMENT would
require that TO-INSERTION be sensitive to the same environ
mental facts as VERB-AGREEMENT. In either case, both rules
must be late enough to allow for all possible changes in the
subject. VERB-AGREEMENT cannot be cyclical, as there has
been no application of VERB-AGREEMENT in an embedded sen
tence prior to the removal of the subject by EQUI-NP-DELE
TION or one of the RAISING rules, and we have already seen
that these rules apply on the cycle of the next higher sen
tence, after the cycle is completed on the embedded lower
sentence. VERB-AGRESMENT cannot apply in the cycle on the

137
next higher sentence, for it then could not apply to the
highest sentence. The remaining alternative is for
VERB-AGREEMENT and TO-INSERTION to be post-cyclic.
Assuming that VERB-AGREEMENT and TO-INSERTION are
post-cyclic, I will turn to the question of whether post-cyc
lic rules are subject to the restriction against downward
insertion of morphological material. I have argued above
that the principle of the cycle requires that cyclical rules
affect proper constituents. Although embedded constituents
may also be affected, no cyclical rule may affect only embed
ded constituents. Post-cyclic rules, however, are not tied
to a particular sentence within the overall structure, but
apply to the whole structure. In these circumstances, it is
logical that post-cyclic rules not distinguish between pro
per and embedded constituents. TO-INSERTION would therefore
be a rule which inserts to before any predicate in a string
which has not undergone VERB-AGREEMENT.
In the subsection on the Cycle and Complementation just
above I stated that the assumption that nominalization occurs
on the cycle of the embedded sentence allows a simple
rule-ordering explanation for the restrictions on producti
vity of derived nominal complements. The analysis of infi
nitival complements I have given above leads to another
indirect argument for that conclusion.
As was mentioned above, an infinitival complement
results when the subject of an embedded sentence is removed
transformationally. The subject noun phrases of gerundive

138
nominal complements are sometimes deleted by EQUI-NP-DELE
TION without creating infinitival complements. Deletion on
Agent-subject identity has occurred in 4.22, and on Dat
ive-subject identity in 4.23*^
4.22a) John anticipated winning the race.
b) Bill delighted in telling tall stories.
c) Mary enjoyed reading Russian novels.
4.23a) Mark imitated Jane putting on a girdle.
b) Alice watched Jerry running the race.
c) Mike worried about Mary walking through town.
If the gerundive nominal complements are formed before
EQUI-NP-DELETION deletes their subjects, there is no possibi
lity of the deletion leading to the creation of an infiniti
val complement. In any case, while infinitival complements
are always the result of such deletions, gerundive nominal
complements are not, as many retain their subjects- If ger
undive nominal complements are formed by the application of
GERUNDIVE NOMINALIZATION on the cycle of the embedded sen-
tence (the cycle before the application of EQUI-NP-DELETION),
the desired results are obtained.
Norainaliz atio n
I have presented arguments that the rule of COMPLEMENT
IZER PLACEMENT not only violates the principle of the cycle,
but is not necessary to account for that-complements and
infinitival complements. All of the rules needed to derive
that-complements and infinitival complements either apply on
the cycle above the complement, or are post-cyclic rules.
On the other hand, I have argued that the rules of
DERIVED NOMINALIZATION and GERUNDIVE NOMINALIZATION apply on

139
the cycle of the nominal complement, so that COMPLEMENTIZER
PLACEMENT is not needed to account for them either. One
major objection possible to this view of rules of nominaliza-
tion is that the transformational derivation of nominal com
plements from underlying embedded sentences must account for
the restrictions on the types of complements predicates in
higher sentences may take. If the rules of nominalization
apply on the cycle of the nominal complements, they must
look up into the next higher sentence to see if the predi
cate of that sentence takes nominal complements. This
implies a major change in the concept of the cycle. It
would seem to me, however, that allowing rules of nominaliza-
ticn to include a higher predicate in their structural des
criptions is preferable to the insertion of morphological
material into a lower sentence. In addition, having rules
of nominalization apply on the cycle of the nominal comple
ment allows a simple explanation of the fact that many rules
do not apply in the derivation of derived nominal comple
ments, an explanation not possible with COMPLEMENTIZER PLACE
MENT. Even if the Lexical Hypothesis is adopted, however,
the above arguments would still apply to gerundive nominal
complements.
Complementation as a Process
There are two basically different ways of handling the
derivation of complements. One way is to write a single com
plex rule which accounts for all the details of structure
characteristic of a particular type of complement. The rules

140
in Leos (1963) are excellent examples of this approach. The
second way to handle the derivation of complements is to
write several fairly simple rules each accounting for only
part of the structure characteristic of a particular type of
complement. Any independent motivation of the rules helps
to strengthen such an analysis. This approach may be used
to profit elsewhere, as with Chomsky's (1970) analysis of
passivization as the two rules of AGENT-PQSTPOSING and
NP-PREPOSING.
I have already presented an analysis of the derivation
of infinitival complements as being due to the deletion of
the subject of an embedded sentence by either a RAISING rule
or EQUI-NP-DELETIONs followed by the post-cyclic application
of TO-INSERTION. Derived nominal complements are formed by
the application of DERIVED NOMINALIZATION, followed by the
rules accounting for the absence of auxiliaries, presence of
a preposition (i.e. of) before direct objects, the prenomi-
nal position of adjectives corresponding to adverbs, the for
mation of a possessive from the subject, or the insertion of
another determiner if there is no subject, and the morpholo
gical modification of the former predicate. Gerundive nomi
nal complements are formed by the application of GERUNDIVE
NOViINALIZATIQN followed by the last two rules listed above
which apply to derived nominal complements. Thus, the rule
producing possessive noun phrases applies in the derivation
of both derived nominal complements and gerundive nominal
complements, while the rule suffixing -ing to a former

141
predicate applies to some derived nominis and all gerundive
nominis.


NOTES
~ Chomsky (196569) defines 'subject-of' as, "for Eng
lish... the relation holding between the NP of a sentence of
the form NP-Aux-VP and the whole sentence," and 'object-of*
as "the relation between the NP of a VP of the form V-NP and
the whole VP."
^ AGENT-PREPOSING moves the Agent into a position imme
diately to the left of the predicate.
J The derivation of 4.6c from 4.7 c involves the raising
cf Bill from subject of Sg to object of watch in S^, the pro-
ncminalization and reflexivization of Bill object of watch
under identity with Bill subject of S^, and, finally, the
deletion of Bill subject of under identity with Bill
Agent of the highest sentence.
4
Actually, that-complementation does not share this
characteristic of COMPLEMENTIZER PLACEMENT, since it involves
the insertion of that as a proper constituent of the higher
sentence adjoined to the embedded sentence.
3 The subjunctive mood also requires the insertion of for.
^ Ross (1972c) calls forms like Jane nutting on a
girdle (in 4.22a) accusative-ing complements. However,
Fraser (1970) correctly pointed out that such forms are not
complements, but rather a noun phrase (Jane) plus a comple-
ment (putting on a girdle) .
142

CHAPTER FIVE
CONCLUSIONS
In this study I have shown that derived nominal comple
ments,. when properly defined, have a regular, if limited,
productivity. The fact that derived nominal complements
have an internal structure like that of noun phrases while
other types of complements have an internal structure more
like, that of sentences is shown to have a simple explanation.
Ordering DERIVED NOMINALIZATION before most cyclical rules
allows a simple explanation of the facts of derived nominal
complements.
However, the simple explanation I have presented here
V
is not really possible in the framework assumed by Chomsky.
In particular, I have assumed underlying structures more
abstract than those in Chomsky's theory, including the
predicate-initial analysis of underlying structures before
the application of cyclic rules. While, as Chomsky said,
the question of whether or not derived nominal complements
can be described in a transformational framework is an
empirical one, the decision as to how abstract the underly
ing analysis is to be is not empirical. By choosing a more
abstract analysis, I have been able to handle derived nomi
nal complements In a transformational framework.
143

144
The various assumptions I have made about the nature of
underlying structures also leads to other interesting
results. I have shown that an abstract, predicate-initial
analysis of the underlying structure of English eliminates
the need for any rule of extraposition, without adding any
rule to the grammar. Contrary to Chomsky's claim that a
transformational analysis of derived nominal complements
would involve the complication of the transformational com
ponent, my analysis results in a simplification of that com
ponent. My analysis also does without a base component in
the sense of Chomskys theory. As far as they can be com
pared, however, the underlying structures that I assume are
no more complicated than the deep structures of Chomskys
theory, and I do not need any distinct deep structures for
derived nominal complements, as Chomsky does.
One interesting consequence of the analysis of trans
formations that I have proposed here is that no rule which
moves a noun phrase moves it to the right. All transforma
tions acting on noun phrases are leftward movement rules.
This study, of course, can only be viewed as a prelimi
nary one, as there remain many problems in the description
of derived nominal complements, but I feel that I have shown
how this construction can be handled within transformational
theory, and while there may be an irreducible remainder of
exceptions, it is better to capture generalizations than it
is to catalog irregularities, as the Lexical Hypothesis does.

BIBLIOGRAPHY
Albury, Donald H. 1971. In defense of the transformational
origin of derived nominis. Paper presented to the
Sixth Meeting of the Southeastern Conference on Lin
guistics, November 4, Atlanta, Georgia.
. Forthcoming. English as a VOIS language.
Bresnan, Joan W, 1970. On complementizers: toward a syn
tactic theory of complementizer types. Foundations of
Language 6.297-321,
Chomsky, Noam. 1965. Aspects of the theory of syntax.
Cambridge, Mass.: The MIT Press.
. 1970. Remarks on nominalization. Readings in Eng
lish transformational grammar, ed. by Roderick A,
Jacobs and Peter S. Rosenbaum, 184-221. Waltham,
Mass,: Ginn.
Emonds, Joseph. 1969. Root and structure-preserving trans
formations. Cambridge, Mass.s MIT Dissertation.
Fillmore, Charles J. 1968. The case for case. Universals
in linguistic theory, ed. by Bmmon Bach and Robert T.
Harms. 1-88. New York: Holt, Rinehart and Winston.
Fraser, Bruce, 1970. Some remarks on the action nominali
zation in English. Readings in English transforma
tional grammar, ed. by Roderick A. Jacobs and Peter S.
Rosenbaum, 83-98. Waltham, Mass.: Ginn.
Katz, Jerrold J., and Paul M. Postal. 1964. An integrated
theory of linguistic descriptions. Cambridge, Mass.:
The MIT Press,
arsky, Paul, and Carol Kiparsky. 1971. Fact. Semantics
ed. by Danny D, Steinberg and Leon A, Jakobovits. Cam
bridge: Cambridge University Press.
off, George. 1968. Deep and surface grammar. Blooming
ton, Ind.: Indiana University Linguistics Club (Mimeo).
3

146
. 1970, Irregularity in syntax. New York: Holt,
Rinehart and Winston,
Lakoff, Robin. I98. Abstract syntax and Latin complemen
tation, Cambridge, Mass.! The MIT Press.
Lee, Paul Gregory. 1970. Subjects and agents. Columbus,
Chios OSU Dissertation.
Lees, Robert B. 1963. The grammar of English nominaliza-
tions. The Hague: Mouton.
McCawley, James D. 1970. English as a VSO language.
Language 46.286-99.
Menzel, Peter. 1969. Propositions, events, and actions in
the syntax of complementation. Los Angeles: UCLA Dis
sertation.
Milsark, Gary, 1972. Res doubl-ing. Linguistic
Inquiry 3.542-49.
Newmeyer, Frederick J. 1971. The source of derived nomi
nis-* in English. Language 4?,786-96,
Pizzini, Quentin A. 1972. A transformational analysis of
infinitival and gerundive complements in English. La
Jolla, Cal.: UC San Diego Dissertation.
Postal, Paul M, 1971. Cross-over phenomena. New York:
Holt, Rinehart and Winston.
Rosenbaum, Peter S. 1967. The grammar of English predi
cate complement constructions. Cambridge, Mass.: The
MIT Press.
Ross, John Robert. 19o7. Constraints on variables in syn
tax. Cambridge, Mass.: MIT Dissertation,
. 1972a. Act. Semantics of natural language, ed, by
Donald Davidson and Gilbert Harman, 70-126, Dordrecht
D. Reid.el,
. 1972b. Doubl-ing. Linguistic Inquiry 3.61-86,
1972c. Nouniness. Forum Lecture presented to the
LSA Linguistic Institute, June 15, Chapel Hill, N.C,
Smith, Carlotta. 1972. On causative verbs and derived norm
inals in English. Linguistic Inquiry 3.136-38.

147
Stockwell, Robert P., Paul Schacter and Barbara Hall Partee.
1973- The major syntactic structures of English. New
Yorks Holt, Rinehart and Winston.
Vendler, Zeno. 1968. Adjectives and norninalizations. The
Hague: Mouton.

BIOGRAPHICAL SKETCH
Donald Herbert Albury was born July 22, 1943, at Miami,
Florida, In June, 1961, he was graduated from North Miami
Senior High School. In December, 1964, he received the
degree of Bachelor of Arts with a major in English from the
University of Florida. From 1965 to 1967 he worked for the
Trailways Bus System and the University of Florida. From
1967 until 1969 he served in the Adjutant General Corps of
the United States .Army, including service in Vietnam. Fol
lowing his discharge from the Army in 1969, he enrolled In
the Graduate School of the University of Florida, where he
received the degree of Master of Arts with a major in Speech
in December, 1970, and has pursued his work toward the degree
of Doctor of Philosophy.
Montes.
Donald Herbert Albury is married to Virginia Elenor
They have one daughter, Rebecca Lynne. He is a
member of the Southeastern Conference on Linguistics, the
Linguistic Society of America, the Modern Language Associ
ation, and. the American Dialect Society.
143

I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy,
lerqn Casagrande* Chairman
/Associate Professor of Romance
Languages and Linguistics
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy,
Edward Charles Hutchinson
Associate Professor of Speech
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
Assistant Professor of English
Florida State University

I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
Assistant Professor of Spanish
and Portuguese and Lin
guistics
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
Associate Professor of Speech
and Linguistics
This dissertation was submitted to the Graduate Faculty
of the Program in Linguistics in the College of Arts and
Sciences and to the Graduate Council, and was accepted as
partial fulfillment of the requirements for the degree of
Doctor of Philosophy.
March, 1974
Dean, Graduate School

2 3. 9 6.a
¡L
i .



4q
On the other hand, the derived nominal complements in
2.79 which correspond to the sentences in 2*75 and gerund
ive nominal complements in 2.77 > are perfectly acceptable.
2.79a) John's vain attempt to win the race
b) John's hurried departure on the bus
c) the existence of a god
d) the occurrence of a riot yesterday
e) the presence of some wine in the bottle
f) John's possession of a car
g) Bill's usual residence in a hotel
It seemsj therefore, that it is possible for a Germanic
(or native) verb to have no corresponding nominal in a
derived nominal complement when there is a closely synony
mous Latinate verb with such a corresponding nominal. The
transformationalist position may be maintained in the face
of these examples in at least two ways verbs like try,
leave. be, have and live can be marked in the lexicon to
block DERIVED NOMINALIZATION, which ignores the existence of
the Latinate synonyms} or, some sort of lexicaX alternation
may be posited, with the Latinate option being mandatory in
derived nominal complements. I prefer the second position,
end I will discuss some different evidence for such lexical
alternation for the synonyms of be, in Chapter Three.
Action Nominis and the Head Noun Action
in Chapter One I noted that derived nominal complements
did not occur as complements of the head noun action. If
action nominal complements are a subclass of derived nominal
complements, then they should not occur as complements of
action either. The examples in 2.80 show that this is
indeed the case.


76
3.103a)
b)
c)
d)
3 104a)
b)
c)
d)
3.105a)
b)
c)
d)
Failure
Peter's (action of) insulting Tom was com
plained about by Tom.
^Bill's (action of) leaving Peggy v/as testified
about by Jerry.
(The fact of) Mary's participation in the
crime was attested to by Mike.
(The event of) Alice's disappearance was
reported on by David.
^Peter's (action of) insulting Tom's being
complained about by John
^Bill's (action of) leaving Peggy's being
testified about by Jerry
*(the fact of) Mary's participation in the
crime's being attested to by Mike
*(the event of) Alice's disappearance's being
reported on by David
^Peter's (action of) insulting Tom's complaint
about by John
^Bill's (action of) leaving Peggy's testimony
about by Jerry
*(the fact of) Mary's participation in the
crime's attestation to by Mike
*(the event of) Alice's disappearance's report
on by David
of both AGENT-PROPOSING and NP-PREPOSING to
apply in the derivation of sentences and gerundive nominal
complements results in unacceptable forms, as in 3*106 and
3.10?, but derived nominal complements to which neither rule
has applied are acceptable, including one with testimony
about (from testify about) as in 3-108.
3.106a) *Was complained about by John Peter's (action
of) insulting Tom.
b) *Was testified about by Jerry Bill's (action
of) leaving Peggy.
c) -"Was attested to by Mike (the fact of) Mary's
participation in the crime.
d) *Was reported on by David (the event of)
Alice's disappearance.
3.107a) -being complained about by John Peter's (action
of) insulting Tom
b) -being testified about by Jerry Bill's (action
of) leaving Peggy


99
I have no explanation for this fact other than the sugges
tion that some unknown rule has applied in the derivation of
the sentences in 3-173b and c which has not applied in the
derivation of the sentence in 3-173a, and that this unknown
rules blocks DERIVED NOMINALIZATION. On the other hand,
there are derived nominal complements corresponding to sen
tences with command without regard to whether the sentences
have undergone EQUI-NP-DELETION. I will show below that
DERIVED NOMINALIZATION blocks the application of EQUI-NP-DE-
LETION for only some underlying predicates, so that there is
nothing troublesome about its failure to do so for command.
That the unacceptability of 3*1751 is linked to the applica
tion of EQUI-NP-DELETION is shown by the acceptability of
my forcing: of the issue corresponding to I forced the issue,
which has not undergone EQUI-NP-DELETION. 'Thus, it is not
the case that DERIVED NOMINALIZATION is blocked idiosyncrati-
cally for force. The examples in 3*175 do not provide clear
evidence on the interaction of EQUI-NP-DELETION and DERIVED
NOMINALIZATION. To find such evidence we must turn to the
Agent-subject condition on EQUI-NP-DELETION.
The sentences in 3*176 result from the deletion of the
subject of the embedded sentence when it is identical to the
Agent noun phrase of the next higher sentence. The corres
ponding gerundive nominal complements are given in 3*177*
There are no acceptable corresponding derived nominal com
plements, the expected forms in 3*178 being unacceptable.


35
As was noted, above (cf. 2-33) not all derived nominis
form plurals, so the fact that many apparent derived nomi
nis in -ins do not is not strong counter-evidence.
Paired Nominis
The nominal complements in 2.39a and b (repeated here
in 2.67), cited by Chomsky (1970:214), appear to be derived
nominal complements with nominis formed with -ing which are
parallel to the derived nominal complements in 2.68, which
have nominis formed other than v/ith the suffix -ing.
2.67a) John's refusing of the offer
b) John's proving of the theorem
2.68a) John's refusal of the offer
b) John's proof of the theorem
The derived nominal complements with nominis formed
other than v/ith the -ing suffix in 2.68 readily take prenora-
inal adjectives, as was discussed above, so that we have
derived nominal complements like those in 2.69..
2.69a) John's abrupt refusal of the offer
b) John's brilliant proof of the theorem
The nominal complements with nominis formed with the
-ing suffix in 2.67 seem to less readily accept prenominal
adjectives, as is indicated by the strangeness of the nom
inal complements in 2.70. That this strangeness is not due
to the nominal having an -ing suffix alone is shown by the
normal occurrence of prenominal adjectives v/ith derived nom
inal complements with nominis formed with -ing in 2.48,
2.49, 2.56, 2.57 2.6o and 2.61 above.
2.70a) ?John's abrupt refusing of the offer
b) ?John's brilliant proving of the theorem


12
I do not discuss the problem of the apparent irregular
ities in the semantic relationship between related predi
cates and derived nominis, I will adopt Newmeyer's (1971)
claim that, at the worst, the transformationalist position
is no less adquate on this point than the Lexical Hypothe
sis, since, if there are no regularities, the information
on restrictions on meanings of derived nominis must be
part of the lexical entrys of the underlying predicate with
the transformationalist position, of the underlying predi
cate/nominal with the Lexical Hypothesis.
Chapter Two is a discussion of the surface structure of
derived nominal complements, and arguments for including in
the class of derived nominal complements certain structures
which have not previously been so identified. Chapter Three
discusses the rules which do not apply in the derivation of
derived nominal complements, or which have different condi
tions on applicability in derived nominal complements, but
which apply without restriction to sentences and gerundive
nominal complements. Chapter Four discusses the ordering of
rules which will account for the structure of derived nominal
complements, and the question of which cycle the appropriate
rules apply on. Chapter Five summarizes the thesis, and
presents certain questions of theoretical import.


134
Infinitival Complementation
In Rosenbaums analysis, infinitival complements are
created by the insertion into the embedded sentence of for
preposed to the subject and to preposed to the part of the
predicate which would other wise carry the tense marker. A
later rule would delete the preposes for s of most infiniti
val complements. Thus, the underlying structures in 4.17
are transformed into the intermediate structures in 4.18 by
COMPLEMENTIZER PLACEMENT, and eventually become the sen
tences in 4.19.
4.17a) Tg John want Bill join John's fraternityjg Jg
b) [g Jerry expect [g Jerry win the prizejg Jg
c) [g Mary be anxious [g Mike arrive^ Jg
4.18a) [g John want [g for Bill to join Johns
fraternity]g ]s
b) Jerry expect fo for Jerry to win the
*1 ^2
Prize]S ]S
2 1
c) fo Mary be anxious fo for Mike to arrive"L 1Q
i>l b2 bl
4.19a) John wants Bill to join his fraternity.
b) Jerry expects to win the prize.
c) Mary is anxious for Mike to arrive.
Kiparsky and Kiparsky (1972) offer an alternative to
Rosenbaum's analysis. They propose that infinitival comple
ments result from the failure of the verb to agree with the
subject because the subject is no longer present as such.
Stockwell et al. (1973 s 546ff.) list three ways in v/hich the


77
3.107c) *being attested to by Mike (the fact of) Mary's
participation in the crime
d) *being reported on by David (the event of)
Alice's disappearance
3.108a) the complaint by John about Peter's (action
of) insulting Tom
b) the testimony by Jerry about Bill's (action
of) leaving Peggy
c) the attestation by Mike to (the fact of)
Mary's participation in the crime
d) the report by David on (the event of) Alice's
disappearance
While the head nouns of the embedded nominal comple
ments in the examples above may be deleted, they may not be
separated from the complements, i.e., by moving the head
nouns into surface subject position but leaving the comple
ments at the right end of the structure (cf. With That-Com
plements, below), so that all of the examples in 3.109
3.110 and 3.111 are unacceptable.
3.109a)
b)
c)
d)
3.110a)
b)
c)
d)
3.1Ha)
b)
c)
d)
^Peter's action was complained about by John of
insulting Tom,
^Bill's action was testified about, by Mike of
leaving Alice.
-The fact was attested to by Mike of Mary's
participation in the crime.
*The event was reported on by David of Alice's
disappearance
^Peter's action's being complained about by
John of insulting Tom
*Bill's action's being testified about by Jerry
of leaving Peggy
*the fact's being attested to by Mike of Mary's
participation in the crime
*the event's being reported on by David of
Alice's disappearance
*Peter5s action's complaint by John about
insulting Tom
^Bill's action's testimony by Jerry about
leaving Peggy
*the fact's attestation by Mike to Mary's par
ticipation in the crime
"the event's report by David on Alice's dis
appearance


121
REFLEXIVE applies between two identical noun phrases
dominated by a single 3 node with no intervening 3 node
(clause mates). There is no need to mention predicate in
the structural description of the rule. EQUI-NP-DELETION
applies between two identical noun phrases also, but the
noun phrases require careful definition. One noun phrase
(the one deleted) is the subject of an embedded sentence
(or complement). While subject is defined in terms of the
relation of a noun phrase to a predicate, this particular
occurrence of subject is embedded within the sentence the
rule is applying to, and does not bear on Newmeyer's claim,
which refers to the higher sentence which is being nominal-
ized. In the analysis of this rule presented in Stoekwell
et al. (1973) the appropriate noun of the higher sentence
is defined by case role, which is independent of any mention
of predicate. As long as EQUI-NP-DELETION canbe stated
without mention of the predicate of the higher sentence,
there is no contradiction of Newmeyer's proposal.
The Order of the Rules
The Cyclic Nature of AGENT-PREPOSING
Newmeyer states that DERIVED NOMINALIZATION precedes
all the cyclic rules. As was mentioned in Chapter Three,
the first rule in Chomsky's (1970) analysis of passivization
is AGENT-P03TP0SING (equivalent to the nonapplication of
AGENT-PREP03ING in a predicate-initial analysis).
AGENT-P0STP0SING (or AGENT-PREPOSING) applies freely in
derived nominal complements, even though predicate is


137
next higher sentence, for it then could not apply to the
highest sentence. The remaining alternative is for
VERB-AGREEMENT and TO-INSERTION to be post-cyclic.
Assuming that VERB-AGREEMENT and TO-INSERTION are
post-cyclic, I will turn to the question of whether post-cyc
lic rules are subject to the restriction against downward
insertion of morphological material. I have argued above
that the principle of the cycle requires that cyclical rules
affect proper constituents. Although embedded constituents
may also be affected, no cyclical rule may affect only embed
ded constituents. Post-cyclic rules, however, are not tied
to a particular sentence within the overall structure, but
apply to the whole structure. In these circumstances, it is
logical that post-cyclic rules not distinguish between pro
per and embedded constituents. TO-INSERTION would therefore
be a rule which inserts to before any predicate in a string
which has not undergone VERB-AGREEMENT.
In the subsection on the Cycle and Complementation just
above I stated that the assumption that nominalization occurs
on the cycle of the embedded sentence allows a simple
rule-ordering explanation for the restrictions on producti
vity of derived nominal complements. The analysis of infi
nitival complements I have given above leads to another
indirect argument for that conclusion.
As was mentioned above, an infinitival complement
results when the subject of an embedded sentence is removed
transformationally. The subject noun phrases of gerundive


44
find many compounds like those in 2.86 corresponding to the
agentive nominal complements in 2.87.
2.86a) lion tamer
b) bookkeeper
c) shock absorber
2.87a) tamer of lions
b) keeper of books
c) absorber of shocks
But some nominis in other derived nominal complements
also form such compounds, as in 2.88, corresponding to the
derived nominal complements in 2.89.
2.88a) token payment
b) tax assessment
c) art collection
2.89a) payment of a token
b) assessment of a tax
c) collection of art
Agentive nominal complements have the same surface
structure as derived nominal complements. I see no reason
1* _
to not include agentive nominal complements in the class of
derived nominal complements.
Summary
In this chapter I have defined derived nominal comple
ments in terms of surface structure as those nominal comple
ments which allow a variety of determiners, allow prenominal
adjectives which correspond to adverbs in sentences, have
prepositions with all objects, and allow the nominal to be
pluralized, but which have no auxiliaries. Using this defi
nition, I have then argued that the structures known as
action nominis and agentive nominis are really subclasses
of the class of derived nominal complements.


73
Again, when AGENT-PROPOSING has not applied in the
derivation of a derived nominal complement, the application
of NP-PREPOSING is optional for some nominis and blocked
for others, as indicated by the derived nominal complements
in 3-92 and 3*93*
3-92a) the city's destruction
b) Abby's acquittal
c) *the offer's refusal
d) *the book's criticism
3-93a) the destruction of the city
b) the acquittal of Abby
c) the refusal of the offer
d) the criticism of the book
Finally, if the underlying predicate-initial structure
contains an intransitive predicate,
then NP-PREPOSING
applies obligatorily in the derivation of sentences like
those in 3*9^ and gerundive nominal complements like those
in 3*95 with the forms in 3*96 and 3*97 being unacceptable.
3-9^a) The boat sank suddenly.
b) John arrived.
c) Mary is kind.
d) Bill is friendly.
3.95a) the boat's sinking suddenly
b) John's arriving
c) Mary's being kind
d) Bill's being friendly
3.96a) *3ank the boat suddenly.
b) ^Arrived John.
c) *Is kind Mary.
d) *Is friendly Bill.
3.97a) *sinking the boat suddenly
b) ^arriving John
c) *being kind Mary
d) *being friendly Bill
Again, the application of NP-PREPOSING in the deri
vation of derived nominal complements corresponding to


29
nominal depends on how the nominal is formed. If a causa
tive verb takes a nominalizing suffix of Latin origin
(-tion, -al, -ment), then it has both transitive and intrans
itive derived nominis. If a causative verb does not take
such a nominalizing suffix, then it occurs only intransi
tively in derived nominis," She then observes that, in
general, verbs of Latin origin form derived nominis with
the suffixes of Latin origin, while verbs of what she calls
Anglo-Saxon origin form derived nominis with the suffixes
-th, -ness, or & (null), which can be considered to be
native, as opposed to the suffixes borrowed from Latin.
Smith has, based on the data above, reached the conclu
sion that "the grammar must distinguish at least two classes
of 'causative' verbs: those that do and do not have transi
tive derived nominis." As we have just seen above, that
distinction is to be based on something like a native/bor
rowed- Latin dichotomy. It seems very strange, however, to
state that a part of the lexicon associated with borrowed
(Latinate) derivational suffixes shows more productivity
than another part of the lexicon associated with native
derivational suffixes. This raises the possibility that the
causative verbs associated with native derivational suffixes
also have corresponding derived nominis, but that for some
reason Smith and Chomsky have failed to recognize them.
Using the definition of derived nominal complements I have
developed above, I will now look for derived nominal comple
ments corresponding to sentences with such causative verbs.


66
3.66a) It seems that John is late.
b) It appears that Mildred has fallen down.
c) It happens that Mike is a brilliant student.
3.o7a) -That John is late seems.
b) *That Mildred has fallen down appears.
c) *That Mike is a brilliant student happens.
It would seem that at least some sentences can be
derived only by obligatory application of IT-EXTRAPOSITION.
I have noted above in Chapter One certain arguments for
assuming an underlying predicate-initial order in English.
All extraposed sentences, including those resulting from the
obligatory application of IT-EXTRAPOSITION. have a predi
cate-initial order, if the presence of the semantically
empty pronoun it in subject position is discounted. I pro
pose that there is no rule of IT-EXTRAPOSITION in English,
and that nonextraposed sentences are derived from the struc
tures underlying extraposed sentences. I will discuss the
details of such derivations in the section on NP-PREPOSINC-
below.
Passive Sentences
I will turn next to a problem, which is not raised by
Chomsky's examplesj that of derived nominal complements cor
responding to passive sentences. The sentences in 3*68 have
the corresponding gerundive nominal complements in 3-69 and
derived nominal complements in 3.70.
3.68a) John refused the offer.
b) John criticized the book.
c) The enemy destroyed the city.
d) The jury acquitted Abby.
3.69a) John's refusing the offer
b) John's criticizing the book


120
derived nominal complements than did in the derivation of
sentences, but as I showed in Chapter Three, there is no
other reason to postulate two rules of NP-PREPOSING. I see
no non-ad hoc way around this problem. .Either NP-PREPOSING
violates Newmeyer's proposed constraint, or two rules of
NP-PREP03ING are admitted on the sole basis of different
conditions on application in the derivation of sentences
and of derived nominal complements.
Newmeyer bolsters the argument for his proposal by
pointing to the rules of EQUI-NP-DELETION and REFLEXIVE.
SQUI-NP-DELETION applies to many, but not all, derived nom
inal complements which meet the structural description of the
rule. Applicability seems to be governed by many interact
ing factors, two of which are the previous application of
DERIVED NOMINALIZATION and the predicate nominal!zed, as
mentioned in the section on that rule in Chapter* Three.
There appears to be no restriction on the application of
REFLEXIVE to derived nominal complements, as indicated by
the sentences in 4.1, with their corresponding gerundive
nominal complements in 4.2 and derived nominal complements
in 4.3
4.1a) Bill hates himself.
b) The president promoted himself to generalsimo.
c) John described himself as intelligent.
4.2a) Bill's hating himself.
b) the president's promoting himself to generalsimo
c) John's describing himself as intelligent
4.3a) Bill's hatred of himself
b) the president's promotion of himself to
generalsimo
c) Johns description of himself as intelligent


51
Chomsky's Subcategorization Account
I will first consider Chomsky's (1970s191) claim that a
correct account of the facts of productivity of derived nom
inal complements should be based on subcategorization fea
tures. The pair of sentences in 3-10 and the pair of sen
tences in 3-11 with their corresponding gerundive and
derived nominal complements, present parallel problems.
3*10a) John is easy to please,
b) John is eager to please.
3.11a) John is certain to win the prize.
b) John is certain that Bill will win the prize.
The corresponding gerundive nominal complements in 3-12
and 3*13 and the corresponding derived nominal complements
in 3.14b and 3*154 are all acceptable, while the expected
corresponding derived nominal complements in 3l4a and 315a
are not acceptable,
3.12a) John's being easy to please
b) John's being eager to please
3,13a) John's being certain to win the prize
b) John's being certain that Bill will win the prize
3.14a) *John's easiness to please
b) John's eagerness to please
3.15a) ^John's certainty to win the prize
b) John's certainty that Bill will win the prize
Chomsky (1970:191) attempts to explain the acceptabili
ty of the derived nominal complements in 3>l4b and 3-154., in
contrast with the unacceptability of forms like those in
3.14a and 3-15a, in terms of the subcategorization features
of eager, easy and certain. Chomsky states that eager is
entered in the lexicon with a strict subcategorization


32
The Intransitive verbs are used in sentences like those
in 2.54, which have the corresponding gerundive nominal com-
plenient s in
2.55 and the corresponding derived nominal com-
plements in
2.56 and 2.57.
2.54a)
b)
c)
The light slowly dimmed.
The bridge gradually lowered.
The road suddenly widened.
2,55a)
b)
c)
the light's slowly dimming
the bridges gradually lowering
the road's suddenly widening
2.56a)
b)
c)
the light's slow dimming
the bridge's gradual lowering
the road's sudden widening
2.57a)
b)
c)
the slow dimming of the light
the gradual lowering of the bridge
the sudden widening of the road
The transitive verbs are used in sentences like those
in 2.53 which have the corresponding gerundive nominal com
plements in 2,59 and the corresponding derived nominal com
plements in 2,60 and 2,61,
2.58a)
b)
c)
^ J
John suddenly dimmed the light.
The tender gradually lowered the bridge,
the city widened the road recently.
2.59a)
b)
c)
John's suddenly dimming the light
the tender's gradually lowering the bridge
the city's widening the road recently
2.60a)
0)
c)
John's sudden dimming of the light
the tender's gradual lowering of the bridge
the city's recent widening of the road
2,61a)
b)
c)
the sudden dimming of the light by John
the gradual lowering of the bridge by the tende
the recent widening of the road by the city
All of the derived nominal complements in 2.56, 2,57,
.60 and 2.61 have nominis formed with the -ing suffix,
niy the nominis in 2.52 and 2.53, which correspond to


93
to the subject of the embedded sentence. The case of
require without a Dative noun phrase is represented by
3.173a require with a Dative noun phrase but without appli
cation of EQUI-NP-DELETION by 3.173, require with a Dative
noun phrase and application of EQUI-NP-DELETION by 3.173c
comraand without a Dative noun phrase by 3.173d, command with
a Dative noun phrase by 3.173e, and force by 3.1731. The
failure of EQUI-NP-DELETION to apply for whatever reason pro
duces that-complements from the embedded sentences, as in
3.173a b and d. The sentences in 3.173 have corresponding
gerundive nominal complements, as in 3.174-, but only 3.173a,
d and e have acceptable corresponding derived nominal comple
ments, as indicated by the forms in 3.175.
3.173a) I require that you solve the problem.
b) I require of you that you solve the problem.
c) I require you to solve the problem.
d) I commanded that he solve the problem.
e) I commanded him to solve the problem.
f) I forced him to solve the problem.
3.174-a) my requiring that you solve the problem
b) my requiring of you that you solve the problem
c) my requiring you to solve the problem
d) my commanding that he solve the problem
e) my commanding him to solve the problem
f) my forcing him to solve the problem
3.175a) my requirement that you solve the problem
b) *my requirement of you that you solve the problem
c) *my requirement of you to solve the problem
d) my command that he solve the problem
e) my command to him to solve the problem
f) *my forcing of him to solve the problem
There are no derived nominal complements corresponding
to sentences with require which have a Dative noun phrase
whether or not EQUI-NP-DELETION has applied to the sentence.


2.80a) ^John's action of opening of the door
b) *John's action of sinking of the boat
c) *John's action of dimming of the light
d) ^John's action of refusing of the offer
e) ^John's action of proving of the theorem
Menzel (1970) argues that the head nouns associated
with nominal complements in specific structures define what
the nominal complement is, i.e., a complement of event is an
event, a complement of action is an action, etc. It is
ironic that those structures called "action nominis" cannot
be complements of the head noun action, and thus are not
actions.
Summary
I have shown in this section that the so-called "action
nominis have the surface structure of derived nominal com
plements. That is, the nominal complements sometimes called
"action nominis" share with other derived nominal comple
ments the features of a possible variety of determiners,
prenominai adjectives corresponding to adverbs in sentences,
prepositions with all objects, the possibility of penaliza
tion (although less for action nominis than is so for other
derived nominal complements), the complete absence of auxi
liaries, and the restriction from being the complement of
the head noun action. The fact that action nominal comple
ments have nominis formed with the suffix -ing, as do ger
undive nominal complements, ought to have no bearing on this
classification.


113
argue below ( cf. IT-EXTRA?OSI PION and NP-PkwPOSIKG) that:
PAS SIVIZATION and IT-EXTRAPOSITION are not rules of English,
and that the relationships to be accounted for by those
rules are accounted for by the rule of NP-PREPOSING. Under
this analysis, RAI3ING-T0-03JECT precedes NP-PROPOSING, and
TOUGH-MOVEMENT and RAISING-TO-SUBJECT are not ordered with
respect to NP-PREPOSING.
^ The rule is called EXTRAPOSITION in Rosenbaum (1967).
The name IT-EXTRAPOSITION has come into use to distinguish
it from other rules of EXTRAPOSITION.
"2 The unacceptability of the forms in 3-53 and 3-5^
can be "explained" by the statement that sentential comple
ments are not allowed to form possessives in English. This
is an observation, however, not an explanation. Indeed, this
restriction also applies to derived nominal complements, as
in ^.Dannys departure's imminence, while ordinary noun
phrases of greater length freely form possessives, as in the
King of England's crown and the man next door's son-in-law.
In this case, derived nominal complements behave like embed
ded sentences, not like noun phrases.
13
J It would appear that sentences like those in i are
counter-evidence to this statement.
a) That John is late seems likely.
b) That Mildred has fallen down appears certain.
c) That Mike is a brilliant student happens to be true
However, such sentences are derived by RAISING-T0-SU3JECT
from the structures underlying the sentences in ii.


101
3.133a) John's intending to leave early
b) Mike's desiring to leave early
c) Bill's refusing to leave early
d) Jerry's attempting to leave early
3.184a) John's intention to leave early
b) Mike's desire to leave early
c) Bill's refusal to leave early
d) Jerry's attempt to leave early
It appears from these examples that EQUI-NP-DELETION,
like NP-PREPOSING, can apply to some derived nominal comple
ments s but not to others. The applicability of EQUI-NP-DELE-
TION seems to be governed by a feature on the predicate.
That the conditions on applicability may be more compli
cated is indicated by the fact that while 3.178c, *Bill's
expectation to leave early, seems hopelessly bad, the
related form, Bill's expectation of an early departure, is
much better, A further point is that in the derived nominal
complements in 3.184, all of the derived nominis except
refusal will also take embedded gerundive or deprived nominal
complements to which EQUI-NP-DELETICN has applied, as in
3.185.
3.185a) John's intention of leaving early
b) Mike's desire for leaving early/an early
departure
c) *Bill's refusal of leaving early/an early
departure
a) Jerry's attempt at leaving early/an early
departure
These facts indicate a difference in the application of
EQUI-NP-DELETION to derived nominal complements with embedded
infinitival complements and to those with embedded gerundive
or derived nominal complements. I will have more to say
about that difference in Chapter Four,


125
It can be shown that all of the other rules discussed
above may be ordered after DERIVED NOMINALIZATION The
rules which mention predicate in their structural descrip
tions are blocked if the predicate has been nominalized;
hence, any rule of this sort which does not apply to any
derived nominal complement may be presumed to follow' DERIVED
NOMINALIZATION. EQUI-NP-DELETION is blocked for some
derived nominis, but not for their corresponding predicates.
As was mentioned in Chapter Three, there seem to be several
factors at work blocking EQUI-NP-DELETION in the derivations
of derived nominal complements, but whatever they are, the
simplest explanation for the fact that it is blocked in some
derived nominal complements is that the rule follows DERIVED
NOMINALIZATION.
REFLEXIVE, which does not interact with DERIVED NOMINAL
IZATION, can nevertheless be shown to follow a.\ least one
rule which follows DERIVED NOMINALIZATION. The sentences in
4." have reflexive nouns as objects. The reflexive nouns
have been raised to object position from the embedded sen
tences (in 4.8) underlying the infinitive and gerundive nom
inal complements by RAISING-TO-OBJECT.3
4.7a) John believes himself to be handsome.
b) Mary considers herself to be intelligent.
c) Bill likes to v/atch himself acting in the movies.
4.8a) John believes [V,John is handsome^
b) Mary considers [gMary is intelligent^
c) Bill likes ["q. Bill watch Fc Bill act in the
S1 S2
movies'^ Iq
"1


3.l60a)
b)
c)
3.161a)
b)
c)
3.l62ai)
ii)
bi)
ii)
ci)
ii)
There appeared, to be no hope.
There seemed to he a disturbance.
There happened to be some wine in the bottle.
there appearing to be no hope^
there seeming to be a disturbance
there happening to be some wine in the bottle
17
^there's appearance to be no hope
*there*s appearance of being no hope
^there's semblance to be a disturbance
*there's semblance of being a disturbance
*there's happening to be some wine in the
bottle
*there's happening of being some wine in the
bottle
The forms in 3.162 do not provide any evidence of the
relation between THERE-INSERTION and DERIVED NOMINALIZATION,
however, as the sentences in 3.160 have been derived by RAIS-
ING-TO-SUBJECT from the structures underlying the sentences
in 3.163, which have the corresponding gerundive nominal com-
18
plements in 3.164- and derived nominal complements in 3.165*
3.163a) It appeared that there was no hope.
b) It seemed that there was a disturbance.
c) It happened that there was some wine in the
bottle.
3.164a) it(s) appearing that there v/as no hope
b) it(s) seeming that there was a disturbance
c) it(s) happening that there was some wine in
the bottle
3.1652-) the appearance that there was no hope
b) ?the semblance that there was a disturbance
c) the happenstance that there was some wine in
the bottle
If we take the embedded that-complements of the sen
tences in 3.163 we have sentences in which there has not
been raised from an embedded sentence, as in 3.166. These
sentences have the corresponding gerundive nominal comple
ments in 3,16?. It would seem that the derived nominal


75
Again, I will claim that AGENT-PREP0SING has applied to
predicate-initial underlying structures to yield the sen
tences in 3.100, and their corresponding gerundive nominal
complements in 3.101 and derived nominal complements in 3.102
3.100a)
John complained about Peter's (action of)
b)
insulting Tom.
Jerry testified about Bill's (action of)
c)
leaving Peggy.
Mike attested to (the fact of) Mary's parti
dj
cipation in the crime.
David reported on (the event of) Alices
disappearance.
3.101a)
John's complaining about Peter's (action of)
b)
insulting Tom
Jerry's testifying about Bill's (action of)
c)
leaving Peggy
Mikes attesting to (the fact of) Mary's
d)
participation in the crime
Davids reporting on (the event of) Alice's
disappearance
3.102a)
John's complaint about Peter's (action of)
b)
insulting Tom
Jerry's testimony about Bill's (action of)
c)
leaving Peggy
Mike's attestation to (the fact of) Mary's
d)
participation in the crime
David's report on (the event of) Alice's
disappearance
The verbs complain about, attest to and report on allow
passivization in my dialect, while testify about does not.
That is, AGENT-PREPOSING applies optionally to the first
three predicates and obligatorily to testify about in the
derivation of sentences. If AGENT-PREPOSING does not apply,
then NP-PREPOSING must apply to yield sentences like those
in 3.103, The expected corresponding gerundive nominal com
plements in 3.104 and derived nominal complements in 3.105
are not acceptable (cf. Note 12 above).


tion of action for do in nominal complements, and the dele
tion of the lower of two identical agents (cf. SQUI-NP-DELE
TION in Chapter Three), will account for the above facts
within the transformationalist position.
3
There are extensive arguments in the literature for
cyclical ordering. I am not concerned here with the argu
ments over extrinsic vs. intrinsic ordering, but will merely
state that assuming ordering permits useful generalizations,
as in Chapter Four and Chapter Five.
4
At this point I am rejecting the Aspects model of
grammar. The Aspects model has syntax as the most basic
component of the grammar, with both semantics and phonology
acting as interpretive components of the output of the syn
tactic component. I believe that this model is unrealistic,
even as a model of competence rather than of performance.
Speech is a stream of sound perceived as a linear string of
distinct units. Semantic propositions, on the other hand,
are unordered in any dimensional sense. There is no dimen
sional order implied in the statement that someone is the
agent of such-and-such action. Syntax, then, is that part
of grammar which relates unordered semantic relationships
to linearly ordered phonological strings.
James McCawley, during a discussion at the 1972 LSA
Annual meeting, summed up the predicate-initial hypothesis
in the statement that if there is any linear order so early
in the grammar, it is a predicate-initial order.
^ I will present my arguments for this analysis in
Albury (forthcoming).


ACKNOWLEDGMENTS
I wish to thank first of all my chairman, Jean
Casagrande, who has always encouraged me to approach lin
guistics in terms of what are the facts of language, and,
then, what do the facts say about the theory. He has had
the faith to let me try unorthodox approaches until I could
either justify them or give them up. He has, above all,
helped me "keep the faith" with my own proposals when my
enthusiasm flagged.
I also wish to thank Peter Menzel, who has pointed out
many errors and inconsistencies in my arguments. Of course,
all remaining errors remain my responsibility.
My wife Ginny and daughter Rebecca have had to put up
with me for the last year while I worked on this disserta
tion. I am grateful for their patience and support,
I also wish to acknowledge financial support from the
University of Florida, in the form of a scholarship and a
research assistantship, during the writing of this disser
tation.


130
is also dominated by S^. In other words, a proper consti
tuent is any constituent of a sentence which is not also a
constituent of a sentence embedded in the first sentence.
An embedded constituent of a sentence S is a constituent
which is dominated by a sentence which is dominated by
the sentence S In other words, an embedded constituent of
n
a sentence is a proper constituent of a sentence embedded in
the first sentence.
The transformations discussed here may be divided into
three classes on the basis of their effect on the proper and
embedded constituents of the sentences they apply to. Some
rules affect only proper constituents. These rules may move
proper constituents, as with AGENT-PREPOSING, NP-PREPOSING
and DATIVE-MOVEMENT, create proper constituents, as with
IT-INSERTION and THERE-INSERTION. or delete proper consti
tuents, as with the rule or rules which deleter-indeterminate
subjects or objects to derived sentences like those in 4.14
from the underlying structures in 4.15*
4.l4a) It is easy to please John,
b) John is eager to please.
4.15a) easy [ypEssoraeone please John^g^yp
b) John eager [[^pj^John please someone Jg^jp
Some rules affect both proper constituents and embedded
constituents of the sentences they apply to. The RAISING
rules convert embedded constituents into proper constituents.
Finally, some rules affect only embedded constituents.
There are three subdivisions of this group. There are rules


31
complements which I have adopted. I will therefore assume
for now that -ing is one of the suffixes forming nominis
in derived nominal complements, which happens to be homo-
phonous with the gerundive suffix -ing.
Nominis of Dead.iectival Verbs
The rule of CAUSATIVE FORMATION is not the only rule by
which a verb may be derived from an underlying form with dif'
ferent properties. Smith (1972) mentions that we find the
derived nominal complement the lights dimness, but not
^John's dimness of the light. Dim is homophonous for three
sensess adjective, intransitive verb derived by the rule of
INCHOATIVE FORMATION (cf. G. Lakoff, 1970), and transitive
verb derived by the rule of CAUSATIVE FORMATION. Other
adjective-verb sets related by these rules are low:lower;
lower and wide¡widens widen. The adjectives are used in sen
tences like those in 2*50, which have the corresponding ger
undive nominal complements In 2.51 and the corresponding
derived nominal complements in 2.52 and 2.53
2.50a) The light was dim.
b) The bridge was low.
c) The road was wide.
2.51a) the light's being dim
b) the bridge's being low
c) the road's being wide
2.52a) the light's dimness
b) the bridge's lowness
c) the road's width
2.53a) the dimness of the light
b) the lowness of the bridge
c) the width of the road


Ill
Removal of the subject of a that-complement (here, by
RAISING-TO-SUBJECT) results in the formation of an infini
tival complement, as in 3*25- This will be discussed in more
detail in Chapter Four.
^ This is unacceptable on the reading of *Jerrv's
appearance of opening the door. A reading based on some
thing like Jerry appeared in order to open the door is not
intended.
8
The possessive suffix does not seem to be necessary
with it in gerundive nominal complements, and its absence
may improve such complements for some speakers. Some speak
ers also apparently accept nouns without the possessive suf
fix in gerundive nominal complements. Ross (1972b) calls
such forms accusative-ins: complements. Fraser (1970) dis
cusses apparent accusative-ing complements, and analyzes
them as Object+Complement constructions. The forms in 3-35
seem to be gerundive nominal complements, rather than some
such construction.
o
Fasv. and several other adjectives and predicate
nouns which govern TOUGH-MOVEMENT. present a special problem,
The only satisfying use of a derived nominal corresponding
to easy seems to be in a construction like the ease with
which John is pleased, which corresponds to something like
John is pleased with ease/easily. I have no explanation for
this exception to DERIVED N0MINALIZATION.
10
In a grammar which includes the rules of PASSI7IZA-
TIGN and IT-EXTRAPOSITION. those two rules must be ordered


144
The various assumptions I have made about the nature of
underlying structures also leads to other interesting
results. I have shown that an abstract, predicate-initial
analysis of the underlying structure of English eliminates
the need for any rule of extraposition, without adding any
rule to the grammar. Contrary to Chomsky's claim that a
transformational analysis of derived nominal complements
would involve the complication of the transformational com
ponent, my analysis results in a simplification of that com
ponent. My analysis also does without a base component in
the sense of Chomskys theory. As far as they can be com
pared, however, the underlying structures that I assume are
no more complicated than the deep structures of Chomskys
theory, and I do not need any distinct deep structures for
derived nominal complements, as Chomsky does.
One interesting consequence of the analysis of trans
formations that I have proposed here is that no rule which
moves a noun phrase moves it to the right. All transforma
tions acting on noun phrases are leftward movement rules.
This study, of course, can only be viewed as a prelimi
nary one, as there remain many problems in the description
of derived nominal complements, but I feel that I have shown
how this construction can be handled within transformational
theory, and while there may be an irreducible remainder of
exceptions, it is better to capture generalizations than it
is to catalog irregularities, as the Lexical Hypothesis does.


5
derived nominal complement structures parallel to the struc
tures of embedded sentences. Indeed, there seem to be few,
if any, sentential deep structures in Chomsky's system which
do not have corresponding derived nominal complements.
According to Chomsky (1970:190), In early transforma
tional theory, "there was no other way to express the fact
that the contexts in which refuse appears as a verb and
refusal as a noun are closely related" than in terms of
transformational rules, Chomsky further states that when
contextual features were introduced into the theory (in
Chomsky, 1965) it became possible to separate the lexicon
from the categorical component of the base, and thus to
adopt the Lexical Hypothesis. According to Chomsky (1970:
190), refuse is entered in the lexicon with certain features
specified, but with no specification of the categorical fea
tures £nounj and [verbj. He adds that fairly idiosyncratic
morphological rules would, be involved In deriving forms like
refusal in derived nominal complements.
Derived Nominis as Complements on Nouns
The term complement has been used in ways different
enough to make it worth while Indicating here what I mean by
gerundive nominal complement and derived nominal complement.
On the one hand, anything that completes a structure can be
regarded as a complement. Thus, the underlined parts of the
examples In 1,4 are all complements of the verbs in their
respective sentences.


71
3-79a) The enemy destroyed the city.
b) The jury acquitted Abby.
c) John refused the offer.
d) John criticized the book.
3.80a) the enemy's destroying the city
b) the jury's acquitting Abby
c) John's refusing the offer
d) John's criticizing the book
38la) the enemy's destruction of the city
b) the jury's acquittal of Abby
c) John's refusal of the offer
d) John's criticism of the book
If AGENT-PROPOSING does not apply to the structures
underlying the active sentences and nominal complements in
3-79 3-80 and 3*81, then the rule of NP-PREPOSING applies
obligatorily in the derivation of passive sentences and ger
undive nominal complements like those in 3-82 and 3*83-
3.82a) The city vas destroyed by the enemy.
b) Abby was acquitted by the jury.
c) The offer was refused by John.
d) The book was criticized by John.
3.83a) the city's being destroyed by the*enemy
b) Abby's being acquitted by the jury
c) the offers being refused by John
d) the book's being criticized by John
Either AGENT-PREPOSING or NP-PREPOSING must apply in
the derivation of sentences and gerundive nominal comple
ments. Forms like those in 3-84 and 3-85 in which neither
rule has applied, are not acceptable.
3.84a) *Was destroyed the city by the enemy.
b) *Was acquitted Abby by the jury.
c) *Was refused the offer by John.
d) *Was criticized the book by John.
3.85a) being destroyed the city by the ene;ny
b) "'being acquitted Abby by the jury
c) *being refused the offer by John
d) *being criticized the book by John


39
I find the sentences in 2.75 to be formal and even awk
ward in comparison to those in 2,7^ a difference presumably
attributable to the choice of predicate.
All of these sentences have acceptable corresponding
gerundive nominal complements, those in 2.76 corresponding
to the sentences in 2.7^ and those in 2.77 corresponding to
the sentences in 2.75.
2,76a) John's trying vainly to win the race
b) John's leaving hurriedly on the bus
c) there being a god
d) there being a riot yesterday
e) there being some wine in the bottle
f) John's having a car
g) Bill's usually living in a hotel
2.77a) John's vainly attempting to win the race
b) John's departing hurriedly on the bus
c) there existing a god
d) there occurring a riot yesterday
e) there being some wine present in the bottle
f) John's possessing a car
g) Bill's usually residing in a hotel
Again, the gerundive nominal complements in 2.76 based
on Germanic verbs seem less formal than the gerundive nomi-
Q
nal complements based on Latinate verbs in 2.77.
Finally, there are no derived nominal complements cor
responding to the sentences in 2.7^ and gerundive nominal
complements in 2.76, the forms in 2.78 being unacceptable.
2.78a) '"John's vain trying to win the race
b) *John's hurried leaving on the bus
c) *the being of a god
d) *the being of a riot yesterday
e) *the being of some wine in the bottle
f) ^John's having of a car
g) ^Bill's usual living in a hotel


To
Ginny
and
Rebecca


59
3.4la) *John's easiness to please
b) '"'algebra's difficulty to learn
c) "'this test's fun to take
The sentences in 3-39i which have not undergone
TOUGH-MOVEMENT. have the acceptable corresponding gerundive
nominal complements in 3-42.
3.42a) it(s) being easy to please
b) it(s) being difficult to learn algebra
c) it(s) being fun to take this test
The apparently corresponding derived nominal complements
in 3-43 are not acceptable, but those in 3*44 are acceptable.^
The argument used with the examples in 3-36 and 3*37 applies
here as well.
3.43a) *its easiness to please John
b) *its difficulty to learn algebra
c) "its fun to take this test
3-44a) ?the easiness of pleasing John
b) the difficulty of learning algebra
c) the fun of taking this test
RAISING-TO-OBJECT. RAISING-TO-SUBJECT and.TOUGH-MOVE
MENT do not apply in the derivation of derived nominal com
plements. The three rules share the property of moving a
noun phrase to a higher sentence. They appear to apply at
the same point in the cycle.G. Lakoff (1968) has argued
that there is only one rule of RAISING (IT-REPLACEMENT in
his paper). To collapse the three types of RAISING to one,
Lakoff has to write a complex rule, which includes simul
taneous structural descriptions. I consider such a rule to
be unlikely in a natural language. I will, therefore, sus
pend judgment on whether it is possible to collapse the
RAISING rules.


85
Finally, although the sentences in 3*136, and their
expected corresponding gerundive nominal complements in 3.1*0
are all unacceptable, the derived nominal complements in
3,144, with the same linear order as the forms in 3.136 and
3.143, are acceptable.
3,143a) *its being reported by Tom the fact that the
account was overdrawn
b) *its being revealed by Jerry the fact that
Alice was smart
c) *its being denied by Mary the claim that Peggy
was pregnant
3.144a) the report by Tom of the fact that the account
was overdrawn
b) the revelation by Jerry of the fact that
Alice was smart
c) the denial by Mary of the claim that Peggy
was pregnant
By referring to the examples given for simple noun
phrases and nominal complements, the reader should be able
to convince himself that the patterns of acceptability
exhibited for the examples with transitive predicates and
specified agents given immediately above also hold for forms
with transitive predicates and unspecified agents and for
forms with intransitive predicates.
If we understand NP-PREPOSING to apply either to the
head noun alone or to the head noun plus its that-comple
ment, we see that the rule has the same conditions on appli
cation to that-complements that it has to nominal comple
ments j it applies optionally in the derivation of sentences,
and not at all in the derivation of derived nominal comple
ments. Acceptable gerundive nominal complements occur only
if the head noun is deleted.


135
subject is made no longer available for verb-agreement in
embedded sentences: when it takes the preposition for with
an emotive predicate in the higher sentence;^ when
EQUI-NP-DELETION has applied; and when RAISING-TO-SUBJECT or
RAISING-TO-OBJECT has applied. If the subject is not removed
by one of these means, a that-complement results. Pizzini
(1972) reduces this list by having the for associated with
emotive predicates adjoined to the embedded sentence, and
the subject of the embedded sentence then raised to become
the object of for. By this analysis, for remains a proper
constituent of the higher sentence.
Starting with the underlying structures in 4.20, under
this analysis, the application of RAISING-TO-OBJECT to 4.20a,
EQUI-NP-DELETION to 4.20b and the adjunctiontion of for to
the embedded sentence in 4.20c, with subsequent raising of
the embedded subject to become object of for, gives the
intermediate structures in 4.21. In each case the embedded
sentence has lost Its subject, and the predicate cannot
agree with a subject it does not have. This causes the for
mation of an infinitive, and results eventually in the sur
face sentences in 4,19.
4.20a) !"- want John fq join Bill John's fraternity!q !q
-^1 >-^2 s2 1
b) T- expect Jerry j~q win Jerry the prizel-, !q
al 2 a2 1
c) fq anxious for Mary To arrive Mike!-, !-,
bl b2 2 1
4,21a) Tq John want Bill [q join John's fraternity]]-. ]jq
1 2 2 1
b) [e Jerry expect [q win the prize]]q ]q
1 2 &2 1
c) Mary be anxious for Mike arrive!-, !,,
S1 S2 a2 *1


115
stories, his stories (=John's stories) is a reduced form of
stories which John told,. Hence, it would appear that a RAIS
ING rule could move John out of with John's stories (or,
rather, copy it out, leaving the pronoun) and into subject
position. Lee's suggestion would involve extending the list
of predicates governing RAISING- rules to psychological predi
cates. My suggestion would involve postulating a new RAIS
ING rule which leaves a pronoun behind, and would require
independent motivation for the substitution of with for the
predicate to derive phrases like with his stories-


28
derived nominis. Thus* Roberts conversion to hedonism,
which at some point in its derivation has an underlying
structure like that of the embedded sentence S^ in the tree
in 2.42, and Johns conversion of Robert to hedonism, which
at some point in its derivation has an underlying structure
like that of the whole tree in 2.42, are both acceptable
derived nominal complements. The derived nominal complement
the conversion of Robert, like many other nominal complements
such as the shooting: of the hunters, is therefore interpreta
ble as either a subject nominal complement or as an object
nominal complement.
2.42 SQ
V NP NP
convert to Robert hedonism John
Some verbs listed by Smith which also form derived nom
inis in both transitive and intransitive forms are explode.
divide, accelerate. expand. repeat, neutralize. conclude
and unify. She then points out that all the listed counter
examples share a morphological propertys their derived nomi
nis are formed with suffixes of Latin origin. She then
goes on to claim that there are almost as many verbs of the
type convert as there are of the type grow, and then states
what she sees as the conditioning factor in distinguishing
the two types; "Whether or not a verb has a transitive


57
in the derivation of the sentences in 3.31? which are
derived from the same underlying structures as the sentences
in 3.32.
3.3.1a) John is certain to win the prize.
b) Bill is likely to be drafted.
c) Jerry appeared to open the door.
3.32a) It is certain that John will win the prize.
b) It is likely that Bill will be drafted.
c) It appeared that Jerry opened the door.
The sentences in 331s which have undergone
RAISING-TO-SUBJECT, have the acceptable corresponding ger
undive nominal complements in 3.33* tut no acceptable cor
responding derived nominal complements the forms in 3.34
being unacceptable,
3.33a) Johns being certain to win the prize
b) Bills being likely to be drafted
c) Jerry's appearing to open the door
3.34a) ^John's certainty to win the prize
b) ^Bills likelihood to be drafted
c) *Jerrys appearance to open the door'
The sentences in 332, which have not undergone
RAISING-TO-SUBJECT, have the acceptable corresponding ger-
8
undive nominal complements in 3.35.
3.35a) it(s) being certain that John will win the prize
b) it(s) being likely that Bill will be drafted
c) it(s) appearing that Jerry opened the door
The apparently corresponding derived nominal comple
ments in 3.36 are not acceptable, but the derived nominal
complements in 3.37* which have the same word order as the
sentences in 3.32, but do not have any reflex of it, are
acceptable.


83
structure, and the sentences in 3.135 result if the head
noun is deleted.
3.133a)
(The fact) that the account was overdrawn was
b)
reported by Tom.
(The fact) that Alice was smart was revealed
c)
by Jerry.
(The claim) that Peggy was pregnant was denied
by Mary.
3.134a)
The fact was reported by Tom that the account
was overdrawn.
b)
The fact was revealed by Jerry that Alice was
smart.
c)
The claim was denied by Mary that Peggy was
pregnant.
3.135a)
It was reported by Tom that the account was
overdrawn.
b)
c)
It v/as revealed by Jerry that Alice was smart.
It was denied by Mary that Peggy was pregnant.
The sentences in 3.133 are passive. Sentences like
those in 3.134 have been described as being derived by the
rule of EXTRAPOSITION PROP/I NP (of which I will have more to
say below). The sentences in 3.135 are extraposed (i.e.,
supposedly derived by IT-EXTRAPOSITION).
In no case in which head nouns are retained in surface
structure are both the head noun and the that-complement to
the right of the predicate in sentences. Thus, there are no
acceptable sentences like those in 3.136.
3.136a)
*It v/as reported by Tom the fact that the
account was overdrawn.
b)
*It was revealed by Jerry the fact that Alice
was smart.
c)
*'It was denied by Mary the claim that Peggy
was pregnant.
The sentences in 3.124 do not have any acceptable cor
responding gerundive or derived nominal complements, the
forms in 3.137 and 3.138 all being unacceptable.


86
EXTRAPOSITION FROM NP
Ross (1967) proposed a rule, which he called EXTRAPOSI
TION FROM NP, to account for the relationship of sentences
like those in 3*145 to sentences like those in 3*146.
3-l45a) A gun which I had cleaned went off.
b) He let the cats which were meowing out.
c) He expected someone who I was acquainted with
to show up.
3*l46a) A gun went off which I had cleaned.
b) He let the cats out which were meowing.
c) He expected someone to show up who I was
acquainted with.
Ross stated the rule in the form given in 3*147-
Although Ross does not state the rule in the context of the
cycle, the restrictions he gives on the forms of the varia
ble Y indicate that the rule is cyclic.
3.147 EXTRAPOSITION FROM NP
NpO^ S]Np
1 2
Y
OPT -
3 ==> i, 0, 3+2
EXTRAPOSITION FROM NP is not needed to account for the
relationship of the sentences in 3*145 to the sentences in
3-146 if NP-PRSPOSING is stated to apply to such noun-rela
tive clause combinations in the same manner as with that-com-
plements and their head nouns. Thus, NP-PREPOSING would


72
On the other hand, in the derivation of derived nominal
complements, if AGENT-PROPOSING does not apply, then the
application of NP-PROPOSING is optional for some nominis
(or underlying verbs), giving the paired acceptable derived
nominal complements in 3-86a and b and 3* 87a and b, and
blocked for other nominis, so that the derived nominal com
plements in 3-37c and d are acceptable, but not those in
3.86c and d.
3.86a) the citys destruction by the enemy
b) Abby's acquittal by the jury
c) *the offer's refusal by John
d) *the book's criticism by John
3.87a) the destruction of the city by the enemy
b) the acquittal of Abby by the jury
c) the refusal of the offer by John
d) the criticism of the book by John
If there is no specified agent noun phrase in an under
lying structure,
then NP-PREPOSING applies obligatorily in
the derivation of sentences, as in 3*83 and gerundive nominal
complements, as in 3-89, with the forms in 3-90 and 3-91
being unacceptable.
3.88a) The city was destroyed.
b) Abby was acquitted.
c) The offer was refused.
d) The book was criticized.
3.89a) the city's being destroyed
b) Abby's being acquitted
c) the offer's being refused
d) the book's being criticized
3.90a) *Was destroyed the city.
b) *Was acquitted Abby.
c) *Was refused the offer.
d) *Was criticized the book
3-91a) *being destroyed the city
b) *being acquitted Abby
c) *being refused the offer
d) *being criticized the book


2.63a)
b)
34
the Bismark's sinking
the British Navy's sinking the Bismark
2.64a) the Bismark's sinking
b)the British Navy's sinking of the Bismark
2.65a) the sinking of the Bismark
b) the sinking of the Bismark by the British Navy
Presumably, there is no specification of the form of
the nominal in a derived nominal complement corresponding to
sink in the lexical entry, so that the form with -ing is sup
plied by the gra.mmar for sink intransitive and transitive
just as it is supplied for the nominis in derived nominal
complements corresponding to sentences with the derived
verbs grow, raise, move, dim, lower and widen. One conse
quence of this is that it is possible for the gerundive nomi
nal complement and derived nominal complement corresponding
to a particular sentence to be identical in surface struc
ture, as are the forms in 2.63a and 2,64a. It jvill there
fore not always be possible to decide whether a nominal com
plement is actually a gerundive nominal complement or a
derived nominal complement, unless the context supplies such
information.
One objection which might be made to the identification
of any nominal complements with nominis formed with -ing as
derived nominal complements is that they do not readily form
plurals, However, it is not true that they never form plur
als, as shown by the examples in 2,66.
2,66a) the wanderings of an old mind
b) the leavings of a great feast
c) the makings of a great scholar
d) the cravings of a glutton
e) the comings and goings of the workers


23
2,30a) John's proofs of the theorem
b) John's attempts to scale the wall
c) Agnew's attacks on the press
Not all derived nominis can be pluralizad. The sen
tences in 2,31 have the corresponding derived nominal comple
ments with singular derived nominis in 2,32, but derived
nominal complements with plural derived nominis like those
in 2.33 corresponding to the sentences in 2.31 do not occur.
2,31a) 'The enemy has destroyed the city on three
occassions,
b) The crowd laughed repeatedly.
2.32a) the enemy's destruction of the city on three
occassions
b) the crowd's repeated laughter
2.33a) *the enemy's destructions of the city on three
occassions
b) *the crowd's repeated laughters
Auxiliaries
Another characteristic of the structure of derived nom
inal .complements discussed by Chomsky (1970:189) is the
absence of any auxiliary verbs, Gerundive nominal comple
ments, on the other hand, can have any auxiliary (with the
exception exemplified by 2.35c) except modals. Perfective
aspect can appear in gerundive nominal complements, so that
sentences like those In 2.3^a and b have the corresponding
gerundive nominal complements in 2.35a and b. Progressive
aspect can also appear in gerundive nominal complements, but
only in conjunction with perfective aspect, as in 2.35b, so
that the sentence in 23^'C does not have a corresponding
gerundive nominal complement with being corresponding to a
form of be which is acceptable (cf. 2,35c and d),-^


118
I will not accept Newmeyer's claim as it stands, but will
present arguments below that DERIVED NGMINALIZATION is
cyclic, and ordered very early in the cycle, although not
necessarily first in the cycle.
Newmeyer's proposal that the nominalization of the
underlying predicate blocks the later application of rules
mentioning predicate in their structural descriptions
appears to be well motivated. In Chapter 'Three I showed
that there are no acceptable derived nominal complements
corresponding to sentences which have undergone one of the
RAISING rules, IT-INSERTION, THERE-INSERTION. DATIVE-MOVE
MENT, and/or PSYCH-CHANGE in their derivations.
Each of the RAISING rules raises a noun phrase into a
position defined by its relationship to the predicate of the
higher sentence. In RAISING-TO-QBJECT, the noun phrase is
raised into a position immediately to the right of the pre
dicate of the higher sentence. In RAISING-TO-SUBJECT (and
the similar rules governed by psychological predicates) and
TOUGH-MOVEMENT, the noun phrase is raised into a position
immediately to the left of the predicate of the higher
sentence.
IT-INSERTION and THERE-INSERTION introduce, in appro
priate circumstances, a semantically empty word as the sub
ject of the predicate. That is, these rules insert it or
there Immediately to the left of the predicate.
DATIVE-MOVEMENT involves a switch in position between
direct and indirect objects, and thus has the effect of


42
Agentive Nominis
Agentive nominal complements share the characteristics
of derived nominal complements. The examples in 2.81, cor
responding to the sentences in 2 82, show that such comple
ments may have determiners other than possessive nouns, pre-
nominal adjectives, object prepositions and pluralization.
Agentive nominal complements do not have any reflexes of
auxiliaries.
2.8la) the short-sighted designers of this building
b) the unlucky holders of Imperial Russian bonds
c) the greedy despoilers of the .Earth
2,82a) The ones who designed this building were
short-sighted.
b) The ones who held Imperial Russian bonds
were unlucky.
c) The ones who despoil the Earth are greedy.
Some characteristics of agentive nominal complements
deserve further comment. Agentive nominal complements never
have possessive nouns as determiners which correspond to
underlying agents. When agentive nominal complements do
have possessive nouns as determiners, as in 2.83, the posses
sive nouns correspond to the underlying objects of the verbs
in sentences which correspond to the agentive nominal comple
ments, as in 2.84.
2.83a) Mike's helper
b) General Electric's workers
c) America's educators
2.84a) Someone helps Mike.
b) Some people work for General Electric.
c) Some people educate America.


105
isation of the second occurrence of John, and something like
CAUSATIVE FORMATION or PREDICATE RAISING to incorporate
21
CAUSE into amuse.
The derivation of either the sentence in 3-193a or the
one in 3>190 from the structure in 3-198 can be explained by
allowing John to be optionally moved out of the embedded sen
tence to subject position of the next higher sentence in the
case of 3-193a* with the alternative being to move the
reduced form of the NP dominating 3^, John's stories, into
subject position.
The sentence in 3*199 is repeated in the set of sen
tences in 3*200s which have the corresponding passive sen
tences in 3*201. The sentences in 3*202 which resemble the
passive sentences in 3*201, but have some preposition other
than by preceding the Agent phrase, are found only with verbs
belonging to the set of psychological predicates.
3.200a) John's stories amused the children.
b) The children's antics amused John.
c) The storm frightened Mary.
d) The ceremony pleased 3ob.
3>201a) The children were amused by John's stories.
b) John was amused by the children's antics.
c) Mary v/as frightened by the storm.
d) Bob was pleased by the ceremony.
3*202a) The children were amused at John's stories.
b) John was amused at the children's antics.
c) Mary v/as frightened of the storm.
d) Bob v/as pleased with the ceremony.
The sentences in 3*200, 3*201 and 3*202 have the respec
tive corresponding gerundive nominal complements in 3*203,
3*204 and 3-205*


19
2.l4a) John's eagerness to please is troublesome,
b) John's refusal of the offer was untimely.
c) John's criticism of the book was unmotivated.
The prenominal adjectives in the derived nominal com
plements in 2.9 can also be predicated of those complements,
as in 2,15.
2.15a) John's eagerness to please is overwhelming.
b) John's refusal of the offer was abrupt.
c) John's criticism of the book was unmerciful.
The parallel I have been drawing between the adverbs in
sentences and gerundive nominal complements and the adjec
tives in derived nominal complements suggests that the ad
verbs and adjectives are derivationally related. The fact
that the adverbs are morphologically derived from the adjec
tives by adding the suffix -ly reinforces that hypothesis.
The three-way correspondence between prenominal adjec
tives in derived nominal complements, adjectives predicated
of derived nominal complements, and adverbs in sentences and
gerundive nominal complements suggests that the adjectives
and adverbs have as a common source a higher predicate. If
the higher predicate is the highest matrix predicate (ignor
ing abstract higher predicates) then it may be expressed as
an adverb of the embedded sentence (raised to surface sen-
tencehood) or as a predicate on a derived nominal complement.
If the higher predicate is in turn embedded under a predi
cate which will appear in surface structure, then it may be
expressed as an adverb with a gerundive nominal complement,
or as a prenominal adjective with a derived nominal comple
ment. The higher predicate-prenominal adjective relation-


63
Chomsky's claim that derived nominal complements corres
pond only to base phrase markers creates a problem here if a
predicate-initial analysis is not adopted. Not only are
there derived nominal complements corresponding to active
sentences, as in 3-70, there are also derived nominal comple
ments corresponding to some, but not all, passive sentences,
as in 3-73t and there are derived nominal complements which
do not correspond to any sentences at all (what might be
called "half-passives"), as in 3*7^-
Chomsky (1970202ff.) proposes two transformations
which must both apply in the derivation of passive sentences
from phrase markers parallel to those underlying active sen
tences, one of which also can apply to noun phrases (includ
ing derived nominal complements). Chomsky proposes a rule
of NP-PREPOSING which will transform phrases like those in
3-75 into phrases like those in 3-76. This rul'e thus applies
to noun phrases whether they have a simple noun or a derived
nominal as their head.
3<75a) the picture of John
b) the bottom of the barrel
c) the destruction of the city
d) the murder of John
3.76a) John's picture
b) the barrel's bottom (the literal meaning)
c) the city's destruction
d) John's murder
There are similar phrases which do not undergo the rule
of NP-PREPOSING, and again the phrases can have either a sim
ple noun or a derived nominal as their head. The phrases in
3-77 do not have acceptable corresponding forms like those
in 3-78.


io6
3203a) John's stories amusing the children
the children's antics amusing John
o)
c)
d)
20 4a)
b)
c)
3205a)
b)
c)
d)
the storms frightening Mary
the ceremony's pleasing 3ob
the children's being amused by John's stories
John's being amused by the children's antics
Mary's being frightened by the storm
d) Bob's being pleased by the ceremony
the children's being amused at John's stories
John's being amused at the children's antics
Mary's being frightened of the storm
Bob's being pleased with the ceremony
While there are the derived nominal complements in
208 corresoonding to the sentences in 3-202 which are
acceptable, the forms in 3-206 and 3-20?, which would appear
to be the expected derived nominal complements corresponding
to the sentences in 3*200 and 3-201 respectively, are not
acceptable,
3-206a)
b)
c)
d)
3.207a)
b)
c)
d)
3208a)
b)
c)
d)
*John's stories' amusement of the children
"'"the childrens antics' amusement of John
*the storm's fright of Mary
*the ceremony's pleasure of Bob
Hu-
*the children's amusement by John's stories
*John's amusement by the childrens antics
*Ivlary's fright by the storm
*3ob's pleasure by the storm
the children's amusement at John's stories
John's amusement at the childrens antics
Mary's fright of the storm
Bob's pleasure with the ceremony
Only the pseudopassive sentences in 3-202 have accepta
ble corresponding derived nominal complements. There are no
derived nominal complements based on psychological predi
cates which show nominal-initial order such as we have found
associated with other types of predicates. The forms in
3-209 are not acceptable.


21
Object Prepositions
Another characteristic which.distinguishes derived nom
inal complements from sentences and gerundive nominal com
plements is the presence of a preposition preceding the noun
phrases in derived nominal complements which correspond to
the objects of verbs in sentences, if such object noun
phrases do not already have a preposition. Among the sen
tences in 2.19, corresponding gerundive nominal complements
in 2.20 and corresponding derived nominal complements in
2.21, only the derived nominal complements have the prepos-
3
ition of preceding the object.
2.19a) John refused the offer.
b) John criticized the book.
c) John robbed the bank.
2.20a) Johns refusing the offer
h) John's criticizing the book
c) John's robbing the bank
2.21a) Johns refusal of the offer
b) John's criticism of the book
c) John's robbery of the bank
When object noun phrases are preceded by prepositions
in sentences such as those in 2.22 and in the corresponding
gerundive nominal complements in 2.23 then the same pre
positions also appear in the corresponding derived nominal
complements, as in 2.24.
2.22a) John was amused at the children's antics.
b) John delighted in teasing Alice.
c) John was doubtful about Dick's honesty.
2.23a) John's being amused at the childrens antics
b) John's delighting in teasing Alice
c) John's being doubtful about Dick's honesty
2.24a) John's amusement at the children's antics
b) John's delight in teasing Alice
c) John's doubts about Dick's honesty


79
3-lla) -'being complained, about Peter's (action of)
insulting Tom
b) *being attested to (the fact of) Mary's parti
cipation in the crime
c) *being reported on (the event of) Alice's
disappearance
3-117a) the complaint about Peter's (action of)
insulting Tom
b) the attestation to (the fact of) Mary's
participation in the crime
c) the report on (the event of) Alice's dis
appearance
Again while the head nouns of the embedded nominal com
plements in the examples above may be deleted, they may not
be separated from the complements, so that all the examples
in 3*118, 3*119 and 3*120 are unacceptable.
3*ll8a) ^Peter's action was complained about of
insulting Tom.
b) *fhe fact was attested to of Mary's partici
pation in the crime.
c) *The event was reported on of Alice's dis
appearance .
3* 119a) '^Peter's action's being complained about of
insulting Tom
b) "the fact's being attested to of Mary's parti
cipation in the crime
c) *the event's being reported on of Alices
disappearance
3.120a) "Peter's action's complaint about insulting
Tom
b) *the fact's attestation to Mary's participa
tion in the crime
c) '"the event's report on Alice's disappearance
If the underlying predicate-initial structure contains
an intransitive predicate, then NP-PROPOSING applies obliga
torily in the derivation of sentences, as in 3*121, but is
blocked from applying in the derivation of nominal comple
ments, so that the forms in 3*122 and 3*123 are unacceptable.


NOTES
~ Chomsky (196569) defines 'subject-of' as, "for Eng
lish... the relation holding between the NP of a sentence of
the form NP-Aux-VP and the whole sentence," and 'object-of*
as "the relation between the NP of a VP of the form V-NP and
the whole VP."
^ AGENT-PREPOSING moves the Agent into a position imme
diately to the left of the predicate.
J The derivation of 4.6c from 4.7 c involves the raising
cf Bill from subject of Sg to object of watch in S^, the pro-
ncminalization and reflexivization of Bill object of watch
under identity with Bill subject of S^, and, finally, the
deletion of Bill subject of under identity with Bill
Agent of the highest sentence.
4
Actually, that-complementation does not share this
characteristic of COMPLEMENTIZER PLACEMENT, since it involves
the insertion of that as a proper constituent of the higher
sentence adjoined to the embedded sentence.
3 The subjunctive mood also requires the insertion of for.
^ Ross (1972c) calls forms like Jane nutting on a
girdle (in 4.22a) accusative-ing complements. However,
Fraser (1970) correctly pointed out that such forms are not
complements, but rather a noun phrase (Jane) plus a comple-
ment (putting on a girdle) .
142


108
Conclusion
In this chapter I first argued that sentences which
have undergone any of the RAISING rules do not have corres
ponding derived nominal complements. I then showed that
sentences which have undergone IT-EXTRAPOSITION in their
derivations have corresponding derived nominal complements,
while sentences which met the structural description of
IT-EXTRAPOSITION, but to which the rule did not apply in
their derivation, do not have corresponding derived nominal
complements. I then showed that while all simple active
sentences have corresponding derived nominal complements,
only some passive sentences have corresponding derived
nominal complements, and that for every pair of active and
passive sentences there is a derived nominal complement to
which only the AGENT-POSTPOSING part of Passivization has
applied. I then argued that with a predicate-initial ana
lysis both extraposed and nonextraposed and active and pas
sive sentences can be related, at least in part, by the
same rule, NP-PREPOSING. I have also discussed certain
interesting aspects of NP-PREP03ING, and concluded that the
application of the rule in the derivation of derived nominal
complements is governed by a set of conditions dependent on
different aspects of the environment. Finally, I have
argued that ^-INSERTION, THERE-INSERTION and PSYCH-CHANGE
do not apply in the derivation of derived nominal comple
ments, and that the application of EQUI-NP-DELETION is
blocked in derived nominal complements for certain underly
ing verbs.


91
The application of NP-PREPOSING in this way is not pos
sible in sentences, so that we find the sentences in 3-15^
but the sentences in 3*1.55 are not acceptable.
3*15^a) The librarian discussed novels last week.
b) The president spoke to the nation this morning.
3*155a) *Last week was discussed novels by the
librarian.
b) *This morning was spoken to the nation by the
president
But it cannot be the POSSESSIVE Transformation which
has moved the prepositionless time adverbials to a position
preceding the derived nominis in 3*152, since the forms in
3.155 are not acceptable, indicating that the POSSESSIVE
Transformation (or NP-PREPOSING) is blocked for discussion
and speech in derived nominal complements.
3*155a) ^novel's discussion by the librarian last week
b) *the nations speech by the president this
morning
All time adverbials are subject to fronting in sen
tences, as in 3*157*
3*15?a) Last week the librarian discussed novels.
b) This morning the president spoke to the nation.
The application of a rule of ADVERB-FRONTING to the
structures underlying the derived nominal complements in
3*153 (with subsequent addition of the suffix *.s) will pro
duce the derived nominal complements in 3*152.
Other time adverbials are also subject to ADVERB-FRONT
ING, but do not form determiners of derived nominis. The
time adverbials with prepositions, however, are subject to
deletion of the preposition, and then behave just like other


CHAPTER THREE
THE PRODUCTIVITY OF DERIVED NOMINAL COMPLEMENTS
The Evidence for Lack of Productivity
One of the three reasons Chomsky (19?0:187-88) gives
for rejecting a transformational account of the origin of
derived nominal complements in favor of the Lexical Hypothe
sis is that gerundive nominal complements "can be formed
fairly freely from propositions of subject-predicate form,"
while "productivity is much more restricted" for derived
nominal complements. In support of this statement, Chomsky
(1970:188-89) refers to the example of sentences like those
in 3-1 which have the corresponding gerundive nominal com
plements in 3.2, and the corresponding derived'nominal com
plements in 3-3 while the sentences in 3* 4 have the corres
ponding gerundive nominal complements in 35 but the expec
ted corresponding derived nominal complements in 3*6 are not
acceptable.
3.1a) John is eager to please.
b) John refused the offer.
c) John criticized the book.
3.2a) John's being eager to please
b) John's refusing the offer
c) John's criticizing the book
33a) John's eagerness to please
b) Johns refusal of the offer
c) John's criticism of the book
48


49
3.4a) John is easy to please.
b) John is certain to win the prize.
c) John amused the children with his stories.
3.5a) John's being easy to please
b) Johns being certain to win the prize
c) John's amusing the children with his stories
3.6a) ^John's easiness to please
b) *John's certainty to win the prize
c) *John's amusement of the children with his stories
Chomsky (1970:189) points out that there are acceptable
derived nominal complements like those in 3-7 which superfi
cially resemble the unacceptable strings in 3-6, and which
correspond to the sentences in 38 and gerundive nominal com
plements in 3-9, and comments, "these discrepancies between
gerundive and derived nominal [complements] call for expla
nation. Specifically, we must determine why the examples of
[3.6] are ruled out although those of [3*7] are permitted."
3.7a) John's eagerness to please
b) John's certainty that Bill will win the prize
c) John's amusement at the children's^antics
3-8a) John is eager to please.
b) John is certain that Bill will win the prize.
c) John was amused at the children's antics.
3.9a) John's being eager to please
b) John's being certain that Bill will win the prize
c) John's being amused at the children's antics
In this chapter I will try to answer Chomsky's question
by showing that sentences like those in 3-^ which do not
have acceptable corresponding derived nominal complements,
have undergone certain transformational rules in their deri
vations which sentences like those in 3*8, which do have
acceptable corresponding derived nominal complements, have
not undergone. Chomsky (1970:191) claims that this is so


8
The head noun action never takes a derived nominal com
plement, no matter where the possessive agent is placed, or
even if it is deleted. The examples in 1.9 are all unaccep
table, even though the parallel forms in 1.10 are perfectly
acceptable.
l-9a) *The action of Humphrey's refusal of the offer
caused trouble.
b) ^Humphrey's action of refusal of the offer caused
' trouble.
c) *The action of refusal of the offer caused
trouble.
1.10a) The fact of Humphrey's refusal of the offer
caused trouble.
b) Humphrey's action of refusing the offer caused
trouble.
c) The action of refusing the offer caused trouble.
Menzel (1969851) notes that gerundive nominal comple
ments can also be complements of act. He states that acts
are a subclass of actions, with the restriction that "an act
is an action which the speaker either (a) disapproves of
strongly (in a legal or moral sense); or (b) admires greatly."
He also notes that this distinction does not seem to be
recognised by some speakers.
While derived nominal complements do not occur as com
plements of action, a restricted class does occur as comple
ments of act. The head noun act does not take derived nomi
nal complements with object prepositional phrases. The
examples in 1.11 illustrate the differences between gerund
ive nominal complements and derived nominal complements in


25
variety of determiners; prenominal adjectives instead of
adverbs; pluralization of the nominal; prepositions preced
ing object noun phrases when they do not appear in the cor
responding sentences; and complete absence of auxiliaries,
including copulas and the passive be. The form of the nom
inal in derived nominal complements does not enter into
this definition.
Action Nominis
Chomsky's Analysis
There are nominal complements which satisfy the new
definition of derived nominal complements given above, but
which have what appear to be gerundive nominis correspond
ing to the verb, such as the nominal complements in 2.39-
Nominal complements of this type are called "action nomi
nis in Lees (1963) and Fraser (1970).
2.39a) John's refusing of the offer
b) John's proving of the theorem
c) the growing of tomatoes
Chomsky (1970s214-15) believes that these complements
belong to a third class of nominal complements distinct from
both gerundive nominal complements and derived nominal com
plements. He does not identify them as "action nominis."
Chomsky claims that these nominal complements appear to have
the same internal structure of noun phrases that derived nom
inal complements have, as evidenced by the possibility of a
determiner other than a possessive noun appearing (cf. 2.39c),
but that prenominal adjectives seem quite unnatural in such
complements. He says that the complement is limited in


24
2.34a) John has criticized the book.
b) John had been criticizing the book.
c) John is criticizing the book.
2.35a) Johns having criticized the book
b) John's having been criticizing the book
c) ^John's being criticizing the book
d) John's criticizing the book
Forms of the verb be which appear as copulas in sen
tences with predicate adjectives or predicate nouns, or in
combination with past/passive participles in passive sen
tences, such as those in 2.36, also appear in gerundive nom
inal complements such as those in 2=37* But the derived
nominal complements in 2.38 which correspond to the sen
tences in 2.36 do not have any form of be.
2.36a) John is strong.
b) Alice is beautiful.
c) John is the chairman.
d) The city was destroyed by the enemy.
e) Abby was acquitted by the jury.
2.37a) John's being strong
b) Alice's being beautiful
c) John's being the chairman
d) the citys being destroyed by the enemy
e) Abby's being acquitted by the jury
2.38a) Johns strength
b) Alice's beauty ^
c) John's chairmanship0
d) the city's destruction by the enemy
e) Abby's acquittal by the jury
A Redefinition
I would now like to abandon my first provisional defi
nition of derived nominal complements as those complements
with nominis formed without an -inn suffix, and instead
define derived nominal complements as those complements
which show one or more of the following features: a wide


95
complements corresponding to the sentences in 3.166 are the
simple noun phrases in 3.163.
3.166a) There was no hope.
b) There was a disturbance,
c) There was some wine in the bottle.
3.167a) there being no hope
b) there being a disturbance
c) there being some wine in the bottle
3.168a) no hope
b) a disturbance
c) some wine in the bottle
The sentences in 3.166 all have the surface structure
there-Copula-Predicate Nominal. THERE-INSERTION is presumed
to have applied in each sentence because no noun phrase was
left in subject position. In an analysis assuming underly
ing SVO order, this requires the postposing of the subject
with certain intransitive verbs. In an analysis assuming
predicate-initial order, THERE-INSERTION applies in case no
other rule has moved a noun phrase into subject" position.
In the section on IT-EXTRAPOSITION above I indicated
that ft is inserted in subject position if no noun phrase or
complement is moved into subject position. I will claim
here that ft is the surrogate subject if the predicate has a
sentential complement, or if the predicate is one which
takes no arguments (i.e., weather verbs such as rain, etc.),
and that there is the surrogate subject if the predicate
has at least one argument, but no sentential complement.
The derived nominal complements I have given in 3.168
are really just the noun phrases v/hich occur as predicate
nominis in the sentences in 3*166. There are variants of


69
377a) the algebra of revolution
b) the strategy of war
c) the refusal of the offer
d) the criticism of the book
3-78a) ^revolution's algebra
b) *war's strategy
c) *the offer's refusal
d) *the book's criticism
Chomsky proposed a rule of AGENT-PQSTPOSING to move the
subject noun phrase to a post-verbal position in phrase mark
ers which are to become passive sentences. NP-PREPOSING
would then apply to move the object noun phrase into
pre-verbal position. Chomsky goes on to say that since pas-
sivizability is a property of verbs (i.e., is a governed
process), then derived nominal complements containing nomi
nis corresponding to such verbs can also be passivized. It
would seem that the passivizability of a verb is best expressed
by having AGENT-P0STP03ING apply optionally to structures
with verbs which can appear in passive sentences. That is,
verbs v/hich can appear in passive sentences are marked as
allowing the optional application of AGENT-P0STP0SING, while
verbs which cannot appear in passive sentences are marked as
not allowing the application of AGENT-POSTPOSING. On the
other hand, NP-PREPQSING applies obligatorily in the deriva
tion of any sentence in which AGENT-PQ3TP0SIKG has already
applied, and optionally in the derivation of any noun phrase
(including derived nominal complements) whose head noun (or
nominal) is marked as allowing the rule. A given lexical
entry may therefore be marked differently for AGENT-POST-
POSING and for NP-PREPOSING.


4
simplification of the transformational component. By con
trast, he gives the transformationalist position as the
choice of simplifying the base structure and extending the
transformational apparatus. Chomsky states that the choice
between the two positions is entirely an empirical one.
Chomsky's 1970 study is a presentation of arguments for the
validity of the lexicalist position. I will consider those
arguments at appropriate points below, but for now I will
give a summary of the Lexical Hypothesis as presented by
Chomsky,
The theory within which Chomsky is making his claims
about derived nominis pictures syntax as being strictly
divided between a base component and a transformational
component (the "Aspects81 models cf. Katz and Postal, 1964,
and Chomsky, 1965). The point of connection between the
two components, when all the base (phrase-structure) rules
have applied, but before any transformational rules have
applied, is called deep structure. A third component is
the lexicon, which supplies lexical items to be inserted
in, and only in, deep structure.
The Lexical Hypothesis involves the claim that derived
nominal complements are produced directly by a subset of the
base rules, that derived nominis are supplied directly from
the lexicon in deep structure, and that derived nominal com
plements are not subject to most transformational rules.
Chomsky (1970:195) notes that such a claim implies that
phrase structure rules must introduce an extensive range of


Copvright
197^
By
Donald Herbert Albury


127
NOMINALIZATION. This would mean that derived nominal comple
ments would function as noun phrases for most of the cycle,
while gerundive nominal complements would function as sen
tences for most of the cycle. One of Chomskys (1970) argu
ments for denying an origin as underlying embedded sentences
to derived nominal complements is that they have a noun
phrase-like internal structure, while gerundive nominal com
plements have a sentence-like internal structure. Having
derived nominal complements function as noun phrases through
most of the cycle while gerundive nominal complements do not
would account for the observed differences. Before pursuing
this proposal further, I must turn to a discussion of the
nature of the cycle.
The Cycle g.nd Complementation
COMFLEMENTIZ£R PLACEMENT
An unstated assumption underlying the discussion in the
previous section is that DERIVED NQMINALIZATION (and GERUND
IVE NGMINALIZATION) apply in the same cycle as the other
rules discussed in connection with the examples presented in
Chapter Three. This means that in the derivation of the sen
tences in 4.9 from the underlying structure in 4.10, DERIVED
NOhIMALIZATION (in the case of 4.9a) and GERUNDIVE NOMINALI-
2ATION (in the case of 4.9b) apply to the sentence dominated
by the node S^. Rules which refer to -predicate in their
structural description are blocked if the nominalization
operation in 4.11 has been performed preceding the rule
within the same cycle.


64
With Psychological Predicates
IT-EXTRAPOSITION can also apply to sentences like those
)
in 3*6l. These are active sentences with complements as
subjects. The verbs which permit this construction share a
number of other characteristics, and will be called psycho
logical predicates here (cf. Postal, 197109-54, and the
section Psychological Predicates, below). The sentences in
3.61 are subject to IT-EXTRAPOSITION in Rosenbaum's analysis,
yielding the sentences in 3*62.
3-6la) That John is here surprises me.
b) For Mary to be so late worries Bill.
c) For Jim to leave now would disturb Alice.
3.62a) It surprises me that John is here.
b) It worries Bill for Mary to be so late.
c) It would disturb Alice for Bill to leave now.
Neither the sentences in 3-61 nor the sentences in 3*62
have acceptable corresponding derived nominal complements,
as shown by the unacceptable forms in 3*63 and ^3*64, and
there are no acceptable derived nominal complements with
determiners which are not possessive nouns corresponding to
the extraposed sentences in 36l, as is shown by the unac
ceptable examples in 3-65-
3.63a) that John is here's surprise to me
b) *for Mary to be so late's worry to Bill
c) "'for Jim to leave now's disturbance of Alice
3.64a) *its surprise to me that John is here
b) "its worry to bill for Mary to be so late
c) -its disturbance of Alice for Bill to leave now
3.65a) *the surprise to me that John is here
b) *the worry to Bill for Mary to be so late
c) "the disturbance of Alice for Bill to leave now


It is usually possible to get an agentless passive sen
tence closely corresponding to such agentive nominal comple
ments, as in 2,85a and c, although 2.85b seems odd,
2,85a) Mike is helped,
b)??General Electric is worked for.
c) America is educated. (Not the stative reading.)
However, the sentences in 2,84 (and 2.85) are not the
only ones corresponding to the agentive nominal complements
in 2.83, For instance, on a different reading, 2.83c, Amer
icas educators, corresponds to the educators of America.
The absence of possessive nouns corresponding to under
lying agents in agentive nominal complements is due to the
fact that sentences corresponding to agentive nominal comple
ments always have nonspecified agents. Other derived nomi
nal complements'may also correspond to sentences with nonspe
cific agents, as in the refusal of the offer, which, like
the agentless passive sentence, the offer was refused, cor
responds to an underlying structure of the form SOMEONE
refused the offer.
The prenominal adjectives in agentive nominal comple
ments are not always related to adverbs in corresponding
sentences, Vendler (1968) points out that beautiful in the
beantifu1 dancer may correspond in meaning to beautiful in
the dancer is beautiful, or to beautifully in SOMEONE dances
be autifully,
The nominis in agentive nominal complements also seem
to more readily form compounds with their objects than do
the nominis of other derived nominal complements, Thus, we


36
Chomsky (1970s215) notes that derived nominal comple
ments with nominis formed with the -ing suffix seem rather
clumsy when a derived nominal complement with a nominal
formed other than with the -ing suffix also exists. It is
also these nominal complements which do not readily take a
prenominal adjective. Since I have claimed above that the
suffix -ing in general appears in derived nominal comple
ments when no other means of forming the nominal is speci
fied j it seems possible that pairs of nominis like refusal
refusing and proof:proving represent a misapplication of the
rule supplying the -ing suffix, and the full regularity of
derived nominal complements with forms like refusing and
proving is blocked by the existance of forms like refusal
and proof.
Some Apparent Counterexamples
Chomsky (1970:214) also noted that derived" nominal com
plements with -ing seem to be limited in production since we
do not get forms like those in 2.40, repeated here as 2.71=
2.71a) *the feeling sad
b) *the trying to win
c) *the arguing about money
d) *the leaving
The forms in 2,71 would presumably correspond to the
verb phrases in 2.72.
2.72a) feel sad
b) try to win
c) argue about money
d) leave
Chomsky (1970:186) argues that sentences like 2.73a
have a structure of Noun Phrase-Verb-Predicate parallel to


10?
3.209a) *the amusement of the children at Johns stories
b) '*the amusement of John at the children's antics
c) *the fright of Mary of the storm
d) *tne pleasure of Bob with the ceremony
In Chapter Two I noted that eagerness does not occur
in derived nominal complements with nominal-initial order,
so that *the eagerness of John to please is unacceptable.
The adjective eager describes a mental state, and belongs
to the class of emotive predicates, of which psychological
predicates are a part. The lack of nominal-initial derived
nominal complements seems to be characteristic of emotive
predicates.
Following Chomsky's claim that derived nominal comple
ments correspond to base phrase markers, the sentences in
3.193a, 3*200 and 3.201 do not have acceptable correspond
ing derived nominal complements because certain rules have
applied in their derivations which are not allowed to apply
in the derivation of derived nominal complements. I have
indicated that one rule has applied in the derivation of
3.193a which has not applied in the derivations of 3.202,
and which appears to be related to the RAISING rules,
It is not clear v/hat rules have applied in the deriva
tions of the sentences in 3.200 and 3.201 which have not
applied in the derivations of the sentences in 3,202. I
will assume an ad hoc rule of PSYCH-CHANGE as a way of refer
ring in the next chapter to the pattern of acceptability
of derived nominal complements associated with psycholo
gical predicates.


6
1.4a) John arrived yesterday.
b) I want to leave now.
c) Mike totaled his car.
On the other hand, complement has been used to refer to
nominalized (or otherwise reduced) embedded sentences such
as those cited in 1.2 and 1.3* By nominal complement I will
refer to nominalized embedded sentences which are comple
ments of nouns in noun phrases.
Peter Rosenbaum (1967) argued that the type of comple
ment structure he called Noun Phrase Complementation is a
noun phrase in deep structure, consisting of the pronoun it
as the head noun, and an embedded sentence as a complement
of it. Rosenbaum contrasted Noun Phrase Complementation
with Verb Phrase Complementation, in which an embedded sen
tence is a complement of a verb in deep structure.
Rosenbaum's claim that the pronoun it is present in
noun phrase complements in deep structure now appears to be
wrong. I will present arguments against that claim in the
section on IT-EXTRAPOSITION in Chapter Three.
Peter Menzel (1969) has argued that gerundive nominal
complements (and, less explicitly, derived nominal comple
ments) are complements of one of a certain set of deletable
head nouns in underlying structure. He points out (pp. 77-
3l) that verbs which take gerundive nominal complements,
with a few exceptions such as verbs of belief and say,
declare and claim, also allow the construction Noun + Com
plement, or just the noun as object. The nouns which can
appear in such constructions (i.e., fact, -pro-position, event.


89
Emonds* POSSESSIVE Transformation
Emonds (1969:78-81) argues that NP-PROPOSING does not
apply in the derivation of "passive" derived nominal comple
ments such as those in 3.149,
3-149a) the citys destruction by the enemy
b) Abbys acquittal by the jury
Ke argues that the possessive noun phrases the city's
and Abbys are preposed by a rule he calls the POSSESSIVE
Transformation. First he notes that NP-PREPOSING may pre
pose a noun phrase over a verb-particle predicate, as in
3.150, but claims that the POSSESSIVE Transformation never
does, so that forms like that in 3*151 are blocked.
3.150) The strike was refered to briefly in the report.
3.151) *the strike's brief reference to in the report
Since many derived nominis idiosyncratically block
preposing of object noun phrases, the fact that verbs with
lexical prepositions also block preposing is weak evidence
for the separate existence of the POSSESSIVE Transformation.
A second reason Emonds gives for requiring as a sepa
rate rule the POSSESSIVE Transformation is that NP-PREPOSING
is associated with the occurrence of the passive morpheme
be-en, which never appears in "passive" derived nominal com
plements. As was mentioned in Chapter Two, derived nominal
complements never have any auxiliaries. All that Is needed
is a simple rule or principle that insures that no auxiliar
ies are present in the surface structures of derived nominal


VI11
Frederick Newmeyer's proposal that such rules fail to apply
because DERIVED NOMINALIZATION applies before they do, and
noninalizes the predicate which is a part of the structural
description of those rules, is adopted. It is 'then argued
that the sane rule-ordering arguments that account for the
restrictions on productivity of derived nominis will also
account for the noun phrase-like internal structure of
derived nominis. It Is further argued that the predi
cate-initial analysis eliminates the need for any rule of
extraposition in English, that DERIVED NOMINALIZATION is a
cyclic rule, and that the formation of complements from
underlying embedded sentences in general is a process of the
cycle applying to the embedded sentence, and not of some
higher application of the cycle.


BIBLIOGRAPHY
Albury, Donald H. 1971. In defense of the transformational
origin of derived nominis. Paper presented to the
Sixth Meeting of the Southeastern Conference on Lin
guistics, November 4, Atlanta, Georgia.
. Forthcoming. English as a VOIS language.
Bresnan, Joan W, 1970. On complementizers: toward a syn
tactic theory of complementizer types. Foundations of
Language 6.297-321,
Chomsky, Noam. 1965. Aspects of the theory of syntax.
Cambridge, Mass.: The MIT Press.
. 1970. Remarks on nominalization. Readings in Eng
lish transformational grammar, ed. by Roderick A,
Jacobs and Peter S. Rosenbaum, 184-221. Waltham,
Mass,: Ginn.
Emonds, Joseph. 1969. Root and structure-preserving trans
formations. Cambridge, Mass.s MIT Dissertation.
Fillmore, Charles J. 1968. The case for case. Universals
in linguistic theory, ed. by Bmmon Bach and Robert T.
Harms. 1-88. New York: Holt, Rinehart and Winston.
Fraser, Bruce, 1970. Some remarks on the action nominali
zation in English. Readings in English transforma
tional grammar, ed. by Roderick A. Jacobs and Peter S.
Rosenbaum, 83-98. Waltham, Mass.: Ginn.
Katz, Jerrold J., and Paul M. Postal. 1964. An integrated
theory of linguistic descriptions. Cambridge, Mass.:
The MIT Press,
arsky, Paul, and Carol Kiparsky. 1971. Fact. Semantics
ed. by Danny D, Steinberg and Leon A, Jakobovits. Cam
bridge: Cambridge University Press.
off, George. 1968. Deep and surface grammar. Blooming
ton, Ind.: Indiana University Linguistics Club (Mimeo).
3


124
appropriate contextual information governing the application
of the rule within each embedded sentence. The rule would
also have to be constrained to prevent it from moving an
Agent out of its own sentence. All of this means that if
AGENT-PROPOSING were precyclic, it would have to be formu
lated to apply to possibly many different sentences within
a single tree, but at the same time confine each application
to a single sentence, and also be governed as to whether or
not to apply in any given sentence by contextual information
specific to that sentence. All-in-all, this does not seem
to be a very desirable type of rule. All of the above com
plications may be avoided by assuming that AGENT-PREPOSING
is a cyclic rule.
The Place of Rules of Nominalization in the Cycle
Since, as was mentioned before, AGENT-PREPOSING
involves the mention of predicate in its structural descrip
tion, and can apply freely in the derivations of derived
nominal complements, it must precede DERIVED NOMINALIZATION.
This means that DERIVED NOMINALIZATION is also cyclic. That
this Is so is also shown Indirectly by the existence of the
derived nominal complements in 4.6, which in turn have embed
ded derived nominal complements.
4.6a) Harvey's discovery of Mary's disappearance
b) Bill's search for evidence of John's betrayal
of Alice
c) Jane's reluctance to admit her hatred of Jerry
The same arguments given above to show the undesirabi
lity of making AGENT-PREPOSING a precyclic rule also apply
to DERIVED NOMINALIZATION.


88
Conditions on Application
The conditions on the application of NP-PREPOSING that
I have discussed above fall along two parameters, 'The con
textual parameter opposes sentences and gerundive nominal
complements to derived nominal complements. The noun phrase
parameter opposes simple noun phrases to head nouns of com
plements. The interaction of these parameters is indicated
in the chart in 3.148.
3.148 Conditions on the Application of NP-PREPOSING
(AGENT-PREPOSING has not applied)
simple noun
phrases
head nouns of
complements
in sentences and
gerundive nomi
nal complements
obligatory
obligatory
in derived nomi
nal complements
optional*
blocked
^blocked for specific predicates
In Chapter Four I will argue that when NP-PREPOSING
applies in the derivation of gerundive nominal complements,
they are still sentences. Thus, NP-PREPOSING is obligatory
when it applies within structures dominated by an S, On the
other hand, I will argue that when NP-PREPOSING applies in
the derivation of derived nominal complements, they are
already nominaiized. Thus, NP-PREPOSING is optional or
blocked when it applies within structures dominated by an
NP. I will have more to say about how this distinction is
achieved in Chapter Four.


3-5oa) That Jerry was a bigamist was discovered by John,
b) That you will go is doubted by them.
3-5?a) *that Jerry was a bigamist's discovery by John
b) '-'that you will go's doubting by them
3.53a) the discovery by John that Jerry was a bigamist
b) the doubting by them that you will go
Chomsky suggests that derived nominal complements such
as those in 3-59 are derived by the obligatory application
of a rule called AGENT-POSTPOSING (cf. Passive Sentences
below) to the underlying noun phrases in 3*60.
3*59a) the necessity for John to leave
b) the likelihood that John will leave
3-Oa) Tofor John to leaver's necessity
** w
b) that John will leaver's likelihood
This solution will explain why the examples in 3*51 and
3.59 are acceptable while those in 3-53 and 35^ are not.
AGENT-POSTPOSING seems to be a poor name for a rule which
moves complements in such structures, however, "and there is
a curious problem with forms like those in 3-57 and 3-53.
If the fact that the derived nominal complements in 3-53 are
acceptable but not those in 3*57 is to be explained by the
ad hoc obligatory application of AGENT-POSTPOSING, we are
left with the problem of accounting for the post-verbal
by-phrases in 3-58. I must reject Chomsky's implied claim
that AGENT-POSTPOSING has applied in the derivations of the
derived nominal complements in 3*58, but not in those in
3.591 and in so doing, reject his claim that AGENT-POSTPOS
ING moves any complement within derived nominal complements.


114
iia) It seems that it is likely that John is late.
b) It appears that it is certain that Mildred has
fallen down.
c) It happens that it is true that Mike is a
brilliant student.
Ik .
Ross5 example (1.10).
^ Ross' example (4.ISO).
1 £
x There does not allow the possessive suffix at ail in
gerundive nominal complements (cf. Note 8 above).
I have indicated the alternate forms in 3162 since
Newmeyer paired 3*l62aii with 3-lOa in his discussion of
this point.
-j 0
The form in 3-l65o seems pretty bad to me. The
choice of happenstance as the derived nominal corresponding
to happen seems more appropriate here than the other possi
bility, happening.
^ SQUI-NP-DELETION applies to the end-cyclic subject
of an embedded sentence, as shown by its application to the
derived subjects of embedded passive sentences. Both i and
ii are possible results of applying the rule.
i) Someone intended to leave Mary behind.
ii) Mary did not intend to be left behind by anyone.
20
Lee (I9?0:?0ff.) suggests that the surface subjects
of sentences like John amused the children by telling
stories (non-intentional reading) are derived by a RAISING
rule from the embedded sentence John told stories. The para
phrase John's telling stories amused the children would be
derived by moving the whole embedded sentence into subject
position. In the sentence John amused the children with his


62
352ci) To take this test is fun.
ii) Taking this test is fun.
3-53ai) *that John will win the prize's being certain
ii) ^John's winning the prizes being certain
bi) *to learn algebra's being difficult
ii) *learning algebra's being difficult
ci) *to take this test's being fun
ii) *taking this test's being fun
3o4ai) *that John will win the prize's certainty
ii) ^John's winning the prize's certainty
bi) *to learn algebras difficulty
ii) ^learning algebra's difficulty
ci) *to take this test's fun
ii) ^taking this test's fun
The Lexical Hypothesis predicts that there should be
derived nominal complements corresponding to the sentences
in 3*52f which are supposedly closer to base phrase markers
than the extraposed sentences in 3*49, but the unacceptabi-
litv of the strings in 3*53 and 3*5^ in contrast to the
acceptability of the nominal complements in 3-50 and 3*51
12
is counter to that prediction.
With Passives
By Rosenbaum's analysisf IT-EXTRAPOSITION also applies
to passive sentences with complements in the derived subject
position. Thus, the extraposed sentences in 3*55 are
derived from the phrase markers underlying the passive sen
tences in 3-56. The sentences in 3-56 do not have any accep
table corresponding derived nominal complements, the expec
ted forms in 3-57 being unacceptable, but the extraposed pas
sive sentences in 3-55 have the corresponding derived nomi
nal complements in 3*58.
3-55a) It was discovered by John that Jerry was a
bigamist.
b) It is doubted by them that you will go.


103
Psychological Predicates
Chomsky presents a set of examples which show a unique
pattern of acceptability for derived nominal complements.
The sentences in 3.193 have the corresponding gerundive nom
inal complements in 3.194, hut only 3.193b has an acceptable
corresponding derived nominal complement in 3.195b, with the
form in 3.195a being unacceptable.
3.193a) John amused the children with his stories,
b) John was amused at the children's antics.
3.194a) John's amusing the children with his stories
b) John's being amused at the children's antics
3.195a) *John's amuement of the children with his
stories
b) John's amusement at the children's antics
The unacceptability of 3.195a raises a couple of pro
blems. The first problem is that of instrumental phrases
such as with his stories. The examples from Chomsky which
a*
I have just repeated above involve sentences with the verb
amuse, which belongs to the set of psychological predicates.
The sentence in 3,193a has at least two semantic readings.
One reading involves the presupposition that John intended
to amuse the children, while the other reading does not
include that presupposition. The two readings may be illus
trated by the use of the adverbs cleverly and inadvertently,
as in the sentences in 3.187.
3.187a) John cleverly amused the children with his
stories.
b) John inadvertently amused the children with
his stories.


147
Stockwell, Robert P., Paul Schacter and Barbara Hall Partee.
1973- The major syntactic structures of English. New
Yorks Holt, Rinehart and Winston.
Vendler, Zeno. 1968. Adjectives and norninalizations. The
Hague: Mouton.


132
allow the rule to be governed by the predicate of the higher
sentence. This allows the rule to properly account for the
restrictions on the type of complement a predicate may take,
and for the fact that only embedded underlying sentences
become complements in surface structure. The principle of
the cycle requires that all cyclic rules apply to an embed
ded sentence before any cyclic rule applies to the higher
sentence in which the first sentence is embedded. COMPLE
MENTIZER PLACEMENT reaches down into an embedded sentence to
which the cycle has previously applied, and thus stretches
the principle of the cycle to the breaking point. Chomsky's
(1965s146) claim that no transformation can insert morpholo
gical material into an embedded sentence supports the above
analysis of COMPLEMENTIZER PLACEMENT as violating the prin
ciple of the cycle.
Bresnan (1970) offers, as a solution to the problems
with COMPLEMENTIZER PLACEMENT discussed above, the
Phrase-structure Hypothesis, which is a proposal that com
plementizers are specified in deep structure by a
phrase-structure rule that rewrites 3 as COMP+S, giving deep
structure trees like that in 4.16, and that predicates are
subcategorized for the form of the complementizer which is
lexically inserted under the node COMP.
4, l6
S
0
NP
VP
s
1
Pred NP
COMP 3
2


30
Nominis of Causatives
Consider verbs such as grow, raise and move. These
verbs occur in both intransitive and transitive forms, the
latter being derived by the rule of CAUSATIVE FORMATION.
The verbs in the intransitive form are used in sentences
such as those in 2.43, which have the corresponding derived
7
nominal complements in 2.44 and 2.45.
2.43a)
b)
c)
The tree grew slowly.
The temperature rose rapidly.
The table moved mysteriously.
the tree's slow growth
the temperature's rapid rise
the table's mysterious movement
2.45a)
b)
c)
the slow growth of the tree
the rapid rise of the temperature
the mysterious movement of the table
The verbs in the transitive form are used in sentences
such as those in 2.46, which have the corresponding gerund
ive nominal complements in
derived nominal complements
2.4?, and the corresponding
in 2.43 and 2.49.8
2.46a) John grows tomatoes skillfully.
b) Tom raised the temperature deliberately.
c) Hike moved the table accidentally.
2.47a) John's growing tomatoes skillfully
b) Toms raising the temperature deliberately
c) Mike's moving the table accidentally
2.48a) John's skillful growing of tomatoes
b) Toms deliberate raising of the temperature
c) Mikes accidental moving of the table
;/ ci j
b)
c)
the skillful growing of tomatoes by John
the deliberate raising of the temperature by Tom
the accidental moving of the table by Mike
The derived nominal complements in 2.48 and 2.49 are
completely consistent with the
definition of derived nominal


2 3. 9 6.a
¡L
i .


3
three characteristic properties; A, that productivity is
more restricted for derived nominal complements than for
gerundive nominal complements, B, that semantic relations
between derived nominis and their associated propositions
are varied and idiosyncratic, and C, that derived nominal
complements have the Internal structure of noun phrases.
Having concluded that derived nominal complements cannot be
derived transformationally, Chomsky claims that the phrase
structure rules of English must be complicated to provide
the structure of derived nominal complements directly and
that derived nominis are lexical entries. This is the
Lexical Hypothesis, and amounts to a claim that derived
nominal complements show less regularity than other struc
tures in English, and the limits of regularity have been
found for this part of the grammar of English, If Chomsky
is wrong about the implications of the properties of derived
nominal complements, then his argument for the need for the
Lexical Hypothesis is weakened. I propose to show/ that
Chomskys three properties do not establish the irregularity
of the formation of derived nominal complements, that their
formation Is as regular as other transformational processes,
and that a transformational account of derived nominal
complements is at least as well motivated as the Lexical
Hypothesis.
The Lexical Hypothesis
Chomsky (1970:188) states the lexicalist position as
the choice of extending the base rules, with concomitant


17
possessive nouns corresponding to the subject nouns in the
sentences. Chomsky points out that the two types of nominal
complement differ in that the possessive nouns in derived
nominal complements can be replaced by other determiners, as
is shown by the derived nominal complements in 2.4 (cf. 2-3).
while the possessive nouns in gerundive nominal complements
cannot be so replaced, as is indicated by the fact that forms
like those in 2-5 (cf. 2.2) do not occur. The derived nomi
nal complements in 2.4 appear to be parallel to the derived
nominal complements in 2-3? but the derived nominal comple
ments in 2.6 are closer in meaning to those in 2.3 than the
2
ones in 2.4 are.
2.4a) ?the eagerness to please
b) the refusal of the offer
c) the criticism of the book
2.5a) *the being eager to please
b) *tha refusing the offer
c) *the criticizing the book
2.6a) ?the eagerness to please by John
b) the refusal of the offer by John
c) the criticism of the book by John
Adjective-Adverb Correspondences
A second difference between gerundive nominal comple
ments and derived nominal complements which Chomsky discus
ses is the fact that derived nominal complements can have an
adjective preceding the nominal, while gerundive nominal com
plements cannot. These pronominal adjectives correspond to
adverbs in sentences and gerundive nominal complements, so
that there are sentences like those in 2.7, with the corres
ponding gerundive nominal complements in 2.8, and the corres-


Abstract of Dissertation Presented to the Graduate Council
of the University of Florida
in Partial Fulfillment of the Requirements for the
Degree of Doctor of Philosophy
NOMINAIS AS COMPLEMENTS
3y
Donald Herbert Albury
March, 197^
Chairmans Jean Gasagrande
Major Departments Linguistics
The syntactic construction Anown as the derived nominal
presents problems to a transformational grammar of English.
Noam Chomsky has claimed that derived nominis have
restricted productivity, an internal structure "like that of
noun phrases, and idiosyncratic semantic relationships to
their associated predicates. Chomsky's claim that the Lexi
cal Hypothesis provides a better explanation of the charac
teristics of derived nominis than the transformational pos
ition is rejected. A predicate-initial analysis of the
underlying structure of English and an analysis of derived
nominis as complements of nouns in underlying structure are
adopted. It is then shown that the productivity of derived
nominis is not as restricted as Chomsky claims, and that
the remaining restrictions are due to the failure of cer
tain rules to apply in the derivation of derived nominis.
vi x


CHAPTER FOUR
RULE-ORDERING AND DERIVED NOMINAL COMPLEMENTS
Introduction
In Chapter Three I showed for a number of examples that
whether or not a particular sentence has a corresponding
derived nominal complement depends on whether or not that
sentence has undergone one or more of a certain set of trans
formations in its derivation. Those sentences which have
undergone one of the rules in question do not have corres
ponding derived nominal complements. In this chapter I will
offer an explanation for the above fact.
In Chapter One I stated my assumption that derived nomi
nal complements are derived transformationally'-from underly
ing embedded sentences. In Chapter Two I showed that, with
the exception of the Germanic member of Germanic-Latinate
synonym pairs, all predicates have corresponding derived
nominis. Finally, in Chapter Three I showed that the only
other restrictions on the productivity of derived nominal
complements are due to the failure of certain transforma
tional rules to apply in the derivation of derived nominal
complements. In this chapter I will argue that certain
rules are blocked from applying in derived nominal comple
ments because DERIVED NOMINALIZATION precedes those rules,
and the rules cannot apply to nominalized structures. I
116


96
those sentences, as in 3.169 "to which THERE-INSERT ION has
not applied. These sentences have the corresponding gerund
ive nominal complements in 3.170 and derived nominal comple
ments in 3.171. The forms in 3.172, with the same word
order as the sentences in 3*169, are not acceptable.
3.169a) No hope existed.
b) A disturbance occurred.
c) Some wine was present in the bottle.
3.170a) no hope('s) existing
b) a disturbance('s) occurring
c) some wine('s) being present in the bottle
3.1?la) the existance of no hope
b) the occurrence of a disturbance
c) the presence of some wine in the bottle
3,172a) *no hope's existance
b) *a disturbance's occurrence
ci) *some wine in the bottle's presence
ii) *some wine's presence in the bottle
Certain points emerge with these examples. The occur
rences of be in the sentences in 3.166 are not really the
v
copula, but rather an alternate representation of the exis
tential verbs in 3.169. The only acceptable derived nominal
complements corresponding to the sentences in 3.169 show pre
dicate-initial order. For this set of predicates, then,
NP-PREP03ING seems to apply optionally in sentences, and not
at all in derived nominal complements. This pattern is paral
lel to the application of the rule with an embedded that-com-
plement.
THERE-INSBRTION and PI-INSERTION are both blocked from
applying to derived nominal complements, and are both gov
erned by very similar conditions. If no noun phrase precedes


10
After all the rules have had a chance to apply to 3
the rules would apply in the same order to the next higher
sentence, S-, which is in turn embedded in SQ, After all
the rules have had a chance to apply to 3-, the rules would
3
apply again in the same order to the highest sentence, Sq.
I will also adopt here a predicate-initial analysis of
the underlying structure of English sentences. McCawley
(19?0) presents arguments for such an analysis for struc
tural representations comparable to those in Chomsky (1965).
A predicate-initial analysis is also at least implicit in
the case grammar of Fillmore (1968). The underlying struc
tures in case grammar differ greatly from the underlying
structures postulated in Chomsky (1965). In Fillmore's case
grammar, the noun phrases in deeply underlying structures
are identified by their semantic relationship to the predi
cate of the sentence, and not by any syntactic relationship.
Without adopting case grammar in toto, I will assume that,
before any syntactic rules have applied, there is no ordered
relationship between the constituents of the underlying pro
positions, and that at least some syntactic rules must recog
nize semantic relationships and transform such semantic rela-
4
tionships into word-order syntactic relationships. For con
venience of representation, however, I will assume that
underlying sentences have structures like that in 1.13
before any syntactic rule has applied.^
1.13
S
PRED
NP
NP


112
after RAISING-TO-OBJECT but before TOUGH-MOVEMENT and RAIS
ING -TO-SUBJECT. A sentence like John was believed by Mary
to be a fool is derived by PASSIVTZATION from the structure
underlying Mary believes John to be a fool, which is derived
in turn by RAI3ING-T0-OBJECT from the structure underlying
Mary believes that John is a fool. This indicates that RAIS
ING -TO-OBJECT must precede PAS3IVIZATIQN. The sentence it
is believed by Mary that John is a fool is derived by IT-EX
TRAPOSITION from the structure underlying that John is a
fool is believed by Mary, which is derived in turn by PASSI-
VTZAT'IQN from the structure underlying Mary believes that
John is a fool. This indicates that PASSIVTZATION must pre
cede IT-EXTRAPOSITION. On the other hand, the sentence John
is easy to olease is derived by TOUGH-MOVEMENT from the
structure underlying it is easy to olease John, which is in
turn derived by IT-EXTRAPOSITION from the structure underly
ing to olease John is easy. Similarly, the sentence John is
likely to win the prize is derived by RAISING-TO-SUBJECT
from the structure underlying it is likely that John will
win the arize, which is in turn derived by IT-EXTRAPOSITION
from the structure underlying that John will win the prize
is likely. This indicates that IT-EXTRAPOSITTON must pre
cede TOUGH-MOVEMENT and RAISING-T0-SU3JECT. Assuming trans
itivity for the ordering relationships exemplified above,
the order of the five rules within a single cycle must be?
RAISING-TO-OBJECT, PASSIVTZATION, IT-EXTRAPOSITION and
TOUGH-MOVEMENT and RAISING-TO-SUBJECT. However, I will


146
. 1970, Irregularity in syntax. New York: Holt,
Rinehart and Winston,
Lakoff, Robin. I98. Abstract syntax and Latin complemen
tation, Cambridge, Mass.! The MIT Press.
Lee, Paul Gregory. 1970. Subjects and agents. Columbus,
Chios OSU Dissertation.
Lees, Robert B. 1963. The grammar of English nominaliza-
tions. The Hague: Mouton.
McCawley, James D. 1970. English as a VSO language.
Language 46.286-99.
Menzel, Peter. 1969. Propositions, events, and actions in
the syntax of complementation. Los Angeles: UCLA Dis
sertation.
Milsark, Gary, 1972. Res doubl-ing. Linguistic
Inquiry 3.542-49.
Newmeyer, Frederick J. 1971. The source of derived nomi
nis-* in English. Language 4?,786-96,
Pizzini, Quentin A. 1972. A transformational analysis of
infinitival and gerundive complements in English. La
Jolla, Cal.: UC San Diego Dissertation.
Postal, Paul M, 1971. Cross-over phenomena. New York:
Holt, Rinehart and Winston.
Rosenbaum, Peter S. 1967. The grammar of English predi
cate complement constructions. Cambridge, Mass.: The
MIT Press.
Ross, John Robert. 19o7. Constraints on variables in syn
tax. Cambridge, Mass.: MIT Dissertation,
. 1972a. Act. Semantics of natural language, ed, by
Donald Davidson and Gilbert Harman, 70-126, Dordrecht
D. Reid.el,
. 1972b. Doubl-ing. Linguistic Inquiry 3.61-86,
1972c. Nouniness. Forum Lecture presented to the
LSA Linguistic Institute, June 15, Chapel Hill, N.C,
Smith, Carlotta. 1972. On causative verbs and derived norm
inals in English. Linguistic Inquiry 3.136-38.


3.137a)
*(the fact) that the account was overdrawn's
being reported by Tom
b)
*(the fact) that Alice was smart's being
c)
revealed by Jerry
*(the claim) that Peggy was pregnant's being
denied by Mary
3.138a)
*(the fact) that the account was overdrawn's
b)
report by Tom
*(the fact) that Alice was smart's revelation
c)
by Jerry
*(the claim) that Peggy was pregnant's denial
by Mary
The sentences in 3.134 ho not have any acceptable cor
responding gerundive or derived nominal complements, the
forms in 3.139 and 3.140 all being unacceptable.
3.139a)
*the fact's being reported by Tom that the
account was overdrawn
b)
*the fact's being revealed by Jerry that Alice
was smart
c)
*the claim's denial by Mary that Peggy was
pregnant
3.l40a)
*the fact's report by Tom that the account was
overdrawn
b)
*the fact's revelation by Jerry that Alice was
smart
c)
*the claim's denial by Mary that Peggy was
pregnant
The sentences in 3,135 have the acceptable corresponding
gerundive nominal complements in 3.141 and derived nominal
complements :
in 3.142.
3.141a)
its being reported by Tom that the account
was overdrawn
b)
its being revealed by Jerry that Alice was
c)
smart
it being denied by Mary that Peggy was
pregnant
3.142a)
the report by Tom that the account was over
drawn
b)
c)
the revelation by Jerry that Alice was smart
the denial by Mary that Peggy was pregnant


2
which he considers to be derived by transformations from
underlying embedded sentences. This is an empirical claim,
and may be tested by examining the sentences of English
which contain these two types of complements. The forms of
these complements, and their relation to each other and to
sentences, is illustrated in 1.1, 1.2 and 1.3; the main
clauses in 1.1 having the corresponding gerundive nominis
1
in 1.2 and derived nominis in 1.3*
1.1a) John is usually calm under stress, which pleases
Mary.
b) John completed the assignment early, which
pleased Mary.
1.2a) John's usually being calm under stress pleases
Mary.
b) John's completing the assignment early pleased
Mary.
1.3a) Johns usual calmness under stress pleases Mary.
b) John's early completion of the assignment
pleased Mary.
The underlined words are intended to supply appropriate
contexts for the nominal complements as well as showing how
sentences and nominal complements can be alternate realiza
tions of a common underlying structure. Gerundive nominal
complements often resemble sentences with progressive aspect.
While this may be confusing at times, the intended reading
of such forms as gerundive nominal complements may be made
clear by putting them into a context such as ...is surpris
ing.- or Ann is worried about....
Chomsky's argument that derived nominal complements are
not regular, and cannot be derived from underlying embedded
sentences, as are gerundive nominal complements, Is based on


119
moving an indirect object to a position immediately to the
right of the predicate.
Although I have not said anything about the form of
PSYCH-CHANGE, this rule (or set of rules) involves the
choice of the noun phrase to be subject of the sentence in
surface structure, as well as differences in the surface
form of the predicate.
Each of these rules must refer explicitly to the struc
tural relationships between a noun phrase and a predicate.1
The structural description of each rule must include the
mention of predicate within the sentence to which the rule
is directly applying.
The rule of NP-PREPOSING presents some difficulty at
this point. This rule also moves a noun phrase to a posi
tion immediately to the left of a predicate, thus making it
the subject of the predicate in surface structure. By
Newmeyer's proposal, this rule should not apply in the deri
vation of derived nominal complements. In fact, it does so
apply, although its application is very limited. The appli
cation of NP-PREPOSING in the derivation of derived nominal
complements is limited to simple noun phrases and to a sub
set of predicates, and moreover, is optional even when it
does apply, while the application of NP-PREPOSING to head
nouns is always obligatory in the derivation of sentences,
if the structural description of the rule is met. There
would be no conflict with Newmeyer's proposal if it were
assumed that a different rule applied in the derivation of


NOTES
One exception to this claim is that verbs like be and
have do not function as main verbs in sentences with corres
ponding derived nominal complements, as was discussed in
Chapter Two.
^ TOUGH-MOVEMENT is a RAISING rule. It raises the
object of an embedded sentence to subject of the next higher
sentence. The other RAISING rules are RAISING-TO-OBJECT,
which raises the subject of an embedded sentence to object
of the next higher sentence, and RAISING-*TO-SUBJECT, which
raises the subject of an embedded sentence to subject of the
next higher sentence. These three RAISING rules are dis
cussed in detail in Stoekwell et al. (1973)- TOUGH-MOVEMENT
is also discussed in Postal (1971)- The relation of IT-EX
TRAPOSITION and the RAISING rules to derived nominal comple
ments is discussed in the next two sections.
3
An indefinite object of please has been deleted.
Jj.
John as the subject of olease has been deleted by
EQUI-NP-DELETION (discussed in the section under that name
below). The unspecified object of olease has also been
deleted.
^ This derivation is questionable. See the section on
IT-EXTRAPOSITION below.
110


109
In this chapter I have shown that the adoption of a
predicate-initial analysis of underlying structure allows
the statement of restrictions on the occurrence of derived
nominal complements in terms of the application of certain
rules being blocked in the derivation of derived nominal
complements from underlying embedded sentences. I will
discuss how such rules are blocked in Chapter Four.
The analysis in this chapter shows that, contrary to
Chomsky's claim, the restrictions on the productivity of
derived nominal complements can be easily stated within a
transformational framework. In Chapter Four I will discuss
how the noun phrase-like internal structure of derived nom
inal complements can also be accounted for in a transforma
tional framework, based on the analysis given in this
chapter.


139
the cycle of the nominal complement, so that COMPLEMENTIZER
PLACEMENT is not needed to account for them either. One
major objection possible to this view of rules of nominaliza-
tion is that the transformational derivation of nominal com
plements from underlying embedded sentences must account for
the restrictions on the types of complements predicates in
higher sentences may take. If the rules of nominalization
apply on the cycle of the nominal complements, they must
look up into the next higher sentence to see if the predi
cate of that sentence takes nominal complements. This
implies a major change in the concept of the cycle. It
would seem to me, however, that allowing rules of nominaliza-
ticn to include a higher predicate in their structural des
criptions is preferable to the insertion of morphological
material into a lower sentence. In addition, having rules
of nominalization apply on the cycle of the nominal comple
ment allows a simple explanation of the fact that many rules
do not apply in the derivation of derived nominal comple
ments, an explanation not possible with COMPLEMENTIZER PLACE
MENT. Even if the Lexical Hypothesis is adopted, however,
the above arguments would still apply to gerundive nominal
complements.
Complementation as a Process
There are two basically different ways of handling the
derivation of complements. One way is to write a single com
plex rule which accounts for all the details of structure
characteristic of a particular type of complement. The rules


102
DAT IVE -*M0 VEM EN T
Newmeyer (1971) points out that sentences which have
undergone DATIVE-MOVEMENT in their derivations do not have
corresponding derived nominal complements. The sentences
in 3*186, which have undergone DATIVE-MOVEMENT, have the cor
responding gerundive nominal complements in 3*18?, but the
expected corresponding derived nominal complements in 3*188
are not acceptable,
3.186a) John gave Mary the book.
b) Bill told Alice the story.
c) Jane bought Jim a soda.
3*187a) John's giving Mary the book
b) Bill's telling Alice the story
c) Jane's buying Jim a soda
3.188a) ^Johns gift to Mary of the book
b) ^Bill's telling to Alice of the story
c) *Jane's buying for Jim of a soda
The sentences in 3*189 which have not undergone DAT
IVE-MOVEMENT, have the corresponding gerundive-nominal comple
ments in 3*190 and derived nominal complements in 3*191*
The nominal-initial derived nominal complements correspond
ing to this set of sentences are given in 3*192.
3*189a) John gave the book to Mary.
b) Bill told the story to Alice.
c) Jane bought a soda for Jim,
3*190a) John's giving the book to Mary
b) Bill's telling the story to Alice
c) Jane's buying a soda for Jim
3-191a) John's gift of the book to Mary
b) Bill's telling of the story to Alice
c) Jane's buying of a soda for Jim
3.192a) the gift of the book to Mary by John
b) the telling of the story to Alice by Bill
c) the buying of a soda for Jim by Jane


140
in Leos (1963) are excellent examples of this approach. The
second way to handle the derivation of complements is to
write several fairly simple rules each accounting for only
part of the structure characteristic of a particular type of
complement. Any independent motivation of the rules helps
to strengthen such an analysis. This approach may be used
to profit elsewhere, as with Chomsky's (1970) analysis of
passivization as the two rules of AGENT-PQSTPOSING and
NP-PREPOSING.
I have already presented an analysis of the derivation
of infinitival complements as being due to the deletion of
the subject of an embedded sentence by either a RAISING rule
or EQUI-NP-DELETIONs followed by the post-cyclic application
of TO-INSERTION. Derived nominal complements are formed by
the application of DERIVED NOMINALIZATION, followed by the
rules accounting for the absence of auxiliaries, presence of
a preposition (i.e. of) before direct objects, the prenomi-
nal position of adjectives corresponding to adverbs, the for
mation of a possessive from the subject, or the insertion of
another determiner if there is no subject, and the morpholo
gical modification of the former predicate. Gerundive nomi
nal complements are formed by the application of GERUNDIVE
NOViINALIZATIQN followed by the last two rules listed above
which apply to derived nominal complements. Thus, the rule
producing possessive noun phrases applies in the derivation
of both derived nominal complements and gerundive nominal
complements, while the rule suffixing -ing to a former


65
Psychological predicates appear to be the only predi
cates which form extraposed sentences which do not have
acceptable corresponding derived nominal complements.
Further Problems with the Rule
The acceptable derived nominal complements in 3-51*
3.58 and 3-59 all show the same order of major constituents
as the extraposed sentences in 3*^6, 3-^9 and 3-55- There
is no subject pronoun it in these derived nominal comple
ments. This fact can be explained by having rt in extra
posed sentences supplied by a rule of IT-INSERTION. which
does not apply to derived nominal complements (cf. Dummy
Sub.iect Insertion, below) The arguments for not having it
present in deep structure are well summarized in Stockwell
et al. (19?3s52?ff.). Given the absence of it in underlying
structures, the derived nominal complements in 3-51 3-58
and 3-59 may be taken as corresponding directly- to the extra
posed sentences cited above. If Chomsky's claim that derived
nominal complements correspond only to base phrase markers
is at all correct, then extraposed sentences would appear to
be more basic than nonextraposed sentences.
Although this conclusion goes against a generally accep
ted analysis, there is some independent supporting evidence.
The intransitive verbs seem, appear and happen take subject
13
complements which must be extraposed.'"^ The extraposed sen
tences in 366 are acceptable while the nonextraposed ones
in 3*67 are not.


16
I will temporarily accept Chomsky's implied morphologi
cal definition of derived nominis, and define derived nomi
nal complements as those complements in which nominis
formed without an -ing suffix are found. Starting from this
definition of derived nominal complements, I would like to
build up a detailed description of the structure of these
complements. Chomskys method of comparing sentences, ger
undive nominal complements and derived nominal complements
to illustrate the characteristics of derived nominal comple
ments is useful, and I will adopt it here.
The Surface Structure.of Derived Nominal Complements
Chomsky (1970s187-88) cites the sentences in 2.1 as
having the corresponding gerundive nominal complements in
2,2 and the corresponding derived nominal complements in 2.3.
The gerundive nominal complements have gerundive nominis
corresponding to the verbs in the sentences, while the
derived nominal complements have derived nominis corres
ponding to the verbs in the sentences.
2.1a) John is eager to please,
b) John refused the offer,
c) John criticized the book.
2.2a) John's being eager to please
b) John's refusing the offer
c) John's criticizing the book
2.3a) John's eagerness to please
b) John's refusal of the offer
c) John's criticism of the book
Determiners
One characteristic shared by gerundive nominal comple
ments and derived nominal complements is the presence of


20
ship is also seen in simple noun phrases, so that no new
rule need be postulated to derive prenominal adjectives in
derived nominal complements from higher predicates.
Chomsky (I970sl95) points out that a claim that the
prenominal adjectives in derived nominal complements are
derivationally related to adverbs leads to the prediction
that sentences like those in 2.1 will have corresponding
derived nominal complements like the structures in 2.1?,
which are not acceptable.
2.16a) John refused the offer in a surprising manner,
b) John is sincere to a limited extent.
2.17a) ^John's refusal of the offer in a surprising manner
b) ^Johns sincerity to a limited extent
However, there are structures like those in 2.18 which
include derived nominal complements, and which seem to cor
respond to the sentences in 2.1.
2.18a) the surprising manner of John's refusal of
the offer
b) the limited extent of John's sincerity
Adverbs which are morphologically derived from adjec
tives seem to be included within the scope of structures
which have corresponding derived nominal complements, while
adverbs which are prepositional phrases seem not to be
included within the scope of such structures. It is not
necessary to claim that the prenominal adjectives are
derived from adverbs to support the claim that the prenomi
nal adjectives and adverbs are derived from the same under
lying structures. The restrictions on the inclusion of ad
verbial prepositional phrases within the scope of derived
nominal complements does not invalidate such a claim.




CHAPTER ONE
INTRODUCTION
Statement of Purpose
The question of the extent of regularity in language
has been a recurrent theme in linguistics since ancient
times, when the Analogists of Alexandria disputed with the
Anomalists of Pergamon, This question has recently become
once more prominent in the study of transformational gram
mar. In an important sense, a transormation is an abstrac
tion of a regularity in language. The question of whether a
particular structure in a language can be described trans
formationally is an empirical one, and is equivalent to the
question of whether that structure in that language exhibits
a regular correspondence to another structure of that lan
guage. An assumption of transformational grammar has been
that language is basically regular, and, as a consequence,
transformational grammars have emphasized the regularity of
a language.
Noam Chomsky (1970) has claimed that the English comple
ment construction which he calls the derived nominal consti
tutes a part of English grammar which is not regular in the
sense indicated above, and which cannot be described trans
formationally. He supports his claim by contrasting derived
nominis to gerundive nominis, another type of complement
1


81
Again, while the head nouns of the embedded nominal com
plements in the examples above may be deleted, they may not
be separated from the complements, so that all the examples
in 3.12?, 3.128 and 3.129 are unacceptable.
3,127a) *The fact was unexpected of John's leaving so
early.
b) ^Bill's action was unfortunate of refusing the
offer,
c) *The event was likely of Dick's resignation,
d) *The fact was remarkable of Jimmy's testimony.
3.128a) *the fact's being unexpected of John's leaving
so early
b) ^Bill's action's being unfortunate of refusing
the offer
c) *the event's being likely of Dick's resignation
d) *the fact's being remarkable of Jimmys
testimony
3.129a) *the fact's unexpectedness of John's leaving
so early
b) ^Bill's action's unfortunateness of refusing
the offer
c) *the event's likelihood of Dick's resignation
d) *the fact's remarkableness of Jimmy's
testimony
i*-
The examples above show that either AGENT-PREPOSING or
NP-PREPCSING must apply in the derivation of sentences with
embedded nominal complements, that the only acceptable ger
undive nominal complements with embedded nominal complements
are those to which AGENT-PREPOSING has applied, and that
derived nominal complements with embedded nominal comple
ments to which AGENT-PREPOSING has applied, as well as
derived nominal complements with embedded nominal comple
ments to which neither AGENT-PREPOSING nor NP-PREPOSING has
applied, are acceptable. No other possibilities are accep
table, not is any case in which the head noun is separated
from the embedded nominal complements.


I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy,
lerqn Casagrande* Chairman
/Associate Professor of Romance
Languages and Linguistics
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy,
Edward Charles Hutchinson
Associate Professor of Speech
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
Assistant Professor of English
Florida State University


136
It would appear at this point that jfco is inserted into
the embedded sentence in each case, in violation of Chomsky's
claim that such insertion by a transformation is unnatural.
This would be so if TO-INSERTION were a cyclical rule.
Chomsky (1965s 146) specifically states that no morphologi
cal material can be introduced into a configuration domi
nated by S once the cycle of transformational rules has
already completed its application to this configuration."
I will argue, however, first, that TO-INSERTION is a
post-cyclic rule, and second, that post-cyclic rules are not
necessarily subject to the restriction stated by Chomsky.
TO-INSERTION (or INFINITIVE FORMATION) applies if
verb-agreement is not possible. The easiest way to handle
this fact in the grammar is to order TO-INSERTION after the
rule (or rules) responsible for verb-agreement, with TO-IN-
SERIION blocked by the application of the rule JfERB-AGREE
MENT. Ordering TO-INSERTION before VERB-AGREEMENT would
require that TO-INSERTION be sensitive to the same environ
mental facts as VERB-AGREEMENT. In either case, both rules
must be late enough to allow for all possible changes in the
subject. VERB-AGREEMENT cannot be cyclical, as there has
been no application of VERB-AGREEMENT in an embedded sen
tence prior to the removal of the subject by EQUI-NP-DELE
TION or one of the RAISING rules, and we have already seen
that these rules apply on the cycle of the next higher sen
tence, after the cycle is completed on the embedded lower
sentence. VERB-AGRESMENT cannot apply in the cycle on the


82
<¡ith That-Complements
That-complements also appear to have head nouns in
underlying structure (cf. Menzel, 1969) Like the head
nouns of nominal complements, these head nouns are deletable.
The optional presence of head nouns in surface structure
will again be indicated by the use of parentheses.
As before, I will claim that AGENT-PREPOSING has applied
to predicate-initial underlying structures to yield the sen
tences in 3*130 and their corresponding gerundive nominal
complements in 3*131 and derived nominal complements in 3*132.
3.130a) Tom reported (the fact) that the account was
overdrawn.
b) Jerry revealed (the fact) that Alice was smart.
c) Mary denied (the claim) that Peggy was
pregnant.
3.131a) Tom's reporting (the fact) that the account
was overdrawn
b) Jerry's revealing (the fact) that Alice was
smart
c) Mary's denying (the claim) that Peggy was
pregnant
3.132a) Tom's report (of the fact) that the account
was overdrawn
b) Jerry's revelation (of the fact) that Alice
v/as smart
c) Mary's denial (of the claim) that Peggy was
pregnant
If AGENT-PR3P0SING does not apply, then NP-PREPOSING
applies obligatorily in the derivation of sentences. How
ever, NP-PREPOSING applies in two ways. Either the head
noun and the that-complement are preposed as a unit, as has
happened in the derivation of the sentences in 3-133 or
only the head noun is preposed, in which case the sentences
in 3*13^ result if the head noun is retained in surface


46
The formation of derived nominal complements corres
ponding to sentences with predicate nouns does not seem to
be completely free. Thus, while Johns manhood is possible,
it corresponds to a limited meaning of John is a man, and
Jane's womanhood seems very strange. Nevertheless, many
predicate nouns do have corresponding derived nominis, as
can be seen in the sentences in i with the corresponding
derived nominal complements in ii.
i.a) Joan enjoys being a mother.
b) Ralph was governor recently.
c) Billy is a minor.
d) Stephen was a martyr.
e) John is a member in good standing in the lodge.
f) Sam is legally a pauper.
iia) Joan enjoys motherhood.
b) Ralph's recent governorship
c) Billy's minority
d) Stephen's martyrdom
e) John's membership in good standing in the lodge
f) Sam's legal pauperdom
Chomsky (1970:198-99) argues that the inclusion of verbs
and adjectives in a category of nredicator, as in G. Lakoff
(1970:115ff)j is wrong, since nouns share the same distri
butional properties. The examples above of derived nominis
corresponding to predicate nouns supports Chomsky's conclu
sion. Chomsky further claims, however, that such distribu
tional properties are properties of lexical categories. It
appears to be more correct to say that properties such as the
stative-active distinction and the possession of a corres
ponding derived nominal, are properties of predicates. Such
properties are exhibited by verbs (which are always predi
cates), adjectives (which presumably are always predicates
in underlying structures) and predicate nouns.


133
This Phrase-structure Hypothesis raises its own pro
blems- It requires the ad hoc extension of the principle of
subcategorization features to allow such features to apply
across sentence boundaries. While it is easy enough to
state that a particular predicate takes such-and-such comple
ments in surface structure, the expression of such restric
tions in underlying structure is not simple. The
Phrase-structure Hypothesis requires that a subcategoriza
tion feature on the predicate in Sq in 4.16 govern the selec
tion of a lexical item (the complementizer) within the embed
ded sentence S^. At the same time, the complementizer under
COMP must eventually be inserted into the embedded sentence
S?. Any possible modification of the Phrase-structure Hypo
thesis would still require either that subcategorization fea
tures reach across sentence boundaries, or that complement
izers be inserted into lower sentences. Bresnan's
Phrase-structure Hypothesis hides the problem of COMPLEMENT-
IZER PLACEMENT without resolving it.
I will show below that COMPLEMENTIZER PLACEMENT can be
eliminated from the grammar without resort to anything like
the Phrase-structure Hypothesis. I will first discuss the
formation of infinitival complements, and then return to
nominal complements. As I have already mentioned, the forma
tion of that-complements does not require the insertion of
any complementizer into the embedded sentence, and thus will
not enter into this discussion.


47
7
The corresponding gerundive nominal complements, as
in i, which I find awkward, but still acceptable, are com
pletely unacceptable to at least some speakers. I have no
explanation for this fact.
ia) the tree's growing slowly
b) the temperature's rising rapidly
c) the table's moving mysteriously
More generally, possessive inanimate nouns are often
awkward. Arnold Zwicky (personnel communication) pointed
out that historically inanimates have not always been able
to form possessives in English. The two facts may be
related.
O
Although move is borrowed ultimately from Latin and
the derived nominal corresponding to the intransitive form
of the verb is formed with -ment. move does not appear to
belong to the class of Latinate verbs described by
Smith (1972).
Q
' I v/ill discuss the fact that there does not take the
possessive suffix in gerundive nominal complements in more
detail in Chanter Three.


126
Since REFLEXIVE cannot cross sentence boundaries in its
application, the subjects of the embedded sentences cannot
become reflexive until they have been raised to object of
the predicate in the higher sentence. Hence, REFLEXIVE must
follow RAISING, which follows DERIVED NOMINALIZATION. 'The
fact that REFLEXIVE is not blocked by DERIVED NOMINALIZATION
supports Newmeyer's claim that only rules that mention predi
cate are regularly blocked by DERIVED NOMINALIZATION.
All of the rules discussed above which are blocked by
DERIVED NOMINALIZATION apply freely in gerundive nominal com
plements. GERUNDIVE NOMINALIZATION also changes the predi
cate so that it no longer functions as such. No modals are
present in gerundive nominal complements, and -ing replaces
the tense marker. I see no reason why this cannot be called
nominalization of the predicate. The presence of be and
have as auxiliaries and the absence of the preposition of
following gerundive nominis can be explained by ordering
the rules responsible after DERIVED NOMINALIZATION and
before GERUNDIVE NOMINALIZATION. If be and have are
inserted, nominalization blocks the rule. If they are pre
sent in deeply underlying structures, they are deleted by a
rule that applies to noun phrases, not to sentences. If of
is inserted, the rule applies to noun phrases, not to sen
tences. If of is present in deeply underlying structures,
it is deleted by a rule that is blocked by nominalization.
I have just argued, in effect, that most cyclic rules
are ordered between DERIVED NOMINALIZATION and GERUNDIVE


IT-EXTRAPOSITION
Introduction
The rule of IT-EXTRAPOSITION was proposed by Rosenbaum
(1967) to account for the relationship of sentences like
those in 3*45 To the sentences in 3*46.'*'"*
3-45a) That John left so early is surprising.
b) For John to quit now would be a disaster.
c) That Mary will come to her senses is to be
hoped for.
346a) It is surprising that John left so early.
b) It would be a disaster for John to quit now.
c) It is to be hoped for that Mary will come to
her senses.
In Rosenbaum's analysis, the subject t in the sentences
in 3-46 is present in deep structure as the head noun of
noun phrases with the structure of 3-47. 'The rule of IT-EX-
TRAPOSITION is given in 3.48.
it that John left so early
3.48) IT-EXTRAPOSITION (Optional)
X N S Y
[+PRO(i.e. it)] ==> 1, 2, 4+3
1 2 3 4
If IT-EXTRAPOSITION does not apply, the head pronoun it
is deleted. If IT-EXTRAPOSITION does apply, moving the
embedded sentence to the right end of the next higher sen
tence, the head pronoun it is isolated in subject position,
and no longer subject to deletion.
In this section I will first discuss the derivation of
certain examples via the rule of IT-EXTRAPOSITION, showing


7^
intransitive sentences is'optional, so that the derived nom
inal complements in 3.98 and 3-99 are acceptable.
3.98a) the boat's sudden sinking
b) John's arrival
c) Mary's kindness
d) Bill's friendliness
3.99a) the sudden sinking of the boat
b) the arrival of John
c) the kindness of Mary
d) the friendliness of Bill
The application of NP-PRHPQSING to simple noun phrases
shows a simple pattern. If there is no noun phrase in sub
ject position when NP-PRBPOSING is applicable (as is the
case with all intransitive predicates, transitive verbs with
unspecified agents and transitive verbs with specified
agents to which AGENT-PREPOSING has not applied), then
NP-PREPOSING moves the underlying object noun phrase into
subject position. This movement is obligatory in the deri-
vation of sentences and gerundive nominal complements and
optional in the derivation of derived nominal complements
(with the exception that certain nominis corresponding to
transitive verbs block the application of the rule in
derived nominal complements).
With Nominal Complements
As I indicated in Chapter One, I have accepted Menzel's
(19o9) analysis of nominal complements as complements of a
restricted set of head nouns in underlying structures.
These head nouns ai'e optionally deletable, and in the follow
ing examples will be enclosed in parentheses to Indicate
this optionality.


9
regard to the head nouns action and act. The adjective
cowardly is more appropriate for the complements of act.
1.11 (in the context ...was unexpected/cowardly.)
a) Johns denying the request
b) Johns action of denying the request
c) John's act of denying the request
d) John's denial of the request
e) ^Johns action of denial of the request
f) *John's act of denial of the request
g) John's act of denial
Gerundive nominal complements and derived nominal com
plements are alike in that they both can be complements of
a limited set of head nouns in surface structure, and seem
to always be complements of such nouns in underlying struc
ture, The two types of complement differ in that only ger
undive nominal complements can be complements of action.
Some Assumptions
In this study I will be making certain assumptions
bearing on the structures and rules I will discuss. First
of ail, I will assume that grammatical rules apply in a
fixed order and cyclically. The cyclic application of rules
means that the full set of rules would first apply in order
to the deepest embedded sentence, S?, in 1.12,
1.12
S
0
S
1
S
2


TABLE OF CONTENTS
Page
ACKNOWLEDGMENTS iv
ABSTRACT vii
CHAPTER
IINTRODUCTION 1
Statement of Purpose ..... 1
The Lexical Hypothesis 3
Derived Nominis as Complements on Nouns ... 5
Some Assumptions 9
Outline 11
NOTES 13
IIDEFINING DERIVED NOMINAL COMPLEMENTS . 15
First Definitions 15
The Surface Structure of Derived Nominal
Complements 16
Action Nominis 25
Agentive Nominis 42
Summary 44
NOTES 45
IIITHE PRODUCTIVITY OF DERIVED NOMINAL COMPLEMENTS 48
The Evidence for Lack of Productivity .... 48
Chomsky's Subcategorization Account 51
RAISING Rules 55
v


CHAPTER FIVE
CONCLUSIONS
In this study I have shown that derived nominal comple
ments,. when properly defined, have a regular, if limited,
productivity. The fact that derived nominal complements
have an internal structure like that of noun phrases while
other types of complements have an internal structure more
like, that of sentences is shown to have a simple explanation.
Ordering DERIVED NOMINALIZATION before most cyclical rules
allows a simple explanation of the facts of derived nominal
complements.
However, the simple explanation I have presented here
V
is not really possible in the framework assumed by Chomsky.
In particular, I have assumed underlying structures more
abstract than those in Chomsky's theory, including the
predicate-initial analysis of underlying structures before
the application of cyclic rules. While, as Chomsky said,
the question of whether or not derived nominal complements
can be described in a transformational framework is an
empirical one, the decision as to how abstract the underly
ing analysis is to be is not empirical. By choosing a more
abstract analysis, I have been able to handle derived nomi
nal complements In a transformational framework.
143