JUN 4 1965
THE GRAMMAR OF ENGLISH PREDICATE COMPLEMENT CONSTRUCTIONS
by
PETER STEVEN ROSENBAUM
A. B., Wesleyan University
(1962)
SUBMITTED IN PARTIAL FULFILLMENT OF THE
REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
at the
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
June, 1965
Signature of Author
Department of Modern Languages, May 14, 1965
Certified by
Thesis Supervisor
Accepted by wi"ut-y
Chairman, Departmental Committee on Graduate Students
M ITLibraries
Document Services
Room 14-0551
77 Massachusetts Avenue
Cambridge, MA 02139
Ph: 617.253.5668 Fax: 617.253.1690
Email: docs@mit.edu
http://libraries.mit.edu/docs
DISCLAIMER OF QUALITY
Due to the condition of the original material, there are unavoidable
flaws in this reproduction. We have made every effort possible to
provide you with the best copy available. If you are dissatisfied with
this product and find it unusable, please contact Document Services as
soon as possible.
Thank you.
Due to the poor quality of the original document, there is
some spotting or background shading in this document.
THE GRAMMAR OF ENGLISH PREDICATE COMPLEMENT CONSTRUCTIONS
by
PETER STEVEN ROSENBAUM
Submitted to the Department of Modern Languages on May 14, 1965
in partial fulfillment of the requirements for the degree of
Doctor of Philosophy.
ABSTRACT
A set of phrase structure rules and a set of transformational
rules are proposed for which the claim is made that these rules
enumerate the underlying and derived sentential structures which
exemplify two productive classes of sentential embedding in
English. These are sentential embedding in noun phrases and
sentential embedding in verb phrases.
First, following a statement of the grammatical rules,
the phrase structure rules are analyzed and defended.
Second, the transformational rules which map the underlying
structures generated by the phrase structure rules onto appro-
priate derived structures are justified with respect to noun
phrase and verb phrase complementation.
Finally, a brief treatment is offered for the extension
of the proposed descriptive apparatus to noun phrase and verb
phrase complementation in predicate adjectival constructions.
Thesis Supervisor: Noam Chomsky
Title: Professor of Modern Languages
Go - lo
A CKNOWLE DGEMENT
The author wishes to express his gratitude to
Professors Noam Chomsky and Paul Postal, Department of
Modern Languages, who, giving freely of their time, influ-'
enced virtually every aspect of this thesis. The author
has also profited considerably from many discussions of
English grammar with Professor Edward Klima, Department
of Modern Languages, and students in the Department of
Modern Languages including Barbara Hall, Bruce Fraser,
Thomas Bever, D. Terence Langendoen, John Ross, and Arnold
Zwicky. The author is also grateful to the U. S. Office
of Education for providing the author with Fellowship
funds, under Title IV of the National Defense Education
Act, which enabled him to pursue graduate study in lin-
guistics at the Massachusetts Institute of Technology
and to the IBM Corporation for its support of the authort s
preliminary study of predicate complementation in English
during the summer months of 1964.
TABLE OF CONTENTS
Page
CHAPTER 1.
A.
B.
CHAPTER 2.
A.
B.
CHAPTER 3.
CHAPTER 4.
A.
B.
C.
CHAPTER 5.
A.
B.
CHAPTER 6.
CHAPTER 7.
APPENDIX.
The Results of the Inquiry. . . . .
Phrase Structure Rules. . . . . .
Transformational Rules. . . . . .
A Defense of the Phrase Structure Rules
Noun Phrase Complementation . . ... .
Verb Phrase Complementation . . . .. .
Complementizing Morphemes and Their
Introduction into the Underlying
Structure . . . . . . . . . . . .*.*..
Noun Phrase Complementation . . . . .
Object Complementation. . . . . .
Subject Complemeitation . . . . .
Oblique Complementation . . . . . .
Verb Phrase Complementation . . . .
Intransitive Verbs. . . . . . . . .
Transitive Verbs . . . . . . . . .
Complementation in Adjectival Con-
structions . . . . . . . . . . . .0..*
A Historical Perspective. . . . . . .
Verb Classifications . . . . . . . .
Biography . . . . . . . . . . . . . .
.4
.4
.9
.15
.15
.25
41
59
60
126
145
167
167
171
180
197
220
227
CHAPTER I
The Results of the Inquiry
The aim of the present study is to develop an adequate
framework for describing certain types of sentential comple-
mentation in English. In particular, this study deals with
instances where sentences are embedded in noun phrases (hence-
forth noun p rase complementation) and in verb phrases (verb
2hasecomplementationX In terms of the theory of syntax
developed by Chomsky in his Aspects of the Theory of Syntax1
the descriptive apparatus postulated to explain noun phrase
and verb phrase complementation consists of 1) a set of phrase
structure re-writing rules which generate underlying sentence
structures and 2) a set of transformational rules which map
underlying structures onto new derived structures. The
various considerations brought to bear in this study lead
to the conclusion that an adequate description of noun phrase
and verb phrase complementation contains the phrase structure
and transformational rules which are summarized in the
following pages.
I. Phrase Structure Rules
Operating in conjunction with two basic rules which
expand S (sentence) and PD? (predicate phrase) into NP, AUX,
PDP and VP, ADV respectively, the two phrase structure rules
which are central to the complement systems under discussion
can be stated as follows;
(PP) SPS Rule 1 VP---->V (NP) (PP) p
PS Rule 2 NP----> DET N. (S)
The two phrase structure rules above allow for the generation
of a variety of underlying structures. Most pertinent to the
present study are the following:2
(1)
N P
VP (ADY)
V S
The above underlying structure, an instance of intransi-
tive verb phrase complementation, is generated as the result
of the particular application of PS Rule 1 which expands VP
into V, S.
'-6-
(2)
N P P
Vr (ADY)
V NP
The application of PS Rule 1, in which VP is expanded
as V, NP, S, yields instances of transitive verbphase com-
plementation.
(3)
N PP
V (ADV)
KS
Where VP is expanded into V, PP -, PS Rule 1 generates
instances of oblique verb phrase complementation.
NV (ADV)
N
DST N 6
The underlying structure above, an instance of object
noun phrase complementation, requires the application of PS
Rule 1 through which VP is expanded into V, NP, and the appli-
cation of PS Rule 2 to this NP, (henceforth the underlying
object NP).
(5)
(ADv)
PREP NP
DST N S
--8--
When PS' Rule 2 is applied to the NP produced through
the application of PS Rule 2 which expands VP into V, PP,
the grammar generates instances of intransitive oblique
napasegnlemenalon, as in (5) above.
(6)
(ADV)
NP P
DET N S PREP NP
DET N S
The expansion of VP into V, NP, PP permits the generation
of instances of transitive oblique noun phrase complementation.
In this case, the underlying object NP may be expanded by PS
Rule 2 along with the NP dominated by PP.
(7)"
p p
DST a (.ADV)
-9-
Since the phrtse structure expanding S yields NP, AUX,
VP, PS Rule 2 may apply to the NP in this configuration (hence-
forth the underlying subject NP) to generate instances of sub-
ect noun phrase complementation.
II. Transformational Rules
PS Rules 1 and 2 suffice to generate the most central
structures underlying noun phrase and verb phrase complementa-
tion (predicate complementation collectively) in English. The
transformational rules necessary to the generation of appropri-
ate derived structures are given below. These rules are
strictly ordered, apply cyclically, and are obligatory unless
otherwise marked.
1. Complementizer Placement Transformation -- TCP
A. X N CNP + Y] Z
(-D)
1 2 3 4---->
1,2,[-13+3,4
B. X N NP + PDP Z
F+ D[EJ
1 2 3 -- >
1,2, +3,4
C. X V (NP) NP + PDP Z
+1
1 2 3 4-
1,02,3tE]+4,35
-10-
V
D. x+D NP have + Y
EJbe
1 2 3 4WWWW>
1,2,3, 2 + 4
2. Identity Erasure Transformation -- TIE
(NP) X +D NP Y (NP) z
I 2 3 4 5 6 7 8
(i)3 5 is erased by 2
(11) 5 is erased by7
The following conditions (henceforth the erasureprinciple)
govern the application of the identity erasure transformation.
A NP is erased by an identical NP if and only if there is
a Sa such that
(i) NP. is dominated by S
(ii) NP neither dominates nor is dominated by $a
(iii) for all NPk neither dominating nor dominated by Sa,
the distance between NP. and NP is greater than3jIk
the distance between NP and NP where the distance
between two nodes is defined in terms of the number
of branches in the path connecting them.
3. Subject-Object Inversion Transformation --
X NP AUX V NP S Y
[+soi)
1 2 3 4 5 6 7auam>
1,5,3,4,6, to+2,7
-11 -
4. Passive Transformation -- T4
X NP AUX V (PREP) NP
1 2 3 4 5 6
1,6,3,4,5,$,7,2,9
5. Extraposition Transformation 
- TE
X N S Y
+PRO
NOT LD
1
1,2, p, 4+3
2 3
(usually
by + P
7 8
optional)
y
9--n>
(usually optional)
4-=u.>
6. Optional Complementizer Deletion Transformation 
-- TocD
(usually optional)
X [v a. N (+D p) P Y
(.ADJ) C+PRO]
b. (NP) C-DJ3
1 2
1,2,3,,5,6
3 4 5 6Sm.->
47. Auxiliary Transformation-- IwTAUX
X
1
1,,3+2,4
v
32
y
4=-m>
(in particular where
Af is E+E] and v is
V, have, be, or NP)
- 12-
8. Pronoun Replacement Transformation 
-- TPR
X N (AUX (MAN))
E+PRO) be ADJ
NP L+PR]
1 2 3 4 5
1,7,93,4,5,6,, 8
9. Pronoun Deletion Transformation T P
[+DJ ]NP
6 7
X N
E+PRO)
1 2
1,,3,4,5
a.
b. ADV)
3
S Y (a. has a few exceptions )
(b. is usually optional )
4 53nsa>
In addition to the transformational rules defined above,
there are several others which are partially ordered. The most
essential of these are the preposition deletion transformation
which must precede the pronoun replacement transformation and
the obligatory complementizer deletion transformation which
must follow the pronoun replacement transformation.
10. Preposition Deletion Transformation m- TPPD
X
I
PREP
2
N
+PRO
y
4smn->
y
8=-m->
- 13-
11. Obligatory Complementizer Deletion Transformation 
-- TCD
X [+] VP Y
1 2 3 4s>
1,0,3,4
The discussion to follow is concerned with exploring the
justification of both the rules' and their ordering in the
description of predicate complementation in English. In the
course of this discussion, it will prove necessary to refer to
various additional transformation rules. These rules are not
listed here in part because the present study is not of
sufficiently broad scope to allow for a precise formulation
of these rules and in part because, in comparison with the
rules stated above, these additional rules are not critically
pertinent to the primary goal of this study: to establish a
general framework for describing predicate complementation in
English.
- I
NOTES
1. N. Chomsky, Aspects of the Theory ofSntax (in press).
2. In the discussion to follow the constituent AUX will be
excluded from consideration. Although the behavior of
the auxiliary, including at least Tense and Modal, is a
pertinent dimension of the predicate complement system,
the details of this behavior and its description go beyond
the range of this study. It is unlikely, however, that
a deeper study of the auxiliary in predicate complement
constructions will result in a significant alteration
of the general outlines of the predicate complement
system as developed in this study.
3. When a constituent A is erased by a constituent B,
Assns>$ just in case A and B meet the conditions stipu-
lated by the erasure principle.
4. For a detailed study of the passive transformation, see
J. B. Fraser, "The Passive Construction in English,"
(forthcoming).
5. For discussion of the auxiliary transformation, see N.
Chomsky, Syntactic Structures, The Hague, 1957.
~Ir~
CHAPTER 2
A Defense of the Phrase Structure Rules
In the previous chapter, two phrase structure rules were
shown to enumerate a variety of underlying phrase structures
containing the constituent S. These rules constitute an
assertion that either NP or VP may optionally dominate S in
the underlying phrase structure representation for an infinite
set of derivations. Underlying the formulation of these rules
lies the belief that the laws or rules governing allowable
sequences of words in any natural language can be given in a
simple manner and the empirical hypothesis that the simplest
formulation of the rules best characterizes the varied in-
stances of human linguistic competence. The purpose of this
chapter is to show that diminished simplicity accompanies
analyses of predicate complementation which do not postulate
the recursion of S under the immediate domination of both NP
and VP.
Noun Phrase Complementation
Consider the following pair of sentences which are tradi-
tionally described as active (la) and passive (lb).
(1) a. the little boy took the book
b. the book was taken by the little boy
-15-
There is hardly a simpler formulation of the passive trans-
formation than that according to which the noun phrases
preceding and following the main verb of a sentence are in-
verted with a concurrent insertion of the passive morphemes
be+en and 12. Any speaker of English will attest that it is
just this transformational process (in the literal sense)
which relates sentences (Ia) and (lb). Chemsky has shown,
furthermore, that a passive transformation of this general
form follows as a logical consequence from a general theory
of language seeking to explain the linguistic abilities
possessed by normal speakers of a language in some systematic
fashion.1
But the generality of the passive transformation in its
usual formulation, i.e., roughly the formulation given in
Chapter 1, is at least superficially questionable with re-
spect to certain observed phenomena of which sentences (2a)
and (2b) are instances.
(2) a. Columbus demonstrated that the world is not flat
b. that the world is not flat was demonstrated by
Columbus
It is as easy to believe that the phrase "that the world is
not flat" is an instance of the constituent S as it is to
believe that it is an instance of the constituent NP. For
instance, phrases of this sort contain the constituent
structure common to other instances of S. Under the appropriate
conditions such phrases can be passivized, e.g., I think that
John hit the ball===u>I think thattheball wasjhitby John.
It seems only good sense to assert that any linguistic descrip -
tion which does not postulate that this phrase is an instance
of S at some level of derivation could not achieve empirical
adequacy. But it is not a priori obvious that successful
linguistic descriptions must identify this phrase as an NP
at some level of derivation.
Let us look more carefully at the passive transformation
in terms of the most conservative analysis of the phrase "that
the world is not flat," where this phrase is analyzed solely
as an instance of S. Applying the same considerations which
led us to relate(la) and (lb), one observes that a new formu-
lation of the passive transformation is required according to
which a noun phrase preceding a verb and a sentence following
the verb are inverted. The pair of sentences in (3) suggests
an additional modification by which a sentence preceding a
verb and a noun phrase following the verb are inverted.2
(3) a. that the doctor came at all surprised me
b. I was surprised that the doctor came at all
On the assumption that the phrases presented in the examples
above are instances of S, one is forced to a more elaborate
formulation of the passive transformation, (4), in which
sentences, in addition to noun phrases, can be inverted.
- 18-
(4) X NP AUX V (PREP) P by+P Y
1 2 3 4 5 6 7 8 9====-
1,6,3,be+en,4,5,0,7,- t,9
The diminished generality of the passive transformation
results not from the fact that phrases like "that the world
is flat" must be analyzed as instances of S, but from the
assumption that such phrases are not also instances of NP.
For if these phrases are assumed to be dominated by NP in the
structures underlying (2) and (3), then the original, more
general formulation of the passive transformation is seen to
be entirely adequate. Allowing that S as well as N can be
dominated by NP, (the formulation given by PS Rule 2 in
Chapter 1), the original passive transformation will effect-
ively generate the sentences in (1), (2), and (3). This is
a consequence of the generalization that both N and S may be
instances of NP.
The simpler statement of the passive transformation, made
possible by the assumption that NP may dominate S, receives
powerful support from the fact that the contrary assumption
is simply not confirmed by observation. For instance, English
contains a great many sentences, e.g., (5), in which infinitival
constructions, which must be analyzed as instances of S at
some level, never undergo passivization.
-I
(5) a. 1) 1 tended to think slowly
2) *to think slowly was tended by me
b. 1) John needs to go home early
2) *to go home early is needed by John
c. 1) she began to cry
2) *to cry was begun by her
Observations of this sort indicate that persistence in the
assumption that S cannot be dominated by NP obliges us to pro-
poss further modifications in addition to a more complex state-
ment of the passive transformation. In particular, it becomes
necessary to include in the description a variety of apparently
unsystematic restrictions on the application of the transforma-
tion in (4) so as to prevent the occurrence of sentences (5,a2,
b2, c2). If, on the other hand, NP can dominate S, then we
can view the sentences in (5) as constructions which are in-
stances of something other than noun phrase complementation,
that is, as constructions whose underlying structures are quite
distinct from those in (2) and (3). If NP dominates S in (2)
and (3), but not in (5), then the ungrammaticality of the
sentences (5a2, b2, c2) is an automatic consequence since the
passive transformation in its most general form applies only
to NP's (dominating either N or S) and not to an S which is
not dominated by NP.
The assumption that an NP can dominate S is instrumental
in explaining a great many observed regularities in complement
F 011_m - - - IMMIN I I
--19-
-20-
constructions which are superficially unrelated, that is,
unrelated on the basis of simple inspection. The fact that
the contrary assumption, discarded above for independent
reasons, does not lead to an account of these regularities
provides further support for the hypothesis being proposed
here. By way of example, consider the following pairs of
sentences.
(6) a. 1) Columbus demonstrated that the world is not flat
2) that the world is not flat was demonstrated by
Columbus
3) it was demonstrated by Columbus that the world
is not flat
b. 1) that the doctor came at all surprises me
2) it surprises me that the doctor came at all
For much the same reason that led us to relate the sentences
in (2), or the sentences in (3) for that matter, to a common
underlying form, it is also necessary to relate sentence (6al)
to (6a2,3) and the sentence (6bl) to (6b2). Careful observa-
tion indicates that there is no phrase, e.g., "that the doctor
came at all," which appears before verb phrases like "surprised
me" which cannot appear following the verb phrase. To assume
that the two sentences are not derived from a common source
is to diminish simplicity greatly. We shall return to a pro-
posal for capturing this generalization shortly.
A reconsideration of PS Rule 2, as given in Chapter 1,
reveals that the expansion of NP into the constituents DET
and N is obligatory. The expansion into DET N S is optional.
If the foregoing considerations are correct, if the phrase
"that the doctor came at all" is an S dominated immediately
by NP, then one may question the formulation of PS Rule 2
since no N is observed preceding the phrase in (6b1). One
might consider taking this fact as evidence that the expan-
sion of NP into DET N is optional also. There is one
general fact, however, which advises us that such a conclu-
sion is in error. The sentences in (6) show that the pronoun
fit" may appear in sentence initial position just in case the
"that" phrase appears at the end of the sentence. Further-
more, the pronoun may not appear when the "that" phrase is
in sentence initial position, as -sentence (7) indicates.
(7) *it that the doctor came at all surprised me
These facts are convincingly explained on the assumption that
it is the pronoun which determines the application of a trans-
formational rule which moves the "that" clause to the end of
the sentence. Thus the structure underlying the sentences in
(6b) might be formalized in terms of the following phrase
structure representation.3
(8)
NP PDP
14E S.VP
+PR N
N
surprise
that the doctor came at all
The phrase structure configuration in (8) asserts that
NP has been expanded into a N which carries the pronominal
feature E+PRO] and a complement sentence S. We postulate
that the sentences (6b1) and (6b2) are derived through the
application of one of two extremely general rules defined
upon this configuration. The first of these rules is the
extraposition transformation, defined in Chapter 1, accord-
ing to which an S following a pronoun is extraposed to the
end of the string, thereby yielding (6b2). A second trans-
4
formation, the pronoun deletion transformation, defined in
Chapter 1, deletes the pronoun when it occurs immediately
before S, thus producing sentence (6b1). It is clear that
the sentences in (6a) can be similarly derived.5
m -
Further support for PS Rule 2 is provided by the fact
that both of the transformations just discussed, TE and T,
which are required to explain the relatedness of (6bl) and
(6b2), are independently motivated in other instances. Con-'
sider, for example, the following pair of sentences.
(9) a. I would like for you to be there very much
b. I would like very much for you to be there
On the assumption that the phrase "for you to be there" is an
instance of noun phrase complementation described by PS Rule 2,
the relation of (9a) to (9b) becomes an automatic consequence
of T E, according to which the phrase is extraposed to the end
of the string.
The conclusion that PS Rule 2 is valid is supported by
yet another consideration, specifically, the limitations on the
occurrence oflpseudo-cleft" sentences. The problem posed by
these sentences is that they are not predictable simply from
the inspection of non-cleft sentences. Consider, by way of
illustration, the differences between the following pairs of
sentences.
(10) a. 1) I hate you to do things like that
2) what I hate is for you to do things like that
b. 1) we prefer you to stay right here
2) what we prefer is for you to stay right here
00239M
Wo 10, 1, 1 -_,
c. I) I defy you to do things like that
2) *what I defy is for you to do things like that
d. 1) we tempted you to stay right here
2) *what we tempted was for you to stay right here
Eschewing the details of wh attachment and pseudo-cleft sentence
derivation, we may nonetheless suppose that the derivation of
these sentences depends upon a noun phrase complement structure.
If this conjecture is correct, then the grammaticality of (10a2,
b2) implies that the sentences (10al, bl) are instances of noun
phrase complementation where the phrases "you to do things like
that" and "you to stay right here" are what remains of an under-
lying structure in which these phrases were generated as Sts
under the domination of NP. Independent proof of this claim
stems from the possible extraposition of these phrases as in (11).
(11) a. I hate very much for you to do things like that
b. I prefer very much for you to stay right here
Since the sentences (10c2, d2) are ungrammatical, our hypothesis
leads us to the conclusion that the sentences (10cl, dl) are
not instances of noun phrase complementation, a prediction
which is borne out by the fact that extraposition is impossible
6for these sentences.
(12) a. *I defy very much for you to do things like that
b. *we tempted very much for you to stay right here
I___________
If the hypothesis that the pseudo-cleft sentences dis-
cussed above depend upon the application of PS Rule 2,
producing noun phrase complements, is true, the grammar
thus predicts that the derivation of the pseudo-cleft sentence
will be possible for the constructions mentioned earlier where
the noun phrase complement analysis was proposed. The fact
that the following derivatives of the sentences in (6) are
grammatical confirms this prediction.
(13) a. what Columbus demonstrated was that the world is
not flat
b. what was demonstrated by Columbus was that the
world is not flat
(14) a. what surprised me was that the doctor came at all
b. what I was surprised at was that the doctor came
at all
Verb Phrase Complementation
Our acceptance of PS Rule 1, which variously expands VP
into S, requires 1) a demonstration that PS Rule 2 is inappro-
priate for certain constructions and 2) a valid argument that
PS Rule 1 is .the most general formulation which adequately
accounts for the residue of cases not handled by PS Rule 2.
Toward these ends, consider the following data.
-25-
(15)
The paradigm (15a) exhibits the properties which are gener-
ally associated with noun phrase complementation, i.e., the
passivization of the entire complement sentence ( in this case
"to remain silent") and the occurrence of the pseudo-cleft
sentence. These properties are not, however, observed in the
second paradigm (15b).
The current formulation of transformational theory allows
at least two devices which are sufficiently powerful to prevent
the generation of the ungrammatical sentences in (15b). The
first alternative postulates that (15a) and (15b) have similar
underlying structures but differ in the specification of' the
transformational rules which apply to this structure. In
other words, the verb " tend" is marked in such a way that
both the passive transformation and these transformations
which are instrumental in the derivation of the pseudo-cleft
sentence are not applicable. The second alternative is to
a. I) everyone preferred to remain silent
2) to remain silent was preferred by everyone
3) what everyone preferred was to remain silent
b. 1) John tended to play with his little brother
often
2) *to play with his little brother often was
tended by John
3) *what John tended was to play with his little
brother often
-26-
-2>-
claim that (15b) is simply not an instance of noun phrase
complementation at all, but rather a paradigm whose underlying
form is something like (1) in Chapter 1. In the latter analysis,
the verb "tend" is also marked, but not for the application
of transformations. Rather, the verb has a strict subcate-
gorization marker7 which permits its occurrence before the
constituent S. In this description, the ungrammaticality of
(15b2) and(15b3) follows from the fact that neither the passive
transformation nor the pseudo-cleft sentence transformations
apply to an S which is not immediately dominated by NP.
Of the two alternatives, considerations of optimal
generality suggest that the second, which assumes (13a) and
(15b) to have different underlying structures, seems to be
preferable. In either analysis, a strict subcategorization
marker on the verb is required. In the first analysis, this
marker stipulates that the verb "tend" must occur immediately
preceding a NP which must itself dominate S. In the second
analysis, the strict subcategorization marker simply requires
the verb "tend" to occur immediately preceding the constituent
S. Thus, the first bit of unnecessary complexity which the
first analysis requires is the specification that the NP
following the verb must dominate S. This requirement follows
from the fact that the verb "tend" is otherwise intransitive.8
Since the second analysis asserts that the verb may occur only
before 8, the ungrammaticality of (16) is an automatic conse-
quence and requires no additional feature specification.
(16) a. *I tended the ball
b. *I tended something
Even if it were the case that the two analyses were equally
simple with respect to strict subcategorization markers, we
should prefer the second analysis over the first on yet other
grounds. In the first analysis, where it is assumed that the
verb "tend" occurs immediately preceding a NP, it becomes
necessary to specify a set of transformations which such a
construction cannot undergo. In particular, such a verb must
be marked for the non-application of the passive transformation
and the pseudo-cleft sentence transformations. The relative
simplicity with which these restrictions can be marked on
verbs is not at issue here. The crucial point is that in
the second analysis, which assumes "tend" to be followed by S
in the underlying structure, such a feature specification is
totally unnecessary in the first place since the transforma-
tions under discussion do not apply to an S which is not
dominated immediately by the constituent NP. Thus the first
analysis, treating (15b) as an instance of noun phrase comple-
mentation, fails on two counts. First, this analysis requires
an unnecessarily complex strict subcategorization statement.
Second, it requires a fairly elaborate set of restrictions on
allowable transformations which is totally unnecessary if it
is hypothesized that (1b) is not an instance of noun phrase
complementation.
-
I
It is thus a necessary conclusion that there are a number
of complement constructions which cannot be instances of noun
phrase complementation. That these constructions are most
fruitfully analyzed as instances of verb phrase complementa-
tion follows from various considerations on the operation of
the identity erasure transformation, TBI, as given in Chapter 1.
In the noun phrase complement system, the deletion of the sub-
ject of the complement sentence is often obligatory when it
is identical to the subject of the main or including sentence.
Consider the following cases.
(17) a. I love Bill to play the piano
b. I love to play the piano
c. *1 love me to play the piano
On the basis of this evidence alone, the simplest formulation
of the identity erasure transformation would stipulate that
the subject of the complement sentence is obligatorily deleted
just in case it is identical to the first noun phrase to the
left of the complement sentence in the main sentence. This
statement is consistent not only for many instances of noun
phrase complementation, but also for all instances of. the type
of complementation exemplified by (15bl). Furthermore, this
formulation holds for transitive verbs9 which take the type
of complement found in the same example, as is illustrated
in the following instances.
--
A
-30-
(18) a. they tempted John to leave early
b. we forced John to ignore his work
The above formulation collapses in the face of certain
adverbial constructions representing a degenerate form of the
prepositional locution "in order to."
(19) a. I sold the boat (in order) to save money
b. she took the car (in order) to buy bread
In these cases the implicit subject of the adverbial sentence
is not the first noun phrase to the left in each case, but
rather the second noun phrase to the left. The problem gets
even more complex when we observe instances where identity to
the right is apparently required. Consider, for instance, the
following sentences.
(20) a. can you expect it of him to do what is right always
b. I absolutely require it of you to be here on time
One will readily appreciate that the phrases "it to do what is
right" and "it to be here on timel" are noun phrase complements
and that the sentences in (20) assume their form by virtue of
the application of the extraposition transformation. The noun
phrase to which the subject of the complement sentence must be
identical lies to the right of the complement sentence, thus
irreparably damaging the left identity hypothesis.
U
NP
DET N S
EPRO
love 3 play pianoI
All of these difficulties may be resolved by the adoption
of an extremely general principle governing the application
of the transformation which deletes the subject of complement
constructions. This principle becomes immediately apparent
upon examination of the phrase structures underlying (17),
(19), and (20).
(21)
PDP
mII
-32-
(22
N P PR
DET N S
VP f+PRO
D T VP1 7p
I sell the boat in order I save money
(23)
S
NP tDP
DU N PREP NP
NP
I require you be here on time of you
-33-.
The above phrase structures suggest that an NP. can be
erased by the identity erasure transformation just in case
there is some sentence S. (a complement sentence) such that
1) NP is dominated by Sa 2) NPi neither dominates nor is
dominated by S., and 3) for any NPk which neither dominates nor
is dominated by SQ the distance between NP. and NPk is greater
than the distance between NP . and NP where the distance be-
tween two nodes is defined in terms of the number of branches
in the path connecting them. In (21) and (22), the "erasing"
NPts are the subject NPts of the main sentence. In (23), the
erasing NP is the NP, "you, " in the prepositional phrase of
the main sentence. In all three phrase structures, the
terased"NP happens to be the underlying subject of the com-
plement sentence. The identity erasure transformation
asserts only that the erased NP must be the one which follows
the complementizing morphemes "for" and "POSS," thus correctly
allowing the identity erasure transformation and the erasure
principle which governs its application to range over derived
subjects, as in cases where the passive transformation has
applied to the complement sentence. As shall be seen in the
discussion to foflow, the rasure principle accounts for most
cases of identity erasure in English and has only one ex-
ception. The fact that the connection principle applies with
such remarkable precision to so many cases suggests that the
latter may indeed be a false counter-example, a possibility
which is attested by considerations to be raised later.
Returning to the question of whether the complement con-
structions in (18) and (15b) are instances of verb phrase
complementatinn, we find that the erasure principle provides
a ready answer. On the supposition that the complement
sentences in (18) lie outside the verb phrase, but under the
10domination of predicate phrase, PDP, the erasure principle
breaks down. This follows from the fact that, if the comple-
ment is not dominated by the verb phrase, then the subject NP
of the complement sentence is ambiguously connected with
respect to the main sentence. This is readily seen in terms
of the hypothetical phrase structure (24), where the subject
NP of the complement sentence is equidistant from the subject
and object of the main sentence.
(24)
N PP
ADVP
V NP PREP N
DET N S
+N
+PROJ
NP
tempt John John leave early
The situation becomes worse for sentences like (25) which
have the underlying structure (26),
(25) 1 prevail upon John to go
(26)
N PP
VIP ADVP
V PP PREP NP
D:PT S
"ST .1 S
PREP NP
+N
+-PR]
NVP
I prevail upon John John go
If the erasure principle holds in the case of (26), the grammar
will always make the false prediction that the implicit subject
I of "to go" could be "I" rather than "John" if the subject of the
r(27) b. S
N PDP
PREP NNP VP
prevail upon John John go
am ~
-36-
complement sentence had been "I." If, on the other hand, we
postulate that the complement sentence is a verb phrase com-
plement, derived through the application of PS Rule 1, then
the erasure principle operates correctly. This result is ob-
served in the following phrase structure diagrams where the
structures underlying (18a) and (25) are assumed to exemplify
verb phrase complementation.
(27) a. S
N P
VP
they tempt John John 1eave early
In (27a), the erasing NP is correctly the object NP of the
main sentence. In (27b) the erasing NP is the NP "John" in
the prepositional phrase. Thus, the grammar predicts the
implicit subject of the complement sentences in (27a, b) to
be "John" rather than "they" and "I" respectively.
The cost of not positing the complement sentences "to
leave early" and "to go" to be instances of verb phrase com-
plementation in (27) is the cost of being unable to generalize
the identity erasure transformation. Without assuming verb
phrase complementation, the principle governing identity
erasure would, in any case, become so complex as to be un-
serviceable. Since the analysis of verb phrase complementa-
tion is a consequence of the most general formulation of
erasure principle, the necessity of adopting such a con-
struct as the verb phrase complement seems unavoidable.
The two types of complementation under discussion in
this chapter by no means exhaust the complementation systems
of English. Other kinds of sentential embedding are quite
common. Among these are relativization, as exemplified by
(31), and subordination of various types (32).
(31) the man who arrived yesterday left today
(32) a. she laughed because the joke was funny
b. the glass broke when I threw it against the wall
c. being late to dinner, I got nothing to eat.
U
-M37-
solo
-38-
While it is not at all certain that the complementation
represented by (31) and (32) share no common properties
with noun phrase and verb phrase complementation as defined
by phrase structure Rules 1 and 2, it is nonetheless clear
that noun phrase and verb phrase complementation presuppose
a fairly intricate set of rules which seem at present to play
but a small role in other complement systems. The remainder
of the present work will be devoted to the study of the kinds
of rules which are required to handle noun phrase and verb
phrase complementation (henceforth collectively referred
to as predicate complementation) and the way in which these
rules are implemented in the grammar.
NOTES
1. Cf. Noam Chomsky, SyntacticStructures, The Hague, 1957,
especially Chapter 7; for further discussion of the passive
transformation, see R. B. Lees, The Grammar of English Nominali-
zations, Bloomington, 1960; J. J. Katz and P. M. Postal, An
Integrated Theory of Linguistic Descriptions, Cambridge, 164;
Cbomsky, Aspects of the theory of Syntax. Cambridge, 1965;
and J. B. Fraser, "The Passive Construction in English"
(forthcoming).
2. The derivation of the sentence (3b) includes essentially
two steps. The application of the passive transformation to
the structure underlying (3b) will yield the string "I be+en
surprise by that the doctor came at all.l" The agentive prepo-
sition "by" is subsequently deleted by the application of a
general preposition deletion transformation discussed in
Chapter 4.
3. The constituent PDP is expanded into VP and an optional
Adverbial which usually takes the form of a prepositional
phrase, as in time and place adverbials, e.g., "at dawn" or
"in the woods." Since the present study does not deal with
these adverbials, the constitutat PDP will, for the most part,
dominate VP directly.
4. The present study assumes that the pronoun "it" in
sentences like (6a3) and (6b2) is introduced into the under-
lying structure by the same rules which introduce lexical items
into the underlying structure. The grammar thus requires a
transformational rule deleting this pronoun in the appropriate
environments. Since no transformational rules are formulated
in terms of the phonological matrix of this pronoun, it is
entirely conceivable that the phonological matrix for the
pronoun is introduced by ordered "spelling" rules perhaps of
following form which apply after the transformational rules.
a. PNa, [:~JW...> i t{[
NP
b. it---- >$/ ADV (optional)
5. The derivation of sentence (6a1) does not presuppose the
application of the passive transformation whereas this trans-
formation is requisite to the derivation of sentences (6a2,3).
-4o-
6. It is reasonable to argue that the sentences in (12) are
blocked by virtue of the fact that the adverbial "very much"
is inappropriate for verbs like "defy" and "tempt" in certain
dialects of English. In such dialects, sentences like the
following are ungrammatical.
*1 defy you to do things like that very much
*we tempted you to stay right here very much
But this counter-argument collapses when suitable adverbials
are employed, as is seen in the following pairs.
I defied him to buy a house consistently
*1 defied consistently for him to buy a house
we tempted him to stay home cleverly
*we tempted cleverly for him to stay home
7. Cf. Chemsky, Aspects of the Theory of Syntax, for dis-
cussion of subcategorization.
8. The intransitive verb "tend" must, of course, be distin-
guished from the transitive homonymous verb "tend," as in
"the shepherd tended the flock." It is the former which is
ungrammatical in the sentences in (16).
9. The assumption that the verbs in (18) are transitive
follows from two considerations, one of which has not yet been
discussed. First, the phrases "John to leave early" and "John
to ignore his work" are not instances of noun phrase comple-
mentation. If they were, we should expect sentences like
*"what they tempted was for John to leave early" and *"what
we forced was for John to ignore his work" to be grammatical.
Second, identity erasure is obligatory for all instances of
verb phrase complementation. Thus the grammaticality of the
sentences in (18) implies that "John" in both of these sentences
originates as an object noun phrase.
10. The assumption that complement sentences like "to leave
early" in (24) are under the domination of the PDP is, of
course, more conservative than the assumption that this sentence
is under the domination of SP, the subordinate adverbial phrase,
which is immediately dominated by S. Since the erasure princi-
ple fails, however, even for the most conservative assumption,
there is little reason to investigate other possible analyses
for the sentences in (18).
11. Sentence (28) is actually ambiguous in three ways if one
views the phrase "to answer all questions" as an instance of
subordination as in "they relied upon his intellectual pro-
pensity in order to answer all questions.
~u I
-41-
CHAPTER 3
Complementing Morphemes and Their Introduction
Into the Underlying Structure
One of the properties of predicate complements which dis-
tinguishes them from other types of complements is a unique
set of markers taking the form of single and paired morphemes.
Such markers, including the morphemes that, for, to, POSS, ing,
and others,1 shall be referred to as complementizing morphemes
or simply coMplementizers. Instances of the complementizers
which this study will deal with are given below.
(1) 1 think that Fords are too expensive
(2) I dislike argujl.Z about silly matters
(3) 1 am concerned about Johnt s being so lazy
(4) the king ordered the proclamation to be read
(5) I should like very much for you to reconsider your refusal
Certain mutual inclusions and exclusions in the set of comple-
mentizers exemplified in (1)(5) are immediately apparent. The
complementizer "for" co-occurs only with the complementizer
"to.I" The complementizer "POSS co-occurs only with "ing."
The complementizer "that" occurs alone, never with either "ing"
or "to." Sentences like the following are impossible in English.
U
-42-
(6) *1 anticipated John's to argue with me
(7) *1 can't stand for John being late to supper
(8) a. *I think that John to be late
b. *I think that John being late
It may thus prove convenient to speak of the "that" comple-
mentizer, the "for-to" complementizer, and the "POSS-ing"
complementizer.
There are two in some sense distinct questions which one
might raise with respect to the complementizers. First, what
are the considerations which are involved in determining the
way in which the complementizers are introduced into various
predicate complement constructions? Second, assuming that
the descriptive constructs devised with respect to the first
question achieve a reasonable level of adequacy, what factors
must be taken into consideration in describing the behavior
of the complementizers after they have been introduced into
the underlying structure of a predicate complement construction?
The present study deals primarily with the latter question
although, as shall be seen in the following pages, certain
as yet unproven aspects of complementizer introduction are
presupposed.
Implicit in the term "complementizer" is the idea that
these morphemes are a function of predicate complementation
and not the property of any particular sentence or set of
sentences. Thus there is no structure underlying any declara-
tive sentence in English which cannot, in some other derivation,
be the structure underlying a predicate complement sentence.
This would seem to imply that the rule or set of rules employed
in the introduction of a complementizer into a predicate com-
plement construction must be either a context-sensitive re-
writing rule or a rule with transformational power since it
is necessary to account for the observation that main sentences,
whibh are not themselves predicate complement sentences, as
well as other types of complement sentences do not contain
complementisers of the type under discussion. For instance,
strings like those in (6) are not sentences in English.
(6) a. *that John came early
b. *for John to have done it
c. *John's having done it
The fact that there is some evidence supporting the view
that context-sensitive re-writing rules are unnecessary baggage
in the syntactic component of the grammar does not, however,
imply that complementizer introduction is to be a transforma-
tional process entirely. Viewing a transformational rule as
a "filter, "2 it 'is entirely possible that the complementizers
are derived in the underlying structure through the operation
of context-free re-writing rules. The application of such
rules would thus provide an object which a transformation can
interpret and mark as either well-formed or not well-formed.
The present formulation of the theory, therefore, allows either
~a I
-44-
for the phrase structure introduction of the complementizers
or for a completely transformational introduction. In the ab-
sence of compelling evidence for accepting one formulation over
the other, and since it is probably true that the selection of
either alternative will not effect the discussion of the opera-
tions defined over complementizers one way or the other, the
clarity of the following exposition will profit by arbitrarily
adopting the transformational alternative simply because this
option is probably the most familiar. But it should be kept
in mind that, although this option offers a description of
complementizer introduction, there is as yet little evidence to
suggest that this description is the right one.
There are at least three considerations which play a role
in the introduction of complementizers into the underlying
structure. The first concerns the classification of the com-
plementizers and the notation in which this classification is
framed. A cursory glance at the list of transformational rules
in Chapter 1 reveals that several transformations are sensitive
either to a "for-to" complementizer or to a "POSS-ing" comple-
mentizer. In most cases, the "that" complementizer exemplifies
properties markedly different from the remaining two comple-
mentizers. This generalization can be captured if, in the
grammar, it is more expensive, i.e., less simple, to refer to
just the "for-to" complementizer or to just the "POSS-ing"
complementizer than it is to refer to both together. Toward
- U1,1I II pl I
I] is realized morphemically as
ol
+D is realized morphemically as
+i
+D is realized morphemically as
+IE
"POSS-ing"
This notation provides precisely the correct generaliza-
tion. Should a rule in the grammar have to refer either to
"for-to"or "POSSing" separately, the structural description
will have to list at least three features, namely, (8) or (9),
for the "for-to" and "POSS-ing" complementizers respectively.
(8)
(9)
+C
+DI
j+E
+D6.
On the other hand, should it be necessary to refer to both
complementizers together, the structural description needs to
contain only two features, (10) , since it does not matter
whether the feature "E" is marked "+" or 1-".
this end, we might propose a binary feature hierarchy for
classifying the complementizers where this hierarchy contains
the following redundancy rules.
(7) +OC->E+nT
ftD}---->*13
(10) [ +:[+DI
Furthermore, reference to all three complementizers requires
only one feature, namely, E+], since it does not matter here
whether the feature "D" is "+" or
On the additional assumption that the complementizing
features comprise a feature system distinct from other features
which might be postulated for lexical items, an assumption
which will effect this presentation trivially even if it is
false, it becomes possible to simplify the representation of
the complementizers still further.
(11) 1. that, for-to, POSS-ing +]
2. that 1-D1
3. for-to, POSS-ing [+D)
Keeping this discussion in mind, we turn to the second
issue which plays a role in the formulation of the transforma-
tional apparatus introducing complementizers into the under-
lying structure. This issue concerns the statement of the
restrictions holding between the main sentence and the com-
plementizer of the predicate complement sentence. Inspection
of the following verb phrase complement data readily indicates
that the choice of the complementizer in the verb phrase com-
plement is dependent upon the verb in the main sentence.
low4 Y
(12) a. they prevailed upon me to help out
b. *they prevailed upon me that I help out
c. *they prevailed upon me helping out
(13) a. the noise forced me to stop working
b. *the noise forced me that I stop working
C. *the noise forced me stopping working
There are few, if any, apparent counter-examples to the empiri-
cal claim that the "that" complementizer never functions as
the complementizer of a verb phrase complement.3 The issue
is not so clear cut with respect to the "POSS-ing" comple-
mentizer. One observes, for instance, the following sentences.
(14) a. we heard him running down the street
b. I imagined myself eating at the Ritz
It may be argued, perhaps, that what appears to be an instance
of the "POSS-ing" complementizer in the above complement
constructions is simply a degenerate form of the "for-to" com-
plementizer where the "to" is deleted before the progressive
morphemes "be-ing." On this view, the constructions in (15)
are earlier stages in the derivation of (14).
(15) a. *we heard him to be running down the street
b. I imagined myself to be eating at the Ritz
Although such an analysis appeals to the semantic in-
tuition that the activities defined in the complement sentence
are, in some sense, on-going, the fact that the progressive
morphemes do not occur freely in the complement sentence pro-
vides reasonable evidence for the view that verb phrase comple-
ments may have "POSS-ing" as well as "for-to" complementizers.
Consider, as an illustration, the following sentences.
(16) a. I imagined myself owning a mansion
b. *I am owning a mansion
Since the verb "own" cannot take the progressive form in
general, it is difficult to see how the phrase t"owning a house,
in (16a), could be an instance of the progressive. The verb
fown" is not, of course, restricted with respect to the "POSS-
ing" complementizer, as we observe in the following noun
phrase complement construction.
(17) I dislike their owning such a big car
On the other hand, constructions such as (18) seem to re-
quire a description in which "ing" is, in Fillmore's sense,
a telescoped Progressive4
(18) a. I felt the rope slip
b. I felt the rope slipping
-49-
Since the complementizers seem, in general, not to effect
the semantic interpretation of the complement sentence, it
becomes difficult to explain the difference in meaning between
(18a) and (18b) on the assumption that the "ing" in (18b) is
the complementizer "ing. But this meaning difference would
be accounted for on the assumption that "ing" in (18b) is the
progressive "ing. ,"5
The question of whether or not there are instances of
the "POSS-ing" complementizer in verb phrase complement con-
structions is, in a certain sense, irrelevant to a successful
demonstration that there are restrictions between the verb in
the main sentence and the complementizer in the verb complement
sentence since there is little question but that the "that"
complementizer is so restricted.6 Should it turn out that the
"POSS-ing" complementizer never occurs in verb complement con-
structions, this will mean simply that whatever mechanism
prevents the generation of the complementizer "that" in such
constructions will also prohibit the generation of "POSS-ing"
In the present discussion, it will be assumed that the "POSS-
ing" complementizer does, in fact, appear in verb phrase com-
plement constructions since this complementizer seems to be
necessary for intransitive verb phrase complement constructions.
To say that the selection of either the "for-to" or the
"POSS-ing" complementizer in a verb phrase complement construc-
tion is dependent upon the verb in the main sentence is to imply
a descriptive apparatus which posits that verbs are marked with
particular complementizer features in the lexicon that may
appear in verb phrase complements of the verb in question.
We propose, in other words, that verbs contain features in-
dicating which complementizer is possible in a coordinate verb
phrase complement. In terms of the notation suggested earlier,
this means that a verb may be marked with the features [+D]
[+E] if the verb may have a "POSS-ing" complementizer, with
the features [+D][-E if the verb may have a "for-to" com-
plementizer, and with the feature t D] in the event that the
verb may take either the "POSS-ing" or the "for-to" comple-
mentizer. Furthermore, should a verb be found which takes
all three complementizers, an apparently unlikely event in
a verb phrase complement construction, this will be marked
simply [+].
The complementizer placement transformation, case C,
precludes the possibility of the complementizer "that" being
generated in a verb phrase complement by stipulating that all
verb which take verb phrase complement constructions are marked
at least (+D). The variable "a," ranging over the feature
coefficients, is introduced into the transformation in order
to insure that the coefficient of the feature "E" in the verb
is identical to the coefficiebt of the introduced complement-
izers. Case "1" of this transformation states, for instance,
that if the verb in the main sentence is marked with the
features [+D]-E], then the features [+D])-E] are intro-
duced into the verb phrase complement sentence first
WF
ir
-50-
phrase complement (henceforth the head of the noun phrase
Svowaso
r -51-
immediately preceding an initial noun phrase under the domination
of S. Case D of this transformation asserts that whatever
features are introduced preceding the initial noun phrase of
the complement sentence by cases B and C are duplicated immedi-
ately preceding V3 have, or be under the domination of VP.
Subsequently, these feature clusters are realized morpho-
phonemically as "to" before V, have or be and as "for" elsewhere,
or as "ing" before V, have or be and as "POSS" elsflhere.
Turning to noun phrase complementation, observation reveals
that complementizer selection is somewhat broader, in the sense
that the complementizer "that" is a possibility, but that there
are restrictions nonetheless. Consider, for example, the
following sentences.
(19) a. I think that John will be late
b. *1 think John's being late
(20) a. I want you to hurry home
b. *I want that you hurry home
(21) a. I relish owning catamarans
b. *I relish that I own catamarans
On the surface there seems little doubt that the restrictions
upon complementizer selection in noun phrase complement con-
structions can be handled in a fashion similar to verb phrase
complementizer restrictions. But a question arises as to
whether it is the noun in the noun phrase containing the noun
-52-
complement construction) which specifies the restrictions or,
once again, the verb in the main sentence. The latter alterna-
tive 'has a certain appeal which stems from the fact that if
the former alternative is taken, it will still be necessary to
mark the verb with the same information for purposes of strict
subclassification. In other words, even if a particular com-
plementizer is determined by the head of the noun phrase coma
plement construction, verbs must be marked according to their
capacity to appear either before or after such a complement
construction. Thus the maximum generality would appear to be
preserved if the restrictions are marked on the verb.
Despite its superficial desirability, the specification of
noun phrase complementizer restrictions in the verb of the main
sentence proves to be inadequate on the basis of considerations
which will be examined in some detail later. Briefly, the ob-
jection is as follows: It is necessary to postulate comple-
mentizer features on pronoun heads of noun phrase complement
constructions since the simplest possible formulation of the
preposition deletion transformation is defined over a pronoun
marked either with the feature t-D] or with the feature (-E].
If this hypothesis is valid, then the supposition that the
verb determines the restrictions on noun phrase complementizers
becomes improbable since it allows for the generation of un-
grammatical strings. Imagine the following situation: A "that"
complementizer is selected for the noun phrase complement in
some prepositional phrase within the verb phrase on the basis
of the feature C-D] in the verb of the main sentence. At the
same time, a pronoun in the noun phrase complement construction
is generated with the features f+DJ C[+C. The grammar will
subsequently produce a sentence containing a noun phrase com-
plement with a "that" complementizer before which the intro-
duced proposition will not have been deleted, as in (22).
(22) *1 decided on that John will represent us.
The resulting ungrammatical sentence is a function of the fact
that there is no dependency established between the complementizing
features on the verb and those on the pronoun. The most readily
apparent solution is to make use of the strict categorization
mechanism to insure that verbp and the relevant constituents
are marked for the same set of features. But once this is ac-
complished, the original feature specification, designed for the
verbal determination of complementizer restrictions in noun
phrase complements, becomes redundant since the features on the
pronominal head of the noun phrase complement construction may
be employed in the same capacity. Thus the greatest generality
is achieved if we postulate that the head of the noun phrase
complement is the constituent which specifies the features upon
which complementizer selection for noun phrase complements is
determined. For these reasons the complementizer placement
transformation in Chapter 1 has two cases, a noun case and a
verb case. In the final analysis, there is probably an
adjective case as well, but this topic will be the focus of
much later discussion in Chapter 6.
A final issue which must be taken into account by any
complementiter introduction device is the fact that, under
certain conditions, complementizers preclude the occurrence
of tense and modals. More specifically, modals do not have
a privilege of occurrence in predicate complement constructions
containing either the "for-to" complementizer or the "POSS-ing"
complementizer. For instance, sentences like the following
are ungrammatical.
(23) a. *1 asked John to will hit the ball
b. John will hit the ball
(24) a. *I dislike John's can playing the piano
b. John can play the piano
On the other hand, there appears to be no comparable modal re-
striction on predicate complement sentences containing the
"that" complementizer, a claim supported by the following data.
(25) a. I think that John will hit the ball
b. I suppose that John can play the piano
The modal restriction can be incorporated into the com-
plementizer placement transformation fairly easily, but the
ma * ~-~-----
< C,
.. 1 .00
solution provided by the complementizer placement transforma-
tion in Chapter 1 has its limitations and must be considered
as a rough correspondence to the facts at best, What is not
explained by the formulation of the complementizer placement
transformation postulated for the present work is the fact
that quite often a particular modal interpretation is impli-
cit in a complement construction containing either the "for-
to" or "POSS-ing" complementizer. Consider, for instance,
the following sentences.
(26) a. I expect that John will go
b. I expect John to go
(27) a. I anticipate that John will not want to leave
b. I anticipate Johnfs not wanting to leave
(28) is it possible for John to leave early
The modal "will" is, in some sense at least, an implicit
aspect of the interpretation of the complement sentences in
(26b) and (27b). Similarly, the modal "can" (on one of its
readings at least) is an aspect of the interpretation of the
complement sentence in (28). On the other hand, in the great
majority of the predicate complement constructions there is
no obvious modal interpretation whatever. These hazy facts
suggest that a certain difficulty may await the apparatus
formulated to account for the gross modal exclusion with
r -56-
the "for-to" and "POSS-ing" complementizers, Since all
syntactic and lexical material necessary for the semantic
interpretation of a sentence is included in the underlying
S
structure, we are forced to believe that the modal inter-
pretation of (26b, 27b, 28) stems either from the actual
existence of the modal in the underlying structure of the
predicate complement sentence or, more likely, from some
special, idiosyncratic feature of particular verbs for which
modal interpretation is necessary. There is, as yet, little
evidence on which to base an evaluation of the two alterna-
tives. But, in the event that the first alternative comes
to be favored, it will, of course, become necessary to revise
the complementizer placement transformation significantly.
Perhaps this fact itself is a hint that the second alterna-
tive will turn out to be correct.
NOTES
14, A second major class of complementizers with which
this study does not deal includes the wh complementizer
as in the following cases:
a) I dislike it when you do that
b) I often wonder (about) why he does these things
c) I know where he went
d) everyone understands how he does it
e) what he is doing is useless
Also functioning as complementizing morphemes are if
and whether, as in the following sentences:
a) I doubt if he is going
b) I wonder whether he is going
2. N. Chomsky, Aspects of the Theory of Syntax (in press).
3. A possible counter-example is the following sentence
with the verb "pretend.
I pretended that I was a pilot
The fact that the "that" clause neither passivizes nor
participates in the pseudo-cleft sentence with the
verb "pretend" suggests that this phrase may be a
verb phrase complement.
*that I was a pilot was pretended by me
*what I pretended was that I was a pilot
These data are hardly reliable, however, since there are
dialects of English which apparently accept the pseudo-
cleft sentence.
4+. C. J. Fillmore, "The Position of Embedding Transformations
in a Grammar,o" Report No. 2r, Project on Linguistic
Analysis, Columbus (1963).
5. This observation was suggested to the author by Edward
S. Klima.
6. It should be pointed out in this respect that there
is reason to believe that sentence (14b) is not an
instance of verb phrase complementation in the first
place. These considerations are raised in Chapter 4.
It may thus be the case that transitive verb phrase
complementation allows only the "for-to t complementizer.
CHAPTER FOUR
NOUN PHRASE COMPLEMENTATION
The discussion in Chapters 2 and 3 indicates (1) that
an adequate linguistic description of English syntax must
posit noun phrase complement structures having the properties
characterized by phrase structure Rule 2 and (2) that such
complement constructions are marked by complementizing
morphemes whose privileges of occurrence are specified by
the complementizer placement transformation, TCP.
The present chapter deals with three distinct instances
of noun phrase complementation. We shall be concerned with
two instances of this construct which are characterized by
phrase structure Rule 1, where either the object of the main
verb or the noun phrase in the prepositional phrase following
the main verb may dominate a noun phrase complement construc-
tion. Furthermore, we shall also study the instance of noun
phrase complementation when the dominating noun phrase is the
underlying subject of a sentence, a construction produced by
the re-writing rule which yields NP, AUX, VP on the basis of
the symbol S. Since the transformational rules governing
-60-
these cases are, in part, dependent upon complementizer
selection, (the properties of the various complementizers.
being in some ways distinct in noun phrase complement con-
structions), this investigation ranges actually over nine
cases and not over three.
I. Object Complementation
The term "object complementation" is a mnemonic makin
reference to the instance of noun phrase complementation which
arises through the expansion of VP at least into V, NP. The
application of phrase structure Rule 2 subsequently yields
the string V, DET, N, S. And it is to this particular con-
figuration that attention is directed in the discussion of the
properties of object complementation.
A. The "that" complementizer
It was mentioned in an earlier discussion that noun
phrase complements may contain one of three complementizers
depending upon the particular verb in the main sentence. In
this section, we shall study the transformational rules which
are required in the event that the complementizer determined
by the main verb is the complementizer "that.'" Toward this
end, consider the following paradigm.1
(1) a. they doubt it that you will go
b. they doubt that you will go
c. they doubt you will go
d. they doubt it very much that you will go
e they doubt very much that you will go
f. that you will go is doubted by them
g. it is doubted by them that you will go
In terms of phrase structure Rules 1 and 2, the structure which
all of the sentences in (1) share is specified as follows:
(2)
$
NP PDP
VP
y P
DET N S
+NI NP PDP
AUX V'P
MV
they doubt you will go
-62-
The relatedness of (if) and (Ig) is partially reflected,
therefore, in the fact that both sentences have a common under-
lying source, namely (2). But this fact in itself does not
tell us very much about the derivation of (If) and (1g) since
there are at least two analyses based upon the common under-
lying structure which, at least superficially, yield equally
simple derivations of the two sentences. To settle this issue,
it is necessary to take into consideration other factors be-
sides the common underlying structure. In the first analysis,
the two sentences share only the underlying structure. In the
second analysis, the two sentences not -only share the under-
lying structure, but also an intermediate stage of the
derivation in the sense that both sentences are the result of
certain transformations applying to a string which is itself
the product of a transformational rule having been applied to
the underlying structure. Since so much ultimately depends
upon the decision made here, it will be fruitful to study
these alternatives in greater detail.
Both analyses presuppose the application of the passive
transformation, TP, as formulated in Chapter 1. But here the
similarity ends. In the first analysis, sentence (If) results
from having deleted the pronominal head of the complement
construction prior to the application of the passive trans-
formation. The passive transformation thus applies to a,
string of roughly the following structure.
I(3) they doubt [[that you will go] ]
S NP
Since the passive transformation applies to any constituent
NP such that this constituent conforms to the structural
description specified in Chapter 1, it follows that this
transformation will apply to (3) producing the string ulti-
mately realized as sentence (if). In the same analysis,
sentence (Ig) results from not having deleted the pronominal
head prior to the application of the passive transformation.
According to this consideration, the passive transformation
applies to a string (4a) thus producing the string (4b).2
(4) a. they doubt [[it] Lthat you will go] ]
N S NP
b. CHit] Ethat you will go] ] is doubted by them
N S NP
Subsequently, the extraposition transformation, T, as defined
in Chapter 1, applies to the string (4b) separating the noun
phrase complement from the pronominal head and producing the
string (5) which coincides with (1g).
(5) Cit] ] is doubted by them Ethat you will go]
N NP S
Thus the first analysis differentiates (If) from (Ig) in terms
of the relative ordering of the transformation deleting the
pronominal head of the noun phrase complement construction
and the passive transformation. The former precedes the
latter in the derivation of (If). The latter precedes the
-64-
former in the derivation of (1g). Furthermore, this deriva-
tion of (Ig) presupposes that the extraposition transformation
follows the passive transformation in application. If the
contrary were true, then we should have no mechanism capable
of producing the string (5) after the passive transformation
gives us (4b). The ordering of the passive transformation
with respect to extraposition is also a feature of the second
alternative analysis which we now explore.
In the second analysis, the derivations of (If) and (1g)
both depend upon the application of the passive transformation
to the underlying structure prior to the deletion of the pro-
nominal head of the noun phrase complement sentence. In other
words, the string (4b) is an intermediate stage in the deriva-
tion of thO two sentences. The subsequent derivation of (lf)
depends upon the deletion of the pronominal head, generating
the string (6).
(6) [[that you will go] ] is doubted by them
S NP
The derivation of (1g), in this analysis, is identical to the
derivation proposed in the first analysis, namely, through
the application of the extraposition transformation. The
differentiation of (If) and (Ig) here has nothing to do with
ordering, as was the case with the first alternative examined.
Rather the differentiation is the function of having applied
one of two transformations, the transformation deleting the
-65-
pronominal head of the complement construction, (1f), or
the extraposition transformation, (1g).
Although both formulations seem to be adequate with
respect to the data thus far examined, the consequences of
these formulations are not identical with respect to other
aspects of the object complement system. More specifically,
consider the following sentences (8) in which the object com-
plement is followed by a preposition phrase in the underlying
structure (7).
(7) CNP AUX V DET N S] PREP NP] J
NP PP S
(8) a. nobody expected it of John that he could be so cruel
b. nobody expected of John that he could be so cruel
c. *nobody expected that he could be so cruel of John
What these sentences indicate is that the pronominal head of
the complement construction cannot, in this particular instance,
be deleted before the application of the extraposition trans-
formation since such deletion permits the derivation of (Sc).
But this fact presents grave difficulties for the first
formulation when considered with regard to the possible
passive versions of the sentences in (8).
(9) a. it wasn't expected of John that he could be so cruel
b. that he could be so cruel wasn't expected of John
-66-
The derivation of sentence (9a) is perfectly straightforward.
The passive transformation, which we already know must precede
the extraposition transformation in order of application,
applies to the string (10) to yield the string (11) upon which
the extraposition transformation subsequently applies to pro-
duce the string (12).
(10) nobody expected (it (that he could be so cruel] 3
S NP
(of John]
PP
(11) (it (that he could be so cruel] ] was expected (of
S NP
John] by nobody
PP
(12) (it] was expected (of John] by nobody (that he could
NP PP
be so cruel]
S
The major problem arises with respect to the derivation of (9b).
If it is the case that extraposition is obligatory for noun
phrase complements in the object position of this construction,
and if extraposition is dependent upon the existence of the
pronominal head of the complement construction, and if the
extraposition transformation follows the passive transformation,
it appears that the pronominal head cannot be deleted prior
to the application of the passive transformation. For if
this deletion were to occur prior to the application of the
Si
passive transformation, which we recall is optional, and if
the passive was subsequently not applied, the grammar should
then have generated the ungrammatical sentence (8c).
What all of this means is that the source of the passive
sentence (9b) could not have been simply the phrase "that he
could be so cruel.1" Rather, since the pronominal head cannot
be deleted prior to the application of the passive transforma-
tion, this passive sentence must have originated from the
structure formalized as (it (that John could be so cruel]
SNP
In other words, the derivation of (9b) must follow the second
formulation given earlier in which the first two stages of the
derivation are identical to those specified in (10) and (11).
But in this case, the extraposition transformation is not ap-
plied to the output, thereby yielding (9a); rather the pro-
nominal head of the complement construction is deleted,
correctly giving (9b). What this demonstrates is that the
first alternative formulation of the rules for characterizing
the derivation of (If) and (lg) does not generalize to other
data which the grammar must cover. The second formulation
succeeds where the first fails and, for this reason, is to be
preferred. Consequently, we can specify the partial deriva-
tion of the sentences (If) and (1g) as follows :2
r(13) Sentence (If)
First Cycle -- no operations
they doubt [[it] [you will go] ] [by + P] BASE
N S NP MAN
Second Cycle
they doubt [[it] [that you will go] 3 [by + P] TOP
N SNP MAN 
[[it] [that you will go] ] be+en doubt [by + them] T
N S NP MAN
[[it] [that you will go] 3 be doubt + en [by + them] TTAUX
N S NP MAN
[[that you will go] ] be doubt + en [by + them] TPDS NP MAN
Post Cycle
[[that you will go] is doubted [by + them] N
S NP MAN
(14) Sentence (1g)
First Cycle -- no operations
they doubt [[it] [you will go] [by + P] BASE
N S NP MAN
Second Cycle
they doubt [[it] [that you will go] 3 [by + P] TCP
S NP MAN
[[it] [that you will go] 3 be + en doubt [by + them] T.
N S NP MAN
[it] 3 be + en doubt [by + them] [that you will go] TE
NP MAN S
[it] 3 be doubt + en [by + them] (that you will go] TAUX
N NP MAN S
Post Cycle
[[it] ] is doubted [by + them] [that you will go] M
N NP MAN S
-68-
-69-
Directing attention now to the optional deletion of the
complementizer "that," we find evidence which suggests that
any rule proposed to account for this phenomenon must follow
the extraposition transformation. Consider the following data.
(15) a. 1) it is strange that John isn't here
2) it is strange John isntt here
b. 1) it happens that John discovered the same
thing yesterday
2) it happens John discovered the same thing
yes terday
On the supposition that a "that" deletion transformation pre-
cedes the extraposition transformation, we should have to
account for the fact that if this rule does apply, the ex-
traposition transformation is always obligatory if the comple-
ment construction is in sentence initial position. Consider,
for example, the following sentences.
(16) a. 1) that you will find him is doubted by them
2) it is doubted by them that you will find him
3) *you will find him is doubted by -them
b. 1) that you were broke was known by everybody
2) it was known by everybody that you were broke
3) *you were broke was known by everybody
To explain the non-occurrence of sentences (16a3, 16b3)
requires, first of all, that the extraposition transformation
be divided into an optional case and an obligatory case and,
second, that the obligatory version be sensitive to more
structure than simply the contiguity of the pronominal head
of the complement construction with the complement sentence
itself. In particular, this statement must list information
about the constituency of the complement sentence, namely,
that the complementizer "that" is no longer in the structure.
There are many difficulties connected with this proposal
which require no elaboration here since the greatly increased
cost of this formulation over the one in which a "that" de-
letion transformation follows the extraposition transformation
is sufficient to exclude this proposal from further consideration.
Although the ordering of the "that" deletion transformation
(henceforth case "b" of the optional complementizer deletion
transformation, TC), following the extraposition transforma-
tion is necessary to the explanation of the non-occurrence of
the ungrammatical sentences in (16), the precise formulation
of this transformation presents certain difficulties. Con-
sider, for instance, the following sentences.
(17) a. 1) I doubt that John came yesterday quite seriously
2) 1 doubt John came yesterday quite seriously
3) I doubt quite seriously that John came yesterday
4) *1 doubt quite seriously John came yesterday
-71-
b. 1) nobody expected (it) of John that he could
be so cruel
2) *nobody expected (it) of John he could be so
cruel
c. 1) it is important to John that you are here on
time
2) *it is important to John you are here on time
d. 1) 1 convinced Bill that John was not so bad
2) I convinced Bill John was not so bad
These data suggest that the deletion of the complementizer
"that" may be restricted just to those instances where this
complementizer immediately follows the verb with a NP inter-
vening optionally, or the predicate adjective in the main
sentence. According to this restriction, the grammar would
correctly predict the deletion of the "that" in the following
instances.
(18) a. 1) I think that John is coming
2) I think John is coming
b. 1) it is strange that John came late
2) it is strange John came late
Certain dialectal phenomena suggest that the transforma-
tion deleting indefinite pronouns, which has been discussed
in the literaturek but not listed in Chapter 1, must follow
r
the extraposition transformation but must precede Case "b"
of TOCD. In these dialects, the following deletions of the
complementizer "that" are encountered.
(19) a. 1)
a)
3)
4t)
b. 1)
a)
3)
+)
it
*it
it
it
it
*it
it
it
was
was
was
was
is
is
is
is
thought
thought
thought
thought
importani
importani
importani
importan
by someone that you would come
by someone you would come
that you would come
you would come
t to someone that you do it
t to someone you do it
t that you do it
t you do it
The sentences (19a2, b2) are ungrammatical and are prevented
by the fact that Case "b" of TOCD is not defined on these
strings. Since the appropriate environment is not met in
these sentences, the deletion of the complementizer "that" is
impossible. However, if the transformation which deletes the
indefinite "someone" in (19) is ordered to apply prior to the
application of case "b" of T ,OC, then the "that" complementizer,
now contiguous with the verb or predicate adjective, can be
deleted optionally.
In the past few pages we have been exploring the form and
order of application of the transformational rules required to
generate the sentences in the paradigm (1). Specifically, we
have seen that, aside from the complementizer placement
-73-
transformation, T& ifive transformations are necessary.
These are the passive transformation, the extraposition
transformation, the pronoun deletion transformation, Case "b'
of the optional complementizer deletion transformation, and
the indefinite pronoun deletion transformation. It has been
proposed, furthermore, that four of these rules are critically
ordered as follows:
1. Passive Transformation -- T
t Extraposition Transformation 
-- TE
3. Indefinite Pronoun Deletion Transformation -- unlisted
4. Optional Complementizer Deletion Transformation --
TOD (Case "b")
The last issue to be discussed before the derivation of
the sentences in the paradigm (1) can be fully specified con-
cerns the ordering of the transformation deleting the pro-
nominal head (henceforth to be identified as the pronoun
deletion transformation as defined in Chapter 1). The central
question has to do with whether the pronoun deletion trans-
formation precedes or follows the extraposition transformation.
Two options present themselves. First, pronoun deletion is
optional and precedes extraposition which is obligatory.
Second, pronoun deletion is obligatory and follows extra-
position which is optional. There is evidence which suggests
that the second alternative is preferable to the first, but
let us consider both alternatives in slightly greater detail.
*ot7 40
The first alternative can be illuminated in terms of
the following data.
(20) a. 1) that you could do such a thing bothers me
2). it bothers me that you could do such a thing
b. 1) 1 didn't suspect that you would fail for a
moment
2) 1 didn't suspect for a moment that you would
fail
In the first analysis, the non-application of the optional
pronoun deletion transformation produces the sentences (20a2,
b2), since the extraposition transformation is obligatory. If
the former rule does apply, then the sentences (20al, bl) will
be generated since the necessary environment for the applica-
tion of the extraposition transformation is not satisfied.
The central difficulty with this analysis stems from the fact
that the extraposition transformation may apply vacuously,
5
for instance, in a base string like (21).
(21) I think [Eit] Ethat John is leaving] )
N S NP
Since the variable Y in the extraposition transformation as
stated in Chapter 1 may be null, it follows that the extra-
position transformation, which we recall is obligatory in
this analysis, will apply with a resulting alteration of the
V
, c -
constituent structure. In particular, the vacuous application
of the extraposition transformation will take the complement
sentence from under the.domination of the NP and attach it
to some higher constituent, perhaps S. We observe, however,
that the application of the extraposition transformation has
no effect on the linear sequence of the constituents in the
structural description if the variable Y is null. Since it
still remains necessary, therefore, to delete the pronominal
head, the first atiysis must include an additional pronominal
deletion rule just to cover the cases in which the extra-
position transformation applies vacuously.
The necessity of positing two pronominal deletion trans-
formations for exactly the same purpose is obviated in the
second formulation in which an optional extraposition trans-
formation precedes an obligatory pronoun deletion transforma-
tion. In this analysis, the non-application of the extra-
position transformation necessitates the application of the
pronoun deletion transformation thus generating the sentences
(20al, bl). If the extraposition transformation does apply,
then the sentences (20a2, b2) are correctly generated.
Finally, in the event that the extraposition transformation
applies vacuously, as in (21), the pronoun deletion trans-
formation applies obligatorily to yield (22).
(22) I think that John is leaving
-76-
We thus conclude that the- pronoun deletion transformation
is obligatory ana follows the extraposition transformation
which is optional.
There is a special case of pronoun deletion which is
optional in many instances. Consider, for example, the
following sentences.
(23) a. 1) I dislike it very much that he is always late
2) I dislike very much that he is always late
b. 1) 1 didn't suspect it for a moment that you
would fail
2) 1 didn't suspect for a moment that you wouLd
fail
c. 1) 1 believe it to be true that oculists are
eye-doctors
2) *1 believe to be true that oculists are eye-
doctors
Although there are many unanswered questions about the restric-
tions on pronominal deletions of the sort exemplified by the
sentences in (23), it appears that it is necessary to postu-
late a variation of the pronoun deletion transformation which
is occasionally optional and which allows for the deletion of
the pronominal head just in case an adverbial intercedes be-
tween the pronoun and the complement sentence. This is case
"b" of TPD as defined in Chapter 1.
-77-
a'
There is little evidence which allows us to decide
whether the pronoun deletion transformation, which follows
the extraposition transformation, must precede or follow case
"b" of TOCD. There seems to be no particular problem with
postulating that the deletion of the complementizer "that"
precedes the pronoun deletion transformation. In such a formu-
lation, we observe that the deletion of "that" depends upon
the vacuous application of the extraposition transformation.
We observe that case "b" of TOCD is not defined upon strings
like (24).
(24) a. I think Lit] Lthat John is coming] ]
N SNP
b. I imagined [[it] Lthat my boat sank in the hurri-
N
cane] ]
3 NP
This follows from the fact that the pronominal head is not a
NP, but a N. The application of the extraposition transforma-
tion to these strings results, however, in an environment upon
which case "b" of T is defined.
(25) a. I think L[it] ] Ethat John is coming]
N NP S
b. I imagined L[it] ] Cthat my boat sank in the
N NP
hurricane]
S
The basic virtue of this analysis is that, as we shall
see in the following discussion, it becomes possible to
collapse, albeit only partially, the "that" deletion rule
with with the rules which optionally delete "for" and "POSS."
But this is anything but conclusive justification for the
above analysis.
The order required for the application of the transforma-
tions discussed thus far is as follows:
1. Passive Transformation -- T
2. Extraposition Transformation 
-- TE
3. Indefinite Pronoun Deletion Transformation -- unlisted
4. Optional Complementizer Deletion Transformation --
TOCDb
5. Pronoun Deletion Transformation 
-- TPD
On the basis of these transformations, plus the auxiliary
transformation, the derivation of the sentences (la)-(le) in
the paradigm may be specified in the following manner.
(26) Sentence (1a)
First Cycle -- no operations
they doubt Etit] [you will go] ) BASE
N S NP
Second Cycle
they doubt [Lit] Ethat you will go] TCP
N S NP
(27) Sentence (lb)
First Cycle -- no operations
they doubt C[it ]you will go ] BASE
N S NP
-79-
Second Cycle
they doubt LLit] Ethat you will go] TCP
N S NP
they doubt [[that you will go] ] T
S N P
(28) Sentence (1c)
First Cycle -- no operations
they doubt LUit) Lyou will go ] BASE
N S NP
Second Cycle
they doubt L[it] ] that you will go] 3 T
N SNP C?
they doubt LLit) ] Ethat you will go] TE
N NP S
they doubt Cit] ) you will go] T
N NP s 00Db
they doubt [you will go] T
SFD
(29) Sentence (ld)
First Cycle -- no operations
they doubt [[it] you will go] 3 Lvery much] BASE
N S NP ADV
Second Cycle
they doubt CLit] that you will go]) [very much] TCP
N SNP ADV
they doubt [[it] ] very much] Ethat you will go] TE
N NP ADV S
(The derivation of sentence (le) requires the application of
I case "b" of TFD to generate the following string.)
-80-
they doubt every much] Lthat you will go] TPDb.
ADV S
B. The "POSS-ing" complementizer
Object complement constructions containing the "POSS-ing"
complementizer are similar in certain respects to those con-
structions containing the "that" complementizer, but are
sufficiently different nonetheless to warrant consideration.
Compared with paradigm (1), the paradigm for the "POSS-ing"
object complement constructions is quite impoverished.
(30) a. *everybody prefers it your driving slowly
b. everybody prefers your driving slowly
c. everybody prefers you driving slowly
d. *everybody prefers it very much your driving slowly
e. *everybody prefers very much your driving slowly
f. your driving slowly is preferred by everybody
g. *it is preferred by everybody your driving slowly
One of the more striking aspects of the restrictions on
the sentences in this paradigm is the fact that the pronominal
head of the object complement cannot occur. This should not
be taken as evidence that these sentences are not instances
of noun phrase complementation. We are advised of the spuri-
ousness of this conclusion by the existence of the pseudo-
cleft sentences in (31).
(31) a. what everybody prefers is your driving slowly
b. what is preferred by everybody is your driving
slowly
It would seem more correct to say that the pronominal head of
the complement construction is obligatorily deleted just in
case the complementizer in the complement sentences happens
to be "POSS-ing."
In addition to the fact that the deletion of the pro-
nominal head is obligatory in such constructions, it is also
observed that extraposition is, in general, impossible, a fact
exemplified by (SOd,e,g). This suggests that if the extra-
position transformation is so restricted that it cannot apply
in the event that the complementizer of the complement
sentence following the pronominal head is "POSS-ing," then
the obligatory deletion of the pronominal head becomes an
automatic consequence of the pronoun deletion transformation.
In other words, by preventing the application of the extra-
position transformation, we bring about the obligatory deletion
of the pronoun. Thus, the string (30a) must become the string
(30b).
In addition to the non-extraposition and obligatory pro-
noun deletion which we observe in (29), we observe that the
"POSS" segment of the "POSS-ing" complementizer can be
optionally deleted, as in (29c). Superficially, this
deletion would appear describable as a special instance of
case "b" of the optional complementizer deletion transforma-
tion, T which asserts that the "that" complementizer can
OCD 
.
be optionally deleted when it follows the verb or predicate
adjective immediately. Certain other considerations suggest
that this simplification is specious. Consider what the
following examples demonstrate.
(32) a. 1) I cenvlnc6d Mary that he was honest
2) 1 convinced Mary he was honest
b. 1) I convinced Mary of his being honest
2) *1 convinced Mary of him being honest
We recall that case "b" of TOCD is defined over a verb or
adjective followed by an optional noun phrase followed by
"that." If the data in (32) is representative, it is observed
that the deletion of the complementizer "FOSS" is impossible
in the event that a NP intervenes between the complementizer
and the verb. This suggests the necessity of establishing a
second case of the optional complementizer deletion transforma-
tion in order to handle the "FOSS" complementizer. In other
words, we should establish a two-case transformation like the
following.
(33) X V a. ( a. +D Y
A5J b. (NP)2+,
b.+D
1 2 3 4 5-=-->1,2,3,#,5
-83-
Looking ahead to the "for-to" object complement
constructions, we discover that case "a" in (32) can be
significantly simplified. Consider the following sentences.
(34) a. 1) I would hate for John to lose it
2) I would hate John to lose it
b. 1) I would hate very much for John to lose it
2) *1 would hate very much John to lose it
c. 1) for you to stay here would be impossible
2) *you to stay here would be impossible
d. 1) it was important for you to do that
2) *it was important you to do that that
e. 1) I was embarrassed for you to see the mess
2) *1 was embarrassed you to see the mess
f. 1) preferring for John to leave is not nice
2) preferring John to leave is not nice
The simplest description of the restrictions on the optional
deletion of the complementizer "for" seems to be the following:
The complementizer "for" may be optionally deleted just in
case this complementizer follows the pronominal head of the
complement construction which itself follows the verb in the
main sentence. In other words, we propose a transformation
something like the following.
i.
(35) X V N + C y
A DJ +PRO )+D
I 2 3 45====>l,2,3, ,
The essential difference between case "a" in (33) and
(35) resides in the fact that the latter apparently requires
an intervening pronoun between the complementizer and the verb
or adjective where this intervening pronoun is not essential
for case "ai" of (33). . Since no unfortunate consequences
arise from allowing a pronoun to intervene in case "b" of (33),
it becomes possible to generalize the special cases of the
optional complementizer deletion rule which handle "POSS"
deletion and "for" deletion in terms of a single rule, namely,
case "a"t of the optional complementizer deletion rule as de-
fined in Chapter 1. Since only two complementizing features,
C+C][+DJ, are sufficient to characterize both the "for-to"
complementizer and the "POSS-ing" complementizer, the generali-
zation of the rules which delete these items represents a
significant improvement on any formulation of these rules in
which three distinct cases are postulated.6
This brings us to the last transformation required to
derive the sentences in the paradigm (30), the auxiliary
transformation which inverts the "POSS" complementizer with
the initial NP of the complement sentence and the "ing" com-
plementizer with V, have, or be. The ordering of the auxiliary
transformation is crucial. If we claim that the auxiliary
transformation precedes the optional complementizer deletion
transformation in order of application, it then becomes
necessary to postulate a much more complex optional comple-
mentizer deletion rule. The reason for this is that if the
auxiliary transformation applies, then the environments of
the "FOSS" and "for" are no longer the same. More specifi-
cally, the "for" still precedes the subject noun phrase of
the complement sentence, but the "POSS" complementizer now
follows the subject noun phrase as a consequence of the
application of the auxiliary transformation. Thus we should
have to propose a transformation of roughly the following
form.7
(36) X V N +C NP +C0] Y
[+PRO]( +D ) +.D
1 2 3 4 5 6 7=ss->l,2,3,6,5,
Since the optional complementizer deletion transformation,
TOCD, is considerably less complex if the auxiliary trans-
formation follows it in order of application (since both
complementizers are in the same environment before the appli-
cation of the auxiliary transformation), this ordering is to
be preferred.
A brief summary will, perhaps, be helpful at this point.
We have established the necessity of the following ordered
set of transformations.
-86-
1. Passive Transformation -- T
2. Extraposition Transformation -- T
E
3. Indefinite Pronoun Deletion Transformation -- unlisted
4. Optional Complementizer Deletion Transformation T
OCD
5. Auxiliary Transformation -- TAUX
6. Pronoun Deletion Transformation 
-- TPD
On the basis of the foregoing discussion, the derivations
of the sentences in the paradigm, (30), may be specified in
the following fashion.
(37) Sentence (30b)
First Cycle -- no operations
everybody prefers CCit][you] Ldrive slowly) ] ] BASE
N NP VP SNP
Second Cycle
everybody prefers C[it] EPOSS[you] ring drive slowly] J J
N NP VP S NP
TCP
everybody prefers L[it] L[you + POSS] drive + ing
N NP
slowly] ) ) T
YP S NP AUX
everybody prefers [C[you*POSS] [drive+ing slowly]
NP VP
3)] T
S NP PDa
Post Cycle
everybody prefers [[your] [driving slowly ] M
NP VP S NP
(38) Sentence (30c)
First Cycle O- no operations
everybody prefers [[it] [[you] Edrive slowly] ) ) BASE
N NP VP S NP
Second Cycle
everybody prefers C[it] POSS [you] Cing drive slowly]
N NP VP
) ) Tp
S NP
everybody prefers H[it) Eyou) Cing drive
N NP
slowly)] TOCD
VP S NP
everybody prefers E[it)[Eyou] [drive+ing slowly)
N NP VP
T
S NP AUX
everybody prefers [[[you] [drive+ing slowly] ] ] T
NP VP S NP PD a
Post Cycle
everybody prefers [[[you] [driving slowly] ) ) M
NP VP S NP
(39) Sentence (30)
First Cycle - no operations
everybody prefers [[it] [[you] (drive slowly] ) ) BASE
N NP VP S NP
~-88-
Second Cycle
everybody prefers [[it] (POSS [you] (ing drive slowly]
N NP VP
] ] (by+P] TS NP MAN OP
((it] (POSS (you] (ing drive slowly] ] ] be+en
N NP VP S NP
prefer (by+everybody] T
MAN
((it] [you+POSS] (drive+ing slowly] ] ] be prefer+
N VP VP S NP
en (by+everybody] T
MUXMAN Au
(((you+POSS] (drive+ing slowly] ] ] be prefer+en
NP VP S NP
(by+everybody] T
MAN
Post Cycle
(((your] [driving slowly] ] ] is preferred (by+
NP VP S NP
everybody] M
MAN
Let us now consider the sentences in the paradigm (40),
where the identity erasure transformation has applied to delete
the subject noun phrase of the complement sentence.
(40) a. everybody prefers driving slowly
b. driving slowly is preferred by everybody
Certain considerations strongly suggest that the identity
erasure transformation must precede the passive transforma-
tion in order of application. As evidence supporting this
contention, consider the following sentences.
(41) a. Bill reminded them to greet me
b. they were reminded by Bill to greet me
As we shall see later, these sentences are instances of noun
phrase complementation having the underlying structure given
in (42)
(42)
1qP PDP
NP(
PR[ J
DET N 8
PRO
NP
Bill persuade they of they greet me
It is observed that if the identity erasure transformation
applies prior to the application of the passive transformation,
that is, directly to the structure given in (42), then the noun
phrase "they" in the complement sentence is correctly erased
by the object noun phrase "they" in the main sentence. Let
us suppose, however, that the identity erasure transformation
does not apply until after the passive transformation, that is,
on a derived structure which has roughly the following form.
-90-
(43)
PDP
VP
PP MN
PRP PREP NP
9ET N
+N
P R OI
they be+en remind of they gree me by Bill
We observe that the erasure principle operates incorrectly
with respect to the derived structure (43). In particular,
the initial NP in the complement sentence, "they," is closer
to the NP in the agentive phrase tby+Bill" than it is to the
subject .of the derived main sentence , "they." It thus becomes
extremely difficult to explain the identity of the deleted NP
in the complement sentence with the derived subject of the
main sentence. At best, it would become necessary to propose
considerable modification of the connection principle. At
worst, the convictions governing deletion could not be stated
at all. This difficulty is completely resolved by requiring
the identity erasure transformation to apply before the
passive transformation since, as we see in (42), erasure is
correctly defined at this stage of a derivation.
"-91-
Finally, we consider the ordering of the complementizer
placement transformation with respect to the identity erasure
transformation. Although the evidence is not overwhelming,
there is some reason to suppose that the complementizer place-
ment transformation precedes the identity erasure transformation.
We recall that identity erasure is possible just in case the
complementizer is either "for-to" or "POSS-ing.1" If the com-
plementizer is "that," then identity erasure does not apply.
If we assume that the identity erasure transformation applies
before the complementizer placement transformation, it then
becomes necessary to insure that the former will not apply in
the event that the complementizer introduced subsequently by
the complementizer placement transformation, is "that." This
can be accomplished, perhaps, by specifying that the variable
X which intervenes between the NP to the left of the comple-
ment sentence in the structural description does not include
any segment which is marked C-D]. In other words, we might
revise the formulation of the identity erasure transformation
as follows.
(44) U (NP) x CNP Y] W (NP) z
not Ca,E-D],$] S
1 2 3 4 5 6 7 8---->
(i) 2 is erased by 4
(ii) 7 is erased by 4
IThis formulation is probably adequate for all instances
where the complementizer allowed by the verb or adjective in
the main sentence is exclusively "that." But this formulation
will not work where all three complementizer options are possi-
ble, as in (45).
(45) a. I dislike it that he is so cruel
b. I dislike it for him to be so cruel
c. I dislike his being so cruel
In cases like (45), the pronominal head will be marked simply
[+C] since all complementizer options are possible. Thus the
restriction imposed on the variable X in (44) will not be
sufficient to block the application of the identity erasure
transformation in (45), and an ungrammatical sentence will
result. There well may be some way in which the identity
erasure transformation can be formulated to prevent the
erasure of a NP in the complement sentence when the comple-
mentizer placement transformation follows the identity erasure
transformation. For present purposes, it seems a good deal
simpler to propose that the complementizer placement trans-
formation precedes the identity erasure transformation. Since
the former marks complement sentences, it becomes possible to
define identity erasure in terms of the complementizer
structure assigned by the complementizer placement transforma-
tion. In other words, for the application of the identity
-92-
-93-
erasure transformation it is sufficient to know only that
the introduced complementizer is either "for" or "POSS, "
C+CJL+D]. In this formulation it is clearly unnecessary to
impose any restrictions on the variable X in the structural
description for the identity erasure transformation.
The following ordered set of transformations is thus
proposed.
1. Complementizer Placement Transformation 
-- TCP
2. Identity Erasure Transformation 
--- TIE
3. Passive Transformation -- T
P
4. Extraposition Transformation -- TE
5. Indefinite Pronoun Deletion Transformation -- unlisted
6. Optional Complementizer Deletion Transformation -- T OD
7. Auxiliary Transformation -- TMUX
8. Pronoun Deletion Transformation 
-- TPD
On the basis of this set of transformational rules, plus
the obligatory complementizer deletion transformation the
motivation and ordering of which will be discussed shortly,
we can specify the derivation of the sentences in (39) in the
following fashion.
(46) Sentence (40)
First Cycle -- no operations
everybody prefers CUit] ((everybody] Edrive slowly]
N NP VP
] ] BASE
S NP
- 9 4--
I
slowly] ]
VP S NP
T
CP
everybody prefers
everybody prefers
everybody prefers
everybody prefers
([it] (POSS (ing drive slowly] )]
N VP/S NP
[[it] [POSS (drive+ing slowly)] ]
N VP S NP
((POSS (drive+ing slowly] 3 ]
VP S NP
[[drive+ing slowly] ] ]
VP S NP
TIE
TAUX
T
PD
T
CD
Post Cycle
Meverybody prefers [[(driving slowly] ] ]
VP S NP
(47) Sentence (40b)
First Cycle -- no operations
everybody prefers [[it] [[everybody] [drive slowly]
N NP VP
) ) (by+P
S NP MAN
Second Cycle
everybody prefers [[it] (POSS [everybody] (ing drive
N NP
slowly] 3 ) (by+P]
VP S NP MAN
everybody prefers ((it] (POSS (ing drive slowly] )
N VP S
3 (by+P)
NP MAN
BASE
TCP
TIE
Second Cycle
everybody prefers [[it] (POSS [everybody] [ing drive
NP NP
[Eit] POSS~ing drive slowly] 2 2 be+en prefer
N VP S NP
Cby+everybody] T
MAN
Hit] POSS~drive+ing slowly] 3 ] be prefer+en Eby +
N VP S NP
everybody] TA
MAN
[POSS~drive+ing slowly] 3 ) be prefer+en [by+
VP S NP
everybody] T
MAN
LCdrive+ing slowly) ] ) be prefer+en [by+everybody] TCDVP S NP MAN
Post Cycle
[driving slowly] ) ) is preferred Eby+everybody) M
VP S NP MAN
C. The "for-to" complementizer
The paradigms listing the basic object complement con-
structions with ttfor-tot are far more complete than the para-
digms for the "POSS-ing" complementizer. This difference stems
from the fact that the extraposition transformation is not re-
stricted with respect to the "for-to" complementizer. This
difference notwithstanding, the set of transformations proposed
earlier for similar constructions with the "POSS-ing" and the
"that" complementizers is fully capable of handling the deriva-
tions of all sentences in the "for-to" paradigm.8
-96-
(49) a.
b.
c.
everyone would prefer
everyone would prefer
everyone would prefer
(48) a. everyone would prefer
b. everyone would prefer
c. everyone would prefer
d. everyone would prefer
come early
e. everyone would prefer
f. it would be preferred
early
g. for you to come early
it to come early
to come early
very much to come early
d. to come early would be preferred by everyone
e. it would be preferred by everyone to come early
In terms of the set of transformations proposed earlier,
the derivations of the sentences in the paradigms may be speci-
fied in the following manner.
(50) Sentence (48a)
First Cycle -- no operations
everyone would prefer [[it] CUyouj Ccome early] ] J a
N NP VP SNP
Second Cycle
everyone would prefer ECit] [for~you] Eto come
N NP
early] ] J T
VP S NP
ASE
it for you to come early
for you to come early
you to come early
it very much for you to
very much for you to come early
by everyone for you to come
would be preferred by everyone
CP
-97-
(51) Sentence (48)
First Cycle -- no operations
everyone would prefer [[it] [[you] [come early) 3 ] BASE
N NP VP S NP
Second Cycle
everyone would prefer [[it] [for[you] [to+come
N NP
early] )] T
VP S NP
everyone would prefer [[for[you) [to come early] TPD
NP VP S NP
(52) Sentence (4 8Q)
First Cycle -- no operations
everyone would prefer [[it] [[you] [come early) ]1 ]3ASE
N NP VP S NP
Second Cycle
everyone would prefer [[it] [for [you] [to come early]
N NP VP
T
S NP OP
everyone would prefer [[it] [[you] [to come early] 3 T0 CD
N NP VP S NP
everyone would prefer [[[you] [to come early] ] ] T
NP VP S NP PD
(53) Sentence (48d)
First Cycle -- no operations
everyone would prefer [[it] [[you] [come early] 3 )
N NP VP S NP
very much] BASE
ADV
ISecond Cycle
everyone would prefer ((it] (for (you] (to come early]
N NP VP
] ] (very much] TCP
S NP ADV
everyone would prefer ((it] ] (very much] (for (you]
N NP ADV NP
(to come early] ] TE
VP S
(in the derivation of (48e) case "b" of TPD applies to
the above string)
everyone would prefer (very much] (for (you] (to come
ADV NP
early] ] TPDb
VP S
(54) Sentence (48f)
First Cycle -- no operations
everyone would prefer ((it] [[you] (come early] ] ]
N NP VP S NP
(by+P] BASE
MAN
Second Cycle
everyone would prefer [[it] [for (you] (to come early]
N NP VP
] ] (by+P] TCP
S NP MAN
((it] (for (you] (to come early] ) ] would be+O
N NP VP S NP
[by+everybody] T
MAN I
Io
H[it] 3 would be+en prefer [by+everybody] [for [you]
N NP MAN NP
[to come early]J TE
Vs
[[it] ] would be prefer+or [by+everybody] [for [you]
N NP MAN NP
[to come early] ] T
VP S
Post Cycle
[[it] 3 would be preferred [by+everybody] [for [you]
N NP MAN NP
[to come early] 3 1
VP 3
(55) Sentence (4 8g)
First Cycle -- no operations
everyone would prefer [[it] [[you] [come early] ] 3
N NP VP S NP
[by+P] BASE
MAN
Second Cycle
everyone would prefer [it] [for [you] [to come early]
N NP VP
] [by+P] TCP
SNP MAN
[[it] [for [you] [to come early] ] would be+4n prefer
N NP VP S NP
[by+everybody] T
MAN
[[it] [for [you] [to come early] ] ] would be prefer
N NP VP S NP
+n [by+everybody] TAUX
MAN
-.99-
-100-
[[for [you] [to come early] ] ] would be prefer+en
NP VP S NP
[by+everybody] TPSDa
MAN
Post Cycle
[for [you] [to come early] ] ] would be preferred
NP VP S NP
[by+everybody] M
MAN
(36) sentence (49a)
First Cycle -- no operations
everyone would prefer [[it] [[everyone] [come early] S ] BASE
N NP VP S NP
Second Cycle
everyone would prefer [[it] [for [everyone] [to come
N NP
early] ] TCP
VP S NP
everyone would prefer [[it] [for [to come early] ] ] TIEN VP S NP
everyone would prefer [[it] [[to come early] ] T TC
N VP S NP CD
(If TP had applied in the above derivation, the grammar
would have generated sentence (49b)).
everyone would prefer [[[to come early] J 3 T
VP S NP
(57) Sentence (49c)
First Cycle -- no operations
everyone would prefer [[it] [[everyone] [come early] ] ]
N NP VP S NP
Every much] BASE
ADV
Second Cycle
everyone would prefer [[it] [for [everyone] [to come
N NP
early] 3 3 [very much] T"I
VP S NP ADV k1
everyone would prefer [[it] 'for [to come early) ] 3
N VP S NP
[very much] T
ADV IE
everyone would prefer [[it] ] [very much] [for [to come
N NP ADV
early] ) T
VP E
everyone would prefer [[it] ] [very much] [[to come
N NP AiJV
early] ] TCD
VP S
(In the derivation of (49d) case "b" of T applies to
the above string.)
everyone would prefer [very much] [[to come early] ] TPDb
ADV VP S
(58) Sentence (49e)
First Cycle -- no operations
everyone would prefer [[it] [[everyone] [come early]
N NP VP
] ] [by+P] BASE
S NP MAN
-102-
Second Cycle
everyone would prefer ((it] (for (everyone] (to come
N NP
early] ] 3 (by+P T
VP S NP MAN
everyone would prefer ((it] (for (to come early] ] 3
N VP S NP
(by+P] TIE
MAN
(Lit] (for (to come early) 3 ] would be+en prefer
N VP S NP
(by+everybody] T
MAN P
((it] ] would be+en prefer (by+everybody] (for (to come
N NP MAN
early]3 T
VP S
((it] ] would be prefer+en (by+everybody] (for (to-
N NP MAN
come early] ] T
VP AUX
((it] ] would be prefer+en (by+everybody] ((to come
N NP MAN
early] ] T
v~s CDVP 5C
Post Cycle
((it] 3 would be preferred (by+everybody] ((to come
N NP MAN
early] 3 M
VPS
(59) Sentence (49f)
Firs t Cycle -- no operations
everyone would prefer [[it] [[everyone] Esome early]
N NP VP
] ] Eby+P]
S NP MAN
BASE
Second Cycle
everyone would prefer LHit]
N
everyone would prefer LUit]
N
[for [everyone] (to come
NP
early] ] ] by+P]
VP S NP MAN
Efor Eto come early] ] ]
VP S NP
[by+P]
MAN
Etit] [for Etocome early]
N VP
] I would be+en prefer
S NP
Lby+everybody]
MAN
T
H(it] Efor [to come early] ] ] would be prefer+en
N VP S NP
(by+everyone]
MAN
T
AUX
C(for Eto come early] ] ] would be prefer+en [by+
VP S NP
everyone] TMAN
[E[to come early] ] ] would be prefer+en (by+everyone] TCDVP S NP MAN
Post Cycle
CE(to come early] J] would be preferred Eby+everyone] M
VP S NP MAN
T
C?
T
IE
-104-
D. General Discussion
It will be useful at this point to turn to the analysis
of a different paradigm, one involving an instance of object
complementation which raises several important questions.
The paradigm illustrates a class of verbs, including "be-
lieve," "suspect," "think," and a great many others, which
participates in complement constructions of apparently consider-
able diversity. We observe, first, that a noun phrase complement
analysis must be allowed for the verbs in this class. Consider,
by way of illustration, the following paradigm.
(60) a. I believe that John has convinced Bill
b. I believe it with no difficulty that John has
convinced Bill
c. what I believe is that John has convinced Bill
d. that John has convinced Bill is believed by me
e. it is believed by me that John has convinced Bill
f. what is believed by me is that John has convinced
Bill
If we assume that the phrase "that John has convinced Bill" is
an instance of noun phrase complementation, then the derivation
of the sentences in the paradigm follows as an automatic conse-
quence of the application of the transformational rules discussed
earlier. The reader will readily convince himself on this point.
Paradigm (60) does not, however, exhaust the number of
constructions in which verbs of the "believet" class may appear.
Consider, for example, the following sentences which do not at
all appear to be instances of noun phrase complementation.
(61) a. I believe John to have convinced Bill
b. John is believed by me to have convinced Bill
There would appear to be some virtue in the conclusion that
the sentences in (61) do not represent cases of noun phrase
complementation. In the first place, we observe that a pseudo-
cleft sentence version of (61) is completely ungrammatical.
(62) *what I believe is for John to have convinced Bill
Secondly, the identity erasure rule does not apply whereas the
reflexive rule does.
(63) a. *I believed to have convinced Bill
b. I believe myself to have convinced Bill
These data suggest that the sentences in (61) are not in-
stances of noun phrase complementation, but rather verb phrase
complementation, and that it is necessary to posit two distinct
underlying structures for all verbs in the class under discussion.
But the verb phrase complement analysis for the verb class under
discussion is not without serious difficulties also, difficulties
which become apparent upon careful consideration of other verb
phrase complement constructions. As a paradigm case of
-106-
transitive verb phrase complementation, consider the following
sentence.
(64) I compelled the doctor to examine John
Predictably, if (64) is an instance of verb phrase complementa-
tion, the pseudo-cleft sentence (65) is ungrammatical.
(65) *what I compelled was for the doctor to examine John
As an instance of verb phrase complementation, sentence (64)
has roughly the following underlying form.
(66)
PDP
V
NPS
)PT N PDP
DET JP
compel the doctor the doctor examineohn
Let us suppose now that the grammar had generated "John"
rather than "the doctor" in the object position of the main
sentence. If the passive transformation is subsequently applied
on the first cycle to the complement sentence in (66), this
allows for the derivation of the sentence (67).9
rI
(67) 1 compelled John to be examined by the doctor
We notice immediately that sentences (64) and (67) have an
entirely different meaning. I can compel a doctor to examine
John without compelling John to be examined by a doctor. The
difference in meaning is explained by the grammar by virtue of
the fact that the two sentences have different underlying
structures. In (64), the noun phrase "the doctor" is inter-
preted as the object of the verb in the underlying structure
of the main sentence. In (67), on the other hand, the noun
phrase "John" is the object of the verb in the underlying
structure. It follows from this discussion that a sentence 31,
with the complement an active sentence derived from a structure
like (66), will be synonymous with a sentence S2 with the com-
plement a passive sentence providing that, all other constitu-
ents being identical, the object of the verb in the main
sentence of the underlying structure is identical to both the
subject and object of the complement sentence in the same under-
lying structure.
Thus, if the sentences in (68) are instances of verb
phrase complementation, it should be the case that passive
application to the complement sentence implies a derivation
different in meaning from the derivation in which the passive
has not applied to the complement sentence. But our pre-
diction is not correct. We observe that the application of
the passive transformation does not in the least affect the
-107-
-108-
truth value synonymy of the two sentences.
(68) a. I believe John to have convinced Bill
b. I believe Bill to have been convinced by John
According to the verb phrase complement analysis, the under-
lying structures in (68) differ in that the underlying object
of "believe" in (68) is "John" while the underlying object in
(68b) is "Bill." The sentences should, therefore, be differ-
ent in meaning; but they are not. This fact may be taken to
mean either that our formulation of transitive verb phrase
complementation is incorrect, an event made a priori quite
improbable by the precision with which this formulation holds
in such a wide range of cases, or that the sentences in (68)
are simply not instances of verb phrase complementation.
We are thus led full circle back to the consideration of
the sentences in (68) as instances of noun phrase complementa-
tion. If we are willing to accept the cost of making the
extraposition transformation obligatory for all verbs in
the "believe" class just in case the complementizer is "for-
to," it becomes immediately clear that we not only have the
transformational machinery sufficient to generate the sentence
in (68), but also a ready explanation of the ungrammaticality
of (62) and (63a). Let us consider this derivation in some
detail.
I(69) Sentence (68a)
First Cycle
I believe [[it] CCJohn][have+en convince Bill] 3 BASE1 0
N NP VP S NP
I believe [[it] LJohn] Ehave convince+en Bill ] 33 T
N NP VP S NP
Second Cycle
I believe E[it] Efor CJohn] to have convince+en
N NP
Bill] ] 3 T
VP S NP OP
At this point, we require the obligatory application of the
extraposition transformation. This transformation, applying
vacuously with respect to the string in this instance, takes
the complement sentence from under the domination of the NP
which dominates it in the underlying structure.
I believe L[it] ] [for EJohn] Eto have convince+en Bill] ] TE
N NP NP VP S
The pronoun replacement transformation (which has independent
motivation in the subject complement system) is now defined
and the following string can be generated.
I believe EJohn] (for (to have convince+en Bill] TPR
NP VP S
Finally, the obligatory complementizer deletion transformation
applies.
I believe LJohnJ L[to have convince+en Bill) ] TC
NP VP S
Post Cycle
I believe LJohnJ L[to have convinced Bill) ] M
NP VP S
The derivation of (68b), which is generated from exactly
the same underlying structure as (68a), differs from the deri-
vation of (68a) only in that the passive transformation applies
on the first cycle of the former.
(70) Sentence (68b)
First Cycle
I believe Cit] CCJohn] Chave+en convince EBill]
N NP NP
[by+P) j 3 3 BASE
MAN VP S NP
I believe C[it] LLBill] have+en be+e'n convince [by+
N NP
John] ] ]J 3T
MAN VP S NP
I believe E[it] CBill][ have be+en convince+en [by+
N NP
John) J 
] TAUX
MAN VP S NP A
Second Cycle
I believe ECit] for CBill[ Cto have be+en convinoe+en
N NP
[by+John) ] 3 TCP
MAN VP S NP
I believe ECit] J Cfor CBill] Cto have be+en convince+
N NP NP
en Cby+John] ) TS
MAN VP S
-III-
I believe [Bill] [for [to have be+en convince+en
NP
[by+John) )TR
MAN VP S
I believe [Bill] [[to have be+en convince+en [by+
NP
John] ) ) T
MANVPS C
Post Cycle
I believe [Bill] [[to have been convinced [by+John]
NP. MAN
M
VP S
We observed in an earlier discussion that the passive
transformation must apply before the extraposition transforma--
tion. Preserving this constraint with respect to sentence (61b),
we note the following derivation.
(71) Sentence (61b)
First Cycle
I believe [[it] [[John] [have+en convince Bill] )
N NP VPS
3 [by+P] BASE
NP MAN
I believe [[it] [[John] [have convince+en Bill] 3 )
N NP VP S NP
[by+P] TAU
MAN
____ 
4;at _____
ISecond Cycle
I believe [[it] [for [John] (to have convince+en
N NP
Bill] ] 3 [by+P] TCP
VP S3NP MAN
(it] [for [John] [to have convince+en Bill] J ]
N NP VP S NP
be+en believe (by+I] T
NAN
t] ] be+en believe [by+I] [for [John] [to have
N NP MAN NP.
convince+en Bill] ]
VP S
[Lit] ] be believe+en [by+I] [for [John] [to have
N=_TP MAN NP
convinoe+en Bill ] TAUX
VP S
[John] be believe+en [by+I] [for [to have convince+en
NP MAN
Bill] ] TP
VPS PR
(John) be believe+en [by+I] ([to have convince+en
NP MAN
Bill ] TCD
VP a
Post Cycle
[John] is believed (by+me] [(to have convinced Bill] M
NP MAN VP S
In addition to allowing us to explain the derivation of
the sentences in (60) and (61) from a common underlying source
through the application of an independently motivated trans-
formation, namely, the pronoun replacement transformation,
-112-
-113-
I
this analysis has another virtue. Consider the problem of
the introductory "there" in true noun phrase complement con-
structions.
(72) a. everybody prefers (for) three chairs to be in
the room
b. it was preferred by everybody for three chairs to
be in the room
c. for three chairs to be in the room was preferred
by everybody
d. everybody prefers (for) there to be three chairs
in the room
e. it was preferred by everybody for there to be
three chairs in the room
f. for there to be three chairs in the room was pre-
ferred by everybody
Assuming the existence of a transformation which, in some ap-
propriate fashion, converts a string of the form "three chairs
are in the room" into "there are three chairs in the room,"
every sentence in (72) can be explained on the basis of the
transformations proposed thus far in the present study. The
introductory "there" takes subject position and is treated as
the subject by all of the transformations which apply in general
to noun phrase complement constructions. Consider, however,
the following sentences.
-111+-
(73) a. everybody believes three chairs to be in the room
b. three chairs are believed by everybody to be in
the room
C. everybody believes there to be three chairs in the
room
d. there is believed by everybody to be three chairs
in the room
If we take the position that the sentences in (72) are
instances of verb phrase complementation, we are forced to
propose either a system in which the underlying object of the
main verb can be a dummy or otherwise unspecified noun or
that "there" is everywhere a possible expansion of NP. Making
a long story short, we see immediately that the existence of
the introductory '"there" in sentences like (73) can be ex-
plained as an automatic consequence of the pronoun replacement
transformation which, under the appropriate conditions, takes
the subject of the complement sentence, no matter what that
subject is, and substitutes it for the pronominal head.
It is of interest to observe the operation of the pronoun
replacement rule when the depth of embeddings is increased.
Consider the following sentence for example.
(74) I believe John's eating to be messy
(75) Sentence (74)
First Cycle -- no operations
1 .11 "IMM-9 "a III ...
I believe [[it] [[[it] [[John] [eat] 3 3][be
N N NP VP S NP
messy] ] 3 BASE
VP S NP
Second Cycle
I believe [[it] [[[it] [POSS[ John] [ing eat] ] I
N N NP VP SNP
[be messy] ] ] TCP
VP S NP
(Observe that extraposition cannot apply on this cycle
since extraposition is impossible for the "POSS-ing" comple-
mentizer.)
I believe [[it] [[[it] [[John+POSS] [eat+ing] ] ]
N N NP VP S NP
[be messy] ] ] T
VP S NP
I believe [[it] [[[[John+POSS] [eat+ing] ] 3 [be
NP VP S NP
messy] 3 ] T
VP S NP
Third Cycle
I believe [[it] [for [[[John+POSS] [eat+ing] ] [to
N NP VP S NP
be messy] ] ] TC
VP S NP
I believe [[it] ] [for [[[John+POSS] [eat+ing] ) ]
N NP NP, VP S NP
[to be messy] ] T
VPS a
(Notice that extraposition must have applied above since extra-
position is obligatorily defined on complements with "for-to"
-116-
complementizers for verbs of the "believe " class.)
I believe (((John+POSS] (eat+ing) ] [for [to be
NP VP S NP
mo ]TPR
VP S
I believe (((John+POSS] (eat+ing] 3 3 [to be messy ] TCDNP VP S NP VP S
Post Cycle
I believe (((Johnts] [eating] 3 3 ((to be messy] 3 M
NP VPS NP VPS
A curious situation arises when the complementizer for the
most deeply embedded sentence is "that." By reconstructing the
above derivation we observe that if extraposition had applied
on the second cycle we would have correctly generated a
sentence like (76).
(76) I believe it to be true that John is honest
If, on the other hand, the extraposition transformation had
not applied on the second cyole, the grammar would have
generated the quasi-grammatical sentence (77).
(77) 1 believe that John is honest to be true
The status of (77) is not entirely clear and one wonders if
its strangeness might not be due to whatever produces the
strangeness of multiple embeddings of various sorts. A simi-
lar strangeness is observed in sentences containing only the
"that" complementizer.
K ________
-117-
(78) a. I believe that it is true that John is honest
b. I believe that that John is honest is true
In any event, the peculiarity of these sentences is not ex-
plained by the analysis being proposed here since both the
sentences in (78) and those in (76) and (77) are generated as
the consequence of the application of the same set of rules.
It is another interesting fact about this analysis that
it makes it almost possible to eliminate the optional comple-
mentizer deletion transformation. Consider again the following
pairs of sentences.
(79) a. I hate for you to do these things
b. I hate you to do these things
We see that the above analysis affords an excellent explanation
of the deletion of "for" in (79b). Let us suppose that the
extraposition transformation applies vacuously to the moun
phrase "it for you to do these things." The pronoun replace-
ment transformation is now defined producing the string "you
for to do these things." By the obligatory complementizer
transformation this becomes "you to do these things." It
thus seems possible to do away with the optional complementizer
deletion transformation.
This proposal has two rather serious drawbacks. In the
first place, it will be necessary to allow extraposition of
complement sentences containing the "POSS-ing" complementizer
I
r
-118-
I
since, otherwise, we shall still need an optional complementizer
deletion transformation to handle the deletion of "POSS." But
relaxing the restriction on extraposition, which may eventually
be necessary to some extent, results in many of the ungrammati-
cal sentences presented in the paradigm (30). A second draw-
back is equally as serious. Let us consider several examples
of a verb for which complementizer deletion is obligatory, e.g.,
"want."
(80) a. I want very much for John to go
b. I want John to go very much
c. *I want for John to go very much
d. *I want very much John to go
Our proposal neatly explains the grammaticality of (80S) and
the ungrammaticality of (80d). Since the pronoun replacement
transformation is blocked by the adverbial intervening between
the pronominal head and the complement sentence, the subject
of the complement sentence will not replace the pronominal
head; consequently the "for" will not be contiguous with the
VP and will not be deleted. But we observe that we can only
explain the non-occurrence of (8c on the assumption that
extraposition is obligatory for the verb "want." This,
however, leaves us with no explanation of the grammaticality
of (80b) since there is no indication whatever that extra-
position has applied here. Thus we cannot rely on our
-119-
proposal to explain all of the facts relating to complementizer
deletion. It seems that the optional complementizer is still
necessary.
But there is an interesting claim here nonetheless.
The grammar tells us that sentences like (80b) may have two
distinct derived constituent structures but one underlying
structure. This follows from the fact that (80b) may have
been derived either through the application of the comple-
mentizer deletion transformation or through the vacuous appli-
cation of the extraposition transformation, and the subsequent
application of the pronoun replacement and the obligatory
complementizer deletion transformations. There is some recent
psychological evidence which supports the plausibility of this
circumstance.11
In closing the discussion of object complementation it
would not be beside the point to mention the several types
of restrictions which seem to limit the application the trans-
formations involved in the complement system. These restric-
tions, which are idiosyncratic to particular verbs, will
require description in the grammar and the purpose of the
following discussion is to identify some instances of this
phenomenon. One of the more common restrictions has to do
with the application of the optional complementizer trans-
formation, as we have just seen. Consider the following
pairs of sentences.
III I Imp
-120-
(81) a. 1) 1 would love for you to have it
2) I would love you to have it
b. 1) 1 can't bear for them to see me this way
a) I cant t bear them to see me this way
c. 1) *1 dont t want for anybody to see me this way
2) I don't want anybody to see me this way
The foregoing considerations lead us to believe that this
restriction can be explained by requiring the obligatory
application of the optional complementizer deletion trans-
formation for verbs like "want."
A second restriction has to do with the identity erasure
transformation. We find instances where certain verbs taking
object complement constructions require erasing and erased
noun phrases to be identical. This restriction, which is also
the property of all verb phrase complements, was not a property
of the object complement constructions discussed earlier. Thus
both of the sentences in (82) are grammatical.
(82) a. I prefer for you to do it
b. I prefer to do it
But certain other verbs, e.g., "promise," do not have this
freedom, a claim supported by the following data.
(83) a. *1 promise for you to bring money
b. I promise to bring money
Still other verbs, e.g., "require, " "say," demand obligatory
non-identity of the erasing and erased noun phrases.
(84) a. 1) I said for you to go
2) *1 said (for me) to go12
b. 1) I require for you to have your hair cut
2) *1 require (for me) to have my hair cut
At the present it is not clear how these restrictions are going
to be stated in the grammar, although there seems little doubt
that the system in which these restrictions are stated is not
likely to be adequate if it does not handle identity restric-
tions in the same way for both verb phrase and noun phrase
complementation.
The verb "promise" is unusual in still another respect.
Sentences like (85) constitute the major exception to the
13erasure principle.
(85) I promised John to bring the money
The erasure principle predicts that the implicit subject of
the complement sentence will be "John" rather than "I." It is
quite doubtful that a principle which holds remarkably well for
such a considerable number of cases will fail as the result of
this one counter-example. The verb "promise" is peculiar in
many respects and there is every reason to interpret this
result as advice to look more deeply into the analysis of
IN III leis NOW I IN POP
-121-
-122-
this particular verb for we are likely to find that the
problem lies not with the erasure principle but with out
analysis of the constructions in which this particular verb
appears.
A final restriction concerns the application of the
passive transformation to object complement constructions
containing the "for-to" complementizer. Consider, by way of
illustration the following data.
a. 1) a)
b)
2) a)
b)
b. I) a)
b)
2) a)
b)
c. 1) a)
b)
z) a)
b)
everybody loves John
John is loved by everybody
everybody loves for you to sing
*for you to sing is loved by everybody
everybody dislikes John
John is disliked by everybody
everybody dislikes for you to sing
*for you to sing is disliked by everybody
everybody prefers John (over Bill)
John is preferred over Bill by everybody
everybody prefers for you to sing
for you to sing is preferred by everybody
What these data demonstrate is that while passivization is
possible in (86a, b) only if the object noun phrase does not
dominate a complement sentence, passivization is possible for
the verb "prefer" (86c) in either case. This fact suggests
(86)
-12)-
that, although the passive transformation itself may not have
been sensitive to the internal constituency of a NP, the
restrictions on the application of this transformation which
are idiosyncratic to particular verbs must be sensitive to
the internal structure of NP's. At the present time there
are several mechanisms which one might propose to state these
idiosyncratic restrictions in the grammar, but there is no
motivation for any other than beyond the fact that they may
work. Until much more work can be done in this area, it is
pointless to take up any more time in the discussion of what
might be a correct formulation. There is simply too little
evidence to establish any motivated system for handling
these restrictions and, for purposes of the present study,
this question is appropriately left open.
In still other cases, passivization is obligatory. Con-
sider sentence (87).
(87) John is said to be honest
In other cases passivization is usually preferable
(88) a. John was rumored to be honest
b. John was alleged to be guilty
These sentences, analyzed as instances of noun phrase comple-
mentation similar to the analysis for the "believe" class of
verbs discussed earlier, present no derivational difficulty.
ran honest man
But it seems reasonably clear that the verbs in question will
require some system of markers which advise of the obligatori-
ness of the passive transformation.
It is somehow ironic that the class of verbs which
historically was the first to be investigated for the purpose
of developing the analysis being presented here should turn
out to be the most recalcitrant. The problem is that the
criteria employed for differentiating verbs in terms of
underlying structure seem to be inadequate at this point.
The class of verbs includes "expect," "desire," and a few
others. Consider the following paradigms.
(89) a. everybody expects (for) me to do what is right
b. it is expected by everybody for me to do what
is right
c. what everybody expects is for me to do what is right
d. what is expected by everybody is for me to do
what is right
(90) a. everybody desired (for) me to be an honest man
b. it was desired by everybody for me to be an
honest man
c. what everybody desired was for me to be an honest
man
d. what was desired by everybody was for me to be
-124-
-125-
It seems beyond question that a noun phrase complement analysis
must be assigned to the sentences in the above paradigm. In-
sofar as the sentences in (91) and (92) have the same inter-
pretation as those in (89) and (90), we are led to believe
that they follow the analysis for the verbs of the "believe"
class, namely, that extraposition has applied to be followed
by the pronoun replacement transformation and the obligatory
complementizer deletion transformation.
(91) a. I expect myself to do what is right
b. I am expected to do what is right
(92) a. I desired myself to be an honest man
b. I was desired to be an honest man
But one wonders what to make of sentences like those in
(93) and (94) which, in addition to requiring an object comple-
ment analysis, also have a prepositional phrase.
(93) a. I expect (it) of myself that I will do what is right
b. I desire (it) of myself that I be an honest man
(94) a. I exp4ot it of myself to do what is right
b. I desire it of myself to be an honest man
These sentences present no derivation problem whatever. The
problem is that there is a sense in which the sentences in (93)
and (94) have the same semantic interpretation as those in (91)
and (92). The open question is whether the burden of explanation
-126-
is to be placed on the semantic component at this point or
whether we should propose an analysis in which the sentences
in (91) and (92) should be derived from the structures under-
lying those in (93) and (94). The considerations raised in
the present study offer no way of resolving this question.
In terms of these considerations, the sentences in (89),
(90), (91), and (92) share a common underlying structure
which differs from the structure underlying the sentences
in (93) and (94). Progress in this area will no doubt depend
upon distinctions which are far subtler than those which the
criteria for differentiating underlying structures proposed
in this study lead us to identify.
II. Subject Complementation for Intransitive Verbs
The term "subject complementation" is a mnemonic referring
to the instance of noun phrase complementation which arises
through the expansion of S into NP, VP. The application of
phrase structure Rule 2 subsequently yields the string DET, N,
s, VP, and it is to this configuration that the discussion in
the next few pages is devoted.
A. The "that" complementizer
The number of intransitive verbs in English taking subject
complements is comparatively small, but structures of this
general type, the underlying structures of which are exempli--
fied by the phrase structure diagram in (95), raise questions
-127-
(95)
DET
PDP
NP VP
Representative sentences employing the complementizer "that"
are the following.
a.
b.
it turns out that John is right
it happened that John came early
The derivation of the sentences in (96) presents no
difficulty which the transformational apparatus outlined and
discussed earlier cannot solve. In particular, these sentences
are generated simply as a function of the application of the
extraposition transformation to the underlying structure given
in (95), as we see in the following derivation.
(97) Sentence (96b)
First Cycle -- no operations
(96)
a
of such importance that it will prove instructive to examine
them in some detail.
-128-
Cit] (John] (came early] I] happened BASE
N NP VP S NP
Second Cycle
Cit] (that (John] [came early ] J happened T
N NP VP S NP CP
[(it] J happened (that (John] (came early] ) T
N NP NP VP
The one novel feature of the constructions in question
is that the extraposition transformation appears to be obliga-
tory since, as we see below (98), the cognate sentences in
which the extraposition transformation has not applied are
ungrammatical.1
(98) a. *that John is right turns out
b. *that John came early happened
Since it is not, in general, true that extraposition is always
obligatory for a noun phrase complement in sentence initial
position, and we have noted many such instances in the earlier
discussion, it becomes necessary apparently to postulate some
instruction, probably incorporated into the lexical entry for
the appropriate verbs, which stipulates that the extraposi-
tion transformation must apply if the structural conditions
for its application are met. And in the case of constructions
like (95), these conditions are always satisfied.
Although this verbal marker or instruction may be
necessary there are certain considerations which suggest that
it is not sufficient. In particular, when verbs of the type
under discussion are followed by a complement sentence, the
occurrence of a complement sentence containing a "that" com-
plementizer in subject position is completely appropriate
and extraposition is not obligatory, as we see in (99).
(99) a. 1) that John is right turns out to be the case
2) it turns out to be the case that John is right
b. 1) that John came early happened to annoy Bill
2) it happened to annoy Bill that John came early
The data in (99) suggests that the verbal marker regulating
the application of the extraposition transformation must be
considerably more complex than that originally envisioned.
It must now consist of the assertion that the extraposition
transformation is obligatory just in case the verb is followed
by the sentence boundary "#," This additional complexity
should, perhaps, be taken as an indication that something
is wrong with this course of action. And, indeed, there are
certain facts which this explanation does not account for.
The explanation of the optionality of the extraposition
transformation which is based upon the marker system crumbles
in the face of more careful consideration of the status of
the complement sentence following the verb "happen,V for
instance, in (99b). At first blush, we should consider the
phrase "to annoy Bill" as an instance of verb phrase comple-
mentation since it does not participate in the variant
-am9-
-130-
constructions, such as the pseudo-cleft sentence which we
have come to associate with noun phrase complement construc-
tians. Thus, for example, neither of the following sentences
constructed on the assumption that the phrase "to annoy Bill"
is a noun phrase complement are grammatical.
(100) a. *what that John came early happened was to annoy
Bill
b. *what it happened was to annoy Bill that John came
early
In the face of this evidence, we are inclined to believe that
the phrase "to annoy Bill" must, in actuality, be an instance
of a verb phrase complement. But this alternate conclusion
is not without serious difficulties also, difficulties which
become apparent upon careful consideration of other instances
of verb phrase complementation.
As a paradigm of intransitive verb phrase complementation,
consider the following sentence.
(101) the doctor condescended to examine John
As we would expect, if (101) is a correct paradigm, the pseudo-
cleft sentence of (101) is impossible.
(102) *what the doctor condescended was to examine John
As an instance of verb phrase complementation, the sentence (101)
has roughly the following underlying structure.
-131-
(103)
N PP
DET N V
N ppp
IP r
DET N V NP
the doctor condescend the doctor examine John
Let us suppose, now, that the grammar had generated "John"
in the subject position of the main sentence instead of "the
doctor.1" If the passive transformation is subsequently ap-
plied to the complement sentence in (1p3). This will allow
for the derivation of the sentence (104).
(104) John condescended to be examined by the doctor
The sentences (101) and (104) have an entirely different mean-
ing. This difference is explained by the grammar by virtue
of the fact that the two sentences have different underlying
structures. In the former, the noun phrase "the doctor" is
interpreted by the semantic component as the underlying sub-
ject of the main sentence. In the latter, the noun phrase
"John" is so interpreted. The fact that in (101) the com-
plement sentence is active while in (103) it is passive is
-132-
of no consequence since both have the common underlying
structure given in (103). It follows from this discussion
that a sentence S1, derived from a structure like (103), will
be synonymous with a sentence S2 providing that, all other
constituents being identical, the subject noun phrase of the
main sentence is identical to the subject noun phrase and
the object noun phrase of the complement sentence, as is
the case in the following sentences.
(105) a. the doctor condescended to examine himself
b. the doctor condescended to be examined by himself
The semantic readings of all sentences based upon the structure
in (103) will be different unless these conditions are met.
The difference in meaning between sentences (101) and (104)
exemplifies this conclusion.
Returning to the original question, namely, is the phrase
"to annoy Bill" in (99b) an instance of verb phrase comple-
mentation similar to (101), we note a serious difficulty if
we answer the question in the affirmative. If the phrase "to
annoy Bill" is an instance of intransitive verb phrase com-
plementation, then how can we explain the fact that sentence (99-
bl) and (106) have the same meaning. (Sentence (99bl) is re-
peated here for convenience.)
(99b1) that John came early happened to annoy Bill
(106) Bill happened to be annoyed that John came early
Since, in the verb phrase complementation analysis, the
phrase "that John came early" is the underlying subject of
the main sentence in (99b1) and since the noun phrase "Bill"
is the underlying subject of the main sentence in (106), the
two sentences should be different in meaning. But they are
not. This fact may be taken to mean either that our formu-
lation of intransitive verb phrase complementation is in-
correct or, more probably, that the phrase "to annoy Bill"
in (99b) is simply not an instance of verb phrase comple-
mentation.
If the phrase "to annoy Bill" is neither an instance of
noun phrase complementation nor an instance of verb phrase
complementation, only one possibility remains. This alterna-
tive asserts that the phrase "to annoy Bill" does, in fact,
originate as a noun phrase complement, but that the pronominal
head of this construction no longer exists in a form to which
the pseudo-cleft sentence transformations can apply. Let us
explore this possibility further.
Let us suppose that the underlying subject of the verb
"happen" in (99b) is a noun phrase complement construction in
which the subject of the complement sentence is itself a noun
phrase complement construction. In other words, the under-
lying subject of the phrase "to annoy Bill" is "it that John
came early" and that the entire complement sentence "it that
John came early annoy Bill" is the subject of the verb "happen,
the underlying structure being represented as follows.
-134-
(107) [[itJ[[[it] [John came early] ] [annoy Bill] ) ] happen
N< N S NP VP S3NP
Eschewing complementizer insertion for the time being, let us
assume that on the second cycle, which applies to "it that
John came early," the only transformation which applies is
the pronoun deletion transformation thus yielding a string
like (108).
(108) [[it] [[[John came early] ] [annoy Bill] ) 2 happened
N SNP VP S NP
On the third cycle, the extraposition transformation must apply,
(this rule being obligatory for verbs like "happen" as we have
seen earlier), yielding the string (109).
(109) [[it] ) happened ([[John came early) ] [annoy Bill]
N NP 'SN? VP S
At this point, the pronoun replacement transformation, moti-
vated in the discussion of object complementation, substitutes
the noun phrase subject of the complement sentence for the
pronominal head in the noun phrase which is the subject of
"happen." The application of this transformation produces
(108), that is, the string coinciding with (99b). The com-
plementizers are added here for the sake of clarity.
(110) [[that John came early] ] happened [[to annoy Bill] 3
S NP VP S
As we see in the formal derivation (113), if the extraposition
transformation has been applied on the second cycle and if
all subsequent transformations remain constant, we should
have generated (111), the string coinciding with (99b2).
(111) (it] ] happened [[to annoy Bill] Ethat John came early] J
N NP VP S S
This formulation has several advantages., In the first
place, it allows for an explanation of the occurrence of the
sentences in (96) and (99) and the non-occurrence of the
sentences in (98). The extraposition transformation is always
obligatory. But complement constructions may end up in
sentence initial position as the result of the pronoun replace-
ment transformation. Second, this formulation explains the
non-occurrence of pseudo-cleft sentences in (100) as a
function of this same transformation. Similarly, it explains
another pseudo-cleft sentence phenomenon which we have not
yet considered. Implicit in this analysis is the claim that
the pronoun "it" in sentences (99a2, b2) is not the pronominal
head of the entire underlying subject, that is, the first
"it" in "it it that John came late annoy Bill." Rather,
this analysis stipulates that the pronoun in the sentences
is actually the pronominal head of the subject of the com-
plement sentence "it that John came late annoy Bill." This
prediction is confirmed by the fact that the pseudo-cleft
sentence based upon the contrary assumption is ungrammatical
(112).
...- 136-
(112) *what happened was to annoy Bill that John came early
Only one pseudo-cleft sentence is possible, namely, the one
based upon the pronominal head of the complement "that John
came late."
(113) what happened to annoy Bill was that John came early
A third virtue of this analysis is that it offers an
explanation of why subject complements containing the POSS-
ing" complementizer are impossible in (114a), but possible
in (114b).
(114) a. *John's coming early happened
b. John's coming early happened to annoy Bill
We can say that verbs like "happen" simply carry the restric-
tion that complement subject may not contain the "POSS-ing"
complementizer. This restriction effectively blocks (114a)
in our analysis but not (114b) since the complement containing
the "POSS-ing" complementizer in this sentence is analyzed to
be the subject of the verb "annoy," a verb which accepts
"POSS-ing" complementizers in subject complements. The com-
plement sentence "Johnt s coming early" only seems to be the
subject of "happened" in the underlying structure. As we
have seen, this complement sentence attains the subject
position only by virtue of the application of the pronoun
replacement transformation.
-137-
For all of these reasons, the pronoun replacement
transformation possesses great credibility and we may
specify the derivations of the sentences under consideration
in terms of this transformational rule (plus the obligatory
complementizer deletion transformation which is a necessary
consequence of this formulation) and others which have been
discussed previously. Clearly, the pro'noun replacement
transformation must follow the extraposition transforma-
tion in order of application since it is the latter which
establishes the appropriate environment for the application
of the former.
(115) Sentence (99bl)
First Cycle -- no operations
[[it] [[[it] [John came late] 3 [annoy Bill] 3
N N S NP VP S NP
happen BASE
Second Cycle
[[it] [[[it] [that John came late] 3 [annoy Bill]
N N SNP VP
3 3 happen TCP
S NP
[[it] [[[that John came late] J [annoy Bill] J ]
N S NP VP S NP
happen TF"
-138-
Third Cycle
[[it] Efor C[that John came late] ] Eto annoy Bill]
N S NP VP
I I happen TCP
S NP
Lit] I happen Cfor [[that John came late] ] Eto
N NP SNP
annoy Bill] 3 T
VP S E
Ctthat John came late] 3 happen [for Eto annoy Bill] ] TPR
S NP VP S
[[that John came late] 3 happen ([to annoy Bill] TCD.S NP VP s
(116) Sentence (99b2)
First Cycle -- no operations
L[it] [[[it] John came late] ] [annoy Bill] ] ]
N N VNP VP S NP
happen BASE
Second Cycle
Lit] ECMit] Ethat John came late] [annoy Bill]
N N S NP VP
] ] happen TCP
3 NP
[it] [[[itI ] [annoy Bill] Ethat John came late]
N NNP VPI S
I ] happen T
S NP
Third Cycle
[[it] [for Cit] [ Cto annoy Bill] Ethat John came
N N NP VP,
labe] ] happen T,
S S NP
---Ao
-139-
[[it] ] happen Efor [[it] 3 Eto annoy Bill] Ethat
N NP N NP VP
John came late] 3 T
S S E
L[it] I happen Efor Eto annoy Bill] Ethat John came
N NP VP
late] J T
S SPR
[Cit] 3 happen [[to annoy Bill] Ethat John came
N NP VP
late] TCD
S S
B. The "for-to" complementizer
Subject complement constructions containing the "for-to"
complementizer present much the same type of problem with
intransitive verbs as do the subject complement constructions
containing the "that" complementizer. Consider, by way of
example, the following sentences.
(117) a. 1) *for John to find gold happened
2) *it happened for John to find gold
3) John happened to find gold
b. 1) *for John to be unhappy appeared
2) *it appeared for John to be unhappy
3) John appeared to be unhappy
That the 3) sentences are, in fact, the result of the applica-
tion of the pronoun replacement transformation and not in-
stances of verb phrase complementation is a conclusion which
-140-
follows from considerations identical to those raised in
connection with the "that" complementizer earlier. We note,
for example, that the "a" and "b" sentences of (118) are
identical in meaning, a result which is the converse of that
predicted by the hypothesis that these sentences are instances
of verb phrase complementation.
(118) a. John happened to mention Bill
b. Bill happened to be mentioned by John
If we assume that the underlying structure of the sentences
in (118) consists of the subject complement construction "it
for John to mention Bill," then the synonymy of the sentences
in (118) is an automatic consequence since these sentences
differ only in their transformational derivation. More
specifically, the passive transformation has been applied
in the case of (118b) before the application of the identity
erasure transformation, but the passive has not been applied
in the case of (118a).
We observe, furthermore, that the fact that the pronoun
replacement is obligatory explains the non-occurrence of
pseudo-cleft sentence, for instance, in (119).
(119) a. *what happened was for John to find gold
b. what happened was that John found gold
-141-
Since the pronoun replacement destroys the environment upon
which the transformation deriving the pseudo'-cleft sentence
are defined, the derivation of (119a) is impossible.
In terms of this analysis, then, the derivation of
sentences (119a3, b3) (92a3, b3) proceeds as follows:
(120) Sentence (117a3)
First Cycle -- no operations
E[it] [[ohn] Efind gold] 3 3jhappened
N NP VP S NP
Second Cycle
(it] Efor [John] [to find gold] 3 3 happened
N NP VP S NP
[[it] happened Efor [John] [to find gold] ]
N NP NP VPS
CJohn] happened Cfor Lto find gold] ]
NP VP S
CJohn] happened [[to find gold]
NP VP S
BASE
CP
TE
TPR
T
III. Subject Complementation for Transitive Verbs
Except for the non-application of the pronoun replacement
transformation, subject complementation for transitive verbs
does not differ from subject complementation for intransitive
verbs. For this reason the paradigm includes instances of
subject complementation with all three complementizers. The
derivation of these sentences is entirely a function of rules
motivated in other parts of the complement system.
-142-
(121) a. 1) that you came early surprised me
2) it surprised me that you came early
b. 1) for you to find me this way embarrasses me
2) it embarrasses me for you to find me this way
c. 1) Johnis playing the bugle annoys me
2) *it annoys me John's playing the bugle
(122) Sentence (121al)
First Cycle -- no operations
Hit] [you came early] 3 surprised me BASE
N S NP
Second Cycle
L[it] Cthat you came early) ] surprised me TCP
N S NP
C[that you came early] 3 surprised me T
S NP CD
(We observe that the extraposition transformation is
not obligatory for subject complements if the main verb is
transitive.)
(123) Sentence (121a2)
First Cycle -- no operations
CUit] [you came early] 3 surprised me BASE
N S NP
Second Cycle
Cit] (that you came early] ] surprised me TCP
N S NP
Cit] ] surprised me Ethat you came early] TE
N NP S
-143-
(124) Sentence (121bl)
First Cycle -- no operations
[[it] [[you] (find me this way] ] J embarrasses me BASE
N - NP VP S NP
Second Cycle
[it] (for [you] (to find me this way] ] ] em-
N NP VP S NP
barrasses me T
CP
(for [you] (to find me this way] ] ] embarrasses me T
NP VP S NP
(125) Sentence (121b2)
First Cycle -- no operations
[[it] [[you] (find me this way] ] ] embarrasses me BASE
N NP VP S NP
Second Cycle
[[it] (for (you] (to find me this way] ] ] embarrasses
N NP VP S NP
me TCP
[[it] ] embarrasses me [for (you] [to find me this
N NP NP
way] ] T
VP S
(126) Sentence (121ld)
First Cycle -- no operations
((it] [[John] (play the bugle] ] ] annoys me BASE
N NP VP S NP
Second Cycle
((it] [POSS [John] (ing.Play the bugle] ] ] annoys me TCP
N NP VP S NP
C[it] [[John+POSS] play+ing the bugle) ] ] annoys me TAUX
N NP VP S NP
CEJohn+POSS) play+ing the bugle] 3 ] annoys me T
NP VPSNP PD
Post Cycle
ECCJohn's] [playing the bugle] 3 ] annoys me M
NP VP S NP
(We observe that sentence (1212) is impossible since
the extraposition transformation is not defined on complements
containing the "POSS-ing" complementizer. Thus, the pre-sentence
pronoun deletion transformation must always apply yielding
sentences like (121c1)).
IV. Oblique Noun Phrase Complementation
A. Intransitive Oblique Noun Phrase Complementation
There are several instances of noun phrase complementation
in English where the selectional restrictions on the comple-
mentizer of an object complement are in complementary distri-
bution with the restrictions on oblique noun phrase complements.
Consider, by way of illustration, the following paradigm.
(127) a. I decided that John shall represent us
b. *I decided on that John shall represent us
c. I decided for John to represent us
d. *I decided on for John to represent us
e. *I decided John's representing us
f. I decided on John's representing us
On the assumption that the verb "decidel" has two analyses, one
in which this verb takes a prepositional phrase and one in which
it takes a direct object, one is forced to the unfortunate con-
clusion that the lexical entry for "decide" contains not only
two strict sub-classificational features, but furthermore that
the restrictions on the pronominal head of the complement con-
struction as object of the verb are entirely distinct from the
restrictions on the pronominal head in the prepositional phrase.
In other words, the restrictions imposed upon the pronominal
head as object must include the features +C][-D] and C+C][+D]
[-E] where the first cluster represents the complementizer
"that" while the second represents the complementizer "for-
to." The restrictions on the pronominal head when it appears
in a prepositional phrase must include the features L+CL+DJ
[+E], the cluster representing the "POSS-ing" complementizer.
In the following pages an attempt will be made to show
that the lexical representation of such verbs as "decide"
can be greatly simplified if it is assumed that the underlying
structures for the grammatical sentences in (127) are identi-
cal. If it can be shown that the prepositional phrase analysis
underlies all three cases, then the necessity for positing the
complex array of restrictions proposed above completely dis-
appears.
The problems posed by the introduction of the preposition
in the structure underlying the sentence (127f) are very
similar to those discussed earlier in Chapter 3 in relation
to the introduction of the complementizing morphemes. More
specifically, we are confronted with two apparently distinct
options -- the introduction of the preposition through the
application of phrase structure rules and the introduction
of the preposition transformationally by a rule which is
sensitive to the structure given in (128) and to the features
on the verb which determine which particular preposition or
set of prepositions are appropriate to particular verbs.
a.
(128) s
N PDP
VP
PREP NP
The issues which bear on the decision of which mechanism
for introducing prepositions will not be discussed here since
the way in which the prepositions is most adequate are intro-
duced into the phrase structure has little if any effect on
the general form of the rules employed in the description of
the predicate complement system.
Having introduced the constituent PREP into the structure
underlying sentence (127f), we observe that the transformations
proposed earlier are fully adequate to the task of generating
the correct derived structure.
(129) Sentence (127f)
First Cycle -- no operations
I decided (PREP [(it] [[John] (represent us] J 3 3 BASE
N NP VP S NP PP
Second Cycle
I decided (PREP [[it] (pOSS [John] (ing+represent
N NP
us ] TCP
VP S NP PP
-147-
mu
I decided CPREP CUit] [[John+POSS1 Crepresent+ing
N NP
us] ]] I T
VP SNP PP AUX
I decided CPREP [[[John+POSS] Crepresent+ing us] ]
NP VP S
] ] T
NP PP
Post Cycle
I decided CON [[[John's] Crepresenting us] ] ] 2 M
NP VP S NP PP
The conclusion that the complement constructions in the
remaining grammatical sentences in the paradigm (127) originate
as noun phrase complements in prepositional phrases stems from
the considerations of simplicity alluded to earlier. If the
claim is made that all of the grammatical sentences in (127)
have an underlying prepositional phrase analysis, then 1) it
is necessary to state for the verb "decide" only that it occurs
before prepositional phrases (rather than before both preposi-
tional phrases and noun phrase) and 2) it is totally unnecessary
to specify any features of the pronominal head in the preposi-
tional phrase except for the feature [+CJ since all comple-
mentizer combinations are possible. Empirical justification
arises when we examine the passive constructions and pseudo-
cleft sentences which are related to the sentences in (127).
-149-
(130) a. 1) that John shall represent us was decided
(on) by me
2) what I decided (on) was that John shall
represent us
b. 1) for John to represent us was decided (on)
by me
2) what I decided (on) was for John to repre-
sent us
Required by this analysis is a preposition deletion rule
which deletes the preposition just in case it appears before
a pronominal head which is marked either C+C][-D], i.e., the
"that" complementizer, or L+CIC-E1, i.e., the "for-to" com-
plementizer. This transformation is defined in Chapter 1 as
the preposition deletion transformation, T . Since the
deletion of the preposition depends upon the contiguity of
the preposition with the pronominal head, it consequently
follows that for the dialects in which the preposition is
not deleted under passivization the preposition deletion
rule must follow the application of the passive transforma-
tion. If the preposition deletion transformation precedes
the passive transformation, then the preposition will invari-
ably be deleted. The following data suggests, however, that
such preposition deletion is rather uncommon.
S.
r
I
-150-
(131) a. 1) they marveled that the plane flew at all
2) that the plane flew at all was marveled at
by them
3) *that the plane flew at all was marveled by them
b. 1) everyone rejoiced that you were happy
2) that you were happy was rejoiced at by
everybody
3) *that you were happy was rejoiced by everybody
c. 1) John insisted that you be here on time
2) that you be here on time was insisted on by
John
3) *that you be here on time was insisted by John
Thus, preposition deletion must follow the passive transforma-
tion and must precede pronoun deletion. Furthermore, it is
likely that the preposition deletion transformation must pre-
cede the pronoun replacement transformation and must follow
the optional complementizer deletion transformation. In
terms of these transformations, we can specify the derivation
of sentence (127a) in the following fashion.
(132) Sentence (127a)
First Cycle -- no operations
I decided EPREP [[it] [John shall represent us ] J BASE
N S NP PP
-151-
Second Cycle
I decided [PREP [[it] [that John shall represent us]
N S
3] C
NPP p
I decided [[[it] [that John shall represent us] ] ] T
N S NP PP PPD
I decided [[[that John shall represent us] 3 3 T
S NP PP
With the application of the passive transformation, we
are able to derive sentences like (131c2)
(133) Sentence (131c2)
First Cycle -- no operations
John insist [PREP [[it] [you be here on time] ] ]
N S NP PP
[by+P] BAS2
MAN
Second Cycle
John insist [PREP [[it] [that you be here on time]
N S
] ] Lby+P] TC
NP PP MAN OP
[[it] [that you be here on time] ] be+en insist
N S NP
[PREP] [by4(dnE] T
PP MAN
[[it] [that you be here on time] 3 be insist+en
N S NP
[pREP] Eby+John] TAUX
PPI' MAN
r -152-
C[that you be here on time] 3 be insist+en LPREP]
S NP PP
[by+JohnJ T
MAN
Post Cycle
[[that you be here on time] was insisted Con]
S NP Pp
Eby+John] M
MAN
An important consequence of the fact that the preposition
deletion transformation follows the application of the passive
transformation is that the same preposition deletion trans-
formation accounts for the obligatory deletion of the agentive
preposition "by." Consider the following examples.
(134) a. 1) for you to see this mess embarrasses me
2) *i am embarrassed by for you to see this mess
3) 1 am embarrassed for you to see this mess
b. 1) to have you visit my home honors me
2) *1 am honored by to have you visit my home
3) 1 am honored to have you visit my home
As is seen in the following derivation, sentences (134a2, b2)
are effectively blocked by the obligatory application of the
preposition deletion transformation.
(135) Sentence (134a)
First Cycle -- no operations
[[it] [[you] [see this mess] I I embarrasses me
N NP VP S NP
by+P] BASE
MAN
Second Cycle
[[it] Efor Eyou] Eto see this mess] 2 1 embarrasses
N NP VP S NP
me Eby+P] TCP
MAN
I be+en embarrass Eby+K[it] Efor Eyou] Eto see this
N NP
mess] ]] ] T
VP S NP MAN
I be embarrass+en [by+[[it] Efor [you] Eto see this
N NP
mess] ]]] T
VP S NP MAN AUX
I be embarrass+en [[[it] [for [you] [to see this
N NP
mess] ] ] ]T
VP S NP MAN
I be embarrass+en [[[for [you] [to see this mess] ]
NP VPS
T
NP MAN
Post Cycle
I am embarrassed [[[for [you] [to see this mess] ]
NP VPS
] ] M
NP MAN
- 154-
This analysis provides a method for deciding on the
derivational status of the morpheme "for" in places where
there is no a priori reason to identify this morpheme as
either the preposition "for" or the complementizer "for."
Consider, for instance, the following sentences.
(136) a. I hope for you to come on time
b. I hope that you will come on time
Superficially, these sentences appear to be instances of ob-
ject complementation. Consideration of the pseudo-cleft
sentences in (137) indicates, however, that a prepositional
phrase analysis is, perhaps, more appropriate.
(137) a. what I hope for is for you to come on time
b. what I hope for is that you will come on time
Since we know from earlier considerations that the preposi-
tion deletion transformation is obligatory when the preposition
precedes a pronominal head which is marked either L+CJL-DJ or
E+CJL-E], the conclusion that the "for" in the sentence (136a)
is the complementizer "for" follows as a consequence. This
morpheme could be a preposition only at the expense of making
the preposition deletion rule less general, that is, by making
it optional for verbs like "hope" when the complementizer is
"for-to." Consider now the derivation of the sentence (136a)
in terms of the analysis in which the preposition deletion
transformation is obligatory.
' r
(138) Sentence (136a)
First Cycle -- no operations
I hope (PREP [[it] (you] [come on time] ] ] 3 BASE
N NP VP S NP PP
Second Cycle
I hope (PREP [[it] (for (you] (to come on time] ]
N NP VPS
3 ] T7
NP PP OP
I hope [[(it] (for (you] (to come on time] ] 3]JI T
N NP VP S NPPP PPu
I hope ([[for (you] [to come on time] ] ] ] T
NP VP S NP PP
Despite the fact that the optional complementizer deletion
transformation is impossible for verbs like "hope," it is not,
in general, impossible for prepositional phrase complement con-
structions, as we see in the following examples.
(139) a. 1) I thought about the world's coming to an end
2) 1 thought about the world coming to an end
b. 1) I worried about the world's coming to an end
2) 1 worried about the world coming to an end
Thus, the prepositional phrase noun phrase complements present
optional complementizer deletion options in the same way as do
object complement constructions.
There are many other prepositions which participate in
constructions of the type under discussion. All of them seem
to share the same properties as "on" and "for." Thus we shall
not take the time to illustrate any of the derivations, but
it is nonetheless clear that the rules proposed earlier are
fully adequate for the generation of the sentences in the
following paradigms as well as many others.
(140) a, John persisted in reading
b. John persisted to read
(141) a. I marveled at your being so late
b. I marveled that you were so late
(142) a. I admit to being lazy
b. I admit that I am lazy
(143) a. I boasted of being strong
b. I boasted that I was strong
B. Transitive Oblique Noun Phrase Complementation
The necessity of positing transitive oblique noun phrase
complement constructions, i.e., where the phrase structure
rules of Chapter 1 produce the configuration given in (144),
stems, once again, from consideration of the pseudo-cleft
sentences. Let us compare, for example, the two sentences
in (145) with respect to this construction.
r
-U
-157-
144)
N PPP Ip
V lP PP
PREP P
DET
(145) a. I reminded John to visit his ailing mother
b. I defied John to visit his ailing mother
(146) a. what I reminded John of was to visit his ailing
mo ther
b. *what I defied John (of) was to visit his ailing
mother
The fact that the pseudo-cleft sentence is possible for (145
and not (145b) suggests that underlying the former is a noun
phrase complement construction. Furthermore, the occurrence of
the preposition "of" in (146a) suggests that this noun phrase
complement originates as the noun phrase of a prepositional
phrase.
This analysis gains additional support from the occurrence
of the "that" complementizer in sentences containing the verb
-158-
"remind" but not in sentences containing the verb "defy."
(147) a. I reminded John that he should visit his
ailing mother
b. *I defied John that he should visit his ailing
mother
We might reason that "defy" is a verb which simply does not
allow the "that" complementizer, but this judgment would not
help to explain the fact that the pseudo-cleft sentence is
as possible for (147a) as it is for (145a), as we see in (148).
(148) what I reminded John of was that he should visit his
ailing mother
Furthermore, if we adopt the position suggested by the facts,
namely that (145a) and (147a) are instances of preposition
phrase complementation where the verb is transitive, it be-
comes possible to simplify the strict subcategorization
information required by the verb. More specifically,
sentences like (149) indicate that verbs like "remind,"
"convince," "persuade, " and others may occur before a noun
phrase object which itself occurs before a prepositional
phrase.
(149) a. I reminded John of the fact that he is late
b. *I defied John of the fact that he is late
-159-
Assuming that sentences like (14a) and (147a) have a prepo-
sitional phrase in the underlying structure, we require no
additional strict subcategorization information to be in-
cluded in the lexical entry for the verb. In short, there
is every advantage to be gained by making this assumption.
We observe in the following derivations that the trans-
formational apparatus employed thus far is fully capable of
generating sentences (1 4 5a) and (147a).
(150) Sentence (14a)
First Cycle no operations
I reminded [John] [PREP [[it] [[John] [visit his
NP N NP
mother] ] ] ] BASE
VP S NP PP
Second Cycle
I reminded [John] [PREP [[it] [for [John][to visit
NP N NP
his mother] ] ] ]TCP
VP S NP PP
I reminded [John] [PREP [Cit][for [to visit his mother]
NP N VP
]] ] T I
S NP PP IE
I reminded [John] [[[it] [for [to visit his mother]
NP N VP
] ] T
S NP PP
mu
-160-
I reminded tJohn] U[[for Ito visit his mother] ] ] ] Tr
NP VP S'NPPP
I reminded UJohn][Lto visit his mother] ] I ] TCD
NP VP S NF PP
(151) Sentence (147a)
First Cycle -- no operations
I reminded [John] CPREP ECit][he should visit his
NP N
mother] ] ] BASE
S NP PP
Second Cycle
I rerbinded LJohn] EPREP Cit][that he should visit
NP N
his mother] ] ] T
S NPPP OP
I reminded LJohn] [[Cit][that he should visit his
NP N
mother] I T T'
S NP PP
I reminded LJohn][C[that he should visit his mother]
NP S
] T
NPPP P
There are several constructions in English which resemble
those discussed above in certain respects, but which also pre-
sent a couple of curious problems. Consider, by way of illustra-
tion, the following sentences.
(151) a. I prevented the doctor from examining John
b. I prevented John from being examined by the doctor
lommommopp- I i mm
rl-
-161-
It will be observed immediately that sentences (151a) and
(151b) have the same truth value synonymy, a fact which could
not be explained on the assumption that the sentences in (151)
are instances of transitive oblique noun phrase complementation
as are the sentences (145a) and (149a). On this assumption,
the underlying structures, and hence the semantic interpreta-
tion, of the sentences in (151) are different. In (151a),
the underlying object of "prevent" is "the doctor,t " while in
(11b), the underlying object is "John.
The second problem which arises with respect to (151) is
that a transitive oblique noun phrase complement analysis pre
dicts incorrectly the grammaticality of the pseudo-cleft
sentences in (152).
(152) a. *what I prevented the doctor from was examining
John
b. *what I prevented John from was being examined by
the doctor
Both of these problems can be resolved if it is assumed
that the morpheme "from" in (151) is not an instance of PREP,
but a complementizer of the basic form "from-ing.1" Consider
the following derivation in terms of this assumption.
(153) Sentence (131a)
First Cycle -- no operations
I prevented [[it] ECJohn] [go] ] ] BASE
N NP VPS NP
- Er
-162-
Second Cycle
I prevented [[it][from [John] [ing go ] TCPN NP VP SNP
Since extraposition is blocked in the event the complementizer
is "POSS-ing" but not if the complementizer is "from-ing,"
extraposition can apply to the above string. It would appear,
in fact, that extraposition is obligatory as it is for the
"for-to" complementizers for the "believe" class of verbs.
I prevented [[it] ] [from [John] [ing go) 3 T
N NP NP VP S B
I prevented [[it] ] (from [John] [go+ing] ] TAUX
N NP NP VP S
I prevented (John] [from [go+ing] ] T
NP VP S PR
If we prevent the obligatory complementizer deletion trans-
formation from applying to the "from" complementizer, the
grammar correctly generates the following string.
Post Cycle
I prevented [John] [from [going] 3 M
NP VP S
This analysis has several virtues which offset, perhaps,
the cost of the required restriction on the obligatory comple-
mentizer deletion transformation. First, the synonymy of the
sentences in (151) is explained by virtue of the fact that the
two sentences do not differ in their underlying structures
except for the constituent marking the obligatory passive
transformation which does not affect the semantic inter-
pretation. Second, the non-occurrence of the pseudo-
l63-
cleft sentences in (152) is explained in precisely the same
way as the non-occurrence of such sentences in constructions
containing main verbs of the "believel" class, that is, the
application of pronoun replacement transformation destroys
the environment on which the pseudo-cleft sentence transforma-
tions must be defined. Similarly, this analysis explains the
introductory "there" phenomenon in sentences like (154).
(154) a. Wyatt Earp prevented there from being trouble on
the range
b. shelters will not prevent there from being great
destruction
Finally, this analysis allows us to explain the synonymy of
the sentences in (155) with those in (151).
(155) a. I prevented the doctor's examining John
b. I prevented Johnts being examined by the doctor
In the sentences in (155) the complementizer is "POSS-ing"
rather than "from-ing" but the underlying structures of the
two pairs of sentences are identical in every respect. Their
semantic interpretations, therefore, must be the same.
The introduction of a new complementizer, "from-ing,"
seems to be necessary as there is no other immediately ap-
parent analysis which can account for as wide a range of facts
with similar economy. But this conclusion raises a host of
Elm
r-P"
-164-
new questions concerning the status of other putative prepo-
sitions and concerning the adequacy of the descriptive apparatus
proposed in this study in the light of whatever new comple-
mentizers might be discovered in following up this tentative
analysis. These questions go far beyond the range of the
present study. There is no independent justification at the
present time for the complementizer "from-ing" other than
that such a postulation accounts for the facts thus far
inves tigated.
NOTES
1. It is unlikely that all of the paradigm sentences will
be judged grammatical in a given dialect. These
sentences simply illustrate the range of possible
sentences based upon this complement construction which
the author has observed in several dialects of English.
2. For the sake of convenience, the pronominal head of a
noun phrase complement construction, +NRj will be repre-
+PRO
sented as Eit2 in all subsequent derivations. Similarly,
complementizers will be spelled out in the derivations
rather than specified in terms of features. In the
derivations beginning with (13), the abbreviation, M,
stands for morphophonemic rules which will not be
elaborated in this study.
3. There are certain exceptions to this formulation. Con-
sider sentences like the following:
a. I ask that John be allowed to come
b. *I ask John be allowed to come
It is perhaps not beside the point to note that verbs
like "ask" belong to that small class of verbs which
are exceptions to the erasure principle, as we see below.
a. I asked John to go
b. I asked of John to be allowed to go
That matter is discussed briefly later in Chapter 4.
4. Cf. N. Chomsky, "Current Issues in Linguistic Theory,"
in The Structure of Lnguae, eds. J. A. Fodor and J.
J. Katz, Englewood Cliffs, 1964.
5. This follows from the fact the extraposition transforma-
tion is blocked in the event that the complementizer is
"POSS-ing. "
6. The reader can readily test this assertion by comparing
the number of complementizer features required by any
three case formulations with the number required in the
statement of the optional complementizer deletion trans-
formation given in Chapter 1.
-166-
7. This transformation would never apply in the event
that structural indices 4 and 6 are not present.
Furthermore, it could never be the case that both
4 and 6 would be present in the structural description.
8. The remarks in note 1 of this chapter hold with respect
to the "for-to" paradigm.
9. In this instance the derived subject is the erased NP.
10. The origin and interpretation of the "have" in this
derivation will not be discussed, but it appears to be
a reflection of the past tense required by this class
of verbs with the "for-to" complementizer.
li. Cf. T. Bever and P. Rosenbaum, "The Psychological
Verification of Linguistic Rules," (forthcoming).
12. Observe that the subject of this complement sentence
can be deleted only when it is a pronoun like "someone.
13. In addition, consider sentences like the following:
a. I demand of you to be allowed to come
b. I ask of you to be allowed to come
c. I request of you to be allowed to come
r
CHAPTER 5
Verb Phrase Complementation
The reasons for postulating the existence of verb phrase
complement constructions in English are the converse of those
requiring the postulation of noun phrase complementation.
Verb phrase complement constructions, generated by phrase
structure Rule 1 in Chapter 1, require a derivational apparatus
which makes use of the same set of transformations that we
have seen to be necessary to the system of noun phrase com-
plementation. In a sense, therefore, the next few pages con-
stitute a review of the transformational machinery discussed
thus far as these rules apply to a different underlying
structure. It is a measure of the adequacy of these rules
that they generalize, as we shall see, to the more common
verb phrase complement constructions of English.
I. Intransitive Verb Phrase Complementation
In proposing noun phrase complementation, we had reason
to refer to pairs of sentences like the following.
(1) a. 1) Bill prefers to stay here
2) what Bill prefers is to stay here
3) to stay here is preferred by BillI4) what is preferred by Bill is to stay here
-167-
-168-
b. 1) Bill condescended to stay here
2) *what Bill condescended was to stay here
3) *to stay here was condescended by Bill
4) *what was condescended by Bill was to stay here
We were able to explain the fact that all four of the sentences
in (la) are grammatical by assuming that the phrase "to stay
here" was a noun phrase complement. On this assumption
passivization is perfectly appropriate since this phrase is
dominated by a NP in the underlying structure. Furthermore,
we can understand the grammaticality of the pseudo-cleft
sentences of both the passive and active versions of this
sentence on the assumption that the pseudo-cleft sentence
transformations apply to instances of NP. In explaining
the non-occurrence of sentences (lb2,3,4), we can say either
that the passive transformation and the pseudo-cleft sentence
transformations do not apply for verbs such as ''condescend
(assuming a noun phrase complement analysis for (Ib)) or we
can adopt the far more economical position that (lb) is
simply not an instance of noun phrase complementation in
the first place in which case the non-application of the
passive transformation and the pseudo-cleft sentence trans-
formations follow as a necessary consequence. Taking the
latter alternative, we were also able to demonstrate that
the phrase "to stay here" must be an instance of verb phrase
complementation for to assume the contrary would make it
-169-
difficult if not impossible to arrive at a reasonably simple
formulation of the identity erasure transformation. In other
words, we were able to show that the underlying structure for
the sentence (lb) was roughly the following configuration.
(2)
NP 
AMV
Bill condescend Bill stay here
It is immediately apparent that the derivation of (11)
from the underlying structure (2) requires no rules in addi-
tion to those already motivated. The derivation can be
specified in the following fashion.
(3) Sentence (Ib)
First Cycle -- no operations
Bill condescendedtEBill] [stay here] J BASE
NP VP S
-U r -170--
Second Cycle
Bill condescend (for (Bill] (to stay here] ] T
NP VPS C
Bill condescended (for (to stay here] ] T
VPzsE
Bill condescended ((to stay here] ] T CD
VP S
There is little more to say about the analysis of intransi-
tive verb phrase complementation except to point out that the
derivation of all verb phrase complement constructions depends
upon the obligatory identity of the erasing and erased noun
phrases, in this case the subject of the complement sentence
and the subject of the main sentence. Thus, for example,
sentences like (4) are impossible.
(4) *I condescended Bill to go
We recall, however, that the identity of erasing and erased
noun phrases is not always a necessary condition for the deri-
vation of noun phrase complements, a fact attested by the
sentences in (5).
(5) a. I hate for John to to
b. I hate to go
Indeed, in the system of noun phrase complements we even find
cases where non-identity seems to be a necessary condition for
derivation.
I r- -W170-9
-171-
(6) a. I said for John to go
b. I said (for someone) to go
c. *1 said for me to go
The necessary identity of erasing and erased noun phrases
also holds for certain fairly major classes of noun phrase
complementation. For instance, there is no instance of a
prepositional noun phrase complement for a transitive verb
where identity of erasing and erased noun phrases is not a
necessary condition for derivation. As an example, consider
the following sentence.
(7) a. 1) *1 persuaded John for Bill to come
2) I persuaded John to come
b. 1) *1 reminded John for Bill to visit his ailing
mother
2) I reminded John to visit his ailing mother
There may well be an explanation for the necessary identity of
erasing and erased noun phrases in the structures just presented,
but this issue will not be taken up in the present study.
II. Transitive Verb Phrase Complementation
Given the rules already presented, the derivation of
transitive verb phrase complement constructions is quite
straightforward. We can establish the existence of such con-
structions by comparing the following sets of sentences.
-172-
a ) 1) somebody prefers for John to do the work
2) what somebody prefers is for John to do the
work
3) for John to do the work is preferred
4) what is preferred is for John to do the work
b. 1) somebody trusts John to do the work
2) *what somebody trusts is for John to do the work
3) *for John to do the work is trusted
4) *what is trusted is for John to do the work
Applying the same reasoning to the paradigm (8) as was just
applied to the paradigm (1), we arrive at the conclusion that
the sentence (8bl) is an instance of transitive verb phrase
complementation having the underlying structure given in (9).
The non-occurrence of sentences (8b2,3,4) is thus predicted.
(9)
S
N PDP
V N
IN VP
somebody trust John ohn do the work
I-.
It seems reasonably clear that the "to" deletion transformation
F -173-
On the basis of the underlying structure (9), we can specify
the derivation of (8bl) in the following fashion.
(10) Sentence (8bl)
First Cycle -- no operations
somebody trusts [John] [[John] [do the work] ] BASE
NP NP VP S
Second Cycle
somebody trusts [John] [for [John] [to do the work] ] T0 P
NP NP VP S
somebody trusts [John] (for [to do the work] ] T
NP VP S IE
somebody trusts [John] [[to do the work] ] T
NP VP S CO
This derivational procedure is common to all verb phrase
complement constructions. There is certain additional apparatus
required to handle those cases in which the complementizer "to"
is either obligatorily or optionally deleted, as in (11), but
the details of this operation are sufficiently transparent so
that it is probably unnecessary to go into further detail on
this phenomenon.
(11) a. 1) 1 helped John to build the house
2) 1 helped John build the house
b. I let John go
c. I made John go
-174-
will have to be dependent upon some sort of verbal marker
since the deletion of the "to" is an idiosyncratic property
of an extremely small number of verbs.
III. Prepositional Verb Phrase Complementation
The necessity of positing underlying structures of this
type follows from a consideration of the paradigm below.
1
(12) a. 1) they prevailed upon John to represent us
2) what they prevailed upon John for was to
represent us
b. 1) they hit upon John to represent us
2) *what they hit upon John for was to represent us
These data suggest a derivation for sentence (12b1) which
follows the pattern set by the derivations of the verb phrase
complement constructions in the preceding two sections.
(13) Sentence (12bl)
First Cycle
they hit [PREP John] [[John] [represent us] ] BASE
PP NP VP S
Second Cycle
they hit EPREP John] [for EJohn] Eto represent us) ] TCP
PP VP S
they hit EPREP John] Efor Eto represent us] ] TIEPP VP S
-175-
they hit LPREP John] L[to represent us] ] T
PP VP S3CD
Post Cycle
they hit [upon John] [[to represent us] ] M
PP VP S
IV. Special Problems
The class of verbs which includes such items as "taste,'t
"smell1, ""feel " and others seems to require certain additional
apparatus.2 Consider, for instance, the following sentences.
(13) a. the meat tastes salty to me
b. the milk smells good to me
c. the batter feels lumpy to me
We observe, however, that such sentences as those in (14) have
an entirely different status from those in (13).
(14) a. *the meat tastes salty to the tree
b. *the milk smells good to the pencil
c. *the batter feels lumpy to the spoon
The restrictions which are imposed upon the noun phrase in
the prepositional phrases in (13) and (14) we observe to be
identical to those imposed upon the subjects of the following
sentences.
(15) a. 1) I taste the meat
2) I smell the milk
3) 1 feel the batter
rb. I)
a)
3)
-176-
*the tree tastes the meat
*the pencil smells the milk
*the spoon feels the batter
The statement of the restrictions noted in the above cases
can be generalized, that is, stated only once in the grammar,
if we assume that the sentences in (13) are derived from
sentences in which the noun phrase observed in the prepositional
phrases of (13) originate as the underlying subjects of these
sentences. In other words, we propose that the sentences in
(13) are actually instances of verb phrase complementation and
that, in the course of the derivation, the subject and object
of the main sentence are inverted. We may specify this deri-
vation in the following fashion.
(16) Sentence (13a)
First Cycle -- no operations
I taste [the meat] [[the meat] [be salty] ]
NP NP VPS
Second Cycle
I taste [the meat] [for [the meat] [to be salty] ]
NP VPS
BASE
I taste [the meat] [for [to be salty] ]
VP S
The identity erasure transformation must precede the subject-
object inversion transformation. Otherwise, the erasure prin-
ciple will fail.
T1
TCP
IE
Cthe meat] taste Efor Lto be salty] 3 to+me T
NP VP S
The passive transformation must follow the subject-object
inversion transformation since if it does not, the grammar
generates sentences like "the meat was tasted by me to be
salty." Note furthermore that since the subject-object in-
version transformation places the subject noun phrase to the
right of the verb phrase complement sentence, the non-
application of the passive transformation follows as a con-
sequence.
Ethe meat] taste [Eto be salty] ] to+me T
NP VP S
[the meat] taste [[salty] ] to+me T
NP VP S TB
The transformation TT which was not defined in Chapter 1
plays a role in a variety of complement constructions. Thus,
for example, one finds sentences like "I consider John to be
a fool" and "I consider John a fool."
Post Cycle
Cthe meat] tastes CEsalty] ] to+me M
NP VP S
The subject-object inversion transformation is not without
its difficulties but it does appear to be a tentative approxi-
mation of a rule which is clearly necessary in some form. It
-178-
is not inconceivable that a rule of this general type will
have an important extension in the analysis which attempts to
relate the sentences in the following pairs.
(17) a. 1) I value the book
2) the book is valuable to me
b. 1) I benefit from your kindness
2) your kindness is beneficial to me
But the development of strong motivation for formulating a
general subject-object inversion transformation does not lie
within the scope of the present work and the employment of
this rule to handle cases like those in (13)-(16) is to be
taken more as suggestive than as definitive.
flr
one179-"n
NOTES
1. We observe that (12a) is actually an instance of
oblique noun phrase complementation since there is
no other analysis which could explain the occurrence
of the preposition "for" in the pseudo--cleft sentence
(12a2).
2. This problem was suggested to the author by Paul
Postal.
-180-
CHAPTER 6
Complementation in Adjectival Predicate Constructions
The behavior of predicate complement constructions in
adjectival structures is, in many ways, remarkably similar to
the behavior of these same constructions in the verbal struc-
tures discussed in earlier chapters. Certain of these
adjectival constructions will be given brief consideration
in this chapter. The analyses provided should be taken more
as suggestive than as definitive since it will become immedi-
ately clear that a fairly wide range of phenomena have not
been taken into consideration. For instance, there is little
doubt that a more insightful analysis will reveal many in-
stances where adjectival complement constructions are probably
to be considered as derivationally related to cognate verbal
complement constructiOns. The purpose of the brief exposi-
tion in this chapter is simply to show that many of the
transformations discussed earlier are equally applicable in
cases where the predicate of the sentence is, in its derived
structure at least, an adjective.
I. Prepositional Noun Phrase Complements
A great many adjectives take propositional phrases which
may contain a noun phrase complement construction. Although
-181-
very few participate in a complete paradigm, allowing all
three complementizer possibilities in the complement sentence,
among this class of adjectives one finds the adjective "scared."
Consider the following paradism.
(1) a. I am scared of leaving home at this time
b. I am scared that I will not be able to do the work
c. I am scared to find out the truth.
Evidence supporting the position that the b" and "c" sentences
in the paradigm are instances of prepositional noun phrase com-
plementation as well as the "a" sentence consists in the fact
that the pseudo-cleft sentences relating to the sentences in
the paradigm are grammatical.
(2) a. what I am scared of is leaving home at this time
b. what I am scared of is that I will not be able to
do the work
c. what I am scared of is to find out the truth
The fact that the sentences in the paradigm bear an obvi-
ous relation to the verbal sentences in (3) does not detract
from the necessity of positing structures of this type since,
as the pairs in (4) attest, there is no immediately apparent
verbal counterpart to the great majority of adjectives in this
class.
r
-*---*-*-KWR" , I I IN
-182-
(3) a. leaving home at this time scares me
b. it scares me that I will not be able to do the work
c. it scares me to find out the truth
(4) a. 1) I was aghast at Johnt s hitting Bill
2) I was aghast that John hit Bill
b. 1) I am aware of Johnis having left
2) 1 am aware that John left
The derivation of the sentences in this paradigm is perfectly
straightforward and, as we see in the following illustration,
includes only the transformational machinery which is requi-
site to the derivation of predicate complement constructions
in verbal sentences.
(5) Sentence (4bl)
First Cycle -- no operations
I am aware LPREP [[it] [[John] [have left] 3 3 3 BASE
N NP VP S NP PP
Second Cycle
I am aware CPREP [[it] EPOSS [John] Cing have left]
N NP VP
S NP PP
I am aware [PREP [it] [[John+POSS] (have+ing left]
N NP VP
T
SNP??? AUX
I am aware [PREP [[[John+POSS] [have+ing left] ] ] 3 ] T
NP VP S NP PP
Post Cycle
I am aware [of [[[Johnts] [having left] M ] N
NP VP S NP PP
(6) Sentence (4ba)
First Cycle -- no operations
I am aware (PREP [[it] [[John] Cleft] ] 3 BASE
N NP VP S NP PP
Second Cycle
I am aware [PREP [it] [that [John] Cleft] 3 3 TCP
N NP VP S NP PP
I am aware [[[it] [that [John] Cleft] ] ] 3 T
N NP FP S NP PP
I am aware [[[that [John] Cleft] ] ] ] T
NP VP S NP PP
Paralleling verbs like "hope, " which require the prepo-
sition "for" in the underlying structure, one finds adjectives
like "eager, " "anxious," "ready,' and several others. Consider,
for example, the following sentences.
(7) a. I am eager for John to get going
b. I am anxious for you to see my etchings
c. I am getting ready to take the examination
The necessity of postulating an underlying preposition phrase
analysis follows, once again, from the data provided by the
pseudo-cleft sentence constructions.
-183-
-IS4-
(8) a.
b.
c.
what I am eager for is for John to get going
what I am anxious for is for you to see my etchings
what I am getting ready for is to take the examination
There are a great many other adjectives which take preposi-
tional noun phrase complements with prepositions other than
those mentioned but they present sufficiently few additional
difficulties that further study is not necessary at this time.
II. Subject Complementation
Subject noun phrase complementation is as productive a
process in adjectival constructions as it is in verbal con-
struction. Consider, by way of illustration, the sentences
in the following paradigm.
(9) a. 1)
2)
3)
b. 1)
2)
3)
c. I)
2)
3)
that John decided to fight is admirable
that he wants to do it is reasonable
that nothing works here is peculiar
for him to want to go is admirable
for you to wish to make money is reasonable
for him to say that was peculiar
John's being prompt is admirable
his wanting to go is reasonable
it* being so cold in here is peculiar
Thpre are two bits of evidence which support the view that the
above sentences constitute instances of noun phrase complementa-
tion. First, we observe that extraposition is possible in the
r
event that the complementizer is either "that" or "for-to.
Extraposition, as we have seen much earlier, depends upon
the contiguity of the constituent S with the pronominal head
of the complement construction.
(10) a. 1) it is admirable that John decided to fight
2) it is reasonable that he wants to do it
3) it is peculiar that nothing works here
b. 1) it is admirable for him to want to go
2) it is reasonable for you to wish to make money
3) it Was peculiar for him to say that
Second, the pseudo-cleft sentence construction is possible for
all sentences in the paradigm (9).
(11) a, what is admirable is that John decided to fight
b. what is reasonable is for you to wish to make
money
c. what is peculiar is his wanting to go
The derivations of the sentences discussed above parallels the
derivations proposed for subject complementation in verbal
constructions and there seems little reason to go through
these derivations here.
Interestingly enough there are several adjectives like
"likely" which parallel verbs like "happen" in requiring, at
I least in the case of the "for-to" complementizer, obligatory
-186-
extraposition and pronoun replacement. Consider, for example,
the following series of sentences.
(12) a. we are likely to be ready on time
b. *for us to be ready is likely
c. *it is likely for us to be ready
We could explain the grammaticality of (12a)in terms of a
verb phrase complement analysis, but in doing this we would
be 1) losing in economy and 2) failing to explain certain
semantic facts. Concerning the first point, it is clearly
more expensive to assign to distinct analyses to adjectives
like "likely, " a subject complement analysis for cases like
(13) and a verb phrase complement analysis to handle (12a).
(13) a. that we will be ready on time is likely
b. it is likely that we will be ready on time
Secondly, if we assume a verb phrase complement analysis for
(12a), we fail to explain the truth value synonymy of the two
sentences in (14).
(14) a. the egg is likely to be broken by John
b. John is likely to break the egg
If, on the other hand, we assume a subject complement analysis,
this synonymy is explained automatically since the two sentences
L.
-187-
differ only in the application of the passive transformation
to a common underlying structure in the case of (14a). Thus,
the derivation of the sentence (14a) might be given as follows.
(15) Sentence (12a)
First Cycle
[[it] [[John] (break the egg [by+P] J 3 3 be likely BASE
N NP MAN VP S NP
[[it] [[the egg] [be+en break [by+John ]be likely T
N NP MAN VP SNP
[[it] [[the egg] [be break+en [by+JohnJ ] 3 3 be likely T AUX
N NP MAN VP S3NP
Second Cycle
[[it] [for [the egg] [to be break+en [by+John] 3 3 3
N NP MAN VP S NP
be likely T
[[it] 3 be likely [for [the egg] [to be break+en [by+
N NP NP
John] ] 3 T
MAN VPS E
[the egg] be likely [for [to be break+en [by+John]
NP MAN
T
VP S PR
[the egg] be likely [[to be break+en [by+John] ] ]T
NP MAN VP S CD
Post Cycle
[the egg] is likely [[to be broken [by+John] ] 3 M
NP MAN VP S
U Er
F"
-188-
In terms of certain other subject complement constructions
the generality of the pronoun replacement once again becomes
an issue. Consider, for instance, the following sentences.
a.
b.
c.
a.
b.
c.
for John to go is important
it is important for John to go.
*John is important to go
for you to see a doctor would be worthwhile
it would be worthwhile for you to see a doctor
*you would be worthwhile to see a doctor
We have two ways of explaining the non-occurrence of the "c"
sentences in (16) and (17). We might say, simply, that these
particular adjectives do not undergo the pronoun replacement
transformation and establish an appropriate system of verbal
markers accordingly. There is a second possibility which is
slightly more appealing because of its greater generality.
We observe that it is just the adjectives in question which
apparently.allow a prepositional phrase following the adjective
in the underlying structure.
a. 1) it is important (to somebody) for John to go,
2) it is important to John to go
b. 1) it would be worthwhile for the doctor for
you to see him
2) it would be worthwhile for you for you to
see the doctor
3) to see the doctor would be worthwhile for you
(16)
(17)
(18)
-189-
These data suggest the possibility of explaining the non-
occurrence of the "c" sentences in terms of the order of the
deletion of the propositional phrase. Although the details
of this deletion go beyond the scope of the present study,
it is clearly a strong possibility that the pronoun replace-
ment transformation may precede whatever transformation deletes
the prepositional phrase in which case the former is automati-
cally blocked. Even though the propositional phrase deletion
rules may subsequently apply, the pronoun replacement trans-
formation no longer can.
III. Verb Phrase Complementation
There are several adjectival complement constructions
which, on the basis of the criteria discussed in this study,
seem to call for a verb phrase complement analysis. Consider,
for example, the following sentences.
(19) a. 1) John was happy to leave early
2) *what John was happy (PREP) was to leave early
b. 1) John was unable to see things clearly
2) *what John was unable was to see things clearly
c. 1) you are welcome to come
2) *what you are welcome is to come
The non-occurrence of the cleft sentences does not prove, how-
ever, that these adjectival constructions do not belong to the
same class as those discussed with respect to the adjective
-190-
"likcely." In other words, the pseudo-cleft sentence might
be obviated by the application of the pronoun replacement
transformation. But this possibility is negated because of
the fact that, as we observe in (20), the sentences in each
pair do not have the same truth value synonymy, a fact which
is explained by assuming distinct underlying structures.
(20) a. 1) the doctor was happy to examine John
2) John was happy to be examined by the doctor
b. 1) the nurse was unable to attend to John today
2) John was unable to be attended to by the
nurse today
We thus conclude that the grammatical sentences in (19)
are instances of verb phrase complementation and to be derived
in exactly the same manner as the intransitive verb phrase
complement construction discussed in Chapter 5.
IV. Special Problems
One of the most recalcitrant problems in the predicate
complement system concerns the extremely productive class of
adjectives including such itets as "clever, " "wise, " "honor-
able," and at least a couple of hundred others. The specific
problem concerns the relatedness of the sentences in (21)
with the sentence(22).
UE
I
-191-
(21) a. to leave early was wise of John
b. it was wise of John to leave early
(22) John was wise to leave early
Certain possibilities suggest themselves, but, as we shall see,
none of them are without difficulties.
One of the more interesting analyses makes use of the
pronoun replacement transformation. What makes the use of
this transformation possible is the curious set of identity
erasure options which exists for this class of adjectives.
More specifically, we observe the following possibilities.
(23) a. it was wise of John for him to leave early
b. it was wise of John to leave early
c. it was wise for John to leave early
In other words, we apparently have the option of deleting
either the "erasing" noun phrase in the main; sentence or the
"erased" subject noun phrase in the complement sentence.
This property is common to another class of adjectives, the
class containing adjectives like "difficult," as in (24).
(24) a. it was difficult for John for him to pass the exam
b. for John to pass the exam was difficult
c. to pass the exam was difficult for John
Returning to the sentences in (23), we observe that if the
-192-
option of deleting the connected noun phrase in the main
sentence is taken, the pronoun replacement transformation
is defined (pn the string "it was wise Efor John to leave
early]") and we may generate sentence (22) accordingly.
A second possibility, although one which is beset by
difficulties more severe than simply a lack of knowledge
concerning the rules governing connection erasure options,
attempts to relate (22) and perhaps the sentences in (21)
to the following sentence*-.
(25) John was wise in leaving early
To generate (22) on the basis of the structure underlying
sentence (25) requires us to posit that (22) is an instance
of prepositional complementation. In this view, the only
difference between (22) and (25) is that the complement
sentence in the former contains the "for-to" complementiser
while the latter contains the "POSS-ing" complementizer.
Several unfortunate findings ensue from a careful study of
this analysis. First, we find that the noun phrase comple-
ment construction proposed here is the first in this entire
study for which the pseudo-cleft sentence is impossible, as
we see in (26). One is naturally reluctant to give up what
has proven to be such a powerful generalization.
(26) *what John was wise in was to leave early
aw- -193-
If we make a deeper claim and assert that the prepositional
phrase structure in (25) underlies (21) and (22) and that an
inversion rule- similar to that discussed in Chapter 5
establishes a structure from which the sentences in (21) and
(22) are subsequently derived by the transformational apparatus
already motivated, we can perhaps explain the non-occurrence
of (26). But this involves us in other messy problems such
as explaining why this inversion rule is obligatory if the
complementizer is "for-to" and optional otherwise and how
it happens that the proposition "in" is necessarily deleted
so as to generate (27a) and not (27b),
(27) a. leaving early was wise of John
b. leaving early was wise of John in
In short, there is no way of relating the sentences in
(21) and (23) which follows as a consequence of any of the
considerations raised in the present study. Such a proposal
seems to raise more problems than it solves. Furthermore,
it seems to be the case that the derivation of (22) is more
appropriately handled in terms of the structure underlying
the sentences in (21) than in terms of the structure under-
lying (25). It is quite possible that a broader study of
these phenomena will yield an insightful analysis in which
the constraints governing the classification of complement
constructions suggested in this study will prove to be
-194-
artificial, And it is no doubt with respect to the adjectives
under discussion that future research may be expected to yield
more informative results.
The last construction which deserves mention is the class
of adjectives including "difficult," "easy," and several others.
These adjectives demand the formulation of an entirely new
transformation, one which takes the final noun phrase in the
verb phrase of the complement sentence and substitutes it
for the pronominal head after the application of the extra-
position transformation. Consider, for instance, the following
sentences.
(28) a. for John to hit Bill is difficult
b. it is difficult for John to hit Bill
c. Bill is difficult for John to hit
The transformation required to generate sentence (28c), which
probably depends upon the prior application of whatever rule
deletes the erasing NP in the derivation of sentences like
(22), has roughly the following form.
(29) W L[it] AUX ADJ Et+o] NP AUX [V (PREP) NP] X Y
N iP VP S
1 2 3 4 5 6 7 8 9 10 11 12
n=n==>l1,l0,3,4,5,6,7,8,9#,11,12
This summary of predicate complementation in adjectival
structures is by no means complete. The reason for this brief
-195-
discussion is to show that, with respect to predicate comple-
mentation, verbal and adjectival constructions share many
properties in common. It would not speak well for the
rules proposed in this study if they cannot be generalized
to predicate complementation in both verbal and adjectives
constructions. It may be the case, however, that the general
adequacy of the rules for adjectival complementation is more
a function of shallow depth of the present study of adjectival
complementation than it is a true reflection of the near
symmetry of the systems of verbal and adjectival comple-
mentation. More substantial study of adjectival complementa-
tion is required before the two systems can be rigorously
compared. The most conservative assertion, namely, that the
rules required to handle predicate complementation in verbal
constructions also handle many cases of predicate complementa-
tion in adjectival constructions, is probably the strongest
claim which can be made at the present time.
r
-196-
NOTES
I. Certain liberties are being taken in the assumption
that words such as "likely, " "sure" and "certain"
are adjectives. There are grounds for arguing that
these items are somewhat peculiar adverbs, This fact
is not overly consequential should it turn out that
lexical items must be marked in some fashion for the
application of the pronoun replacement transformation.
The present study suggests certain ways of avoiding
having to mark lexical items for the application of
pronoun replacement, e.g. ordering of prepositional
phrase deletion and pronoun replacement. We should
not be surprised, however, if subsequent research
shows that lexical markers for pronoun replacement
are absolutely necessary.
- Er
CHAPTER 7
A Historical Perspective
The most provocative work on the predicate complement
constructions of English falls into two major categories: the
so-called traditional approach, best characterized by the work
of Poutsma and Jesperson, and the generative transformational
approach including primarily the work of Chomsky, Lees, and
Fillmore. The traditional view of complementation is based
upon descriptive methodology underlying which is the goal of
explicating the structure of sentences in English in terms
of the relation obtaining between words and an abstract logical
structure implicit in every grammatical English sentence.
Perhaps it is true, as Lees suggests,1 that the traditional
approach to language study differs from the transformational
approach only to the extent that the scholars involved in
traditional linguistics did not have access to the formal
apparatus allowing the recursive specification of the well-
formed sentences in a natural language. But it is indeed a
moot point whether the mere availability of such techniques
would have been sufficient to insure that the traditionalists
would come to view explanation rather than simply description
as the primary goal of linguistic inquiry. Whatever else a
-197-
-198-
traditional grammar may contain notwithstanding, the deficiencies
of the traditional approach can be traced ultimately to the fact
that the goal of linguistic inquiry was, for the traditionalists,
not a matter of justification but a matter of description.
Our interest in the traditional approach to the problem of
complementation stems from the fact that the traditional gram-
marians considered it important for a linguistic description to
take cognizance of a fairly wide range of introspective lin-
guistic data, in particular, judgments about relations obtaining
between words in sentences, about the identification of parts
of speech and the role played by the parts of speech in sentences,
about the constituency of sentences as these constituents ful-
fill the logical conditions imposed upon sentences, and so forth.
The art consisted not in providing a framework in which these
human intellectual abilities might be explained, but in simply
observing how various arrangements of words contribute to the
meaning of sentences. To take a specific example, both Poutsma
and Jespersen acknowledge that though an infinitive clause may
express an action or state, the person or thing with which the
action or state is associated is usually not indicated in any
way in the infinitive clause. For both grammarians it was
sufficient simply to indicate the deletion. The specification
of which word or clause in the "head-sentence," to use Poutsmats
term for the main sentence, is the same as the deleted word of the
clause in the complement sentence was never taken to be an
issue. Thus, it is never taken to be of interest that the
11
speech and larger grammatical units which operate as logical
constituents.
-199-
implicit subject of the complement sentence in "John promised
to gol" is the subject of the head-sentence while the implicit
subject of the complement sentence in "we defy you to go" is
the object of the head-sentence. In other words, subject
deletion in "for-to" and "POSS-ing" complements was, in the
traditional view, an important fact but not a fact that re-
quired explanation.
The central concern of a traditional approach to comple-
mentation is perhaps best understood in terms of the tradition-
alistst views on the constituency of sentences. Assuming that
well-formed sentences in a language consist of two levels of
constituents, logical and grammatical, the traditional gram-
marians sought to "discover" grammatical constituents on the
basis of intuitions about logical constituents. Logical con-
stituency refers to a definition of the sentence according to
which a sentence is said to consist minimally of a subject
and a predicate. A predicate may consist of a verb and its
object. Furthermore, subjects, objects, and verbs (where no
distinction is drawn between the logical status of a verb and
its grammatical status) may be modified by additional constitu-
ents variously referred to as adjunctomodifiers, and occasion-
ally, complements. A grammatical analysis consists first in
showing the constituency of these logical categories and
second in deriving a classification of the various parts of
r
- U
-200-
This methodology leads to some very interesting results.
In the first place, Poutsma observes a strong functional simi-
larity between subordinate clauses with "that" and infinitive
and gerundive clauses. In particular, he notes that all three
clauses may appear as the subject of a verb. Similarly, they
may appear as the object of a verb. Since all three construc-
tions have the same logical constituency, he reasonably assumes
that these constructions share a common feature in the language.
This observation, correct though it may be, left Poutsma with
a conceptual problem which his grammatical system did not
allow him to solve, namely, there was no way in which he could
express the notion "common feature," Insofar as notions like
subjoct, " "object," and so forth are logical structures ex-
pressing merely the relation between words in well-formed
sentences, it is impossible to offer a unique logical character-
ization of the three clauses under study. In one instance, the
three clauses are subjects, in another they are objects. Indeed,
one finds in both Poutsma and Jespersen a considerable prolifera-
tion of such mappings. Poutsma, for example, refers to adverbial
clauses of place, time, cause, reason, consequence, inference,
purposes, concession, disjunctive concession, quality, attendant
circumstances, degree, alternative agreement, proportionate
agreement, restriction, and exception.2 In each case of ad-
verbial modification, Poutsma observes the three types of
clauses as being an instance of any single logical constituent.
IThe most that Poutsma could say was that the three construc-
tions are instances of "clause." This alternative presents
several difficulties which we shall examine shortly. It is
immediately clear, however, that this classification, regard-
less of any other deficiencies, asserts that there is no
grammatical relation whatever between the clause and the noun.
These entities are logically related insofar as they can play
the same role in the sentence, but where it is possible to
say with assurance that two items have the grammatical
property of being nouns or clauses, there is no grammatical
property which both nouns and clauses share.
In many cases, Poutsmat s classificational scheme leads
to analyses which, in terms of the earlier presentation, seem
to be substantially correct. For instance, "that" clauses
in sentences like "I think that John should go" are analyzed
as objects of the verb. In effect, this coincides with the
earlier analysis in which the NP dominating "it that John
should go"t is interpreted as the object of the verb "think"
in the underlying phrase structure configuration. Similarly,
the infinitive and gerundive phrases in sentences like "I
swore never to divulge our secret" and "I suggest gaining
permission first" are correctly analyzed as verbal objects.
The major problem in Poutsmats analysis arises from the fact
that his clausal analysis of the three complement construc-
tions is taken overly seriously. Clauses, Poutsma hypothesized
-201-
-202-
correctly, are actually sentences which are embedded in other
sentences and which perform thereby some logical role. But
it is the property of all sentences that the subject is in
the nominative case while the object is in the accusative
case. It followed, therefore, that a phrase like "him to do
that" in the sentence " I would like him to do that" could not
be an instance of an infinitive clause, the reason being that
the clause is a sentence and all sentences have nominative
subjects. Thus, in the sentence, "I would like him to do
that" it is only the infinitive phrase "to do that" which is
the clause., The pronoun "him" is actually the object of the
sentence. This reasoning led Poutsma to postulate the sof-
called "accusative with infinitive" construction where the
infinitive could modify the object or the subject. In other
words, the complement constructions in "I caused him to go"
and "I wanted him to go" are identical. Furthermore, the
infinitive clause is, in each case, said to modify the object.
It is also interesting that in sentences like "I would hate
for him to do that" the "for" is claimed to be a preposition
in the prepositional phrase "for him" while "to do that" is
of course, an infinitive phrase modifying the object of the
prepositional phrase.
We see then that for every infinitive construction where
a noun, pronoun, or nominalization intervened between the main
verb and an infinitive clause, the same analysis was given,
-2O3-
this analysis consisting of something like the verb phrase
complement analysis discussed earlier. This comparison is,
perhaps, a bit lenient since the assumption is being made
that no special force should be attributed to Poutsma's in-
sistence that the infinitive in accusative with infinitive
constructions modifies the object. In any case, difficulties
enough confront Poutsmats analysis. For example, can it be
that the two sentences "I want John to go" and "I want to go"
represent distinct constructions; the former is an accusative
with infinitive construction while the latter consists of a
verb and an infinitive clause as object. This would appear
to be the case. At one point2 Poutsma suggests that when
the accusative is a reflexive pronoun it is sometimes dropped
after some verbs. But this assertion is of little help since
it ultimately says nothing more than that there is really no
such thing as an infinitive as object construction since it
could be argued that the reflexive pronoun is dropped in
every case.
Jespersen's nexus theory constitutes an attempt to remedy
the difficulty Poutsma finds himself in. Unfortunately, this
effort is a perfect instance of throwing out the baby with
the bath water. Nexus is "a combination implying predication
and as a rule containing a subject and either a verb or a
predicative or both. Besides these a nexus may contain one
or more objects, often a direct and an indirect object."4
-2 4-
The application of this view to the complement system
effectively summarized by Jespersen in the following fashion.
If we compare the following sentences,
(1) they judged me a happy man
(2) I believe him as honest as myself
(3) this will make her happy
(e) they elected Tom their chief
(5) he slept himself sober
(6) 1 want this done at once
(7) I believe him to be an honest man
(8) this will make the watch go
(9) I want this to be done at once
and if we ask in each case what is the object of the
verb, many grammars say that in the first six it is
the word placed immediately after the verb, and the
rest is called a "complement" of the object, or an
tadjective or noun used predicatively of the object"
(Sonnenschein); in the sentences (7) to (9) some
writers speak of the infinitive as one of two objects.
The correct analysis is that all these are analogous
and contain not two objects (as in "I gave (made) her
a ring"), but only one, which is a nexus containing the
same two parts as a nexus that forms a complete sen-
tence or clause; compare with (2) "I believe that he
is as honest as I" and with (8) "this will have the
effect that the watch goes." In (5) it is particularly
easy to see that it is wrong to look upon the first
part (himself) as the real object of sle t: the resu4 t
of his sleeping is that he became sober himself").
Jespersen can be taken as asserting that the mere fact
that a pronoun following the verb in infirtive constructions
must take the accusative case in insufficient evidence on
which to discard the clausal analysis for infinitive con-
structions. Since the accusative case is the property of any
pronoun following a verb and can be explained with no reference
to the logical constituency of the pronoun, there is no reason
-205-
whatever to assume that the noun "Tom" in sentence (4) above,
for instance, is not actually the underlying subject of an
infinitive clause taken as object. But where Jespersen's
view avoids the inconsistencies observed in Poutsma's analysis,
it is accompanied by a host of other difficulties. First, it
now becomes extremely difficult to offer a coherent explana-
tion of irffinitive clause subject deletion. Why should it
be the case,' for example, that the subject is deleted in "I
want to be a virtuous man" but not in "I believe myself to be
a virtuous man"? Why should it be the case that a sentence
like "what I want is to be a virtuous man" is grammatical but
"what I believe is myself to be a virtuous man" is not. In
preserving the clausal analysis to the exclusion of such con-
structions as Poutsmat s accusative with infinitive, Jespersen
effectively forfeited the ability to explain a wide range of
differences which seem to accompany simple nexus constructions.
He can and does list a good many of these differences, but
never considers very seriously the possibility that these
differences might actually have an explanation.
In fairness to both Jespersen and Poutsma, it must be
pointed out that the sort of explanation whose absence is
being claimed was simply not the issue involved in traditional
linguistic analysis. It is difficult for the contemporary
linguist to keep this in mind since both Jespersen and Poutsma
say so many things that are either right or close to being
-206-
right. For instance, both linguists recognized the phenomenon
referred to earlier in this text as extraposition. That the
sentences "it is strange that John left" and "that John left
is strange$ are related by an implicit rule was not doubted.
Both of them recognized that prepositions are often suppressed
before infinitive and "that" clauses, e.g., 'I am aware of
Johnts being honest" and "I am aware that John is honest."
One can cite many instances of such insights. But the fact
remains that these insights are the result of attempts to
determine the way in which parts of speech, which, for all
purposes, seem to include such entities as clauses as well
as words, perform logical functions in the sentence. Thus
the "that" clause in "it is strange that John came" performs
the role of subject even though it is not in the normal sub-
ject position. Since the "that" clause in "I am aware that
John is honest" performs exactly the same logical function
as the gerundive clause in "I am aware of John's being honest,"
preposition suppression follows as a logical consequence.
The fact that traditional approaches to complementation con-
tain so much that is appropriate is a testimony to the goals
of traditional description. But it seems no tribute to the
traditional grammars to assert that certain of their results
are confirmed by research in transformational grammar since
one can hardly agree with the postulates that lead to these
results. Furthermore, it is just these postulates that lead
-207-
to so many conclusions which are simply false. In sum, the
traditional grammarians did not use the data to get at the
facts.
It is quite true, as both Poutsma and Jespersen thought,
that clauses may play the role of an object in a sentence
containing a transitive verb. But it is not the case, as
both imply, that noun objects differ from clausal objects.
This was a necessary conclusion, however, in a descriptive
system which postulated a logical constituency rather than
a grammatical constituency for sentences. Both clauses and
nouns could be objects, subjects, and many other things, but
the traditional grammarians found no way to express this
generalization. This generalization is automatic in a
description which posits a grammatical constituency for
sentences since we may say that both nouns and clauses are
instances of noun phrases and whatever affects a noun phrase
affects both nouns and clauses. Thus we see that, in this
case, the traditional grammarian who posits that a "that"
clause may be an object of a verb is giving a correct ob-
servation for the wrong reason. A noun phrase may be the
object of a verb and insofar as we can prove that noun phrases
may dominate "that" clauses, such clauses may be interpreted
as verbal objects.
When Poutsma devises the accusative with infinitive
analysis for a class of constructions, we find that his
-208-
judgment is correct in many cases, e.g. "I defied John to go,
but for entirely the wrong reason. The validity of this analysis
has nothing to do with the accusative form of the noun following
the verb, but rather with a variety of considerations raised
earlier in this study. Poutsmafs formulation is incorrect as
often as it is correct and, once again, for reasons which his
view of linguistic inquiry could not offer. Similarly,
Jespersents nexus hypothesis produces the right analysis for
a small class of items, but not for any reason which he pro-
posed. And the analysis fails for so many cases, in the
sense that nexus covers such a wide range of different phenomena
which just happen to look the same, that it becomes extremely
difficult to think of the few correct instances as a virtue
of the traditional approach to the phenomenon of predicate
complementation. It thus seems inadvisable to devote further
consideration to specific aspects of the traditional analysis
of complementation. The traditional grammarians were extremely
diligent. They present much data that is quite relevant to
the construction of a grammar for the complement system. But
it does not lie within the realm of theoretical possibility
for a traditional approach to give us this grammar.
The most extensive treatment of the phenomenon referred
to in this study as noun phrase complementation is found in
6
Lees' The Grammar of English Nominalizations. Lees correctly
identified "that," "for-to, and "POSS-ing" clauses as instances
-209-
of noun phrase complementation, but was more concerned with
the role played by such strings in the derivation of complex
nominal structures than with the various underlying structures
and transformations required to handle complementation in
general. Lees$ work is extremely insightful and it is pertinent
to discuss those aspects of his discussion of nominalization
which are relevant to the present study.
The noun phrase complement constructions discussed in
this study are instances of what Lees calls "factive" nominals
and "action" nominals. Sentence (1) is an instance of the
former while sentence (2) is an instance of the latter.
(1) for him to have eaten vegetables was a great surprise
(2) to eat vegetables is healthful
In his discussion of factive nominals, Lees clearly recognized
the necessity of an extraposition transformation, a proposition
suppression transformation, and a generalized transformation
which had the effect of embedding the nominal into a noun
phrase. Certain features of Leest discussion are not quite
right however. In his analysis of the passivization of "com-
plement-type" sentences, like "he was persuaded to work," we
observe that he assigns a verb phrase complement analysis to
the active form. In other words, "X persuaded Y to work,"
has the structure NP-AUX-V-NP-S. As we have seen earlier in
this study, there is considerable justification for the analysis
-210-
in which the phrase "to work" is an instance of a noun phrase
complement dominated by a proposition phrase. Leest analysis
was probably conditioned by Chomskyts early work in which
verb phrase complementation was assumed to be a more productive
phenomenon in English than it actually is].
One of the most interesting aspects of Leest analysis
concerns the formulation of a "second passive" transformation
to handle the derivation of sentences like (3).
(3) a. they were believed to have seen him
b. he was thought to be rich
In Leest analysis, the second passive transformation operates
on a string like "I think that he is rich" to generate the
string "he is thought to be rich by me." Leest was forced
to postulate this additional transformation since the "regular"
passive is obliged to apply to the highest level NP to the
right of the verb in the phrase structure. Thus, the regular
passive would always yield the string "that he is rich is
thought by me" and never "he is thought to be rich by me "
since the pronoun "he" is necessarily dominated by a higher
NP. We now know that the passive sentence "he is thought to
be rich by me" does arise through the application of the regular
passive transformation, at least in part. The passive trans-
formation produces the string "it for he to be rich is thought
by me." Through the application of two independently motivated
u-I'll 211-
transformations, the extraposition transformation and the
pronoun replacement transformation, we generate first "it
is thought by me for he to be rich" and second "he is thought
by me for to be rich." By the later application of the com-
plementizer deletion transformation, we derive the string
"he is thought byrme to be rich." It thus appears that Lees'
second passive transformation is probably unnecessary.
In his discussion of gerundive nominals, Lees raises
an interesting question concerning certain types of restric-
tions which are apparently imposed upon the subject of the
complement sentence. Consider, for example, the followings
(4) a. 1) swimming there is great fun
2) *his swimming there is great fun
b. 1) dressing oneself is fun
2) *dressing himself is fun
Lees suggests that these restrictions may have something to
do with the "action"-"factive" distinction, but it might also
be the case that these restrictions are to be explained by
certain more general considerations. We observe that "fun"
may take a propositional phrase, as in (5).
(5) a. swimming there is great fun for him
b. swimming there is great fun for somebody
It is abundantly clear that the subject of the complement
sentence is the same as the noun phrase in the prepositional
phrase.
-212-
(6) a. (his) swimming there is great fun for him
b. (somebody's) swimming there is great fun for somebody
Now let us suppose that in these constructions, as in all cases
of verb phrase complementation for instance, identity between
the subject of the complement sentence and the noun phrase in
the propositional phrase is obligatory. Since these two noun
phrases are connected, it follows that the identity erasure
transformation, which also must apply, will erase the subject
of the complement sentence giving the sentences in (7).
(7) a. swimming there is great fun for him
b. swimming there is great fun for somebody
Finally, we know that pronominals like "somebody" can be
deleted, giving thereby sentence (8).
(8) swimming there is great fun
We see that sentence (8) can have only one source; when the
subject of the complement sentence is "somebody" and when
the noun phrase in the prepositional phrase is "somebody.
Furthermore, since the identity erasure transformation is
obligatory, sentence (4a2) is automatically blocked. If the
noun phrase in the propositional phrase in the underlying
structure for this sentence has been "somebody," then the
identity erasure transformation would have blocked. If the
noun phrase had been "he," the grammar would correctly have
generated (7a).
-213-
These same considerations explain the restriction on (4b2).
It is clear in this sentence that the underlying subject of
"dressing himself" must have been "he." Thus, sentence (4b2)
is blocked for exactly the same reason as (4a2).
Not all of the restrictions Lees raises can be handled
in this manner however. Consider the following.
(8) a. eating vegetables is fashionable
b. *his eating vegetables is fashionable
There is no obvious independent motivation for postulating an
underlying prepositional phrase analysis in this instance,
although this possibility is not entirely out of the question.
In any case, insofar as the action-factive distinction does
not offer a reasonable explanation for the fact that the sub-
ject of the complement sentence in (8a) can and must be the
unspecified pronoun "somebody" to the exclusion of everything
else, one suspects that a deper explanation will be forth-
coming. But short of stating that "fashionable" is an ad-
jective which requires an unspecified pronominal subject in
the underlying structure of the complement sentence, the con-
siderations raised in this study seem to offer little help.
An examination of a few of Lees' analyses lead one to
suspect that he overlooked certain relevant facts. Thus, for
instance, one finds Lees assigning a common derivation to such
pairs of sentences as (9) and (10).
(9) the man is reluctant to go
(10) the man is clever to go
As we have seen earlier, however, there are a great many
differences between these sentences. Consider, for instance,
the following:
(11) a. 1) it is clever of the man to go
2) to go is clever of the man
b. 1) *it is reluctant of the man to go
2) *to go is reluctant of the man
These data suggest that sentence (10) does not have the
same derivation as (9), but rather something quite different.
Perhaps, as suggested earlier, the derivation of (10) depends
upon the application of the pronoun replacement transforma-
tion. This could not, however, be the case for (9).
With respect to sentences (12) and (13), Lees correctly
points out the similarity between what we have called verb
phrase complementation and the complementation observed in
these sentences.
(12) he is willing to leave
(13) he is free to leave
There is little question but that sentence (13) is an instance
of verb phrase complementation but it is not at all clear that
the same may be said for (12). Consider for instance the
following;
-215-
(i1) a. 1) 1 am willing for you to leave whenever you
are ready
2) what I am willing for is for you to leave
whenever you are ready
b. 1) *1 am free for you to leave whenever you are
ready
2) *what I am free for is for you to leave whenever
you are ready
The cost of insisting that sentence (12) is an instance of
verb phrase complementation is the cost of losing the generali-
zation that all verbs phrase complement constructions require
obligatory identity of the connected noun phrases. Further-
more, to the extent that (14a2) is grammatical, it is clear
that sentence (12) is a noun phrase complement construction
and that the complement originates in a prepositional phrase
in the underlying structure. Thus there is considerable
virtue in asserting that these two sentences are not the same.
One could point to several other instances similar to the
above. Their enumeration seems a little betide the point
since it is perfectly clear that the above criticisms of Leest
work imply nothing more than Leest sometimes overlooked
relevant data. The general adequacy of his theoretical frame-
work is not at issue.
I
-216-o
What the present study tells us is that it is not the
transformational machinery by which noun phrase complementa-
tion differs from verb phrase complementation. Indeed, the
set of transformations required in the derivation of these
constructions is the same for both. The only difference be-
tween the two resides in the underlying structure, produced
by the phrase structure rules. The major burden in the
grammar of the predicate complement constructions falls,
therefore, not on the transformational rules, but on the
phrase structures rules, in particular PS Rules 1 and 2
which produce the full range of predicate complementation
in English.
1 8Fillmoret s work on complementation is less extensive
than Leest, but interesting nonetheless since it was Fillmore
who first directed serious attention to complement sentence
markers. His discussion of the telescoped progressive is
particularly noteworthy. On another topic, however, Fillmore
overlooked certain critical facts and was led to an
analysis which is quite suspect. In Fillmoret s analysis of
the verb "believe," for instance, he postulates that the
sentence (15) has an underlying structure in which the noun
phrase "the butler" is verbal object.
(15) the detectives believe the butler to have been murdered
Fillmore never discusses these considerations which lead to the
-217-
formulation of an underlying structure or, in his terms,
the two terminal strings which are combined through an embed-
ding rule to generate (15). It is thus difficult to say much
beyond simply asserting that the proposed structure of the
two terminal strings in this instance fails on two grounds.
First, this formulation does not permit us to relate the
sentence (15) with (16).
(16) the detectives believe that the butler was murdered
Secondly, Fillmorels analysis leaves us without an explana-
tion of the synonymy of the sentences in (17) and the non-
synonymy of the sentences in (18).
(17) a. the detectives believe the butler to have been
murdered by the cook
b the detectives believe the cook to have murdered
the butler
(18) a. the detectives forced the DA to interrogate the
butler
b, the detectives forced the butler to be interrogated
by the DA
These facts are explained automatically on the assumption that
the sentences in (17) are instances of noun phrase complementa-
tion while those in (18) are instances of verb phrase comple-
mentation. Under these analyses, the non-synonymy of the
-218-
sentences in (18) is predicted from the fact that the underlying
verbal object is different in (18a), i.e., "the DA," from the
underlying verbal object in (18b), i.e., "the butler." In (17),
the verbal object is the entire noun phrase complement con-
struction and the only difference between the two sentences
is that the passive has applied to the complement sentence in
(17a) but not in (17b).
This historical summary does many injustices to both the
traditional and transformational linguists. The work of both
groups warrants more intensive study. But the general character-
istics of both approaches become fairly clear even on a cursory
examination. The traditional approach to the study of predi-
cate complementation is one of description without justifica-
tion. The transformation approach, on the other hand, is one
of description with justification where the essential goal is
to provide a description which successfully explains the data
exemplifying predicate complementation. The present work is
in the latter tradition and can be construed as confronting
earlier transformational descriptions of predicate comple-
mentation with new data. In this sense, the rules proposed
in the present study do not represent a grammar of English
which differs in any crucial respect from earlier formula-
tions in the transformational tradition. Rather the present
study constitutes a new synthesis of the rules in an adequate
English grammar which differs from earlier formulations only
insofar as it succeeds in providing an account of a new
collection of syntactic phenomena.
-219-
NOTES
1. R. B. Lees t"Transformation Grammars and the Fries
Framework, Readings in Applied English Linguistics,
Ed. H. B. Allen, New York, 1964, p. 137.
2. H. Poutsma, A Grammar of Late Modern English, Groningen,
1904, pp. 433-533.
3. Ibid., pp. 576.
4. 0. Jespersen, A Modern English Grammar on Historical
Princples, IV, London, 1956, p. 5.
5. Ibid, p. 7f.
6. R. B. Lees, Grammar of English Nominalizations, Bloomington,
1960, especially Ch. 3.
7. N. Chomsky, "A Transformational Approach to Syntax,"
in Third Texas Conference on Problems of Lingaistic
Analysis in English, ed. A. AHill, Austin, 192.
8. C. J. Fillmore, "The Position of Embedding Transformar
tions in a Grammar, " Report No. 2r- "roject on Linguistic
Analy is ,Columbus, 1963.
APPENDIX
Verb Classifications
The lists following provide classifications of English
verbs in terms of the complement structures in which particular
verbs may participate. The lists are representative, but by
no means complete.
I. Object Noun Phrase Complementation
A. "that" complementizer
accept
acknowledge
admit
advocate
affirm
allege
announce
answer
appreciate
ascertain
assert
assume
believe
certify
charge
claim
comment
complain
confess
confide
conjecture
contend
decide
declare,
decree
deduce
demand
deny
deplore
desire
discern
disclose
discover
dislike
divulge
doubt
dream
emphasize
estimate
expect
explain
fancy
fear
feel
figure
find
foresee
forget
gamble
gather
guarantee
guess
hate
hear
hint
hypothesize
imagine
imply
indicate
infer
insinuate
insist
intimate
intuit
know
learn
like
love
maintain
mean
mention
mind
mutter
neglect
note
notice
object
observe
order
own
perceive
point out
preach
predict
prefer
prescribe
presume
profess
promise
pronounce
prophesy
propose
prove
postulate
realize
reason
recall
reckon
recognize
recollect
recommend
regret
reiterate
remark
remember
reply
report
request
require
resent
reveal
rumor
say
see
sense
specify
state
stipulate
submit
suggest
suppose
surmise
suspect
swear
teach
testify
theorize
think
tolerate
understand
verify
wager
write
-220-
-221-
B. "for-to" complementizer
1. optional extraposition
bear
demand
desire'
dislike
expect
fear
hate
intend
like
loathe
love
promise
prefer
prescribe
request
require
want
2. obligatory extraposition
acknowledge
admit
affirm
assume
attest
believe
concede
conceive
conclude
confess
conjecture
consider
declare
deny
fancy
grant
guess
hold
imagine
judge
know
maintain
perceive
presume
proclaim
pronounce
prove
realize
recognize
remember
represent
show
state
suppose
suspect
think
understand
C. "POSS-ing" complementizer
abhor
admire
advocate
allow
avoid
await
bear
cherish
consider
contemplate
countenance
criticize
curse
defend
denounce
deny
deplore
deprecate
deride
disavow
discredit
discuss
dislike
dread
endorse
endure
entail
eschew
evaluate
examine
fear
forget
glorify
hate
imagine
intend
justify
like
love
mind
miss
neglect
notice
overlook
pardon
postpone
praise
preach
preclude
prefer
protest
publicize
question
recommend
reconsider
regret
relish
remember
repudiate
renounce
resent
resist
ridicule
risk
savor
stand
suggest
surtive
value
veto
vindicate
welcome
II. Subject Noun Phrase Complementation
A. intransitive verbs
1. "that" complementizer
appear
came to pass
happen
matter
seem
turn out
2. "for-to" complementizer
appear seem
chance turn out
happen
B. transitive verbs -- all complementizers
alarm
amaze
anger
annoy
appeal
arouse
astonish
astound
attract
awe
baffle
bedevil
befuddle
beguile
bemuse
benefit
bewilder
bolster
boost
bore
bother
calm
charm
cheapen
cheer
comfort
compliment
concern
deafen
defame
delight
fascinate
depress
disconcert
discourage
disgrace
disgruntle
disgust
dishear ten
dishonor
dismay
displease
disquiet
dissatisfy
distress
disturb
elate
embarrass
enchant
enrage
exasperate
exhaus t
exhilarate
fluster
frighten
gall
gladden
gratify
harm
hearten
help
horrify
humble
humiliate
hurt
insult
interest
irritate
madden
mortify
nauseate
nettle
outrage
overawe
overwhelm
pain
please
relieve
sadden
satisfy
scare
shame
sicken
soothe
startle
stupefy
suit
surprise
sustain
tempt
terrify
torment
trouble
unnerve
unsettle
upset
worry
III. Intransitive Oblique Noun Phrase Complementation
A. "that" complementizer
admit (of)
ask (for)
conceive (of)
decide (on)
hope Ifor)
insist (on)
pray (for)
rejoice (at)
wish (for)
B. "for-to" complementizer
ache (for)
aim (for)
arrange (for)
aspire (to)
beg (for)
care (for)
consent (to)
decide (on)
hope (for)
look (for)
long (for)
lust (for)
persist (in)
plead (for)
plot (for)
pray (for)
strive (for)
struggle (for)
succeed (in)
thirst (for)
wait (for)
wish (for)
yearn (for)
C. "POSS-ing" complementizer
approve (of)
arrange (for)
bank (on).
beware (of)
blush (at)
boast (about)
brag (about)
check (on)
comment (upon)
conceive (of)
consent (to)
concentrate (on)
cope (with)
correspond (to)
count (on)
decide (on)
delight (in)
depend (on)
disapprove (of)
dispense (with)
dwell (upon)
elaborate (on)
engage (in)
escape (from)
gamble (on)
gloat (over)
gloss (over)
guard (against)
harp (on)
indulge (in)
inhere (in)
inquire (into)
insist (upon)
intrude (upon)
jeer (at)
joke (about)
know (about)
laugh (at)
long (for)
muse (on)
participate (in)
persist (in)
plan (on)
pore (over)
protest (against)
provide (against)
puzzle (over)
react (against)
rebel (against)
reflect (on)
rejuice (at)
reminisce (about)
scoff (at)
slur (over)
sneer (at)
subsist (on)
succeed (in)
talk (of)
think (about)
weary (at)
wince (at)
wonder (about)
write (about)
'I
IV. Transitive Oblique Noun Phrase Complementation
A. "that" complementizer
advise NP (of)
apprize NP (of)
assure (NP (of)
convince NP (of)
inform NP (of)
notify NP (of)
persuade NP (of)
reassure NP (of)
remind NP (of)
tell NP (of).
warn NP (of)
B. tfor-totf complementizer
advise NP (of)
coax NP (into)
coerce NP (into)
convince NP (of)
drive NP (to)
entice NP (into)
force NP (into)
notify NP (of)
persuade NP (of)
remind NP (of)
warn NP (of)
C. "POSS-ing" complementizer
absolve NP (of)
accuse NP (of)
bully NP (into)
cajole NP (into)
caution NP (about)
coax NP (into)
coerce NP (into)
convict NP (of)
cure NP (of)
deter NP (from)
discourage NP (from)
dissuade NP (from)
entice NP (into)
exclude NP (from)
fool NP (into)
force NP (into)
goad NP (into)
lecture NP (about)
pester NP (into)
prevent NP (from)
prod NP (into)
prohibit NP (from)
provoke NP (into)
remind NP (of)
save NP (from)
scare NP (into)
suspect NP (of)
trick NP (into)
V. Intransitive Verb Phrase Complementation
A. "for-to" complementizer
begin
cease
commence
condes cend
continue
dare
decline
endeavor
B. "POSS-ing" complementizer
cease-
conmence
complete
continue
finish
quit
recommence
I.
fail
get
grow
hasten
manage
proceed
refuse
start
I,
I
A-225-
VI. Transitive Verb Phrase Complementation
A. "for-to" complementizer
admonish
allow
appoint
assist
bribe
bring
beseech
bestir
cause
challenge
charge
choose
coax
command
commission
compel
defy
detail
direct
empower
enable
encourage
enjoin
entreat
exhort
force
help
impel
implore
incite
induce
inspire
instruct
invite
lead
let
make
motivate
oblige
order
permit
predispose
prompt
schedule
stimulate
tempt
train
trouble
trust
urge
warn
B. Progressive
apperceive
behold
catch
detect
discern
feel
find
glimpse
keep
notice
observe
overhear
pass
perceive
see
show
watch
witness
C. "POSS-ing" complementizer
imagine
picture
remember
visualize
(There is virtue1 perhaps, in asserting that
a sentence like I imagine myself being tall"
is an instance of object noun phrase comple*-
mentation along with "I imagine that I am tall."
This assertion requires, however, that we allow
extraposition for "POSS-ing" complementizers.
Although we find cases where complement sentences
containing "POSS-ing" are extraposed from sentence
initial osition, as in "it is useless trying to
do that, 'such extraposition is not, in general,
possible in sentence initial position. Subse-
quent research may show, however, that vacuous
application of extraposition is possible every-
where for all complementizers, in which case
the above classification will have to be revised.)
I
VII. Oblique Verb Phrase Complementation
bank (on)
impose (upon)
presume (upon)
prevail (upon)
rely (upon)
I-
BIOGRAPHY
The author was born in New York City, New York, in 1940
and received his primary education there. In 1954, he moved
to Harrison, New York, where he received his secondary educa-
tion. He entered Wesleyan University in 1958 and majored in
history and Russian. The author received an A. B. with
High Honors in General Scholarship and Distinction in
Philosophy in June, 1962.
In September, 1962, the author entered the PhD in
Linguistics program at the Massachusetts Institute of Tech-
nology and has studied linguistics as a graduate student in
the Department of Modern Languages on a NDEA Fellowship
since then.
The author enrolled in the EdM for General Purposes
program at the Harvard Graduate School of Education in 1964
with the intention of studying the education applications
of linguistics. In June, 1964, the author was invited to
become a member of the Board of Editors of the Harvard
Educational Review.
The professional experience of the author is limited
to summer employment at the MITRE Corporation, in 1963,
and at the IBM Corporation (Thomas J. Watson Research
Center), in 1964. At both places, the author did gramatical
research on English.
MMOW
17
-228-
The author is a member of Phi Beta Kappa, Phi Delta
Kappa (honorary education fraternity), and is the recipient
of a Woodrow Wilson Fellowship and an NDSA Fellowship (Title
IV). He is a member of the Linguistic Society of America,
the Linguistic Circle of New York, the National Council of
Teachers of English, and the Conference on College Composi-
tion and Communication. He has a publication forthcoming
entitled "On the Role of Linguistics in the Teaching of
English," Harvard Educational Review, (Summer, 1965).