226 61 3MB
English Pages 154 [156] Year 1975
Linguistische Arbeiten
12
Herausgegeben von Herbert R Brekle, Hans Jürgen Heringer Christian Rohrer. Heinz Vater und Otmar Werner
Ljle Jenkins
The English Existential
Max Niemeyer Verlag Tübingen 1975
ISBN 3-484-10492-x © Max Niemeyer Verlag Tübingen 1975 Alle Hedíte vorbehalten. Ohne ausdrückliche Genehmigung des Verlages ist es audi nicht gestattet, dieses Buch oder Teile daraus auf photomechanischem Wege zu vervielfältigen. Printed in Germany
V PREFACE
This study arose out of earlier work (Jenkins 1972; 1974a/b) on the English existential.
The analysis of such constructions
has been the object of many investigations in traditional grammar as well as in the past decade and a half of work within the framework of generative grammar.
The theoretical studies of
the existential construction within this framework has been completely dominated during the past ten years by one form or another of what we may term the transformational the most recent study, see Milsark
(1974).
analysis; for
The pervasive influ-
ence of the transformational analysis of the existential construction is reflected in the fact that such analyses appear in numerous basic introductions to transformational grammar and syntax spanning over a decade; see, e.g., Roberts (1964) and, more recently (Grinder and Elgin 1973; Akmajian and Heny, in press; Bach 1974). The present study defends an entirely different analysis of the English existential formulated within the framework of the Extended Standard Theory (Chomsky 1970b; 1972) which involves no special transformations for this construction, as in previous analyses.
Thus we are arguing for the "null hypothesis" which
posits no new transformations not needed independently elsewhere in the grammar of English.
We will call this theory of the Eng-
lish existential the phrase structure analysis.
The theoretical
implications of the phrase structure analysis for linguistic theory are discussed. I am grateful for comments and discussion to John R. Ross, Mike Brame, Joe Emonds, Richard Carter, participants of the Colloque Franco-Allemand
(Paris, 1973) and particularly to Noam
Chomsky and Richard Kayne who read and provided comments on various versions of this work.
L.J.
VII CONTENTS
0.
1.
2.
3.
INTRODUCTION:
THE ENGLISH EXISTENTIAL AND OF LANGUAGE
1
0.1.
The E n g l i s h E x i s t e n t i a l
1
0.2.
The English Existential and Language Universale
3
THE PHRASE STRUCTURE ANALYSIS
8
1.1.
The C o m p e t i n g H y p o t h e s e s
1.2.
The D i s t r i b u t i o n of there
1.3.
The A g r e e m e n t a n d the P r o g r e s s i v e C o n d i t i o n s
17
1.4.
There
8
21
1.5.
Conclusion
in D e e p S t r u c t u r e
as a D e t e r m i n e r
11
24
THE PHRASE STRUCTURE ANALYSIS AND CLEFT REDUCTION
26
2.1.
Introduction
26
2.2.
The Reduced Cleft Analysis
27
2.3.
O t h e r C a s e s of C l e f t R e d u c t i o n
31
2.4.
Conclusion
36
THE TRANSFORMATIONAL ANALYSIS
40
3.1.
Tfcere-insertion
40
3.2.
TTzere-insertion a n d the C y c l e
40
3.3.
Tftere-insertion a n d D e r i v e d C o n s t i t u e n t S t r u c t u r e
42
3.4.
T A e r e - i n s e r t i o n a n d the F e a t u r e
43
3.5.
There-insertion
3.6.
(In)definite
and Semantic Features
(Agency, G e n e r i c i t y , a n d 4.
UNIVERSALS
(In)alienability)
Conclusion
S T A N D A R D Τ / / M E - I N S E R T ION
45 50 52
4.1.
Introduction
52
4.2.
T h e T w o be s A r g u m e n t
53
4.3.
The Semi-modal Argument
60
4.4.
The Predicate Nominal Argument
68
4.5.
Other Arguments
69
4.6.
Conclusion
73
Vili 5.
Τί/ΕΛΈ-INSERTION A N D S E M A N T I C I N T E R P R E T A T I O N 5.1.
Introduction
5.2.
The E
5.3.
T h e L e f t m o s t be
83
Rule
85 Condition
5.4.
The S e m i - m o d a l
5.5.
The D e f i n i t e n e s s
5.6.
The P r e d i c a t e
5.7.
Extraction
5.8.
Other
5.9.
Stress
86
Restriction
89
Restriction
91
Restriction
97
Arguments
104
Arguments in
107
Existentials
THE STRUCTURE-PRESERVING 6.1.
Introduction The Optional/Obligatory
6.3.
The Semi-modal
6.4.
The P r e d i c a t e
6.5.
C r i t i c i s m of the Formulation
Tff£7?E,-INSERTI0N
112 112
Argument
Argument
115 117
Restriction
118
Structure-preserving 122
IN G E N E R A T I V E
7.1.
Introduction
7.2.
C r i t i c i s m of M c C a w l e y ' s
L O C A T I V E A N A L Y S E S OF
109
A N A L Y S I S OF
6.2.
THERE- - I N S E R T I O N
83
SEMANTICS
130 130
Analysis
ïff£7?i>INSERTION
8.1.
Introduction
8.2.
The L o c a t i v e A n a l y s i s of
8.3.
The L o c a t i v e A n a l y s i s of K i m b a l l
8.4.
The Case Grammar
132 138 138
Analysis
Kuno
138 139 141
CONCLUSION
142
BIBLIOGRAPHY
143
O.
INTRODUCTION:
0.1.
THE ENGLISH EXISTENTIAL AND UNIVERSALS OF LANGUAGE
The English Existential
In this study we will investigate the grammatical properties of such existential sentences in English as 1)
There are ghosts.
2)
There is a man in the garden.
3)
There is a man standing on the corner.
4)
There is a man being beaten up outside.
We will defend a hypothesis about the analysis of these constructions which we may call the "null hypothesis".
That is, we
will assume and try to prove that no special additional syntactic (phrase structure or transformational rules) are required for existential constructions that are not already independently required elsewhere in the grammar of English.
This approach will
be contrasted with a variety of other approaches which involve the postulation of special rules to generate such existential sentences as in 1)—4).
These other approaches, which I often
refer to as varieties of the transformational
analysis generally
involve the postulation of a transformational rule known as "thereinsertion"
(see Chapter 3).
Since in our analysis no special transformation inserting there is required, but rather there is generated directly in the base by the phrase structure rules, we could properly term our analysis the phrase structure analysis of the English existential.
This contrast between the transformational analysis vs.
the phrase structure analysis of the English existential should not suggest that the phrase structure analysis of such constructions as l)-4) does not itself involve any transformations.
In
fact, as we shall see, in the derivation of sentences like 1)—4) in the phrase structure analysis, transformations (e.g., Affix
2 Placement and Agreement) are involved here as well.
But what
is crucial is that no special transformational rule such as "tTzere-insertion" which is specially postulated to account for such examples as l)-4) is required under the phrase structure analysis.
Thus we are claiming that if you were to postulate
the phrase structure and transformational rules necessary for English, leaving the English existential constructions out of consideration, these rules would suffice as well for an account of the whole grammar including the existential constructions.
This
strategy is essentially that adopted by F. R. Higgins (1973) in his investigation of the English pseudo-cleft construction, where he also defends the "null hypothesis" for that construction. If our hypothesis that the English existential constructions are base-generated is correct, then we would expect to find that the structures in l)-4) correspond to base structures which are independently required for non-existential sentences.
Observ-
ing no more than that existential sentences seem superficially to be some sort of copula sentences and noting that copula sentences are of the form NP-ie-Pred(icate), we would then expect existential sentences to occur in such base-generated copula-predicate structures and to exhibit their properties. qualification:
(With one
Pred must be of the form NP-(X); i.e., there is
no *there is red, *there is in the garden—thus an NP that existence is predicated).
it is only of
And, as we will see, these
expectations are fulfilled. Corresponding to such constructions as 1) there are ghosts (there-be-HV) we have base-generated copular constructions such as 5)
Al and Max are ghosts.
And corresponding to 2) there is a man in the garden
(there-
ie-NP-Locative, in one reading), we have base-generated copular constructions as 6)
John is a bother in the garden.
Furthermore, corresponding to 3) (and similarly 4)) there is a man standing on the corner
(there-be-NP-S), we have the
base-generated copular constructions like
3 7)
That's a man standing on the corner. There's John standing on the corner.
In the course of this study we will document the fact that the existential constructions have the expected properties of such copular structures as the NP-fee-NP structure in 5), the locative structure in 6), and the cleft structures in 7).
This is pre-
cisely the state of affairs that we would expect if the nullhypothesis (our phrase structure analysis) for the English existential is correct.
If this were not the case, then we would have
to complicate the base to prevent the existential structures from occurring in and having the syntactic properties of such copular structures as 5)-7).
Of course, whatever differences might remain,
we would hope they could be explained on the basis of independently motivated principles; e.g., principles of semantic interpretation, etc. On the other hand, under the transformational analysis, it is an accident that the structures generated by t^ere-insertion correspond to independently generated base structures.
That is,
it would be just as possible to expect the output of theve-ixisertion in some hypothetical language (pseudo-English) to be the following sentences 8)
There are ghosts.
9)
There a man is in the garden.
10)
(* in English)
There is standing a man on the corner.
(* in English)
Under the phrase structure analysis, the impossibility of 9) and 10) in English is not accidental, but a necessary consequence of the hypothesis. 0.2.
The English Existential and Language Universals
In this study we will be discussing the relation between such pairs of sentences as follows la) lb)
Elephants are in Africa. There are elephants in Africa.
2a) 2b)
A man is standing on the corner. There is a man standing on the corner.
It has often been argued in transformational literature that such synonymous pairs of sentences are related by transformation
4 as in B a k e r and M c C a w l e y
(1973), Burt (1970).
Thus,
the structure u n d e r l y i n g inserts the
there
noun phrase a man
(1971), Emonds
(1970), P e r l m u t t e r
for e x a m p l e , 2b) m i g h t derive
2a) by a t r a n s f o r m a t i o n a l
after be.
Below w e w i l l term this
the
analysis
analysis.
We w i l l argue, however, that there is no rule of there-insertion but rather that there
ing 2a), but is roughly the
w h e r e the sequence a man
in deep
2a)-2b),
structure.
2b) is not the structure
underly-
following
Np[there]Np-be~Npta
of a cleft s e n t e n c e .
transformational
relating pairs like la)-lb) and is generated directly
Thus the structure u n d e r l y i n g
ture
from
rule which
m o r p h e m e in subject p o s i t i o n , p o s t p o s i n g
the transformational
3)
(1970),
man]Np~s[standing
standing
on
the
corner
on the
corner]g
is the
remnant
We w i l l call this analysis the phrase
struc-
analysis.
Before continuing, we would
like to comment on the
cance of such issues as to w h e t h e r there is a rule of
tfcere-insertion or n o t .
signifi-
transformational
N o t i c e that w h a t is at
stake
is not simply w h e t h e r w e h a v e one rule m o r e or less added
to
the g r a m m a r of E n g l i s h .
post-
W h a t is of real interest w h e n w e
ulate a rule like there-Insertion
in E n g l i s h is not only
w e g e n e r a t e the correct subset of E n g l i s h e x i s t e n t i a l
that
sentences
and their structural d e s c r i p t i o n s , but above all the formal perties that such a rule has.
That
is, research
tional g e n e r a t i v e g r a m m a r , w h i c h has recognized of the linguist
in
pro-
transforma-
the primary
task
to give an account of language a c q u i s i t i o n ,
shown that all the rules in a g e n e r a t i v e syntactic, and s e m a n t i c — t o be subject
has
grammar—phonological,
to highly
restrictive
universal c o n d i t i o n s on their form and function and w a y s in w h i c h they are organized of this extremely
and interrelated.
For an extended
important topic, see Chomsky
to take the e x a m p l e of syntactic t r a n s f o r m a t i o n s , ough for the linguist
discussion
(1971).
Thus
it is not
to, say, list the t r a n s f o r m a t i o n s
of
enEng-
lish that he d i s c o v e r s , but it is of p r i m a r y i m p o r t a n c e for him to d e t e r m i n e by studying this list w h a t c o n s t r a i n t s on the that these rules h a v e and on the w a y that they function and terrelate that he can r e a s o n a b l y p o s t u l a t e to b e u n i v e r s a l
form into
5 all languages and which he can thus asciri.be to the innate lan— guage learning mechanism which the child brings to language acquisition.
Thus the linguist is in essence providing a defin-
ition of "possible transformation" for human language.
Similar
comments can be made for phonological and semantic rules.
In
the case of syntactic transformations, some possible universal constraints on their form are that they are structure-independent, do not depend on semantic relations and meanings, that their structural descriptions are quantifier-free, etc.
For
examples of how the functioning of transformational rules are constrained, see Chomsky
(1971).
To return to the example involving the putative transformation of tftere-insertion, let us review a few of the cases in which it has played a role in discussion involving conditions on rules, possibly universal: A.
Emonds
(1970) gives a formulation of there-insertion
to
explain a number of facts, which he claims provides support for his "structure-preserving" hypothesis.
This hypothesis argues
that transformations in all languages fall into two major classes (leaving aside his class of "minor" rules) , those like
there-
insertion in English, which are structure-preserving, and others like Subject-aux Inversion in English, which are not.
For a
critical discussion of the structure-preserving treatment of tftere-insertion, see Chapter 6. B.
Chomsky
(19 70a) points out that the existence of a rule
of tfrere-insertion would be an argument that transformations be permitted to introduce phrase structure. C.
Ross (1969) argues that the existence of a rule of
there-
insertion has consequences for the controversy as to whether deleted anaphora are to be accounted for by a transformational rule (such as VP-deletion) or by an interpretive rule.
He claims
that a rule of t/zere-insertion argues against the interpretive position. D.
See also the discussion below in Chapter 1.
Bresnan
(1970) uses the existence of a rule of
there-in-
sertion to argue for the interpretive position for pronouns as against a transformational rule of Pronominalization for a different class of examples. E.
McCawley
(19 70; 1972) argues that his particular formu-
6 lation of tftere-insertion taken together with the hypothesis that deep structures have underlying
V(erb)-S(ubject)-O(bject)
order allow on the one hand tighter constraints to be imposed on the class of "possible transformations" than the standard proposal of verb-second order in deep structure in English and provide evidence on the other hand that the semantic deep structures that the linguist postulates converge with the logical structures postulated by logicians
(e.g., the system of Luka-
siewicz; see Chapter 7. F.
Bresnan
(1970; 1971) cites data from Ross to argue that
the application of tftere-insertion must be cyclic. cussion of the relevance of there-insertion
For a dis-
to the universal
principle of cyclical rule application, see Grinder
(1972) and
the discussion in Chapter 3. These examples suffice to show how the existence of one rule like iTzere-insertion can be crucial to determining what general conditions, possibly universal, we must ascribe to the theory of language; viz., that certain classes of transformations are structure-preserving, that transformations can introduce derived phrase structure, that anaphora should or should not be described by syntactic transformations, and that certain rules must be applied cyclically.
If, however, as we will argue in this study,
there is no transformational rule of t/ze^e-insertion, then one piece of evidence for each such postulated universal condition disappears, although the universal might be independently supported by o£her evidence.
Thus in A. and B. above, one might
seek other evidence, independent of tTzere-insertion, to support the structure-preserving hypothesis or the assumption that transformations may introduce phrase structure.
Similarly, in the
case of C. and D., the argument for or against interpretive theories of anaphora must be dropped or else reformulated.
In Ε.,
one could seek other arguments for underlying V-S-0 order independent of the facts about the existential there. And in F. one might try to show that other rules require cyclical application in the grammar. In this study, it will become clear that to describe the English existential within the transformational analysis would require a theory of transformations with great expressive power.
7 That is, the putative rule of there-insertion
would require that
we extend the standard definition of a transformation as a structure-dependent operation such that we allow the statement of quantifiers in structural conditions, for example.
And some accounts
of tfrere-insertion would require reference in the structural conditions of the transformations to such semantic properties as the feature + definite
(or the presence or absence of an ar-
ticle in the underlying subject noun phrase) and to semantic properties of .the underlying predicate.
We wish to avoid all
such undesirable extensions of the expressive power of the theory of transformations.
As we will see, the phrase structure
analysis of the existential there avoids these difficulties in a principled way and hence enables us to keep the theory of transformations very restrictive, which is our ultimate goal if we wish to explain the facts of language acquisition. For these reasons, any proposal concerning the existence and the correct formulation of even a single rule
(like
there-Lnser-
tion) must be given careful attention by the linguist if he is to hope to attain deeper insights into the universals of language that make the acquisition and use of language possible and in what follows we must always keep such linguistic universals in mind in evaluating a particular linguistic analysis of the English
existential.
1.
THE PHRASE STRUCTURE ANALYSIS
1.1.
The Competing Hypotheses
In the next few chapters we will examine constructions of the following form, all of which contain the existential 1)
There are ghosts.
2)
There are ghosts in that haunted house.
3)
There is a woman standing on the corner.
4)
There arc some new houses being built.
5)
There appeared a hatless
there^
stranger.
The analysis of such constructions has been the subject of considerable study and debate, both in traditional grammar and during the past decade and a half of transformational grammar. Syntactic analyses from a rich variety of viewpoints have been adopted during the history of the study of English syntax, as is witnessed from a small handful of the studies of the existential there·. Fillmore Burt
Jespersen
(1968); Emonds
(1971); Kuno
(1940); Poutsma (1970); Allan
(1971); Kimball
(1916); Lyons
(1967; 1968);
(1971); McCawley
(1973); Baker
(1970);
(1973).
We will explore the hypothesis that the deep structures of l)-5) all contain an underlying there·, the deep structure of 1) would be
1
There are some grammatical differences between existential and demonstrative (as in there goes the rocket) there other than the difference in semantic interpretation. For example, there (demonstrative) is stressed in there's Bill, but there (existential) is not in *thêre's a God. With pronouns there (existential) allows only the word order be-pronoun as in well, there's Mike, Bill, Pete, and then there's you, but not *there you is/are. With there (demonstrative) the order is pronoun-be as in there she goes (but not *there goes she.) Furthermore, there (existential) may be embedded as in tii.ll said that there is a God but not there (demonstrative) as in Bill said that there +gces/tient the rocket.
9
Thus the deep structure of 1) is almost identical to its surface structure, which results from the automatic application of several low-level syntactic rules
(Affix Placement and Agreement).
We will call this hypothesis the phrase
structure
analysis,
since under this hypothesis, the there in l)-5) is generated in deep structure by the phrase structure rules. An alternative to our phrase structure analysis is the transformational insertion
analysis is assumed.
in which a transformational rule of thereVarious versions of the transformational
analysis of there-insertion
may be found in many of the studies
(within the framework of generative grammar) cited above.
One
version of the transformational analysis posits the following underlying forms for 1) and 2) 7) Ghosts be. 8) Ghosts are in that haunted house. A transformational operation of there-insertion
is applied to
introduce the there morpheme to yield the correct surface forms 1) and 2). In defending the phrase structure analysis we will argue against a wide variety of analyses proposed within the trans2 formational framework. The alternative transformational ana-
2
Another conceivable hypothesis is that in l)-5) some instances of the existential there are generated in the base and some are introduced by transformation. As far as I know, such a mixed phrase structure and transformational account of the English existential has not been argued for in the literature; any such account would have to provide some way to avoid the difficulties with the transformational account.
10 lysis is discussed in the preceding paragraph is essentially that formulated in Burt
(1971).
Another formulation of the transformational analysis of thereinsertion is that of Emonds
(1970) which is offered in support
of his "structure-preserving"
hypothesis.
Still another version of the transformational analysis of there-insertion,
within the framework of generative semantics
is to be found in McCawley
(1970), where he argues that the deep
structure of 9) is closer to 11) than to 10) . 9)
There are ghosts in that haunted house.
10) Ghosts are in that haunted house.
(=2)
(=8)
11) Are ghosts in that haunted house. Still other analyses seek to relate constructions containing the existential there with other structures.
For example, com-
pare
Ross
12)
Ghosts are in that haunted house.
13)
There are ghosts in that haunted house.
(=10)
14)
That haunted house has ghosts in it.
(=9)
(1967) has proposed for sentences like 12)-14) that the
structure underlying 13) is derived from the structure underlying 12) by the standard rule of transformational
there-inser-
tion as described above and that the structure underlying 14) is derived from the structure underlying 13) by a rule of
"there-
replacement". Fillmore
(1968) has proposed deriving sentences like 12)-14)
all from one common underlying case representation, within the framework of his "case grammar".
One can easily imagine still
other variant analyses which relate the structures in 12), 13), and 14) . We will demonstrate that the phrase structure hypothesis we are proposing accounts for the distribution of surface forms containing the existential there in a natural way and furthermore accounts in a principled way for the absence of surface
3
Fillmore's particular variant involves rules of copying and pronominalization. See Chapter 8.
11 forms of tfrere-constructions containing predicate nomináis, the semi-modals be going
to and be to, etc.
In fact, such gaps of
occurrence will provide the crucial evidence for selecting the phrase structure analysis for there over the transformational analysis of tftere-insertion.
By carefully examining proposals
concerning the English existential within various theoretical frameworks—the structure-preserving theory semantics
(McCawley), and others
(Emonds), generative
(Kuno, Fillmore, Kimball), it
is shown that in each case the phrase structure analysis is empirically more adequate than these alternative analyses.
The
consequences of this analysis for general syntactic theory are discussed. 1.2.
The Distribution of there in Deep Structure
Let us first examine the simplest structures as in 1)
There are ghosts.
and ask what kinds of rules (phrase structure, transformational) are required to generate such structures.
The simplest as-
sumption about 1) would appear to be that its derived structure consists of the morpheme there followed by the present tense form of the verb be followed by a noun phrase ghosts', i.e., the following structure 2)
there-Aux-be~ N p[ghosts] N p
Moreover, it has been noted
(see Burt (1971)) that there acts
as a noun phrase with respect to many transformations; e.g., 3) Is there any hope?
(Interrogative)
4) There's no hope, is there?
(Tag question)
5) There is believed to have been a revolution. 6) There seems to be something brewing.
(Passive)
(tt-replacement)
Thus a more accurate representation of the derived structure of 1) would be 7)
Np[there]Np-Aux-be-Np[ghosts]Np
The derived structure 7) is near, but, of course, not identical to surface structure since the transformations of Affix Place-
12 ment and Agreement have not applied. Let us now ask what the deep structure of 1) is.
We will
make the minimal assumption that the deep structure of 1) is the derived structure in 7); viz., the assumption that the deep structure of 1) is practically the same as its surface structure (except, as we have noted, for low level rules such as Affix Placement and Agreement). assumption to be true.
There is no a priori reason for this
In general it is the case that deep struc-
tures are distinct from surface structures, the two being related only through a sometimes long chain of complex transformational operations.
But clearly we prefer an analysis which adds
no new transformation(s) to the grammar of English unless the addition of such transformation(s) simplifies the grammar.
But
the burden of proof is on the one who proposes adding such transformational rule(s) to the grammar.
Assuming, then, that 7)
is for all practical purposes the deep and surface structure of 1); i.e., that there ave ghosts has the deep structure NPAux-be-NP, notice that not only do we need not add any rules to the transformational component to generate
there-sentences,
but we need not add any new rules to the base either since we already have rules which generate base structures of the form NP-be-NP; viz., the rules which produce sentences like 8)
These
(things) are pictures.
9)
John is a farmer.
10) That is my mother. To make this last point more clear, observe that we must have rules of the following form (or the equivalent) in any grammar: 11)
S Aux Τ VP Pred
NP+Aux+VP Τ (M) (have+en) (be+ing) Pres, Past ÍV+NP \ \be+Predf NP
NP
i(Det)N) (NP+S j
Det
a, the
Ν
John, farmer,
ghost
13 These rules 12)
4
will then generate the underlying structure for
John is a farmer.
(=9)
as follows
The obligatory rules of Affix Placement and Agreement apply to yield the correct surface form 12). Now let us consider the derivation of there are ghosts. assume the following underlying
Thus we enter there in 11). 4
5
in the lexicon as a Ν (oun); i.e., Ν
(But see 1.4 below.)
We
structure
there
Again Affix Placement and Agreement""
These rules are only for purposes of illustration. For a more empirically accurate formulation of the lexical insertion rules in terms of features, see Chomsky (1965). See below for the statement of agreement (1.3).
14 apply to give 1). In fact, it seems that all we need to say about there in the grammar of English is what is said in all traditional grammars; viz., that existential there occurs in subject position before a copula with predicate nominal. We place the following condition on the occurrence of there in deep structure (as a first approximation) 15)
Deep Structure** Condition:
there may occur only in the context Aux-be-NP
The Deep Structure Condition (henceforth DS Condition) reflects the fact that there has a very restricted distribution in English, as has often been noted. Thus we get there are ghosts, but not *I like there,
*there
is pretty,
etc.
Thus since the deep structure in 14) can be segmented in the following way 16)
there-Aux-be-NP
the DS Condition is met and 14) will yield a well-formed surface structure. On the other hand, suppose we generate the following deep structure 17)
there Since the there in 17) does not satisfy the DS Condition; i.e., does not occur in the context Aux-be-NP, it is starred as ungrammatical 18) 6
*I like there.
We do not consider idioms here such as there arose a great clatter,
there
ensued
a controversy,
etc.
See also 3.5.
15 Notice that with the phrase structure hypothesis we can state the restricted distribution of there quite easily; viz., as the DS Condition 15).
The statement of the distribution of there
becomes more complex if stated at any other level than that of deep structure. 19)
Thus consider
There seems to have begun to be a commotion.
If we were to state the distribution of there at the level of surface structure, we would have to modify the DS Condition 15) to accommodate cases like 19).
However, consider the deep struc-
ture of 19)
Here too the DS Condition 15) is satisfied and the deep structure is well-formed since there occurs in the following context 2D
Aux[for-to]Aux-be-NP[a
commotion] N p
We will see below that the distribution of there can be accounted for under the transformational hypothesis of i/zere-insertion only if we permit a list of completely ad hoc conditions on the structural description of the rule (see Chapter 4).
16 Notice again that the phrase structure hypothesis accounts in a natural way for the fact that there acts as a noun phrase with respect to many transformations; e.g., 22)
Is there any hope?
(Interrogative)
(=3)
23)
There's no hope, is there?
24)
There is believed to have been a revolution. (=5) (Passive)
25)
There seems to be something brewing.
(Tag question)
(=4)
(=6) (•i ^-replacement)
Our hypothesis accounts for this by generating there as a deep structure noun phrase. Observe that a great variety of what appear to be complex noun phrases containing relative clauses and other complements are generated in predicate position after there 20)
There are
p [some
27)
There are
Np[women
28)
Thero are
Npl
29)
There is
30)
There was
Np[a
31)
There are
Np[too
32)
There is
N p [nothing
33)
There is
Np
34)
There is
N p [nothing
a
Np[a
[a
people who don't like beer] N p . who are always in the kitchen] N p .
l Q t of people willing to help] N p . woman in the garden] N p . man with a hat o n ] N p . many people owning yachts] N p . to sayl N p .
secret to his success] N p . to it/something to thatl N p .
In many of the cases above, the noun phrases are generated by rules needed independently elsewhere in the grammar; e.g., compare 35)
Np[Some
people who don't like beer] N p
„„[A man with a hat on]„_ NP NP Np[A
woman in the garden] N p
are/ is waiting to see you. At first glance, then, it would seem attractive to propose that the constituent in the context there-Aux-be is always a noun phrase
CNP), which, furthermore, in most cases is generated as
17 a noun phrase (NP) in other contexts elsewhere in the grammar by rules independently needed.
Thus we would say that whatever
rule or rules generate the noun phrases some people who don't like
beer,
a woman
in
the garden,
etc. in the context
N p
[
are/is waiting to see you would generate these noun phrases in the context there-Aux-be~ Np [
l N p in exactly the same way.
However, there is considerable evidence to be discussed in Chapter 2 which shows that we must postulate at least the two following structures for the English existential 36)
there-Aux-be-NP
37)
there-Aux-be-NP-S
there
where in 37) the S is a remnant of a reduced cleft. Before considering this evidence we will turn to a few other problems. 1.3.
The Agreement and the Progressive Conditions
Observe that the rules as we have given them allow the generation of 1) *There is unicorns. 2) *There are a unicorn. alongside of the well-formed 3) 4)
There are unicorns. There is a unicorn.
That is, assume that number is introduced by context-free rules of the base® 5)
Ν
-»- [Δ, α plural, ...], α + or -
Then we will generate trees of the form
7 8
Example 34) in no other *nothing to See Chomsky
provides an instance of a noun phrase occurring context than after there s *I have nothing to it, it was discovered, etc. See 3.5. (1965), p. 232.
18
Suppose that the lexical insertion rules substitute there [N, 0 plural, ...] for the first occurrence of Δ, and unioovn O plural, ...] for the second.
[N,
Then, after Affix Placement and
Agreement have applied, we will get the ill-formed 2) (similar comments apply for the derivation of 1). Thus we need a condition which rules out strings of the form 1), 2)
7)
Condition on Agreement: *
NP -beNP [a plural] [-a plural]
This condition will rule out 1), 2) while permitting 3), 4). But there is independent evidence that such a condition is needed in the grammar.
Observe that anomalous strings of the
following form are generated in the syntax 8)
*Two men are a doctor.
9)
*A doctor is two men.
alongside of the well-formed 10) Two men are doctors. 11) A doctor is a man. Thus quite independently of the there construction, condition 7) is needed to rule out 8), 9). the form
Thus we allow structures of
NP -beNP to be generated in the syntax, [aplural] [-aplural]
19 Then condition 7) will rule out the inappropriate 8), 9) at the 9 same time ruling out 1), 2) with no further statement. Our decision to allow 8), 9) to be generated in the syntax is supported by the following sets of examples 12)
Two halves are a whole.
13)
A whole is two halves.
14)
*Two halves is a whole.
15)
*A whole are two halves.
16)
Two fourths are a half.
17)
A half is two fourths.
18)
*Two fourths is a half.
19)
*A whole are two halves.
20)
John and Mary are the happy couple.
21)
The happy couple is John and Mary.
22)
*John and Mary is the happy couple.
23)
•The happy couple are John and Mary
(Note that 14) , 18) are acceptable to some speakers—compare also two and two is four, but this does not affect our main point.) Observe that we have the configuration 24) 9
NP -beNP [a plural] [-a plural]
In Ross's (1969) discussion of the rule of VP-deletion and the controversy over whether a transformational rule of deletion or an interpretive rule is involved, he regards as "particularly troublesome" for interpretive rules examples of the following nature: A) Some people think there are no such rules, but there are/*is. He argues that: to generate sentences like A), 1. one will be forced to duplicate the there-insertion transformation in the phrase structure rules and 2. that number agreement must be formulated such that *he are fat is semantically anomalous. However, given the phrase structure analysis of there, neither 1. nor 2. follows. That is, as we show, there is good reason for generating there is/ are (NP) freely in the syntax. Secondly, there is no way that *he are fat can be generated since he is marked [-plural] , yielding he is fat by ordinary number agreement. Thus *he are fat is syntactically anomalous rather than semantically anomalous in this analysis. Thus given the phrase structure hypothesis for there, examples like A) provide no evidence against interpretive rules, but are neutral with respect to the choice between VP-deletion and a rule of interpretation.
20 in 12), 13), 16), 17), 20), 21) as can be seen by the agreement witli be.
This shows that we must generate structures of the
form 24) in the syntax, whereas 14), 15), 18), 19), 22), 23) are excluded in the syntax.
To see this consider
25)
*Three fourths are a half.
26)
*A half is two wholes.
Thus, were one to allow 12), 13), 16), 17) in the syntax while excluding 25), 26), one would have to permit arithmetic computations in syntax to generate the correct
sentences.
Observe that condition 7) is stated most generally at the level of deep structure.
Thus consider
27)
There are/*is believed to be agents in the garden.
28)
There *are/is believed to be an agent in the garden.
The starred example in 27) has the following deep structure 29)
*Δ believes
g[Np[there]Np-Aux-beNp[agents
[-plural]
in the
[+plural]
garden]Np]s (Assume agents
in the garden
of illustration.)
is an underlying NP for purposes
The structure in 29) is then ruled out by
condition 7) in the same way as 1), 2), 8), 9) above w e r e . 1 0 Turning to another problem, notice that the phrase structure rules 11) in 1.2 generate the following set of sentences 30)
a God.
There was has been had been may be *is being •was being
They also generate the structure underlying 31)
*There was being
Np[some
women in the k i t c h e n ] N p
(assuming again, for the moment, that some women
in the
kitchen
is an underlying NP).
10 For a discussion of how such examples might be handled in the transformational analysis, see Chomsky (1970a).
21 Thus there must be some condition in the grammar which rules out the starred examples 30), 31).
That is, in certain cases,
the string NP-Aux-be-NP must be starred as ungrammatical.
Let
us state this condition as 32)
The Progressive Condition: NP-Aux-be-NP Aux may not contain Prog (ressive)
We observe that such a principle is needed in the grammar quite independently of 30), 31).
Thus consider the following
sentences 33)
*John is being a doctor.
34)
*Mary is being an Italian.
Thus we n'eed condition 3 2) on other grounds to rule out 33) , 34); then the starred examples in 30), 31) will be similarly ruled out by this independently motivated principle.'''"'" 1.4.
There as a Determiner
Recall that we assumed above that the rules introducing there into deep structure were the rules NP
Ν, Ν
-»- there.
Thus
we assumed that there had the status both of an NP and of an N. While we presented some evidence (see 3)-6) in 1.2) that there is an NP, we did not justify the assumption that there is a noun There is in fact some evidence in favor of treating there as a determiner
(like the demonstrative determiner that) and thus
that the rules introducing there would then be NP Det
there.
(Det)N,
We will review this evidence.
It has often been pointed out that, in standard treatments of tfcere-insertion in which there is inserted as an NP, it is a complete accident that there and not, say, spaghetti or banjo, is the inserted item.
Emonds (1970) points out that if we con-
11 Condition 32) must be modified to allow cases where the second NP (the predicate nominal) is [-stative] as in John is being a fool. A further condition is that the first NP (the subject nominal) be [+animate]. Thus we have neither *a ston is being an ohjeat/a fool nor *there is being a God/a fool. Actually these modifications must be stated in still more general form to rule out predicate adjective constructions as *John is being tall while allowing John is being foolish.
22 strain the theory of grammar so that phrase nodes never immediately dominate terminal elements; i.e., for example, there is no rule like NP
there and furthermore require that every
morpheme X which is transformationally inserted under an empty node of category C also has a regular lexical meaning in which it is also of category C, then it is possible to argue that there is a determiner.
Emonds points out that the locative adverbs
there and here are noun phrases as in 1)
He brought the book in there (here).
since there/here are objects of the preposition in.
Thus note
that we do not have particle movement in 1) 2)
φ He brought in the book there (here).
Thus there and here are noun phrases.
But, following Emonds,
if phrase nodes do not directly dominate terminal elements, there is no rule NP
there/here·, rather there/here are, say, either
determiners or nouns. An argument that they are determiners is that they exhibit 12 the same contrast as the determiner pair this/that. An argument that they are not nouns is that there contains an initial voiced th (like this/that), which does not normally occur initially in English nouns. Another argument implicit in Emonds relates to some facts that he observed about the particle right in English, which occurs, roughly, with prepositional phrases of space and time 3) Come right into the room. 4)
Come right after lunch.
Notice that it does not occur in general with noun phrases 5)
*John saw right the/that man.
However, it may occur with a noun phrase in the special case where it contains a demonstrative determiner
(this/that) and
it has a spatial or temporal interpretation.
12 Note also the substandard use of here/there as an expansion of the determiner pair this/that as in this here book and that there house. See also the discussion in Lyons (1973).
23 6)
Come right this way.
7)
Come right this minute.
N o t i c e , t h e n , t h a t if w e t r e a t the l o c a t i v e a d v e r b s as noun phrases and furthermore
(demonstrative)
the grammaticality of the following 8)
I p u t it r i g h t
sentence
there/here
determiners,
is
explained
there/here.
A s s u m i n g t h e n t h a t t h e l o c a t i v e a d v e r b s there
a n d here
are
of t h e c a t e g o r y o f d e t e r m i n e r a n d , a s s t a t e d a b o v e , t h a t
every
m o r p h e m e X w h i c h is t r a n s f o r m a t i o n a l l y
emptv
inserted under an
n o d e of c a t e g o r y C a l s o h a s a r e g u l a r l e x i c a l m e a n i n g it is a l s o of c a t e g o r y C, t h e n i t f o l l o w s t h a t w i l l i n s e r t there
under the category
We will later reject Emonds's ter
in w h i c h
there-insertion
determiner.
tfcere-insertion a n a l y s i s
6), b u t m e n t i o n in p a s s i n g t h a t t h e r e is i n o u r
analysis
a s w e l l a n e x p l a n a t i o n for t h e f a c t t h a t t h e m o r p h e m e
there
c u r s in e x i s t e n t i a l
The
c o n s t r u c t i o n s a n d n o t spaghetti.
lanation rests on the fact that there are syntactic for considering derlying
the existential
there
there,
a s in
(pointing); viz., a cleft structure
2).
Thus,
in o u r a c c o u n t , i t is n o a c d i d e n t t h a t the
tial
there
and the demonstrative
there
(Chapter
have the same
c a l s h a p e , s i n c e t h e y o r i g i n a t e in t h e s a m e k i n d of
un-
there's
over
existenphonologi-
underlying
We m a y capture the similarities between the
tial, demonstrative, and locative
exp-
arguments
John
structure.
oc-
to o c c u r in t h e s a m e
structure as the d e m o n s t r a t i v e there
(Chap-
existen-
there s by g e n e r a t i n g t h e m
all
as determiners w h i c h are furthermore dominated by NP. A n o t h e r a r g u m e n t t h a t there
h a s t h e s t a t u s of a d e t e r m i n e r
may be seen by considering the following facts. has pointed out that the sequences P o s s is u n g r a m m a t i c a l . 9) 10)
this
{that,
Ross
As he notes, we have neither
*One p e t a l of t h i s ' s f e l l *That's execution was
(1972)
these,
those)of
off.
flawless.
R o s s p r o p o s e s a d d i n g a c o n s t r a i n t to t h e g r a m m a r w h i c h o u t a l l s u c h s e q u e n c e s of d e m o n s t r a t i v e d e t e r m i n e r s by the possessive
morpheme
rules
followed
24 11)
Any surface structure analyzable as X - this )- Poss - Y that thesei those' is ungrammatical.
But observe that corresponding to 12)
There appeared a hatless stranger,
we do not have such nominalizations as 13)
*There's appearance of a hatless stranger
as would be compatible with the lexical hypothesis and the assumption that there is generated in the base by R. Kayne, personal communication).
(as pointed out
But if we assume that
there has the same status as the demonstrative determiner tkis, that3
etc., then the same constraint
(e.g., 11)) that rules out
9), 10) will also serve to rule out 13). A final argument that there does not have the status of a Ν(oun) can be seen from considering the fact that there is not interpretable as are other nouns by EQUI. 14)
John can be chairman without
That is, we have
(his) being
(hypocritical. (a hypocrite.
Here the embedded PRO may be interpreted.as John and can be optionally deleted.
However, there does not similarly allow in-
terpretation as an embedded subject in 15)
There can be peace
(without there being a war. (•without being a war.
(as pointed out by R. Kayne, personal communication). In the following chapters we will not be concerned about the further phrase structure of there other than the fact that it is an NP and hence will give no further detail than 1.5.
Np[there]Np.
Conclusion
To sum up then, we have shown that a wide range of facts can be accounted for by generating there by the phrase structure rules of the base subject to the Deep Structure Condition in 1.2).
(15)
Two other conditions, needed to account for facts in-
25 dependent of the existential construction, serve to rule out a number of otherwise deviant sentences.
We also gave some ev-
idence that there might have the status of a determiner as well as that of a noun phrase in deep structure.
2.
THE PHRASE STRUCTURE ANALYSIS AND CLEFT REDUCTION
2.1.
Introduction
In the last chapter we showed how simple existential there constructions like 1) are generated by means of phrase structure rules 1)
There are ghosts.
In this chapter we wish to consider the analysis of more complex constructions such as 2)
There is a man in the gar-den.
3)
There is a man standing on the corner.
4)
There is a man being beaten up outside.
5)
There is a Mustang parked on the corner.
In particular we will be interested in determining the correct constituent analysis of the italicized strings in 2)-5).
We
will find a number of arguments in favor of considering the italicized portions of the sentences 2)-5) to be the remanants of cleft sentences; i.e., reduced clefts in surface structure. Thus the there construction will be analyzed as a reduced cleft that is derived by a rule of Cleft Reduction which is independently required in the grammar to derive cleft constructions such as that1 s/there rs Mary walking toward the railroad
station.
Furthermore, a new analysis is given for verbs of perception (see, hear, feet, etc. as in John saw Mary walking toward the railroad station) making use of the rule of Cleft Reduction. This rule applies also in the complements of abstract nouns (the sight, picture,
thought, idea of Mary walking toward the rail-
road station), thus showing this grammatical process to be of great generality in the syntax of English.
27 2.2.
The Reduced Cleft Analysis
Consider again 1)
There is a man in the garden.
(=2 in 2.1)
The rules of the grammar will generate the NP a man in the garden in the context there-Aux-be
in exactly the same way that
the NP is generated in the subject position of 2)
Np[
l N p is waiting to see you.
In this view then the italicized phrase a man τη the garden in 1) is a so-called "reduced relative" noun phrase and whatever the rules are that generate such noun phrases in sentences like 2) will also generate them in existential tTzere-constructions. However, there is considerable evidence that the assumption that the italicized phrase in 1) is simply a "reduced relative" as in 2) is inadequate for explaining a number of grammatical properties of the existential there constructions.
In order to ex-
plain these properties, we will show that it is necessary to consider the italicized phrase a man in the garden to be the remnant of a cleft sentence; i.e., a "reduced cleft."
The ques-
tion as to whether the phrase can ever be syntactically ambiguous; i.e., whether it can ever be both a "reduced cleft" and a "reduced relative" will be put aside for now. Turning now to the evidence for the reduced cleft analysis, observe that the normal intonation center in 1) as in 3)
There is a man in the garden,
is parallel to cleft sentences as in 4) It's John who is in the garden. Pursuing a suggestion made by Chomsky, if we regard 3) to be in fact a reduced cleft, we could explain the facts about the intonation.''"
1
Thus, we assume the deep structure of 3) to be
Personal communication. He notes that one might also be able to account for the distinction between the purely existential forms like there are ghosts and quantified expressions like there are prime numbers between one and ten (note: Φ*prime numbers are between one and ten) which have the logical form of quantifier: thing quantified.
28 5)
There-Aux-be-NP-S
where NP is the focus of the cleft and S is α man is in the garden (reduced in surface structure to the remnant in the garden). In order to support this analysis, it is first necessary to motivate a rule of "cleft reduction" in the grammar of English. Consider the following construction 6)
That's Belmondo that just came.
Observe that 6) is parallel to cleft sentences like the following both with respect to structure and intonation contour 7)
It's Belmondo that just came.
That is, in both cases we have the structure 8)
That/it-be-fôcus-that S
There is syntactic evidence that 6) is, in fact, a cleft sentence as can be seen by considering that the phrase that just came cannot be analyzed as either a restrictive or a non-restrictive relative.
On the one hand, proper nouns do not permit
restrictive relatives 9)
«Belmondo that just came lit up a cigarette.
On the other hand, non-restrictive relatives do not permit the use of the relative pronoun that 10) Belmondo,
(*that) ( who f
just came, lit up a cigarette.
Hence 6) must be considered an ordinary cleft sentence parallel to 7) . 2 Continuing to examine that-clefts,"^ compare now 11) That's Belmondo that/who is next to my mother. 12) That's Belmondo next to my mother. 2 3
Perhaps slightly more restricted as to choice of focus however . We restrict attention here to that-clefts. A more detailed analysis would also take into consideration constructions with demonstrative there as in Look! There's John standing on the corner/over there /being beaten up, etc., which provide further support for our analysis. For some reason, the cleft reduction discussed below does not seem to apply as freely in ¿¿-clefts.
29 It is plausible to assume that 12) is derived from 11) by a rule which deletes that/who is (relative pronoun-Tense-ie) . call this rule Cleft Reduction.
Let us
It applies in a number of con-
texts 13)
That's Belmondo that/who is standing on the corner.
14)
That''s Belmondo standing on the corner.
15)
That1's Belmondo that/who is being beaten up by that guy.
16)
That' s Belmondo being beaten up by that guy •
17)
That' s our Mustang that/which is parked on the corner
18)
That' s our Mustang parked on the corner.
Having motivated the rule of Cleft Reduction, let us return to the sentences with existential there as in 19)
There is a man in the garden.
Let us assume that underlying this sentence is a cleft sentence 4 of roughly the following form 20)
there-Aux-be-NP-S
that is, 21)
there-Aux-be-a man-that/who is in the garden
The same rules that assign the intonation center to Belmondo in 22)
That's Belmondo in the garden,
assign the correct stress to a man in 23)
There is a man in the garden.
And, similarly, the same rule of Cleft Reduction 5 has applied 4
5
We will not discuss here the possibility of a deeper structure such as N p [there-S] N p -Aux-be-NP (or N p [there] N p -Aux-beΔ-S) since it has no bearing on what follows. We leave open the question whether the rule.of Cleft.Reduction is obligatory in the case of there-constructions (it seems preferable in the case of demonstrative there—Look! There's John (?*who is) standing over there. See also footnote 3). In this view there's a man who is standing on the corner would be an ordinary relative clause construction rather than be parallel to that's a man who is standing on the corner.
30
in the derivation of both 22) and 23) (viz., relative pronounTense-fce
0 ) .
Similar comments apply to the derivation of each of the following sentences 24)
There's a man standing on the corner.
25)
There's a man being beaten up outside.
26)
There's a Mustang parked on the corner.
In each case the intonation center is correctly assigned to the focus noun phrase
after be and in each case Cleft Reduction
has applied, deleting relative pronoun-Tense-ie. There is another fact which follows from this analysis; viz., observe that in the following sentence the intonation center does not fall on someone as it does on a man in 23)^ 27)
»There's someone in the garden. (except with contrastive stress)
But observe that the corresponding that-cleft is also ungrammatical 28)
*That's someone (who is) in the garden.
Additional confirmation for this analysis is provided by sentences like 29)
There is a girl who knows you standing on the corner.
Observe that the italicized phrase in 29) could not possibly be derived as a noun phrase since reduced relatives like standing on the corner do not follow full relatives like who knows you 7 in English. Thus we do not have 30)
*A girl who knows you standing on the corner is waiting for a bus.
But in our analysis 29) is considered to be a reduced cleft parallel to 31) 6
7
That's a girl who knows you standing on the corner.
There is, of course, the possibility that all of these facts are part of some more general phenomenon and may be related to other facts about the destressing of redundant elements, like the contrast hard work matures scholars vs. hard work matures people (see Chomsky, 1970b). Pointed out by N. Chomsky, personal communication.
31 Observe, furthermore, that the following construction (see Perlmutter, 1971) 32)
There's getting herself arrested for Mary to consider.
might be derived parallel to the following cleft sentence 33)
It's getting herself arrested that Mary has to consider.
Notice that the stress contour is the same in 32), 33) with stress on the focus getting herself
arrested.
Noie, finally, that it may be necessary to slightly revise the statement of the Deep Structure Condition (15) in 1.2), the Condition on Agreement (7) in 1.3) and the Progressive Condition (32) in 1.3) to account for the case where there occurs in cleft structures, depending on what deep structures we assume for clefts (see also Akmajian, 1970); we will henceforth assume that the appropriate revisions have been made. 2.3.
Other Cases of Cleft Reduction
Consider now verbs of perception
{see, hear, feel, etc.) as
well as verbs like find, catch, have, want, etc.
Notice that
the complements after such verbs correspond to reduced thereand
that-clefts 1)
I saw John standing on the corner.
2)
I saw John in the garden.
3)
I saw John being beaten up outside.
4)
I saw a Mustang parked on the corner. Q
Notice that by assuming that Cleft Reduction
applies in the
derivation of l)-4), we can account for the fact that see, hear, etc. have precisely the same set of complements as reduced clefts. Thus 1) derives from 5) 8
I saw John-(that/who was) standing on the corner.
As pointed out by R.. Kayne (personal communication) , certain constructions in French involving, e.g., voir 'see' as in je l'ai vu qui sortait du cinéma might be amenable to the analysis under discussion (the phrase qui sortait du cinéma is neither a restrictive nor a non-restrictive relative clause)
32 w h e r e John
that/who
was
the i t a l i c i z e d p h r a s e 6)
T h a t ' s John
standing
on the
corner
is p a r a l l e l
to
in (that/who
was)
standing
on the
corner.
9
T h e r e is a d d i t i o n a l c o n f i r m a t i o n for o u r a n a l y s i s from a c o n s i d e r a t i o n of s u c h p h r a s e s as with
a hat
phrase does not occur with proper nouns
on.
Observe that this
(without pauses)
7)
The m a n with a hat on came
8)
*John w i t h a h a t o n came
in.
9)
I bumped into a man with a funny hat on.
10)
* I b u m p e d into J o h n w i t h a funny h a t on.
in.
(except w i t h the r e a d i n g I had a hat onc o m p a r e *John with a funny hat on was bumped into/pushed by me) B u t o b s e r v e t h a t w e g e t s e n t e n c e s of t h e 11)
form
I saw J o h n w i t h the h a t on.
w i t h the p h r a s e with
the
hat
on m o d i f y i n g the p r o p e r n o u n
John.
B u t o b s e r v e t h a t w e a l s o g e t the i d e n t i c a l s e q u e n c e in c l e f t sentences 12)
T h a t ' s J o h n w i t h the h a t on.
N o t i c e t h a t o u r a n a l y s i s a c c o u n t s for t h e c o m p l e m e n t s
for
the c l a s s of v e r b s u n d e r d i s c u s s i o n w i t h o u t the n e e d of any ad h o c rule of and E m o n d s
fee-deletion
as p r o p o s e d , e.g., in R o s e n b a u m
o b s e r v a t i o n t h a t the s e c o n d , b u t n o t t h e first, of the two s e n t e n c e s h a s a p r o g r e s s i v e
9
(1967)
(1970); this r u l e w a s p o s t u l a t e d to a c c o u n t for the
13)
I saw J o h n d i e .
14)
I saw J o h n
following
interpretation
dying.
A p a r t from c o n s t r u c t i o n s like I saw John open the door, w h i c h w e a s s u m e are i n f i n i t i v e c o n s t r u c t i o n s w i t h d e l e t e d to. In some c a s e s the to a p p e a r s in s u r f a c e s t r u c t u r e : I made John open the door v s . John was made to open the door. Observe t h a t s o m e v e r b s of p e r c e p t i o n do n o t . a l l o w t h i s c o n s t r u c t i o n : *T spotted John cross the street v s . I spotted John crossing the street (noted by V e n d l e r , 1964) n o r do t h e v e r b s find, catch, etc. In the case of the c o n s t r u c t i o n s u n d e r d i s c u s s i o n t h e f a c t s a b o u t the p l a c e m e n t of i n t o n a t i o n c e n t e r s e e m to b e m u c h m o r e c o m p l e x t h a n f o r the s i m p l e c l e f t s a n d the j u d g e m e n t s are too u n c e r t a i n to p r o v i d e any d i r e c t s u p p o r t for o u r a n a l y s i s .
33 Thus it was assumed that 14) derived from 15)
* I saw John be dying.
with subsequent deletion of be from the progressive infinitive (Emonds, 1970). However, in our analysis, the underlying form for 14) contains the progressive 16)
I saw John-that/who is dying
with subsequent deletion of relative pronoun-Tense-fce by the independently needed rule of Cleft Reduction. Finally, observe that the often discussed ambiguity in (see Chomsky, 1957) 17)
John saw the boy studying in the library,
with the first interpretation as in 18)
The boy studying in the library was seen by John,
and the second interpretation as in 19)
The boy was seen studying in the library by John.
is accounted for by two different underlying structures.
The
first, corresponding to 18), is 20)
John saw NP
where NP is the reduced relative the boy studying
in the
library.
The second corresponds to 19) 21)
John saw NP-S
where NP is the boy and S is the remnant of an underlying who is studying
that/
in the library after Cleft Reduction.
Notice that in reduced cleft structures of the form 21), both the object NP and noun phrases inside the reduced S may be questioned 22)
Who did you see standing on the corner?
23)
What corner did you see John standing on?
10 The fact that the reduced clefts containing statives are sometimes unacceptable may be due to semantic factors. That is, we get I oan't imagine John weighing ISO pounds but not ?That's John weighing 150 pounds.
34 Observe further that we have similar possibilities in questioning in the parallel existential there sentences 24)
What kinds of people are there living in Africa?
25)
There are some kinds of people living in Africa.
26)
?What kinds of cities are there people living in?
27)
There are people living in some kinds of cities.
from
and
from
However, in the case of existential t^ere-constructions, the constraints on ωΤζ-movement are much heavier.
Thus the NP in
the sequence there-be-NP-S may not always be questioned 28)
?*What is there in that box?
29)
?*Who is there in the kitchen?
although sentences like 28), 29) can be improved with the addition of a complement modifier. 30)
What is there in that box that you are trying to hide from me?
31)
Who is there in the kitchen that you don't want me to meet?
32)
What is there in Italy that attracts you so much?
33)
What is there in this drink that makes me so dizzy?
34)
What is there about her that fascinates me?
35)
Who is there in Paris that could make Mary fall in love?
And with phrases such as how many, what kind(s) of, what else/ otherj there-questions
are more acceptable
36)
How many states are there in the United States?
37)
What kinds of trees are there in Africa?
38)
What other explanation could there be? (Jespersen, 1940)
39)
Who else could there be?
and as an example of questioning from S in the sequence therebe-NP-S, we have 26) above.
35
Furthermore, there are heavy conditions on relativization (see also Ross 1967:266) as can be seen in 40)
*The book that there was on the table was red.
With certain expressions like uproar,
turnout,
etc. relative
clauses with there are more acceptable 41)
I was amazed by the uproar (commotion, problems, turnout, number/quantity of people) that there was/were (at the circus) .
and for extraction of a noun phrase from the complement S, compare 42)
This is a problem which there are a lot of people working on. (Ross 1967)
Moreover, with only, all the, what(ever), like biggest,
and superlatives
relative clause-like structures also appear
43)
That was the only problem that there was in getting the car.
44)
This is the only bottle that there was in the icebox.
4 5)
John and Mary are the only people there are that can help us.
46)
He took all the/whatever money there was.
47)
What(ever) butter there was, I gave to Bill.
48)
John and Bill were the biggest players that there were.
49)
Money is the most useful thing that there is.
Notice that in general, quite apart from the ifcere-construction under consideration, there are stringent conditions placed on ωΤΐ-movement after the copula be 50)
The football coach that John saw/*is went home.
(see Emonds 1970).
Again, though, such constructions as 50)
are acceptable in certain contexts 51)
I'll never be the football coach that John is.
It has been suggested that such constructions as 51) do not involve ω/ι-movement at all
(see Wasow 1972) , but are the result
of transformational deletion.
Part of the evidence that wfc-move-
ment is not involved in 51) is seen by observing that the relative pronoun who is excluded in 51)
36 52)
* I'11 never be the football coach who John is.
Similarly, note that in 45) and 48) who is impossible 53)
*John and Mary are the only people who there are that can help us.
54)
*John and Bill were the biggest players who there were.
The latter facts suggest that some of the constraints on questioning and relativization in the there-construction
may be re-
lated to more general constraints on «^-movement involving the copula be. To conclude this discussion we give the following examples which illustrate another context in which Cleft Reduction occurs; viz., in the complements of abstract nouns 55)
The sight, picture, idea, thought, etc. of John with a helmet on in bed with the flu sitting in the gutter being beaten up by that guy locked behind bars dismays me.
Again, the complements after John in 55) cannot be derived from restrictive or non-restrictive relative clauses as can be seen by a consideration of 56)
The sight of him in bed with the flu dismayed me.
57)
The thought of it (the cake) being eaten dismayed me.
where he, it do not in general permit relative clauses
2.4.
58)
*He, who is in bed with the flu, just phoned.
59)
* It, that/which is being eaten, tastes horrible.
Conclusion
We have given reasons for postulating underlying structures 11 Further indication of how complex the facts are concerning extractability from existential constructions can be seen from some data on Danish existentials (Erteschik 197 3) . From der er en mand som hav ktfbt det hus 'there is a man who has bought that house' one gets the topicalized construction det hus er der en mand som har krfbt 'that house is there a man who has bought'.
37
for the existential there of the form 1)
there-Aux-be-NP-S
where NP is the focus of a cleft structure parallel to such cleft sentences as 2)
That's
(there's) John standing over there.
We postulated the structure 1) alongside of the structure thereAux-be-NP, which we assume underlies such sentences as there are
ghosts. We showed how the postulation of underlying cleft structures
could explain parallels between existential
tftere-constructions
and cleft structures; viz., the fact that they have the same range of complements and the same intonation contours and the fact that both allow the occurrence of what superficially appear to be sequences of full relative-reduced relative otherwise forbidden in English.
sequences
Moreover, the oft-noted phone-
tic similarity between the existential there and the demonstrative there is accounted for by the postulation of the same underlying structure for both. We furthermore demonstrated that the one rule of Cleft Reduction which we required in our analysis was in fact needed not only to derive the existential t^ere-constructions, but was required in the derivation of a wide range of cleft structures and in the derivation of the apparently unrelated complement structures of verbs of perception
(and other verb classes) and
in the complements of abstract nomináis.
Thus the rule of Cleft
Reduction discussed in this chapter is by no means restricted to clefts, but is a grammatical process of great generality in the syntax of English. Moreover, we indicated that at least some of the mysterious restrictions on questioning and relativization in
there-construc-
tions are of a more general nature and independently necessary for the description of other copular constructions in English. One more word is in order concerning the facts of extraction. In Chapter 5 (5.7.) we again take up the discussion of another possible avenue of explanation for the facts about extraction from existential constructions after we have introduced the
38
notion of "idiom reanalysis rule," as it is called by Milsark (1974) . Although in 2.2. (see also footnote 4) we left unspecified the exact nature of the underlying structure of the cleft construction in general, we feel that a promising approach is offered along the lines of Higgins
(1973) who argued for the null hypoth-
esis for the pseudo-cleft construction in English.
Thus he pro-
poses, as we do for the existential construction, that the deep and surface structure of the pseudo-cleft construction is essentially identical in that no special pseudo-cleft transformation is needed.
The only necessary transformations in the derivation
of this construction are those required elsewhere in the grammar. It may well be that one -can extend this kind of analysis to the cleft constructions of the type that we are considering, so that here too deep and surface structures are "essentially"
identical
(say, NP-¿e-Pred-S, where the focus Pred is filled in deep and surface structure). A final word is necessary about the Deep Structure Condition discussed in 2.2.
As it is formulated there, it requires there
to occur before predicate NPs and in cleft structures.
But as
we pointed out in the Introduction, if it is the case that there is base-generated, then we would expect it to occur in all copula-Predicate structures.
That is, it would lead to a complica-
tion of the phrase structure analysis if this were not the case, since some way would have to be found to specify which copulaPredicate structures there could occur in and which it couldn't occur in. That this expectation is borne out is seen when we consider the third major type of copula-Predicate structure illustrated in the following 3)
John is a hypocrite at home.
4)
John is a hypocrite on Sundays.
which are of the form 5) 6)
NP-Aux-be-NP-Loc NP-Aux-be-NP-Time
That the constituents labelled Loc and Time are not
(neces-
sarily) part of the preceding NP can be seen by their preposa-
39 bility 7)
At home John is a hypocrite.
8)
On Sundays John is a hypocrite.
If the phrase structure analysis is correct, then we would e x p e c t — i n fact, it must be the case under the null h y p o t h e s i s — that there occur also in such structures as 5), 6).
That this
is so is seen by comparing 9), 10) with 11), 12) 9) There is a man in the garden. 10) There is a concert at six. 11) In the garden there is a man. 12) At six there is a concert. As can be seen from 11), 12) the same possibility of preposability is available.
Hence the structures are of the type in 5),
6) .
What this means then is that there occurs in all types of copula-Predicate structures.
Of course, Predicate is in fact
restricted to the form NP-(X), i.e., there is no 13) *There is red. 14) *There is in the garden. However, in Chapter 5 we will see that there is motivation for a rule of semantic interpretation, the E rule, which provides a semantic interpretation for structures of the form NP-Aux-beNP-X, such that existence is predicated of the class denoted by NP.
Thus it is reasonable to assume that 13) and 14) are marked
deviant by the E rule, since there is no NP to be interpreted. Thus, in effect, what we have called the Deep Structure Condition reduces to the statement that there may occur in any and all copula-Predicate structures, with any gaps of occurrence being explained by independently required principles, say, of semantic interpretation.
And this, of course, is exactly what
the null hypothesis predicts. In the following chapters we turn to criticism of the various transformational analyses of the English existential.
3.
THE TRANSFORMATIONAL ANALYSIS
3.1.
There-insertion
Most transformational analyses of the English existential there assume that the structure underlying 2) is derived from the structure underlying 1) 1)
A man is in the garden.
2)
There is a man in the garden.
Thus the transformation inserts the morpheme there in place of the noun phrase subject which is postposed after be.
If this
transformation of "there-insertion" does not apply, the structure underlying 1) is yielded as output.
In other words, the
transformation is optional.
3.2.
There-insertion
and the Cycle
Furthermore, it has been argued that the transformation of there-insertion
must be cyclic.
in Wasow (1972).
Several such arguments are found
Wasow points out that in the derivation of
such sentences like 2) from sentences like 1) 1)
I think that a man is in the garden.
2)
I think that there is a man in the garden.
we must apply the transformation on the internal cycle bedded sentence).
(the em-
For, if we were to wait until the external
cycle, including the matrix sentence, we would violate general constraints on the operation of transformational rules.
One
such general constraint, known as "insertion prohibition"
(Chom-
sky 1965) forbids the insertion of morphological material
(in
this case there) into a domain which has already been processed on a cycle.
Obviously, in this case, if t^ere-insertion were
delayed until the matrix cycle, "insertion prohibition," would be violated.
41 Chomsky
(1971) has recently proposed replacing this general
constraint of "insertion prohibition" on transformations by a number of other conditions, including the principle of the "strict cycle."
The principle of the strict cycle prohibits a rule from
involving solely a domain which has already been passed in the cycle.
Thus in 1), we could insert there on the internal cycle,
but again we cannot delay insertion until the external matrix cycle, since the transformation would have effects involving solely the internal cycle; viz., the insertion of there and postposition of the subject noun phrase after be, thus violating the principle of the "strict cycle".
If, as Chomsky suggests,
the principle of the strict cycle is restricted to major transformations; i.e., transformations involving phrase categories like noun phrase
(NP), the non-cyclic application of
there-ins-
ertion would be prohibited, since this transformation is a movement transformation involving the movement of a major category (the subject NP). Furthermore, there have been language specific arguments for making i^ere-insertion cyclic.
An argument due to Ross
(Bres-
nan 1971) is based on the interaction of tfeere-insertion with passive.
It is noted that there are derivations in which
there-
insertion follows passive as in the derivation of 3)
While you watch, there will be a pig roasted.
and derivations in which passive follows tftere-insertion as in the derivation of 4)
There was proved to be mercury in the bottle.
Similar data due to Ross is found in Bresnan
(1970).
It is
claimed that t^ere-insertion must be cyclic to account for such data as 5)
There was believed to have been a tiger shot.
6)
There was a tiger believed to have been shot.
7)
PThere's a man from Berkeley believed to be proving there to be no real numbers.
In Grinder
(1972) similar arguments based on data from Ross
and observations by Kimball for tftere-insertion being cyclic are discussed involving the interaction of t^ere-insertion with
42 Raising.
In particular Grinder is interested in showing that
this data provides only what he calls "secondary motivation" (i.e., the data is consistent with the principle of the cycle) rather than "primary motivation" for the principle of the cycle. We will, however, not go further into these issues, since, if as we will argue, there is no rule of t^ere-insertion, then the data on there shows nothing at all about the cycle and one is forced to look to other rules to confirm or disconfirm the principle of the c y c l e — f o r general discussion of this problem see Kimball
(1972) and Grinder
(1972).
Thus, summarizing the discussion above, it has generally been assumed that there-insertion
is an optional, cyclic trans-
formation with the effect of inserting the there morpheme into subject position, while postposing the subject noun phrase after he. 3.3.
r^ere-insertion and Derived Constituent Structure
It has often been objected that the transformational analysis provides no explanation for why the morpheme there is inserted.
In Kimball's
(1973) words:
"It fails to explain why there
is inserted and not, say, chop suey ; or, whether the there that is inserted is the same as the deictic there, or what relation the two bear to each other."
As we showed in Chapter 2, the
phrase structure analysis was able to provide an explanation for these facts in that reasons were given for assuming that the existential there appears in underlying cleft structures of the same form as the demonstrative or deictic there.
However, we
also presented an analysis of Emonds in 1.4 which suggested how a " tTzere-insertion" analysis might be able to relate the existential there to locative there by means of a general constraint on the operation of insertion transformations, so that this particular point may not be an insurmountable obstacle for the proponent of a t^ere-insertion transformation. Another problem noted
(Chomsky 1970a) in connection with the
postulation of a t?zere-insertion transformation is how the there is assigned the correct constituent structure.
That is, in Chap-
ter 2 we saw that there acts syntactically like a noun phrase with respect to many transformations.
Thus we have
43 1)
Is there any hope? (Interrogative)
(=3) in 1.2.)
2)
There's no hope, is there? (Tag question)
3)
There is believed to have been a revolution. in 1.2.) (Passive)
4)
There seems to be something brewing. it-replacement)
(=4) in 1.2.) (=5)
(=6) in 1.2.)
We were able to describe this range of data quite easily, by assuming that there is directly generated in the base as an NP (i.e., dominated by the NP).
Chomsky (1970a) showed that by
treating phrase nodes as feature complexes, it is possible to formulate there-insertion
so that there is introduced with the
appropriate information that it is a noun phrase.
Thus, in ef-
fect, the use of features enables transformations to introduce structure into trees.
There are several other cases
1970a) which suggest the need for such devices.
(Chomsky
On the other
hand, if we succeed in showing on independent syntactic grounds that there is no transformation of t^ere-insertion, then one piece of evidence in favor of allowing transformations to introduce phrase structure has been eliminated. 3.4.
Tftere-insertion and the Feature
(In)definite
Another restriction placed on the operation of
there-inser-
tion is the requirement that the subject noun phrase be indefinite (see Burt 1971)."'"
Thus we have 2) from 1) but not 4) from
3) 1)
A man is in the garden.
2)
There is a man in the garden.
3)
The man is in the garden.
4)
?There is the man in the garden.
Thus sentence 4), but not 2), would be ruled out by restricting tTzere-insertion to indefinite noun phrase subjects.
However,
it is not at all clear that such sentences as 4) are to be ruled out by the grammar. 1
In fact there are perfectly good interpre-
For a discussion of a number of constructions in English that similarly seem to require noun phrases with an indefinite interpretation, see Postal (1970).
44 tations for 4) even though somewhat different than for 2).
Com-
pare 5)
Who do we have to help us clean up the garage? Well, there's John and Mary. And then there's the man in the garden.
Notice also the occurrence of a definite noun phrase in well, there's John and Mary
in 5).
However, observe that if we sim-
ply removed the restriction on t&ere-insertion to indefinite noun phrases, thus deriving 4) from 3), we would be making the false prediction that 4) and 3) have the same meaning. Further examples illustrating the normal use of definite noun phrases in tTzere-constructions are given in Borkin and Peterson (1972) 6)
There was the usual argument in class today.
7)
There is the same narc at all Mary's parties.
8)
There was the expected silence when Sam stopped singing.
9)
There was the most surprising odor in the closet today.
Note, for example, that 7) has much the same interpretation as 10) The same narc is at all Mary's parties. while 11) has a different interpretation than 12); viz., the interpretation discussed in the last paragraph
R.
11)
There is the fat man at all Mary's parties.
12)
The fat man is at all Mary's parties.
Lakoff suggests
(in Borkin and Peterson 1972) that the ap-
propriate semantic distinctions must be stated in terms of such features as factivity and negativity. from Perlmutter 13)
Note one final example
(1970b)
In England there was never the problem that there was in America.
Whatever the correct solution turns out to be that characterizes the acceptability of 6)-9) and 13) in contrast to sentences like 5) and 11) with their different interpretations, it seems unlikely, given our knowledge of the formal properties of transformational rules, that such semantic distinctions should or
45 or could be stated within the structural descriptions of particular transformations like tfcere-insertion. ues of approach are open.
Two possible aven-
The proponent of tftere-insertion might
try to find general conditions on the applicability of transformations that can explain the facts in 3)-13). we have argued, there is no there-insertion
Or else, if as
transformation, then
there will be generated in the base freely with definite and indefinite noun phrases, and the task of stating the appropriate range of semantic distinctions discussed above will be assigned to the semantic component. It has been suggested by Halle
(Emonds 1970:36) that one
might generalize the there-insertion
transformation for the case
that the subject noun phrase is definite by inserting it in the latter case to derive, for example 14)
It's that old friend of mine coming down the street.
Thus in effect we would have an it-insertion transformation combined with a tfeere-insertion transformation, the first inserting it in place of a postposed definite noun phrase and the second inserting there in place of a postposed indefinite noun phrase.
However, this would still leave unexplained the origin
of the tTzere-construction occurring with definite noun phrases as in 4), 5) .
Furthermore, an -it-insertion transformation would
be susceptible to the same criticism as we will make below in connection with the tftere-insertion transformation in Chapter 4. 3.5.
TTzere-insertion and Semantic Features and
(Agency, Genericity,
(In)alienability)
Other facts which indicate that the application of
there-in-
sertion (Jepends on semantic factors in contrast to the widelyheld assumption that the constructions with the existential there are synonymous with the underlying source constructions can be seen by considering the following facts from Kirsner
(1973).
With certain intransitive verbs (see further discussion below), as in 1), tfeere-insertion can apply to yield 2) 1)
A tattered banner waved from the flagpole.
2)
There waved from the flagpole a tattered banner.
46
Now compare 3) with the questionable 4) 3)
A bearded student waved from the roof. (wave=beckon)
4)
*There waved from the roof a bearded student. (except when wave^beckon)
Apparently the acceptability of the tTzere-construction depends on whether the underlying subject has the feature of agency as in 4), where the there-construction
is unacceptable, or of non-
agency, as in 2), where the sentence is acceptable. Similarly 5) is ambiguous, with the agentive interpretation, as when the soldiers voluntarily remained, and the non-agentive interpretation of be left.
However, 6) is unambiguous having
only the non-agentive interpretation
(examples also from Kirs-
ner 1973) 5)
Five soldiers remained in the camp.
6)
There remained five soldiers in the camp.
Again, it is undesirable to require that tTzere-insertion refer to such semantic features as "agency" since, in general, particular transformations seem to be otherwise stateable independent of the meaning of the categories to which they apply.
And there
is no obvious general constraint on transformations to account for such cases.
For example, one could not constrain movement
transformations from applying to "agentive" noun phrases, as this would falsely exclude 7)
Will a bearded student be likely to wave from the roof?
8)
Are five soldiers expected to remain in the city?
where the noun phrases a bearded student and five soldiers with the agentive interpretation have undergone the application of various movement transformations. There is other data which indicates that, if there is a transformation of there-insertion,
then it would have to be sensitive 2 to semantic information of a very subtle sort. For example, Kimball (197 3) has noted that the class of intransitive verbs which permit there-insertion
can in part be roughly said to be
characterized by the semantic notion "coming into being". 2
See also Hellberg
In
(1970) for a discussion of some Swedish data.
47 Kimball's words:
"the existential there can appear with a sen-
tence if it expresses coming into being of some object, where this coming into being can include coming into the perceptual field of the speaker". consider the contrasts 9)
Compare also Jespersen
(1940).
Thus
(from Kimball 1973)
There ran a man
Í into the room ] from the building. ( *around the track.
10) There rose a green monster from the lagoon. •There sank a green monster into the lagoon. (acceptable perhaps if said by an underwater observer) 11) There began a riot. •There ended a riot. In the last example, the coming into being is into the "timespace" of the speaker.
For other examples of the class of verbs
permitting the existential there, see Borkin and Petersen
(1972).
Again, it seems that the statement of tfrere-insertion would require semantic information of the sort which is not in principle available to the statement of particular
transformations.
Notice, furthermore, that the tfcere-insertion transformation would have to be sensitive to the generic/specific semantic reading of sentences.
That is, compare
12)
Minors must be in the dorm by midnight.
13)
There must be minors in the dorm by midnight.
Sentence 12) has a generic interpretation; i.e., involving all minors and perhaps a specific interpretation as well of some minors.
This specific reading of some minors is brought out
clearly in sentence 13) where it is the only possible reading; i.e., the dorm must have minors in it by 6:00.
In any case the
sentences 12) and 13) do not have the same range of readings, and it is not clear that a there-insertion
transformation could
in principle be made to distinguish these meanings for the same reasons as above. Another case of the same sort is given in Kimball
(1973)
where he points out the ambiguity in 14)
Condors are in the Andes.
where condors can be interpreted either generically, as all con-
48 dors, or specifically as some condors.
However, sentence 15)
is unambiguous, having only the specific interpretation of some condors 15)
There are condors in the Andes.
A further case (Kuno 1971) where the existential construction has a different range of interpretation than its putative underlying source involves the order of interpretation of quantifiers. For example, in 16)
Many students are in all classes.
the predominant interpretation is that there are many students who attend all classes, a quite different interpretation than in 17)
There are many students in all classes,
which means 18)
All classes have many students in them.
Sentence 16) can be read with the latter interpretation only with great artificiality, if at all. are discussed in Kuno
A number of similar cases
(1971).
Turning now to another phenomenon, notice the ambiguity in 19) 19)
There is a king in France.
In one interpretation, the sentence means a particular king, say from Greece, is in France, as in 20)
A king is in France.
In the second interpretation, the sentence means 21)
France has a king.
Notice that it is impossible to derive both readings from an underlying 20) since the latter has only the first interpretation.
Kimball
(1973) has noted a number of similar contrasts
as in 22) 23)
There is * Space) Corn /
ispace) (corn /
in the manger.
is in the manger.
49 24)
There is
(fire ) in his eye. U sty/ *Fire \ is in his eye. A sty J
25) 26)
There is
27)
ja pain 1 (a splinter/
* A pain \ A splinter)
As Kimball notes
in my arm.
is in my arm.
(observation attributed to M. Pryse), the starred
cases—space, fire, α pain—all while the unstarred cases—corn,
involve "inalienable possession" a sty, a splinter—without
this
interpretation all have underlying full sentence paraphrases. Thus one can derive the inalienable form there is a pain in my arm by tfcere-insertion only at the cost of postulating a nonexistent deep structure *a pain is in my arm, which obligatorily undergoes tftere-insertion.
But this is obviously an artifice
which avoids simply saying that certain phrases associated with the existential there may have a special tation.
(inalienable) interpre-
There are many phrases which have no full sentence para-
phrases and which in our phrase structure analysis would be simply generated with there in deep structure 28)
There's an answer to the question. *An answer is to the question.
29)
There's a solution to the problem. *A solution is to the problem.
30)
There's a secret to his success. *A secret is to his success.
31)
There's a trick to it. *A trick is to it.
32)
There's a catch to it. *A catch is to it.
3 3)
There's a top to the table. *A top is to the table.
34)
There's an explanation to/for everything. *An explanation is to/for everything.
35)
There's a method to his madness. *A method is to his madness.
36)
There's a funny ring to that argument. *A funny ring Is to that argument.
37)
There's a difference between X and Y. *A difference is between X and Y.
50 38)
There's no point to that. *No point is to that.
39)
There's no use in doing that. *No use is in doing that.
40)
There's something to that. •Something is to that.
41)
There's nothing to it. •Nothing is to it.
Not only do the forms 28)-41) have no underlying full sentence paraphrases; some of them, for example 41), are apparently not generated anywhere in the grammar except as existential
there-
constructions 42)
3.6.
*I have nothing to it. •Nothing to it was found/seen by John.
Conclusion
To sum up, we have illustrated a range of sentences involving the existential there, many for which there is no underlying full sentence paraphrase source and a number which have different semantic interpretations than their source sentences. In some of these cases it might be argued that the
there-inser-
tion transformation applies freely, with rules of surface structure semantic interpretation accounting for the meaning.
Thus
in the case of the generic-specific examples 12)-15) in 3.5., one might try to account for the range of interpretation by surface structure rules of interpretation, particularly since it has been argued elsewhere on independent grounds
(Jackendoff
1972) that semantic properties like genericity are properties of surface structure.
Similarly, it is conceivable that such ex-
amples as 16), 17) in 3.5., involving quantifiers, might be handled with a rule of t^ere-insertion with the use of surface structure rules of interpretation, since there is some independent evidence that the proper interpretation of quantifiers depends on surface structure interpretation; see Chomsky Jackendoff
(1970b) and
(19 72).
In fact just such an approach to some of these problems as, for instance, the generic-specific problem, is adopted by Milsark (19 74) who proposes a rule of semantic interpretation on derived (possibly surface) structure
(the E rule).
We will discuss this
51 proposal in detail in Chapter 5. Milsark's discussion establishes clearly that the existential construction is different in meaning from its putative source under a transformational analysis.
Thus the following sentences,
with identical underlying structures, have different interpretations 1)
Unicorns are believed to be in the garden.
2)
There are unicorns believed to be in the garden.
In the second, but not the first, it is necessary that some unicorns exist. Moreover, Milsark's discussion serves to confirm our claim that the nature of the data that we have dealt with in this chapter is semantic rather than syntactic in nature.
Thus we do not
wish to mention such semantic factors in the structural condition of a tftere-insertion transformation for this would vastly increase the expressive power of the theory of transformations. Although Milsark seeks to show that the there-insertion
trans-
formation can be kept free of such semantic restrictions by adding a rule of semantic interpretation on derived structure, we will show in Chapter 5 that the entire range of facts can be accommodated in the phrase structure analysis we have proposed without any iTzere-insertion transformation at all. We turn now in Chapters 4-8 to a criticism of various transformational analyses of the English existential.
4.
STANDARD
THERE-INSERTION
4.1.
Introduction
In the following chapters we will consider various versions of the transformational analysis; viz., the hypothesis that existential there is aot generated in the base, but is inserted by transformation. Most analyses of there-insertion
assume that 1) is derived
from the structure underlying 2) 1)
There is a man in the garden.
2)
A man is in the garden.
The there is then subsequently inserted by a transformation which applies to the structure underlying 2). We will discuss first the formulation of t^ere-insertion in Burt (1971) which is typical of the transformational analysis and which has the advantage of being more explicit than many similar proposals in the literature.
This analysis may be termed
the "standard" analysis of t^ere-insertion. We will proceed in the following way. mational rule of there-insertion mulation of the rule. a wider range of data.
We state the transfor-
and ask what motivated the for-
We then seek to extend the rule to cover We show that the rule of tfcere-insertion
can handle the data only at the cost of considerably complicating the grammar by the addition of ad hoc conditions to the structural description of the rule which are unnecessary in the alternative phrase structure analysis. The rule of tTzere-insertion
1
is as follows
We have modified Burt's formulation of tfcere-insertion slightly in ways irrelevant for the following discussion.
53 3)
TTzere-insertion NP-
A
Aux-ie-W
Β
Aux S.D.
1
2 3
S.C.
there
Condition:
4
2 3+1 4
a) b)
opt
5
1 has an indefinite determiner Be directly follows T(ense)
We present three arguments analysis of tTzere-insertion : 2.
5
2
against this transformational
1.
The Two Be s Argument (4.2.).
The Semi-modal Argument (4.3.).
Argument (4.4.).
3.
The Predicate Nominal
We also consider a rule ordering argument
which is an apparent counter-argument for the transformational analysis (4.5.) . 4.2.
The Two Be s Argument
Observe that the rule of there-insertion has the effect of moving the subject noun phrase after the be, while inserting the morpheme there into the vacated subject position.
It may
apply in the contexts which meet the structural description of the rule (see A. and B.-3) in 4.1.).
Furthermore, there are
two conditions on the application of the rule. The structural description of t/zere-insertion corresponds in the phrase structure (henceforth PS) analysis to the DS Condition (15) in 1.2.).
To see this, consider the fact that we
have 1)
There are ghosts,
but not 2)
*I like there.
In the phrase structure analysis we accounted for this by the DS Condition (15) in 1.2.) which restricts there to the following context 3) 2 3
Aux-be-NP
See Emonds (1970) for a discussion of these arguments and a different proposal for a transformational rule of there-insertion. See also Chapter 6. Condition a)(3) in 4.1.) was discussed above in 3.4.
54 Thus the there in 1) meets this condition, but the there in 2) fails to meet it. In the transformational analysis there may be inserted into a tree with the structural description 4)
NP-Aux-be-W
(=A.-3) in 4.1.)
This structural description is met by the following tree (where W=0, the null string) 5)
S NP
Aux
VP
Τ ghosts Pres
be
but not by the following tree, which underlies 2)
like there Thus it is impossible to derive 2) whereas 1) is correctly generated . So with respect to constructions like 1) there is no evidence available for choosing between the PS analysis and the transformational analysis. Now consider forms containing the Prog(ressive) as in 7)
There is a man standing on the corner.
The PS analysis makes the claim that the be generated in the deep structure of 7) is not the Prog be, but is the copula be. Earlier we pointed out that there were at least two logically
55 possible alternative analyses for phrases like a man standing on the corner with there generated as a subject noun phrase in the base.
One possibility is that a man standing on the corner
is a reduced relative as in 8)
NP [ there ].._-Aux-be-.._ NP NP [a man standing on the corner]..NP
In this analysis the noun phrase a man standing on the corner is generated by exactly the same rules that generate it in NP
_] ρ is waiting
to see
you.
The second possibility is that a man standing on the corner is derived as a reduced cleft as in 9)
N p [there] N p -Aux-be- N p [a
man] N p ~ s [a man is standing
on the corner]c In this analysis the phrase a man standing on the corner is generated by the same rules that generate it in Look! man standing on the corner.
That's a
We presented evidence in Chapter 2
that the reduced cleft analysis could explain, for example, facts about intonation and the existence of constructions with apparent full relative-reduced relative sequences.
Observe that both
in the case of the reduced relative and in the case of the reduced cleft analysis, the DS Condition, appropriately modified (see 2.2.), is satisfied. In the there-insertion analysis being considered 7) has the following deep structure 10)
a man
Pres
be+ing
stand on the corner
Observe that the terminal string in 10) cannot be segmented into 11)
NP-Aux-be-W
(=4)
56 Thus for the case where the relevant be is in the auxiliary, we must state another context for there-insertion.
This is the
reason for Part B. in the structural description of tTzere-insertion (3) in 4.1.).
That is, the terminal string in 10) can be
segmented in accordance with Part B. 12)
a man-.
[Pres-be-ing]
nuX
-stand on the corner ÄUX
NP
X
be
Υ
Ζ
1
2
3
4
5
That is, the phrase marker 10) receives the proper analysis 13), indicated by the bracketing 13)
S
Then there-insertion structure 7).
applies to 13) to yield the correct surface
Notice that Part A. and Part B. of the structur-
al description of there-insertion
cannot be collapsed into a
single generalization. So with respect to constructions like 1) and 7) there has been no additional evidence uncovered for choosing between the 4 PS analysis and the transformational analysis. The statement of iftere-insertion in the transformational analysis requires the specification of two separate environments in the structural description of the rule
(Parts A. and B.-3) in 4.1.), environ-
ments which cannot be collapsed into a single statement.
Sim-
ilarly, in the phrase structure analysis, we allow there to be generated in deep structure in two contexts -in the environment of various nominal structures 4 5
(predicate nomináis, relative clause
We put aside the question of how the transformational analysis handles the data in Chapter 2 until 4.5. below. But this is no complication, since, as pointed out in 2.4., the phrase structure analysis entails that there occurs in all copula-Predicate structures (including copula-Locative).
57 structures) and in cleft structures. Observe next sentences containing in addition the Pass (ive) morpheme be+en as in 14)
There are some new houses being built.
The PS analysis claims that the are in 14) is not the Pass(ive) be, but is the copula be generated in deep structure. Again, given that there is a noun phrase generated in deep structure, two possible analyses present themselves as alternatives. In the f i r s t , the p h r a s e some
new
houses
being
built
is a reduced
relative 15)
„„ [there]., -Aux-be-„„ [some new houses being built].,_. NP NP n NP NP
Thus the noun phrase some new houses being built would be generated by the same rules that generate it in the context N p [ ]Np were
destroyed
by fire
yesterday.
The second alternative is that the phrase some new houses being built is a reduced cleft construction 16)
[there] ..-Aux-be--, [some new houses],..-, [some NP NP NP S new houses are being built]g
NP
In this a n a l y s i s the p h r a s e some
new
houses
being
built
is gen-
erated by the same rules that generate it in such constructions as Look!
That's
(there's)
a new
house
being
built
over
there.
Note again that, both in the case of the reduced relative and in the case of the reduced cleft analysis, the DS Condition, modified to account for cleft structures, is satisfied. In the tfcere-insertion analysis under consideration this sentence will derive (after passive has applied) from the following structure 17) N]
some new houses
Pres be+ing
be+en
build
58 Observe, however, that the terminal string in 17) can be segmented in two ways by Part B. of the structural description of there-insertion 18)
Np[some
new
houses] N p " A u x [Pres-be+ing-be+en] A u x ~build
NP
X
be
Υ
Ζ
1
2
3
4
5
In other words, the phrase marker 17) receives the proper analysis 19), indicated by the bracketing
And, secondly, the terminal string in 17) can be segmented as in 20)
N p [some
new
houses] N P ~ A u x [Pres-be+ing-be+en] A u x ~build
NP
X
1
2
fee 3
Υ
Ζ
4
5
Here the phrase marker 17) receives the following proper analysis, again indicated by bracketing
If tftere-insertion applies to the structure in 19) we get the correct surface form 14).
On the other hand, if it applies to
the structure in 21), we get the ungrammatical 22)
*There are being some new houses built.
59 Thus in order to exclude 22) it is necessary to add some ad hoc condition to the rule.
This is the reason for Condition b) on
tTzere-insertion (3) in 4.1.) 23)
Condition: Be directly follows Τ (ense)
This condition will reject 21) as an input to tTzere-insertion thus preventing 22). The ungrammatical 22) is ruled out in the PS analysis by the Progressive Condition, appropriately modified for clefts (see 2.2.), which excludes Prog from Aux in certain cases as in 24)
* N p [there] N p -are being~ N p [ghosts) N p
25)
* N p [there] Np -are being~ Np [some new houses being built] N p
26)
* N p [there] N p -are being~ Np [some new houses] N p s
[some new houses are being built]g
But observe that such a condition is also necessary in the t^ere-insertion analysis. 27)
That is, to exclude
*There are being ghosts.
one must invoke the same principle as is required in the PS analysis.
Thus alongside of the general principle (the Progressive
Condition) required in either the PS analysis or in the thereinsertion analysis; viz., that in certain cases 28) the
Aux may not contain Prog
ere-insertion analysis has the ad hoc condition b) (3)
in 4.1.) 29)
Condition:
Be directly follows T(ense) (=23)
Thus on grounds of simplicity we must prefer the PS analysis, which does not require Condition 29). lysis of there-insertion,
Even worse for the ana-
however, is the fact that Condition
29) gives precisely the wrong results in some crucial cases in6
What is meant is that the first be after Τ(ense) is selected. Literally interpreted, Condition 23) would not allow thereinsertion to produce there may be someone in the garden since may, not be, follows Τ(ense) in underlying structure.
60 volving semi-modals, which we turn to now. 4.3. The Semi-modal Argument Consider the following sentence 1)
A demonstration is going to be in the gym.
We will consider what happens when iTzere-insertion applies to 1). There are two possible analyses depending whether or not it. is assumed that be going to belongs to the auxiliary or not.' In Case I we assume that it does; in Case TI we assume that it does not. Case If he going to is an Aux, then 1) has the following two possible segmentations under tTzer-e-insertion (3) in 4.1.). Part A. of the structural description yields 2)
NP
[a demonstration] Np - Aux [Pres-be going to] Aux * NP 1
Aux 2
be-in the gym be
3
W 4
Thus Part A. of the structural description permits the following proper analysis for 1) 3)
[a demonstration][Pres be going to][be][in the gym] Part B. of the structural description yields
7
See Jenkins (1972) for a discussion of semi-modals.
61 4)
Np[a
demonstration] N p - A u x [Pres-be-going
to]Aux~
NP
X
be
Y
1
2
3
4
be in the gym Ζ 5 And Part B. of the structural description permits the following proper analysis for 1)
Consider now Condition b) on there-insertion; 6)
Condition:
viz.,
Be directly follows T(ense)
(=23) in 4.2.)
This condition rules out the proper analysis 3) since the relevant be
(see 2) above) does not directly follow Tense·, Pres
is followed by the be of be going
to.
Thus the following gram-
matical sentence is ruled out by condition 6) 7)
There is going to be a demonstration in the gym.
On the other hand, consider now 5).
Here Condition 6) does not
rule out the proper analysis 5) since the relevant be directly follows Pres
(compare 4) above).
Thus the following ungrammat-
ical sentence is generated 8)
*There is a demonstration going to be in the gym.
Thus we get exactly the wrong results.
With Condition b) we
cannot generate the grammatical sentence corresponding to the proper analysis 3), but generate instead the ungrammatical sentence 8), corresponding to the proper analysis 5). Case II: Let us now assume that be going
to does not belong to Aux.
62
Then Part A. of there-insertion
(3) in 4.1.) will analyze 1)
in the following way 9)
Np[a
demonstration] N p - A u x [Pres] A u x -beNP
Aux
be
1
2
3
going to be in the gym W 4 Thus Part A. of the structural description permits the following proper analysis for 9) 10)
[a demonstration]
[Pres]
[be][going to be in the gym]
Note that Condition b); viz., that be must directly follow T(ense), is satisfied by the structure 10), since be directly follows T(ense). 11)
Thus we get the ungrammatical 11) »There is a demonstration going to be in the gym.
(=8)
However, not only do we get the ungrammatical 11), but there is no way to get the grammatical 7), for now Part B. of the structural description of tfcere-insertion cannot apply, since be going to does not belong to Aux. Of course, we can again state the facts by adding another ad hoc condition to Condition 6) (fee directly follows T(ense)) by stating that be going to (and be to—see
below) are except-
ions to this condition, but doing this only brings out more clearly the ad hoc nature of Condition 6) itself, which was devised only for the case where Pass be cooccurs with Prog be. Let us see how the PS analysis accounts for all the facts. The grammatical 12)
There is going to be a demonstration in the gym.
(=7)
63 g if it is a reduced relative, derives from an underlying 13)
XTT,
NP
[there][Pres-be NP Aux
going to]- -beAux
.._[a demonstration in the g y m ] ^ NP NP where there is a deep structure noun phrase satisfying the DS Condition 15) in 1.2. and the noun phrase a demonstration
in
the gym is generated by the same rules that generate it in „_[ NP
]„_ was cancelled. NP The ungrammatical 11) will not be generated from 14)
Np[there]Np-Aux[Pres]Aux-be-Np[a
demonstration
going to be in the gym] N p since the rules above which generate noun phrases of the form a demonstration
in the gym in the context „_[ ]„„ was cancelled NP NP must exclude the noun phrase a demonstration going to be in the gym from the same context; i.e., we do not get 15)
*A demonstration going to be in the gym was cancelled.
nor, for example, 16)
*A nurse going to come will help you.
although we get the syntactically and semantically well-formed 17) A demonstration which was going to be in the gym was cancelled. 18) A nurse who is going to come will help you. which clearly shows that 15), 16) are to be ruled out on syntactic rather than semantic grounds. Similar comments apply if 12) is to be analyzed as a reduc'ed cleft, as the facts about the intonation contour seem to indicate (intonation peak on a 19)
demonstration)
N P l t h e r e l N P ~ A u x [ P r e s _ b e 9oing t o ] A u x ~ b e - N p [ a demonstration] N p - g [ a demonstration is in the gym] g
8
We assume here that be going to belongs to Aux. It is important to note, though, that the phrase structure analysis does not depend on whether be going to belongs to the Aux or not. If we were to assume be going to were, say, an intransitive verb, like seem, the correct sentences will also be generated.
64 where there again satisfies the DS Condition, modified for clefts, and the phrase a demonstration
in the gym in 12) is generated
by the same rules that generate such constructions as in that 's going to be a demonstration
in that gym/on that stage and that's
going to be a fireplace over there. Observe in general that the contexts where we get Cleft Reduction correspond to the contexts where we get Relative Clause Reduction g 20) The man (that/who is) standing on the corner is Belmondo. (Relative Clause Reduction) That's Belmondo (that/who is) standing on the corner. (Cleft Reduction) 21)
The man (that/who is) next to my mother is Belmondo. (Relative Clause Reduction) That's Belmondo (that/who is) next to my mother. (Cleft Reduction)
22)
The man (that/who is) being beaten up is Belmondo. (Relative Clause Reduction) That's Belmondo (that/who is) being beaten up. (Cleft Reduction)
23)
The car (that/which is) parked on the corner is a Mustang. (Relative Clause Reduction) That's a Mustang (that/which is) parked on the corner. (Cleft Reduction)
Notice that there are a number of cases where Cleft Reduction does not apply. Reduction either.
In these cases we do not get Relative Clause One such case involves the semi-modal be go-
ing to 24)
That was John who was going to be here at 5:00.
25)
*That was John going to be here at 5:00. Reduction)
26)
The man who was going to be here at 5:00 didn't show up.
27)
*The man going to be here at 5:00 didn't show up. (Relative Clause Reduction)
(Cleft
Thus it is impossible to generate the ungrammatical form
9
We will speak of Cleft Reduction and Relative Clause Reduction as separate rules for purposes of discussion, but obviously the similarities between the two rules is no accident and calls for explanation.
65 28)
»There's a demonstration going to be in the gym.
since both Relative Clause Reduction and Cleft Reduction are prohibited with be going and 16)).
to as in 25), 27) (compare also 15)
However, nothing prohibits the generation of the gram-
matical 29)
there~ A u x [is going
t o
^ux~be
a
demonstration in
the gym from 13) or 19). Thus in the PS analysis of there 11) is impossible to generate.
But in the tAere-insertion analysis one must add another
ad hoc condition to the grammar to generate the correct sentences.
Furthermore, the ad hoc condition; viz., that only the
second be of be going
to be may be selected in the structural
description of tftere-insertion violates the ad hoc condition 6) that be directly follow T(ense).
Again observational adequacy
is reached only at the cost of complicating the statement of there-insertion
(and hence the grammar).
Since nothing new must
be stated about there in the PS analysis to generate all the correct surface forms, we must on the basis of this evidence select the simpler PS analysis over the tfrere-insertion analysis. Exactly the same point can be made about the semi-modal be to as was made for be going
to above.
Thus consider
30)
There are to be three senators at the conference.
31)
*There are three senators to be at the conference.
Again the there-insertion
transformation
(3) in 4.1.) gener-
ates the ungrammatical 31), but excludes the grammatical 30), unless another ad hoc condition for be to is added to the structural description of
there-insertion.
In the PS analysis 30), if it contains a reduced relative following be, derives from an underlying 32)
Np[there]Np-Aux[Pres-be
to]Aux"be-Np[three
senators
at the conference] NP where there is a deep structure noun phrase satisfying the DS Condition
(15) in 1.2.) and the noun phrase three senators
at
66 the conference in
Np[
]Np
is generated by the same rules that generate it
became ill.
The ungrammatical 31) will not be generated from 33)
N p [there] N p - A u x [Pres] A u x -be- N p [three
senators to be
at the conference] Np since the rules above which generate noun phrases of the form three senators at the conference in the context „T„ [ ]„,„ became NP NP ill must exclude the noun phrase three senators to be at the conference 34)
from the same context; i.e., we do not get *Three senators to be at the conference became ill.
nor, for example, 35)
*Two speakers to come tonight became ill.
although we get the syntactically and semantically well-formed 36)
Three senators who were to be at the conference became ill.
37)
Two speakers who were to come tonight became ill.
Similar comments apply if 30) is to be analyzed as a reduced cleft, as the facts about the intonation contour once more indicate (compare the possible intonation peak on three 38)
N p [there] N p - A u x [Pres-be
senators)
to] A u x ~ b e - N p [three senators]^-
g[three senators are at the conference] g where there again satisfies the DS Condition, modified for clefts, and the phrase three senators at the conference
in 30) is gener-
ated by the same rules that generate such constructions as in that's to be a new hospital over there. In the case of 34) and 35) above we observed that Relative Clause Reduction could not apply With the semi-modal be to.
The
same holds true for Cleft Reduction as can be seen by comparing 39)
That was John who was to be here at 5:00.
40)
*That was John to be here at 5:00.
41)
The man who was to be here at 5:00 didn't show up.
42)
*The man to be here at 5:00 didn't show up. ative Clause Reduction)
(Cleft Reduction) (Rel-
67 Thus it is impossible to generate the ungrammatical form 43)
*There are three senators to be at the conference.
since both Cleft Reduction and Relative Clause Reduction are prohibited with be to.
However, nothing prohibits the generation
of the grammatical 44)
there~ A u x [are to] A u x ~be three senators at the conference
from 32) or 38). Thus in the PS analysis it is impossible to generate the incorrect form 31).
This form is excluded in the t^ere-insertion
analysis only at the cost of adding an ad hoc condition on be to to the structural description of ifcere-insertion (3) in 4.1.). Thus again the simpler PS analysis must be preferred over the t/zere-insertion analysis. To conclude with another example involving the semi-modal be to, compare 45)
There is nothing more to be said (about this matter).
Observe that the standard analysis permits the underlying 46)
Nothing more is to be said (about this matter).
which, however, has a different meaning.
That is, 45) means
that the topic being talked about has been exhausted, while 46) is a request for silence. There is only one source for 45) in the phrase structure analysis; viz., 47) where
there-Aux-be- v [nothing more to be said) v . Λ A
v[
] v is a reduced relative or a reduced cleft,, but does X X not derive from a full sentence paraphrase nothing more is to be said.
That is, we assume that we must independently gener-
ate underlying complex phrases of the form 48)
nothing to do (say), nothing more to be said, nowhere to go
to account for such sentences as 49)
I have nothing to say.
68 50)
There is nothing to say.
51)
There is nothing to be said.
52)
There is nowhere to go.
Observe that we do not have as possible sources the strings 53)
»Nothing is to say.
54)
»Nothing more is to be said. sense above)
55)
»Nowhere is to go.
(except in other
Thus the phrase structure analysis correctly accounts for the fact that the semantic interpretation of 45) differs from 46) (but not the there-insertion
analysis, unless we permit the ad
hoc condition for be to under discussion). 4.4.
The Predicate Nominal Argument
Notice that still another ad hoc condition must be added to the structural description of tTzere-insertion to prevent the derivation of 1) from 2) 1) 2)
»There are some graduate students union members. Some graduate students are union members.
Thus we must in effect require the following condition on t/zö^e-insertion (3) in 4.1.) 3)
Condition:
W may not contain Predicate Nominal
But in the PS analysis it is impossible to generate 1) since the rules of English grammar do not generate the phrase some graduate students union members as either a reduced relative or a reduced cleft.
That is, we do not have, for example, Rel-
ative Clause Reduction to yield 4)
* N p [Some graduate students union members] N p
will
speak tonight. This is excluded on syntactic grounds rather than semantic grounds as is seen by the existence of 5) N p[Some graduate students who are union members] N p will speak tonight.
69 We observed several times above that the contexts where we do or do not get Cleft Reduction correspond closely to the contexts where we do or do not get Relative Clause Reduction.
This
holds true also for predicate nominal constructions as can be seen by comparing 6)
That was Sam and Sue that/who were twins.
7)
»That was Sam and Sue twins.
8)
People that/who are twins usually look alike.
(Cleft Reduction)
9)
»People twins usually look alike. Reduction)
(Relative Clause
Thus it is impossible to generate the ungrammatical 1) since both Relative Clause Reduction and Cleft Reduction are prohibited with the copula be in predicate nominal constructions
(com-
pare 4) , 7) , 9) ) . Thus in the tftere-insertion analysis, the grammar must be complicated by the addition of Condition 3) to the statement of tftere-insertion
(3) in 4.1.), in order to rule out 1), but
in the simpler, and hence preferable, PS analysis, nothing new need be stated at all to generate the correct forms. 4.5.
Other Arguments
At this point we may ask what sort of facts motivated the formulation of a transformational operation of
there-insertion
in the first place to derive tfrere-constructions from full sentence paraphrases as in deriving 1) from 2)
A.
1)
There is someone in the garden.
2)
Someone is in the garden.
SEMANTICS
One piece of evidence on semantic grounds often advanced is the claim that 1) and 2) are synonymous.
In Chapter 3 we gave
a number of examples which disproved this claim.
However, even
assuming synonymity, there are syntactic arguments independent of meaning considerations which show that there is generated in the base rather than inserted by transformation 4.3., 4.4.).
(see 4.2.,
Thus once again we see that any argument for re-
lating two constructions transformationally which is based on semantic paraphrase alone is at best quite weak in the absence
70 of independent syntactic arguments. Β.
AGREEMENT
Turning to possible syntactic evidence for the derivation of 1) from 2) the facts of agreement between be and the following noun phrases might be cited 3)
There is someone in the garden. There are some people in the garden.
4)
*There are someone in the garden. »There is some people in the garden.
But, as we showed above, such examples provide no evidence for choosing between the two hypotheses since such sentences as 4) must be generated in the syntax anyway and are ruled out by the Condition on Agreement (7) in 1.3.) which takes semantic and extra-linguistic information into account. C.
SYNTACTIC DISTRIBUTION
A second piece of syntactic evidence is the fact that in 1) and in a large number of other cases the string after there is a syntactic permutation of a full sentence.
Thus in 1) the string
is someone in the garden is a permutation of 2).
But notice
that this is characteristic of strings which are generated by other transformations, such as Interrogative.
Thus consider
(restricting ourselves to the subset of sentences with be) 5)
Is someone in the garden? (Compare Q is someone in the garden with there is someone in the garden) Is someone standing on the corner? (Compare Q is someone standing on the corner with there is someone standing on the corner) Were any windows broken? (Compare Q were any windows broken with there were some windows broken)
That is, in each case the underlined string which follows the interrogative marker (later deleted) is a syntactic permutation of a full sentence 6)
Someone is in the garden. Someone is standing on the corner. Some windows were broken.
Thus at least for this set of cases there seems as much motivation for deriving t/iere-constructions from sentence permutations
71 as t h e r e is f o r t h e a n a l o g o u s
interrogative
cases.
But the p a r a l l e l b e t w e e n the tfcere-constructions atives collapses when we consider ináis.
Thus corresponding 7)
interrognom-
to
A d e m o n s t r a t i o n is g o i n g t o b e in t h e g y m . S o m e s e n a t o r s a r e to b e a t t h e c o n f e r e n c e . Some graduate students are union m e m b e r s .
we have the 8)
but
and
semi-modals and predicate
interrogatives
Q is a demonstration Q are any senators Q are any graduate
going to be in the gym to be at the conference students union members
not 9)
* t h e r e is a demonstration * t h e r e are some senators • t h e r e are some graduate
going to be in the gym to be at the conference students union members
If i t w e r e t h e c a s e t h a t t h e i n t e r r o g a t i v e s f o r m e d ; i . e . , if w e h a d *are bers?,
any
this w o u l d be evidence
transformation.
graduate
that there
in 8) w e r e
students is n o
ill-
union
mem-
interrogative
T h e o n l y a l t e r n a t i v e w o u l d b e to p l a c e a d
hoc
conditions on the interrogative
transformation
the right forms are generated.
B u t t h i s is p r e c i s e l y w h a t is d o n e
in t h e t r a n s f o r m a t i o n a l
a n a l y s i s o f tfcere-insertion.
c o n d i t i o n s o n t h e o p e r a t i o n of t ^ e r e - i n s e r t i o n to e x c l u d e the ill-formed as g r o u n d s
to g u a r a n t e e
sequences
in 9 ) .
f o r r e j e c t i n g t h e there-insertion
are
Ad
This would analysis
suffice in
i t is
a complete accident that the sentences are all starred
10a) 10b) 10c) In fact,
favor
conditions.
E v e n w o r s e t h o u g h is t h e f a c t t h a t u n d e r t h i s a n a l y s i s
if t h e d i s t r i b u t i o n o f d a t a w e r e a s
hoc
permitted
of the PS a n a l y s i s w h i c h d o e s n o t r e q u i r e the ad h o c
T h u s in t h e t^er-e-insertion a n a l y s i s
that
it w o u l d m a k e no
in
9).
difference
follows
T h e r e is a d e m o n s t r a t i o n g o i n g to b e i n t h e g y m . (well-formed) * T h e r e a r e s o m e s e n a t o r s to b e a t t h e c o n f e r e n c e . There are some graduate students union m e m b e r s , (well-formed)
if t h e d i s t r i b u t i o n o f d a t a w e r e a s in 10)
the
grammar
of E n g l i s h w o u l d be simpler since w e could drop the ad hoc t r i c t i o n s o n tfcere-insertion w e n e e d f o r be
going
to a n d
res-
predi-
72
cate nomináis to exclude 10a) and 10c). In the PS analysis it is no accident that the distribution of data is as in 9), but not as in 10)·
This is expressed in
the PS analysis, which predicts that exactly that set of thereconstructions will be ruled out (see 9)) whose derivation would violate independently needed constraints on the application of Relative Clause Reduction and Cleft Reduction; viz., the fact that neither may apply to the be of the semi-modals be going to and be to nor to the be in predicate nominal constructions. Thus the PS analysis predicts that the distribution of the data cannot be as in 10) (which in this framework would complicate the grammar) but must be as in 9). D.
RULE ORDERING
We consider here a possible argument based on rule ordering which as been taken to show that there must be inserted transformationally. 11)
The argument can be summarized as
The t^ere-insertion transformation must follow the Passive transformation.
Thus it is argued that 12)
There were several soldiers injured in the war.
must be derived from 13)
Δ injured several soldiers in the war by Pass.
However, tftere-insertion cannot apply to the deep structure string 13) since the be morpheme necessary for its application has not yet been inserted by the Pass(ive)
transformation.
Thus, it is
argued, there must be inserted by transformation following Passive . However, this argument can be seen to be fallacious when it is recognized that the be in the string in 12) is the deep structure copula be rather than the passive be.
That is, we have the
structure
10 See Perlmutter (1970). Burt (1971) considers the case where Pass(ive) be is generated in deep structure and gives arguments for the order 1. Passive 2. there-insertion. Her examples pose no additional problems for our analysis.
73 14)
N p [there] N p - A u x [Pres] A u x -be- x [several
injured in the war] x
(note:
soldiers
X = NP or NP-S)
where there is a deep structure noun phrase satisfying the DS Condition and, furthermore, the string several soldiers
injured
in the war results from Relative Clause Reduction or Cleft Reduction. Thus data as in 12) does not argue for the transformational insertion of there.
Such data is handled in the PS analysis
where there is generated in deep structure and, as we have seen, this analysis handles additional data which is beyond the scope of the there-insertion
analysis unless ad hoc conditions on the
rule are introduced into the grammar. It thus appears that there is no evidence—either semantic, based on agreement, on the permutability of sentence constituents, or on rule ordering—from which one can conclude that tTzere-constructions derive from structures underlying their full sentence paraphrases. 4.6.
Conclusion
In this chapter we presented a number of syntactic arguments, including the Two be s Argument, the Semi-modal Argument, and the Predicate Nominal Argument, which argued against deriving the existential construction in 1) from its full sentence paraphrase in 2) 1)
There is a man in the garden.
2)
A man is in the garden.
We showed that we could account for all the facts by considering 1) to be derived by Cleft Reduction in the phrase structure analysis . Let us return now to the question of the range of possible constituents occurring in the context 3)
There is/are
As we have noted, simple noun phrases may occur in this context as in 4)
There are ghosts.
74 However, consider now the status of the italicized portion of 5)
There is a man in the garden.
In Chapter 2 we gave arguments for considering the italicized phrase in 5) to result from the operation of Cleft Reduction. In 2.2 we left open the question as to whether the italicized phrase in 5) might in fact also be a reduced relative; i.e., might derive by the parallel operation of Relative Clause Reduction.
In this derivation, then, a man in the garden would be
generated after there by the same rules that generate it in the context N p [
] N p is waiting to see you.
Then 5) would be struc-
turally ambiguous, with the italicized phrase resulting from either Cleft Reduction or Relative Clause Reduction. One might try to argue against the reduced relative analysis on semantic grounds.
That is, while one might make a case that
some reduced relatives derive from underlying full relative clauses; e.g., 6) from 7) 6)
There are a lot of people willing to help.
7)
There are a lot of people who are willing to help.
such an analysis seems dubious for other cases.
Thus in some
cases there is a full relative clause as a possible source, as 9) for 8) 8)
There is a man in the garden.
9)
There is a man who is in the garden.
but many speakers feel a difference in meaning between the two expressions.
But if this is the case, then here the derivation
of such reduced relative noun phrases as a man in the garden from an underlying full relative clause is questionable.
We can
in this view regard the difference in meaning in 8) and 9) as a piece of evidence that "reduced relatives" are not in fact derived from full relative clauses. (1971).
For similar evidence, see Williams
Thus the existence of non-synonymous expressions like
8) and 9) don't necessarily argue that the phrase a man in the garden is not a reduced relative, but these facts might be interpreted as showing that reduced relatives are not derived from full relatives.
75 It is important in this connection to note that the facts in 8), 9) provide no argument for the transformational analysis of there-insertion
either.
One must keep in mind the fact that
the standard tfcere-insertion proposals in fact always allow two sources for 8) , a full sentence paraphrase as in 10) and a noun phrase source as in 11) 10)
A man is in the garden.
11)
A man in the garden be.
Intransitive structures like 11) must be postulated in order to generate such sentences as 12)
There is a God.
13)
There are women who are always in the kitchen,
which derive from underlying 14) 15)
A God be. Women who are always in the kitchen be.
In fact then, for the there-insertion
analysis too, we must gen-
erate the full range of nominal phrases in the context
Aux-
be, including a man in the garden as in 11) a man in the garden be to which there-insertion
applies, yielding 8).
Thus there is a man in the garden viz., 10) and 11).
(=8) has two possible sources;
As we have already demonstrated, it is the
full sentence paraphrase, source 10), rather than the noun phrase source 11) which entailed the adding of ad hoc conditions to the grammar.
In the phrase structure analysis for there, these
problems do not arise, for in this analysis the sentences there is a man in the garden and a man is in the garden are not syntactically related. Now let us ask what the difference in meaning between there is a man who is in the garden and there is a man in the garden means now for the there-insertion
analysis.
Again this fact can be
interpreted to mean that reduced relative noun phrases like a man in the garden are not derived from underlying full relatives. To see this, consider the fact that we must generate 16) alongside 15) 16)
Women who are in the kitchen be.
76 If "reduced relatives" are derived from underlying full relative clauses, then 17) will derive from 16) by Relative Clause Reduction 17)
Women in the kitchen be.
which will yield 18) by 18)
there-insertion
There are women in the kitchen.
However, as noted, for certain speakers, 18) is not synonymous with 19) (as with 8), 9)) 19)
There are women who are in the kitchen.
Thus in the tftere-insertion analysis too, it would follow that if 18) and 19) are different in meaning, then "reduced relatives" like women in the kitahen do not derive from full relative clauses. Let us sum up the immediately preceding discussion.
We have
assumed that 20/ and 21) are well-formed sentences of English 20)
There is a man who is in the garden.
21)
There is a man in the garden.
(=9)
(=8)
and have noted that some English speakers feel a difference between 20) and 21).
But, we have argued, whatever the facts here
are, they do not bear in any obvious way on the choice between the PS analysis and tfcere-insertion, but may bear ση the different question as to whether 22) is derived from the structure underlying 23) by a rule of Relative Clause Reduction or not 22) 23)
Np[a
man in the garden] N p man
who
"''s
in
5 a r ( ^ e n ^NP
To see this again let us examine the consequences for both the PS analysis and the tftere-insertion analysis of the two independent assumptions 24)
20) and 21) are
25)
22) is (not) derived from 23) by Relative Clause Reduction.
CASE I:
(non-)synonymous.
20) and 21) are synonymous. 22) is derived from 23) by Relative Clause Reduction
In the PS analysis 20) and 21) are both derived from the same
77 underlying
structure
26)
Np[there]Np-Aux-be~Np[a
the
man
a man
21)
involves the application
Reduction.
20)
and
In t h e
Since
21) h a v e
is a c c o u n t e d
t&ere-insertion
27)
N p
t
a
in
garden]sJNp
The d e r i v a t i o n of their synonymy
is
m a n
w
On the other hand,
ho
the same u n d e r l y i n g
Clause
structure,
for.
analysis
i s in t h e
21)
of R e l a t i v e
allows
20)
is d e r i v e d
from
garden]Np-be
two p o s s i b l e
sources; viz.,
27)
w i t h s u b s e q u e n t a p p l i c a t i o n of Relative C l a u s e R e d u c t i o n and 28)
Np
[ a man]Np-Aux-be-pp[in
The deep structures
27)
antic interpretations CASE
lis
20)
the
and
21)
garden]pp
a n d 28) m u s t b o t h r e c e i v e
by the semantic are
identical
synonymous.
for purposes
one-in the g a r d e n ] N p
of i l l u s t r a t i o n ,
is g e n e r a t e d d i r e c t l y
other derivation of reduced relatives w i l l do as w e l l Then derived
20)
26)
Np
27).
[some-
structure
see Williams
f r o m 26)
any
and
21)
(1971). is
a n d 29) m u s t t h e n r e c e i v e
man-in
the
identical
garden]Np interpretations
component.
However,
21)
20)
has the underlying
has again two possible
sources;
struc-
viz.,
ar.d a l s o 30)
N p
[a m a n - i n
the
Here the deep structures cal
from underlying
is d e r i v e d
In t h e t f t e r e - i n s e r t i o n a n a l y s i s 28)
Np
in t h e b a s e , b u t
discussion;
[there]Np-Aux-be-Np[a
by the semantic ture
that
Clause
from
29) Both
for the following
in t h e P S a n a l y s i s
sem-
component.
22) is n o t d e r i v e d f r o m 23) b y R e l a t i v e Reduction. L e t us a s s u m e ,
also
interpretations
garden]Np-Aux-be
27),
28), a n d
by the semantic
30)
all receive
component.
identi-
78 CASE III:
20) and 21) are non-synonymous. 22) is not derived from 23) by Relative Clause Reduction.
In the PS analysis 20) is derived from the underlying structure 26).
21) is derived from the underlying structure 29).
Here the deep structures 26) and 29) must receive different semantic interpretations, corresponding to their difference in meaning by the semantic component. In the tTzere-insertion analysis 20) derives from 27) and 21) has two possible sources; viz., 28) and 30).
Here the semantic
component assigns a different semantic interpretation to 27) than it does to 28) and 30) which are interpreted identically. CASE IV:
20) and 21) are non-synonymous. 22) is derived from 23) by Relative Clause Reduction.
In the PS analysis 20) and 21) both have the underlying structure 26), where 21) derives by Relative Clause Reduction.
How-
ever, now we cannot account for the non-synonomy of 20) and 21) on the basis of deep structural differences, since the deep structures are identical. In the there-insertion
analysis 20) derives from 27) while
21) derives from both 27) by Relative Clause Reduction and 28). But if 21) can derive from 27), then we cannot account for the non-synonomy of 20) and 21) on the basis of deep structural differences, since they have identical deep structures.
On the
other hand, we cannot exclude 27) as a possible deep structure for 21) without complicating the grammar. In all four cases sketched above the same problems arise in either the PS analysis or the tfzere-insertion analysis.
As we
have seen, it makes no difference whether it is assumed that reduced relatives are derived from full relatives or not.
Thus
sentences like 20), 21) do not provide an answer to the original question we posed; viz., whether a man in the garden in 31)
There is a man -in the garden.
(=5)
is ever to be considered a reduced relative. Note that the following sentences might be given as evidence that a man in the garden is not a reduced relative in 31)
79
32)
In the garden there is a man.
33)
On the table there is a book.
34)
It's in the garden that there is a man.
35)
It's on the table that there is a book.
Thus in 32)-35) the locative phrase has been preposed by the operations of Adverb Preposing and Cleft Formation.
However,
this is not normally possible with reduced relative constructions 36)
I'm proud of the man in the garden.
37)
φ In the garden I'm proud of the man.
38)
φ It's in the garden that I'm proud of the man.
However, in this case, it is possible to argue that there is a man ϊη the garden is structurally ambiguous.
It may derive from
either 39)
there-Aux-be~ N p[a man in the garden] N p
40)
there-Aux-be- Np [a man] N p
or
where Loc is a sentence adverbial
Loctin
(i.e., S
the garden] L q c NP+VP+Adv), par-
allel to such constructions as 41)
John is
Np[a
42)
John met
hypocrite] Np
Lqc
[at h o m e l y
and Np[Mary]Np
Lqc[at
the dance] L o c
Notice that we have optional preposing in all of these cases 43)
In the garden there is a man.
44)
At home John is a hypocrite.
45)
At the dance John met Mary.
(=32)
Similar observations hold for sentential time adverbials as in 46)
There is a concert at six. At six there is a concert.
47)
John is a hypocrite on Sundays. On Sundays John is a hypocrite.
48)
John met Mary last night. Last night John met Mary.
80 Thus sentences like 32)-35) do not count as evidence against the analysis of a man in the garden in 49) as a reduced relative 49)
There is a man in the garden.
(=31)
as there is independent motivation for postulating a second structure as in 40).
But as we have no positive evidence in
favor of this assumption either, we will be forced to leave the question open here. However, we were able to present positive arguments for deriving a man in the garden in 49) by Cleft Reduction.
The ar-
guments were based on facts about stress and impermissible sequences of full and reduced relative clauses (Chapter 2).
Thus
we showed how we could explain the stress in 50)
There is a man in the garden.
and the fact that we can get what appears to be a reduced relative after a full relative in 51)
There's a man who knows you
in the garden. standing on the corner.
by assuming that the constituents in the garden, standing on the corner are the remnants of reduced clefts. It would be possible to get around the argument based on stress if one were to permit movement transformations, like there-insertion
in this case, to apply to structures phonetic-
ally interpreted for stress.
Thus notice that the natural in-
tonation center in 52) and 53) is 52)
That fool is
53)
A fôol is
in the garden. standing on the córner.
in the garden. standing on the corner.
If t?zere-insertion were to apply after whatever rules assign the (topical) stress in 53), we could then explain the stress contour in 54) and 50) 54)
There is a fôol
in the garden. standing on the corner.
There is some support for the assumption that movement rules may apply to phonetically interpreted syntactic structures in
81 Bresnan (1971), where she shows that wft-movement must apply after the application of certain rules of English stress; but in her framework the stress rules follow all cyclic transformations.
T/zere-insertion then would be the only cyclic transfor-
mation following stress placement. To sum up then, we have presented a number of reasons for regarding sentences like 55)
There is a man in the garden. There is a man standing on the corner. There is a man being beaten up outside.
to be derived by a process of Cleft Reduction, parallel to the derivation of such cleft sentences as 56)
It's Bill Smith who is standing on the corner. That's Bill Smith (who is) being beaten up outside. There's Bill Smith standing on the corner.
In order to speak more exactly about the underlying form of cleft constructions and the rules of cleft formation, more research is needed on the syntactic properties of cleft constructions such as are illustrated in 56).
Thus, for instance,
note that the focus constituent varies, depending on whether we are dealing with an -it-cleft, a that-cleft, or a
there-cleft.
With an it-cleft, we may have as focus constituents subjects, objects, and various prepositional phrase complements 57)
It's It's It's It's
John that hit Bill. (subject focus) Bill that John hit. (object focus) with John that Mary likes to talk. (PP focus) on Sundays that Bill goes to church. (PP focus)
With that-clefts, on the other hand, the possibilities are more restricted.
In general, for example, prepositional phrase
complements sound unnatural in focus position 58)
That's John that hit Bill. (subject focus) That's Bill that John hit. (object focus) That's with John that Mary likes to talk. (PP focus)
And with there-clefts the possibilities in focus position are still more restricted.
Here objects ^re excluded from focus
position as well unless they have become derived subjects 59)
There's John in the garden. (subject focus) •There's John Peter hit. (object focus) There's John being beaten up by Peter.
82 These properties of cleft constructions as well as the restrictions on Cleft Reduction touched on in Chapter 2 must be accounted for by any theory of cleft formation. Recall finally such expressions as (see 3.5.) 60)
There's an answer to that question. There's a solûtion to that problem.
We might treat such constructions as reduced clefts (observe the intonation peak on answer, solution) note 1).
(see Chapter 2, foot-
It might seem that the problem that we must assume
the unacceptable sources *an answer -is to that question, *a solution is to that problem reappears as in 61)
N p [there] N p -Aux-be- N p [an
answer] N p -
g[an answer is to that question]g However, we may regard such expressions as an answer to that question and a solution to that problem to be generated in the base as reduced cleft structures analogous to base generated nomináis as the bottom of the barrel, the prospects for peace, prolegomena
to any future metaphysics,
a house in the woods,
a man to do the ¿ob; see Chomsky (1970a). 62)
Compare
There's a servant to do the work. There's a top to the table.
In any case the extension of the lexical hypothesis from reduced relative-like structures to include reduced cleft-like structures seems plausible for a wide variety of cases.
5.
TÄ£7?£-INSERTI0N AND SEMANTIC INTERPRETATION
5.1.
Introduction
Another very recent analysis of tftere-insertion which attempts to overcome many of the difficulties pointed out in Chapters 3 and 4 for standard tftere-insertion is proposed in Milsark (1974).
There is also a discussion by Milsark of earlier work
of Jenkins
(1972; 1974a) in which he criticizes what he calls
the "phrase structure hypothesis" and the "cleft reduction hypothesis".
That is, he treats the hypothesis that there
occurs
in deep cleft structures as separate from the hypothesis that there occurs in deep predicate nominal structures and seeks to refute each one on separate grounds.
But as we have made clear
in the preceding chapters, neither of these partial hypotheses alone is adequate to handle all the facts.
What is in fact re-
quired is the phrase structure analysis that we have proposed which insists that there occurs in all copula-predicate deep structures—predicate nominal, cleft, and locative
structures.
Milsark's analysis relies heavily on the use of a semantic interpretation rule called the E rule which applies to derived (perhaps surface) structures, containing the existential His strategy is to leave the syntactic rule of
there.
there-insertion
as unconstrained as possible; i.e., without the many ad hoc conditions which were postulated in the past two chapters.
Of course,
as we saw, without these conditions, the rule of tfeere-insertion will overgenerate, producing many ungrammatical structures. Part of the function of the semantic interpretation rule applying to derived structure is to rule out the set of sentences rectly
incor-
generated.
Thus Milsark states the following as a first to the correct formulation of
there-insertion
approximation
84 1)
T^ere-insertion SD:
SC:
Χ
ΝΡ
1
2
1 there
Y be
Ζ
3
4
5
3
4 2 5
Obviously this general formulation of there-insertion
allows
any occurrence of NP whatever in the sequence NP-Y-be to be placed after any occurrence of a following be.
In order to ex-
clude unwanted derivations, Milsark requires that derived structures containing occurrences of the existential there be interpretable by a rule of semantic interpretation called the E rule 2)
E rule there AUX
(have-en) be NP X
is interpreted:
the class C denoted by NP has at least one member c such that P(c) is true, where Ρ is a predicate and Ρ is the reading of X What is important to note now is that if a derived structure containing an occurrence of the existential there is uninterpretable by this rule, then the structure is ruled out. see below how this rule operates.
We will
Thus the E rule takes over
the function of several of the main conditions proposed on the structural condition of standard
t/zere-insertion.
Before we consider Milsark's analysis, it must be pointed out that the formulation of iTiere-insertion given above as 1) is only a first approximation.
Milsark points out that in a
more highly constrained theory of transformations Chomsky
(see also
(forthcoming)) we might want to prohibit transformations
from performing complex or multiple operations as in the formulation of 1) which involves both movement of the phrase category NP and the insertion of the item there.
In line with such
a more highly restricted theory of transformations Milsark proposes replacing 1) with the two transformational operations 3) and 4) 3)
NP Downgrading SD:
SC:
X
NP
Y
be
Ζ
1
2
3
4
5
1
t
3
42 5
85 4)
Trace Removal SD:
X
t
Aux
SC:
2 •*• there
Y
NP Downgrading has the effect of postposing the subject NP, leaving behind a trace which acts as a bound variable with respect to the moved category NP; see Chomsky go (1974) .
(19 71; forthcoming); Fien-
Trace Removal is a transformational spelling rule
which has the effect of replacing the trace with
there.
However, whether we take the formulation in 1) or the one in 3), 4), it will have little effect on the discussion that follows.
So we will make use of Milsark's first approximation of
the rule of there-insertion
in 1) which is more similar in form
to standard there-insertion
rather than the formulation in 3)
and 4), although all of the remarks to follow apply equally well to the formulation 3) and 4) of t?z e re-insertion.
5.2.
The E Rule
Milsark argues that existential sentences
(sentences contain-
ing an occurrence of the existential there) have a particular interpretation in which existence is predicated of the class of objects denoted by the NP following the verb poses here). insertion
(be for our pur-
Thus 2) derives from 1) bv the application of
there-
(1) in 5.1.)
1)
Collicky cows were on the cobblestones.
2)
There were collicky cows on the cobblestones.
Sentence 2) is then analyzed by the E rule (2) in 5.1.) as 3)
t h e r e A u x [Past] be
N p (collickv
cows]
χ
[on the cobblestones]
Applying the E rule we get roughly the interpretation 4)
The class of collicky cows had at least one member c such that c was on the cobblestones.
The details of the application of the E rule are not critical here (see Part II of Milsark
(1974) for more discussion).
What
is important is that the rule is so formulated such that it applies only to derived structures which can be analyzed as 5)
there Aux (have-en) be NP X
86
Any derived existential structure which cannot be so interpreted is ruled out.
This fact will be important for the discussion
to follow. That it is in particular the presence of there which gives existential sentences their special interpretation can be seen by comparing 6)
Three men were believed to have been in the garden.
7)
There were believed to have been three men in the garden.
8)
There were three men believed to have been in the garden.
Milsark notes that although all of 6)-8) have identical deep structures, only in sentence 8) is it necessary for the three men to have existed in order for the sentence to be true.
He
shows that the E rule can account for the correct interpretation of 8).
The contrast between 6) and 8) shows quite clearly that
the presence or not of the item there is directly linked with the special existential interpretation observed in 8).
Thus
given that there exists a iftere-insertion rule, it seems highly plausible to assume that the existential interpretation present in existential sentences like 2) and 8) is not present in deep structure but is due to a semantic interpretation rule applying to derived structures containing occurrences of
5.3.
there.
The Leftmost he Condition
Let us now return to the restriction we observed in existential sentences in 4.2., which was discussed there as the "two be s argument".
It was observed that we could derive sentences
like 2), but not 3), from 1) 1)
Some houses are being built.
2)
There are some houses being built.
3)
*There are being some houses built.
Thus it must be guaranteed that the be referred to in the structural condition of tfcere-insertion be the first or "leftmost" be in the construction under consideration.
Milsark refers to
this condition as the "leftmost be condition".
As Milsark points
out, it would be most undesirable to have to say in the struc-
87 turai condition of the transformation of there-insertion
that
the leftmost be is to be picked, as this would be tantamount to allowing the use of quantifiers in the structural conditions of transformations, and hence would represent an increase in the expressive power of transformational theory, which we would like to avoid, if at all possible. Milsark points out that we could prevent the generation of 3) while allowing the generation of 2) if we complicate the statement of the structural condition for there-insertion Thus suppose we change the structural condition of
(1) in 5.1.). there-inser-
tion to 4)
X
NP
1
2
AUX
(have-en) be Y
3
4
5 6
Spelling out the material between NP and be guarantees that the first be will be picked out. ution.
Of course, this is an ad hoc sol-
It is just another way of saying that the first or left-
most be in the string must be chosen; or, in Burt's formulation, that the be "directly follows T(ense)"
(see 4.2. and footnote
6) .
Furthermore, Milsark wishes to reject such a solution on metatheoretical grounds. ization"
He assumes a principle of "minimal factor-
(see also Chomsky
(forthcoming)) which, in Milsark's
words, "would disallow the statement of transformations whose structural analyses mention strings of elements which are not directly affected by the operation of the rule, requiring the use of a variable in such cases".
Such a principle would rule
out, e.g., a rule preposing adverbs across transitive verb phrases (but not intransitive verb phrases).
Thus in this more restric-
ted theory of transformations one could not mention such factors as AUX and have-en
(factors 3 and 4 in the revised structural
condition of t&ere-insertion given as 4) above), for example. Rather one would have to reolace factors 3 and 4 by a variable as was the case in Milsark's original formulation of sertion (1) in 5.1.).
there-in-
But then, of course, both 2) and the un-
grammatical 3) above will be generated by this general formulation of tfcere-insertion. Milsark proposes that the E rule (2) in 5.1.) rule out the
88 ungrammatical 3). 5)
That is, in the string
There are being some houses built.
only the first be will satisfy the structural analysis of the E rule, i.e., the first be after there Aux. lowed by being, sis.
But this be is fol-
not NP, as is required by the structural analy-
Thus the string 5) is uninterpretable by the E rule and
hence ruled out.
Observe that, in effect, what Milsark has done
is to build into the structural analysis of the E rule the semantically irrelevant information that the coda NP to be interpreted and the the ve can be separated at most by an AUX followed by an optional occurrence of have-en.
This, of course, pre-
vents the string 5) from receiving a semantic interpretation, but this ad hoc device is precisely what Milsark wished to rule out for structural conditions of transformations on general grounds (the "minimal factorization" principle).
He instead re-intro-
duces the device into the semantic rule so that rather than saying that the string 3) is not generated syntactically because the be in question is not the first be after there AUX, he is saying that the string 3) is generated syntactically but the semantic rule doesn't give it an interpretation because the be is not the first be after there AUX.
It is not really the case
that 3) is being ruled out on the basis of some explanatory semantic principle, but rather an ad hoc device has been eliminated from the syntax to be reintroduced into the semantics. Although Milsark is willing to accept a minimal factorization principle to rule out such ad hoc devices in syntax, he argues that one might not want such a principle for semantic rules, which might involve continuous strings of elements rather than discontinuous terms as in the formulation of t/zer-e-insertion, NP Y be
(1) in 5.1.).
But even if this were the case,
Milsark gives us no reason for believing that the continuous string mentioned in the structural analysis of the E rule must contain there AUX
(have-en) be NP rather than there Aux Aspect
be NP (where Aspect = (Perf)(Prog) = (have-en) (be-ing)) or there Aux be NP, etc.
The latter formulation would be no more comp-
licated than the one Milsark chooses.
The only reason for the
particular formulation of the structural analysis in the E rule
89 is that it happens to work.
But no principled explanation,
semantic or otherwise, is forthcoming for why this particular formulation lacks the Prog (= be-ing) Since the phrase structure analysis can handle the facts under discussion in a principled way, we have an argument in favor of this analysis and against Milsark's transformational analysis.
We will see in Chapter 6 that the "leftmost be condition"
remains a problem in Emonds' formulation of there-insertion
as
it does in Milsark's. 5.4.
The Semi-modal Restriction
We recall that one of the problems for the standard analysis of there-insertion
was to generate 1) while excluding 2)
1)
There is going to be a demonstration in the gym.
2)
*There is a demonstration going to be in the gym.
(The facts are similar in the case of the semi-modal be to.) We discussed this problem under the heading of "the semi-modal argument"
(see 4.3.).
Milsark refers to the exclusion of sen-
tences like 2) as the "semi-modal restriction". analysis of there-insertion
In the standard
it was the case that the subject
noun phrase is generally postposed after the first be if there is more than one, as we saw in the last section.
Thus we get
4), but not 5) from 3) 3)
Some houses are being built.
4)
There are some houses being built.
5) »There are being some houses built. Thus it should be the case that 2) is grammatical, with the noun phrase a demonstration
postposed to the position after the first
be, while 1) should be ruled out since the noun phrase a demonstration is postposed to the position after the second be. ever, the facts are exactly the opposite. confronting standard
How-
This was the problem
there-insertion.
Milsark proposes that the items be going and be to be reanalyzed by "idiom reanalysis rules" as idiomatic units like take advantage
of (see also Chomsky (forthcoming)), which are
thereby insulated from certain syntactic operations such as the
90
insertion of major categories»
Thus in the
CSSG
of 2) the effect
of the idiom reanalysis rule is that it is not possible for the major category noun phrase, a demonstration, to be inserted between be and going since be going is treated now as a unit. To support this position, Milsark points out that be going and be to exhibit exceptional behavior with respect to a number of syntactic operations.
Thus, as we observed in 4.3., the be
of be going to and be to was not deleted in reduced relatives and reduced clefts.
Nor do they occur as complements to the
perception verbs examined in 2.3. 6)
*John saw Mary going to go home.
Finally, as Milsark observes, such be s cannot undergo Gapping 7)
"John was about to leave and Bill (going) to arrive.
He proposes to account for these facts by treating be going and be to as idiomatic units. Observe that if such a solution should turn out to be correct it can be incorporated into the phrase structure analysis that we are arguing for. of our hypothesis.
Thus it has no bearing on the correctness To see this, recall that the explanation that
we gave for the ungrammaticality of 2) was that there exists no reduced relative or reduced cleft structure of the form a demonstration going to be in the gym.
This was because the be
in be going to does not undergo either Relative Clause Reduction (see 8)) or Cleft Reduction (see 9)) 8)
*The man going to be here at 5:00 didn't show up. (= 27) in 4.3.)
9)
*That was John going to be here at 5:00. (= 25) in 4.3.)
Similar remarks apply to be to. All that is left then is to explain why this be does not undergo either Relative Clause Reduction or Cleft Reduction (or Gapping).
If Milsark's explanation that be going and be to are
syntactic units by the idiom reanalysis rule is correct, then this principle will apply in these cases in the framework of the phrase structure analysis.
Thus the data under consideration
provide no evidence for relaxing the most restrictive hypothe-
91 s i s — t h e null hypothesis, which we are defending for the existential construction. 5.5.
The Definiteness Restriction
In 3.4. we discussed the limitation of there-insertion cases where the subject NP is indefinite (-definite).
to
Thus we
have 1)
A man is in the garden.
( = 1 ) in 3.4.)
2)
There is a man in the garden.
3)
The man is in the garden
(= 2) in 3.4.)
4)
?There is the man in the garden
( = 3 ) in 3.4.) (= 4) in 3.4.)
where 4) is acceptable only with another interpretation the "list" reading by Milsark).
(called
Similar facts hold for other
so-called syntactically indefinite noun phrases 5)
»There was (John, John's mother, that man, him) in the garden.
Milsark argues, as we argued in Chapter 3, that, if possible, it would be desirable to exclude syntactic transformations which, for instance, apply when the determiner is -definite, but not when it is +definite.
This would require an extension of expres-
sive power of the theory of transformations, which we would like to avoid, if possible. Milsark proposes that we can avoid this unwanted extension, if we allow the rule of there-insertion
to apply unconstrained
both to definite and indefinite noun phrases and then to rule out the definite case (4) above) ing)
by the E rule.
(except with the "list" read-
To see how this is done, it is necessary
to examine Milsark's analysis of these definite/indefinite cases more closely. In fact, Milsark proposes that the notion of definite/indefinite be replaced by the notion quantificational/non-quantificational (cardinal).
He distinguishes three separate classes of
quantifier types, those that are unambiguously quantificational 6) Quantificational 1.
universal (the, each, all, every, both)
2.
most
3.
three of the, ten of the
92 those that are unambiguously cardinal 7)
Cardinal 1.
a
2.
three, ten3
fifty-two
and those that are ambiguous 8)
(quantificational/cardinal)
Quantificational/Cardinal 1. some (some, some of = quantificational; sm = cardinal) 2.
many (many, many of = quantificational; mny = cardinal)
3.
plural indefinites (e.g., uniaorns)
Thus the expressions in 6) serve as quantifiers over the classes denoted by their noun-phrases.
They set the proportion of
the class under consideration, all of NP, most of NP (many of NP, see 8) ). On the other hand, in 7) the cardinal quantifiers only specify the size of the class involved, one(= a), three,
fifty-two.
In 8) we see that some quantifiers can be ambiguous (in the appropriate contexts). 9)
Thus consider
Many men left.
Many men can have a quantificational interpretation; viz., that out of the class of men under consideration, many left (although perhaps many stayed).
In addition to this interpretation con-
trasting a subset of a set with its complement, we have the cardinal interpretation; viz., that an indefinite, although large, number of men'left.
Here no contrast is implied.
The first
(quantificational) interpretation can be brought out more clearly by stressing many or replacing it by many of.
Similar remarks
apply to some. Consider now the case of the plural indefinites in 8) as in 10)
Typhoons arise in this part of the Pacific.
Here the plural indefinite typhoons has two interpretations
(com-
pare the generic/specific discussion in 3.5. of sentences like aondors are in the Andes).
The first interpretation is that
all typhoons, typhoons in general, arise in the part of the Pacific under discussion.
This universal interpretation is then
93 the quantificational interpretation and falls together with all (alt(of the/the) typhoons). The other interpretation is that some typhoons, but not necessarily all, arise in the part of the Pacific under discussion. This interpretation is the cardinal interpretation and falls together with the some (= sm) interpretation. Observe that the is classed together with the universals in 6). This is motivated by a suggestion of Chomsky's (1975) that the definite determiner be regarded as universal quantification, rather than as naming a unique entity, as in traditional.characterizations of definite descriptions. As we will see directly below, Milsark shows this classification of the with the universals is supported by evidence dealing with the existential construction. It is further argued that what is ruled out in existential constructions are noun phrases with a quantificational interpretation, rather than simply those with a "definite" interpretation as in traditional descriptions of standard t&ere-insertion. To see this let us examine examples from 6), which contains expressions which are unambiguously quantificational. As we expect, existential sentences containing such phrases are ill-formed (apart from "list" interpretations) 11)
*There are
all men I in the garden, the men [ most of the men |
Milsark argues that such expressions as in 12) are ruled out since they involve species of universal quantification like the definite determiner the 12)
»There was
John John's mother that man him
in the garden.
We would furthermore expect that cases as in 7) involving unambiguously cardinal expressions should be acceptable in existential constructions. That this is the case is borne out by the following examples 13)
There is a woman in the garden.
14)
There are three women in the garden.
94 Finally, we would expect in the case of 8), where the expressions are ambiguous between a quantificational interpretation and a cardinal interpretation that only the cardinal interpretation would be permitted in the existential construction.
This ap-
pears to be true as well 15)
16)
Some girls were at the party. a)
= some of (quantificational interpretation)
b)
= sm (cardinal reading)
There were some girls at the party. a)
= *some of (quantificational interpretation)
b)
=
sm (cardinal reading)
Thus 16) has the cardinal reading; viz., that an indefinite quantity of girls were at the party, but not the quantificational reading; viz., that some, but not all, of the girls under consideration were at the party.
As Milsark points out, one
must take care not to confuse an additional reading present in the dialogue in 17) with the quantificational reading 17)
Were there many girls at the party? There were some girls at the party.
The answer in 17) does not mean some but not all of the class of girls under discussion were present even though some can be stressed in the case of the quantificational interpretation, as in 18)
Some girls were at the party. (But not all—others were at home.)
Finally, we would expect the indefinite plural expressions to have only the sm interpretation and to exclude the interpretation of universal quantification.
That this is true can be
seen by comparing 19)
Typhoons arise here. a) = all
(universal/quantificational) interpretation
b) = em (cardinal interpretation) 20)
There arise typhoons here. a) = *all b) = sm
95 In 20) the quantificational interpretation is excluded. A similar explanation can be provided for the difference of interpretation of minors must be in the dorm by midnight and there must be minors in the dorm by midnight
(12) and 13)
in 3.5.). Milsark notes a further fact (pointed out by Horn, 19 72) which fits into the framework sketched above; viz., that of the two senses of any, the universal sense and the polarity sense to some, only the latter may occur in the existential constructions 21)
There isn't anything here.
22)
*There is anything John would do for you.
23)
Is there anything John would do for you?
Let us see now how Milsark proposes to accommodate these facts into the tTzere-insertion framework.
As we have seen, more than
only "definite" noun phrases are excluded in the existential construction.
What are excluded are what Milsark has denoted
"quantificational" noun phrases, a term which subsumes universally quantified noun phrases, hence definite noun phrases.
But rath-
er than put a restriction on the structural condition of the tfcere-insertion rule to the effect that the subject NP must be "-quantified", which would represent an unwanted extension of expressive power in the theory of transformations, Milsark proposes allowing there-insertion to generate freely and have the E rule rule out the unwanted cases. Note first that Milsark proposes an addendum to the E rule, which is italicized below 24)
E rule (revised) there AUX (have-en) be Q NP X
is interpreted:
the class C denoted by NP has at least one member c such that P(c) is true, where Ρ is a predicate and Ρ is the reading of X (and the set of suah members a is of cardinality Q). (Note also the Q introduced into the structural analysis of the rule).
The addendum is proposed to handle cases where the NP
denotes a class of cardinality greater than one, as in 25) which has the reading 26)
96 25)
There were two people in the room.
26)
The class people has at least one member ρ such that ρ was in the room, and the cardinality of the set of such members ρ is two.
Milsark argues that if the NP in the structural analysis of the E rule has a quantificational determiner, then the sentence will receive no interpretation, since the addendum in effect requires that a quantificational determiner be interpreted as an expression of cardinality, which is impossible. Thus 27) would receive the impossible interpretation 28) 27)
There were (the, all) people who hate Chopin on the boat from Poland.
2 8)
The class of people who hate Chopin had at least one member c-h such that c-h was on the boat from Poland, and the cardinality of the set of such members c-h is universal.
The last clause "the cardinality of the set of such members c-^h is universal" is meaningless; hence the sentence receives no interpretation. This would then account for the exclusion of the quantificational interpretation in all the cases discussed above. We have discussed this class of cases in some detail since we feel that it gives a more adequate account of the "definiteness restriction" than has been given in standard tftere-insertion accounts. What is important from our point of view is that if this account is correct, it can with little modification be incorporated into the phrase structure analysis of the existential there that we are proposing. The only necessary modification is a generalization of the E rule; the E rule may have the structural analysis there Aux be Q NP X (where Aux = Τ(M)(Aspect)) rather than the ad hoc formulation in 24) which we saw was necessary in Milsark's account of the "leftmost be condition" in 5.3. In the phrase structure analysis, the E rule can be thought of as applying either at the level of deep or surface structure. If we take the assumption of the null hypothesis that all semantic interpretation (apart from the matter of thematic relations) is determined at the level of surface structure (see Chomsky (forthcoming)), then we may assume that the E rule applies at the level of surface structure.
97 5.6.
The Predicate Restriction
In 4.4. we discussed such sentences as 1)
*There are some graduate students union members.
We argued that in the standard tftere-insertion analysis, there is no way to prevent the generation of 1) from 2)
Some graduate students are union members.
unless an ad hoc restriction is added to the structural condition of tftere-insertion prohibiting predicate nominal phrases from occurring after the copula.
We called this "the predicate
nominal argument". As we showed, such sentences as 1) are ruled out in the phrase structure analysis since the phrase some graduate
stud-
ents union members can arise neither by Relative Clause Reduction nor by Cleft Reduction as one can see by examining the following two sentences (7) and 9) in 4.4.) 3)
*That was Sam and Sue twins.
4)
»People twins usually look alike.
(Cleft Reduction) (Relative Clause
Reduction) As we observe in 3) and 4), the be in predicate nominal constructions cannot be deleted either in reduced relatives or in reduced clefts. As Milsark points out, the restriction required is still more general. 5)
We must also rule out such constructions as *There were some
sm) people tall.
Unless some additional restriction is placed on
there-insertion,
5) will arise from 6)
Some (^ sm) people.
We could revise the ad hoc restriction on the structural condition of tTzere-insertion referred to above so that predicate adjective phrases along with predicate nomináis are prohibited from occurring after the copula.
But this formulation is too rest-
rictive, as examples like 7) discussed by Milsark show 7)
There were some people sick. There were some people hungry.
98 Cases like 7) seem to me slightly odd or stilted (Emonds (1970) does "not always find these sentences completely acceptable"), but they are certainly much better than the completely ungrammatical cases like 5) . The class of adjectives that are acceptable to Milsark in existential sentences of the form 7) are the following 8)
sick hungry tired alert clothed naked
drunk stoned closed open etc.
The classes of adjectives that are unacceptable to Milsark in existential sentences of the form under discussion (see 5)) are 9)
(shapes) (colors) intelligent beautiful
boring crazy etc.
As Milsark points out, the class in 8) can be roughly characterized as adjectives which describe temporary states whereas the class in 9) are adjectives which describe more permanent properties, where, as he points out, the characterization "temporary state" and "permanent property" are variable and depend on such factors as one's knowledge about the real world, etc. Thus we have a characterization which roughly corresponds to the distinction in traditional grammar between essential and accidental properties. Thus we see that the rule of tTzere-insertion must be complicated still further if one is to account for the distinction in 5) and 7).
Thus the rule must be blocked not for all adjective
phrases but only for those that describe properties.
But if we
allow the transformation of there-insertion to refer to such information in its structural condition, prohibiting the rule from moving noun phrases if the predicate contains an adjective describing a property, we are again increasing the expressive power of transformations, much as we would be if we allowed the tfcere-insertion transformation to refer to such information as to the definiteness or nondefiniteness of the noun phrase to be moved (see 5.5.).
And, of course, these extensions of the expres-
99 sive p o w e r of the theory
are e x a c t l y w h a t w e w a n t t o a v o i d ,
l e s s t h e f a c t s f o r c e u s to d o o t h e r w i s e .
Furthermore,
un-
note
that
e v e n if w e p e r m i t t e d t h i s e x t e n s i o n in e x p r e s s i v e p o w e r , w e
would
b e g a i n i n g n o t h i n g in t h e w a y o f e x p l a n a t i o n o f t h e f a c t s .
There
is n o t h i n g
in w h a t h a s b e e n s a i d u p t o n o w t h a t e x p l a i n s
tftere-insertion a p p l i e s
if t h e a d j e c t i v e s d e s c r i b e
n o t if t h e y d e s c r i b e p r o p e r t i e s ,
states,
but
rather than conversely.
is, t h e r u l e w o u l d b e n o m o r e c o m p l e x , for adjectives describing
why That
if w e c a u s e d i t t o b l o c k
s t a t e s l i k e sick,
happy,
etc.
B e f o r e w e t u r n t o the e x p l a n a t i o n o f t h e s e f a c t s p r o p o s e d Milsark,
l e t us n o t e o n e o t h e r f a c t ; v i z . , t h a t p r e d i c a t e
ináis denoting states
(along w i t h o n e s d e n o t i n g p r o p e r t i e s )
excluded from the existential construction. phrases
l i k e drunk
sentences
Thus consider
are noun
w h i c h might be said to describe a state
in
like
10) Observe
by
nom-
A drunk walked down the
t h a t e v e n t h o u g h drunk
adjective 11)
is s t a t e - d e s c r i p t i v e ,
There were some people
Milsark
*There w e r e too many people
Too many people were correlates
icate position.
14)
like
drunks,
from drunks.
this unexpected gap w i t h the fact
noun phrases which can denote states (alcoholic)
T h u s in 14) drunk
has only a property
rather than a state-descriptive John was a drunk at Bill's
like
tall,
etc.
that
l o s e t h i s r e a d i n g in
last
reading party.
predicate
(in 9 ) ) .
The explanation w h i c h Milsark offers for the fact that descriptive
adjectives
pred-
reading
(drunkard)
T h u s all p r e d i c a t e n o m i n á i s b e h a v e like the class of adjectives
the
drunk. sentences
w h i c h o n e m i g h t e x p e c t to d e r i v e 13)
like
drunk
we do not have it in existential 12)
street.
l i k e sick
state-
c a n a p p e a r in e x i s t e n t i a l
structions, while property-descriptive
adjectives
l i k e tall
not, can b e s e e n f r o m an e x a m i n a t i o n of the f o l l o w i n g
data
concan-
100 15) 16) 17) 18)
Sm people were sick. *Sm people were tall. Some people were sick. Some people were tall.
From 15) and 17) we see that state-descriptive adjectives like a-iok can be predicated both of noun phrases with a quantificational interpretation (17)) and noun phrases with a non-quantificational (cardinal) interpretation (15)) .
However, property
adjectives like tall can be predicated only of noun phrases with a quantificational interpretation as in 18), but not with a nonquantificational interpretation as in 16).
But if tall can be
predicated only of noun phrases with a quantificational interpretation and if the E rule rules out existential constructions whose NP satisfies its structural analysis if that NP receives a quantificational interpretation, then it follows that tall and adjectives of its class will be excluded by the E rule from occurring in existential constructions.
However, since adjectives like
siok put no restriction on the noun phrase it is predicated of as to whether it is quantificationally or non-quantificationally interpreted, and since the E rule rules out structures only where the NP has a quantificational interpretation, it follows that siok (and adjectives of its class) will occur in existential constructions, but only with the non-quantificational interpretation. That this prediction is borne out is seen by comparing the following with 15)-18) 19) 20)
There were sm people sick. * There were sm people tall.
21) 22)
*There were some people sick. «There were some people tall.
In 19) we see that siok occurs in the existential construction with sm people which has the non-quantificational reading.
In
21), on the other hand, siok is excluded since some people has a quantificational reading, which is excluded by the E rule. Tall is excluded in 22) for the same reason as s-iak in 21); viz., because of the occurrence of a noun phrase with a quantificational reading, again excluded by the E rule.
20) is excluded not by
101 the E rule, but by the fact observed above (see 16)) that tall (and all property adjectives) exclude noun phrases with a nonquantificational
(cardinal) reading.
Observe that an explanation has been provided for the fact that it is property adjectives and not state-descriptive adjectives that are excluded in the existential construction.
Viz.,
property adjectives can be predicated only of noun phrases with a quantificational r e a d i n g — b u t such noun phrases are excluded by an independently needed principle, the E rule.
Hence the
"predicate restriction" is not an additional ad hoc restriction on the rule of t/zere-insertion independent of the "definiteness restriction", but rather the two are interconnected in the way just outlined.
This can truly be said to be a case of deeper
explanation of the "predicate restriction" on semantic grounds, in contrast to the ad hoc move discussed above to "explain" the "left-most be condition". Moreover, what Milsark is proposing is that just as the "definiteness restriction" can be dropped from the structural condition of there-insertion
(with its function taken over by the E rule),
we can similarly drop the "predicate restriction" from the structural condition of there-insertion
(with its function taken over
by a conjunction of the E rule and the independently needed principle excluding non-quantificational NP from being predicated by property adjectives).
Thus having accounted for the "pred-
icate restriction" on semantic grounds, the rule of
there-inser-
tion is left in a simple form without restriction and allowed to apply freely.
It is seen to be unnecessary to increase the
expressive power of transformational theory by allowing new sorts of conditions to be stated in the structural conditions of transformations . Although Milsark has succeeded in showing how the predicate restriction can be accounted for without complicating the statement of the structural condition of there-insertion
in undesir-
able ways, he claims that the facts connected with the "predicate restriction" actually provide evidence in favor of the transformational hypothesis and against the phrase structure analysis that we are arguing for.
More exactly, he tries to show
that the facts argue against deriving existential constructions
102 from either reduced relatives or reduced clefts. This qualification is necessary since Milsark considers the hypothesis that existentials derive from reduced relatives ("the phrase structure hypothesis") to be separate from the hypothesis that existentials derive from clefts ("the cleft reduction hypothesis") whereas in the phrase structure analysis we are claiming that both reduced relative and reduced clefts are possible sources. But this distinction has little effect on the issue at hand since Milsark gives examples to argue that in both the case of reduced relatives and in the case of reduced clefts there is no syntactic source for such constructions as 23)
*There were many people
siak.
It is argued that the italicized string in 23) some people siak cannot derive as a reduced relative since there are no independently attested reduced relatives of this form occurring in nonexistential constructions. So, for example, we do not get 24)
*Many people sick are in the hospital.
Similarly, Milsark argues that cleft sentences apparently do not "allow any adjectival or AP predicates at all among the range of structures occurring after the postverbal NP". He gives as examples 25)
* It was John sick.
26)
*It was John tall.
From these facts Milsark concludes that neither a hypothesis that derives existential constructions from reduced relatives nor one that derives them from reduced clefts has a plausible source for existentials containing state-descriptive adjectives. The same criticism holds then for our phrase structure analysis which derives existentials from both sources. Note, however, that there is contrary to Milsark's claim, a range of cleft structures which allow the state-descriptive adjectives to occur. Consider the following (said in some appropriate context—say, showing snapshots) 27) 28)
Here's Brigitte Bardot with her clothes on. And that's Brigitte Bardot naked.
103 29)
That's my father
30)
There's my father
sober.
31)
H e r e ' s t h e f r o n t d o o r of H a r v a r d T r u s t o p e n , (said by b a n k r o b b e r s v i e w i n g s n a p s h o t s for a p l a n n e d robbery)
32)
A n d that's the door
drunk.
closed.
N o t i c e , f u r t h e r m o r e , n o t o n l y do s u c h s t a t e - d e s c r i p t i v e
adjec-
t i v e s o c c u r in v a r i o u s c l e f t s t r u c t u r e s , b u t the p r o p e r t y
adjec-
t i v e s a r e e x c l u d e d from s u c h s t r u c t u r e s , g i v i n g us e x a c t l y
the
distinction that w e w a n t 33)
»Here's Brigitte Bardot
tall.
34)
»That's Brigitte Bardot
intelligent.
35)
»That's Brigitte Bardot
(beautiful,
boring).
W e m i g h t go o n to ask w h y t h e c l e f t s c o n s i d e r e d b y M i l s a r k n o t a l l o w t h e s t a t e d e s c r i p t i v e a d j e c t i v e s as in 25) 36)
»It w a s J o h n s i c k .
which would derive 37) by C l e f t
do
above
(= 25))
from
It was John
(who was)
sick.
Reduction.
H o w e v e r , as w a s o b s e r v e d in C h a p t e r 2 (footnote 3) , the
it-
c l e f t s are for s o m e u n k n o w n r e a s o n s o m e w h a t r e s i s t a n t to the r u l e of C l e f t R e d u c t i o n . 39) w i t h t h e it-clefts
T h u s c o m p a r e that-clefts
in 38)
and
in 40) a n d 41)
38)
T h a t ' s J o h n S m i t h w h o is in the g a r d e n .
39)
T h a t ' s J o h n S m i t h in t h e
40)
It's J o h n S m i t h w h o is in the g a r d e n / i n the
41)
? * I t ' s J o h n S m i t h in t h e g a r d e n / i n t h e
garden. army.
T h u s it m a y b e d u e to p e c u l i a r i t i e s of t h e i t - c l e f t w h i c h led M i l s a r k t o a s s u m e t h a t s t a t e - d e s c r i p t i v e do n o t o c c u r in t h e c l e f t s t r u c t u r e s .
army.
construction adjectives
Thus the objection
existential constructions containing such
that
state-descriptives
have no syntactic source under the "cleft reduction
hypothesis"
(and m o r e g e n e r a l l y u n d e r t h e p h r a s e s t r u c t u r e a n a l y s i s )
is r e -
moved . N o t i c e , f u r t h e r m o r e , t h a t the e x p l a n a t i o n t h a t M i l s a r k
gives
104 for the "predicate restriction" carries over to the phrase structure analysis as well.
That is, the fact that state-descriptive
adjectives like siak, but not property adjectives like tall occur in existential constructions
(reduced clefts) follows from
the restriction on property adjectives that they can only be predicated of NPs with a quantificational reading plus the prohibition of the NPs with quantificational reading in existential constructions. Thus assuming that Milsark's analysis is correct, it can be incorporated easily into the phrase structure analysis.
The
explanation thus does not provide evidence against the phrase structure analysis and for Milsark's transformational analysis, as he claims. 5.7.
Extraction Arguments
Milsark presents a separate argument in favor of the transformational analysis which he argues can't be handled in any version of the phrase structure analysis based on the extraction facts discussed in 2.3. 1)
The central cases are examples like
I know many people lem.
2)
(who are) interested in this prob-
*A problem which I know many people
(who are) interes-
ted in 3)
There are many people interested in this problem.
4)
A problem which there are many people interested in
The point is that, in general, ω/ζ-movement out of relative clauses is prohibited in both unreduced and reduced relative clauses in English
(see 1) and 2)).
That is, wft-movement is subject to
the Subjacency Condition; see Chomsky
(1971).
However, in exis-
tential sentences we get extractions in superficially parallel cases in the case of reduced relatives
(see 3) and 4)).
Milsark
argues that this is unaccounted for in the "reduced relative analysis" since the phrase many people
interested, in this
prob-
lem is a reduced relative noun phrase in that analysis and extraction of this problem
should be prohibited by the Subjacency
Condition.
In the transformational analysis, on the other hand,
many people
interested
in this problem
is not a single constit-
uent but rather the noun phrase many people
is a separate cons-
105 tituent from interested in the problem, the two strings having been juxtaposed by tfeere-insertion.
Thus under this analysis,
extraction is permitted. Similar comments apply for the "cleft reduction analysis". If existential constructions are derived as reduced clefts, then one would expect extraction to be possible from reduced clefts which is apparently not the case; compare 6) and 8) with 7) 5) 6) 7) 8)
That's a house situated on that hill. *the hill which that's a house situated on a hill which there's a house situated on *a problem that it's John interested in
The same arguments then hold against the phrase structure analysis which derives existentials both as reduced relatives and as reduced clefts. It should be pointed out that extraction from existential sentences is not a marginal phenomena, but is quite general. This may not appear to be the case at first glance since quite often apparent restrictions on extraction out of existential constructions are actually due to independent factors, as Milsark points out. 9)
Compare these examples
*What was there a cat on?
10) *Who was there a man shot by? 11) *What was a cat on? 12) *Who was a man shot by? As can be seen by comparing 9), 10) with 11), 12), the deviance of the first pair has nothing to do with the fact that they are existential constructions, since the same restrictions hold for the parallel cases in 11) and 12).
Thus, abstracting away from
such cases, extraction out of existential constructions; i.e., the apparent violation of the Subjacency Condition, is quite general. Thus within the phrase structure analysis the problem reduces to explaining why extraction out of relatives is possible in the case under consideration.
One possible explanation is
that an idiom rule is involved (see Chomsky (forthcoming) for further discussion).
That is, that a rule is involved which in
106 effect treats there is NP as a semantic unit.
Thus since there
are many people is reanalyzed as a semantic unit, extraction would be allowed from the remaining string 13)
There are many people interested in this problem.
That is, this problem may be extracted from interested in this problem to yield 14)
a problem which there are many people interested in (= 4))
In fact, just such a solution is proposed by Milsark himself for another case.
Thus he wants to explain why one can prepose
the adverbial in this house in 15) 15)
In this house, there resides a dirty old man.
The phrase in this house should not be extractable since it is inside the verb phrase containing reside and since normally adverbials inside of and which subcategorize verbs are not preposable.
That in this house subcategorizes reside can be seen in 16)
*A dirty old man resides.
That it is not normally preposable is seen in 17)
*In this house, a dirty old man resides.
Milsark suggests that a solution might be to assume the "existence of a class of rules similar to the E rule, whose effect is to reanalyze a given string of formatives as a semantic unit, assigning to it an interpretation as an "operator".
The specific
formulation he attributes to Erteschik (1973) and he goes on to say that such rules delete major constituent boundaries in the strings to which they apply to allow otherwise ungrammatical extractions .
Thus in this case the E rule could be thought of
as erasing the VP boundary, permitting the otherwise unallowed extraction of in this house in 15). However, if this idea is correct, then a similar device would erase major constituent boundaries in 13), permitting the extraction of this problem, as in 14).
Recall, furthermore, that Mil-
sark makes use of a similar "idiom reanalysis rule" to account for the fact that the semi-modals be going and be to are impervious to insertion of major category items (see 5.4.).
Thus the sol-
107 ution to the extraction problem we are advocating is in fact quite compatible with suggestions put forth by Milsark himself to handle independent cases involving the existential construction. Further support for our suggestion is derived from considering facts from Erteschik note 11).
(1973) discussed in Chapter 2 (see foot-
Thus recall that in the comparable existential con-
struction in Danish, we get extraction even from unreduced relatives . 18)
Thus compare Der er en mand som har k^bt det hus. "There is a man who has bought that house". Det hus er der en mand som har k0bt. (that house is there a man who has bought)
Here, in 19), there is quite clearly a violation of the Subjacency Condition.
It is clearly the case that en mand som har
krfbt det hus is a constituent; it is not possible to argue, as is possible for English, that this string arises from juxtaposition by there-i.nserti.oTi..
And since extraction transformations
obey the Subjacency Condition except in a few other special and well-defined contexts, it seems to be quite reasonable to assume again here that a special idiom rule is applying, which has the effect of allowing us to ignore the major constituent boundaries (that define the Subjacency Condition) in order to perform extraction. 5.8.
Other Arguments
Milsark gives a few additional arguments in a section entitled "other arguments concerning passives and progressives" which he claims are beyond the scope of the "phrase structure hypothesis" and the "cleft reduction hypothesis" (hence beyond the range of our phrase structure analysis. The main arguments fall into two categories:
a) cases where
tense, aspect, and adverbials in the matrix are interpreted with embedded clauses and b) cases where some existential sentences have ambiguous structures Examples of a) are 1)
There's (just) been a frog discovered in the toilet.
108 2)
While you watch, there will bé a live pig roasted.
where it is argued that in 1) the punctual aspect (has just) is interpreted with the subordinate phrase containing
discovered
rather than with the matrix verb and similarly in 2) the adverbial while you watch is interpreted with the subordinate roasted rather than with the matrix verb. Observe that a possible solution for such cases is along the lines of that given in Section 5.7., viz., the idiom rule discussed there.
Thus, just as it is possible to extract freely from
embedded phrases out of contexts in cases otherwise not allowed, it should be possible to interpret tense, aspect, and adverbials down into embedded clauses in the same cases.
In fact, a case
discussed by Erteschik (1973), who Milsark quoted in this context, suggests just this solution.
She notes the following con-
trasts, pointed out by Kiparsky 3)
Oddly enough, here is the man who likes Peter. φ Here is the man who, oddly enough, likes Peter.
4)
Oddly enough, there are people who like Peter. = There are people who, oddly enough, like Peter.
and points out that it is precisely in the case of the existential there that the matrix adverbial can qualify the embedded clause. She proposes that the same notion (based on the concept of "semantic dominance" in her theory) which accounts for otherwise unallowed syntactic extractions might also account for these cases of otherwise unallowed adverbial interpretation. Notice that other examples discussed elsewhere by Milsark might be handled in a similar way.
That is, Milsark points out that
in 5) and 6) a man shot and people dancing have no sources as reduced relatives 5)
There was a man shot.
6)
There were people dancing.
That is, there is no 7) 8)
*A man shot walked into the room. «People dancing arrived.
He argues that under the there-insertion
analysis, 5) and 6)
109 have the sources 9)
A man was shot.
10) People were dancing. One possible approach under the phrase structure analysis is to assume as above that the matrix tense is interpreted with the subordinate participles by virtue of the idiom rule.
However,
in the case of 6), at least, another possible syntactic source offers itself—a reduced cleft.
Consider the following cleft
structures 11)
That's a bear dancing.
12)
Those are bears dancing.
where a bear dancing, bears dancing do not otherwise occur in isolation 13)
*A bear dancing arrived. •Bears dancing arrived.
An example of b), the case where some existential constructions are syntactically ambiguous is 14)
There was a man studying Gothic at the party.
which is ambiguous between a description and an action interpretation.
Only in the action case is extraction possible
15)
What were there people studying at the party?
Milsark argues that the "phrase structure hypothesis" can't account for these facts.
But recall that we already have a way
to account for the facts of extraction; viz., by means of an idiom rule (see 5.7.).
Similarly, the ambiguity Milsark notes
could also be ascribed to the effect of the E rule or perhaps ascribed to the alternative syntactic structures (the reduced relative and the reduced cleft structure) available for 14) under the phrase structure analysis.
These two alternatives de-
serve further investigation, but it seems reasonable to assume that the solution lies along these lines.
5.9.
Stress in Existentials In 2.2. we discussed the correlation between stress in exis-
tentials and that in cleft sentences and suggested that similar
110 mechanisms of stress assignment might be responsible for the correlation.
Milsark argues that the stress facts are not as
parallel as we have claimed and thus that the argument is undermined on factual grounds.
The facts that he considers are given
below 1)
It was John in the room.
2)
There was a man in the room.
3)
There was a car on the sidewalk.
4)
It was a car on the sidewalk.
5)
It was John run over by a bread wagon.
6)
There was a man run over by a bread wagon.
7)
It was John talking on the phone.
8)
There was a man talking on the phone.
However, as we have noted on several occasions, in the phrase structure analysis, the existential structures can be systematically structurally ambiguous between reduced cleft and reduced relative structures.
Hence we would expect systematic am-
biguities as to stress contour.
Thus take 7) and 8).
In 7),
a cleft structure, the stress is on John, while in 8), a superficially parallel existential construction, the stress is not on a man, but on the phone.
This non-parallelism,
Milsark claims,
vitiates the claim that existentials derive from cleft structures. However, recall that under the phrase structure analysis, existential constructions may have reduced cleft and reduced relative sources.
Thus a man talking on the phone may be either
a NP reduced relative structure or an NP-S reduced cleft structure.
In the former case, we would expect to find stress placed
by the nuclear stress rule as in a man talking on the phone passed out and as in 8).
With the NP-S structure we expect stress
on the NP by whatever rules provide stress in clefts.
And this
we also find in 9)
There was a gorilla talking on the phone.
The facts are the same with a man in place of a gorilla, but a man may be destressed in this context when read as a semantically empty element like people, somebody, etc., as in there were people talking on the phone.
Ill Of course, in 7) stress falls only on John, since John talking on the phone is not a reduced relative structure. Closer examination of the examples l)-8) reveals similar systematic ambiguities in the possibility of stress placement. Of course, preferred stress in any given case may depend on a varItety of factors, including semantic factors (see Bresnan (1971) on topical stress).
But once we abstract from these factors and
we keep in mind that the existential structures are systematically ambiguous, it seems quite reasonable to assume the stress factors can be accounted for within the phrase structure analysis.
6.
THE STRUCTURE-PRESERVING ANALYSIS OF TSETSE-INSERTION
6.1. Introduction In two important studies of there, Emonds (1969; 1970) noticed the defects implicit in the standard accounts of thereinsertion.
He proposed remedying the standard transformational
account by some very general assumptions about constraints on transformations in syntactic theory; viz., the "structure-preserving" hypothesis.
Since Emonds was able to support the plaus-
ibility of this hypothesis with a wide range of phenomena taken from English and other languages, it is necessary to consider the evidence for a structure-preserving rule of
there-insertion.
We can illustrate the notion of "structure-preserving" with the passive rule which relates sentences like 1) and 2) 1)
Russia defeated Germany.
2)
Germany was defeated by Russia.
It has been argued in transformational literature (see Chomsky (1970a) that 2) is derived from 1) by two operations, an operation of Agent Postposing which postposes the underlying subject noun phrase Russia into the verb phrase and an operation of NP Preposing which préposés the underlying object noun phrase Germany into the vacated subject position. Emonds hypothesizes that transformations fall into roughly two classes—there are "root" transformations which apply only in the root or topmost sentence like Subject-Aux Inversion 3)
Will John come?
4)
* I don't know if will John come.
And there are "structure-preserving" transformations which may operate in embedded clauses (as well as in root sentences), like the operations of Agent Postposing and NP Preposing just discussed
113 5)
Germany was defeated by Russia.
6)
I don't know if Germany was defeated by Russia.
(=2)
Emonds further hypothesizes that all such transformations (like Agent Postposing and NP Preposing) have what he calls the "structure-preserving" property (except for a structurally definable class of so-called "minor movement" rules, which need not concern us here). The property of being structure-preserving (restricting the discussion now to movement transformations), means that a constituent can only be moved to a position which is provided by independently motivated phrase structure rules. To quote Emonds (1970): "This notion of moving a constituent labeled X into a position where a node X is already provided for by the phrase structure rules is the central idea in the definition of a structure-preserving transformation." Thus in the case of Agent Postposing, this rule moves the subject NP into the NP position which is provided by the independently motivated rule PP P-NP, where PP is an expansion of VP. On the other hand, NP Preposing moves an object NP into the NP position provided for by the independently motivated phrase structure rule S NP-TENSE-(M)-VP. Emonds formalizes the notion of structure-preserving by allowing the phrase structure rules to generate empty nodes; i.e., nodes like NP, which do not dominate any terminal elements. He requires only that at some point in the transformational derivation, the empty nodes be filled by terminal elements. Then he constrains movement transformations (non-root ones) to substituting a constituent labeled X (i.e., for example, an NP for an NP), where the second X is empty. Let us illustrate the derivation of Germany was defeated by
Russia (=2) which has the deep structure 7) NP
TENSE
Russia
ed
VP V
NP
PP Ρ
defeat
Germany
by
NP
114
We have generated an empty NP node in the PP in the verb phrase; i.e., pptplbylp
Np[
]
NP ] PP·
The
°Peration
of
Agent Postposing
must now apply, moving the subject NP Russia into the position of the empty NP
The operation of NP Preposing now moves the object NP Germany into the vacated subject position
The vacated object NP node fulfills the requirement that all empty nodes must be filled at some point in the derivation; in this case it was filled in deep structure. The empirical content of the structure-preserving hypothesis is that an explanation is supplied for the fact that in all cases such movement transformations (as Agent Postposing and NP Preposing above) applying in embedded clauses only move constituents to positions provided already on independent grounds by the phrase structure rules and not into positions not so provided, as, for example, into the auxiliary or in front of the verb.
115 We now turn to the three arguments that Emonds advances in favor of his structure-preserving analysis of tfcere-insertion. 6.2.
The Optional/Obligatory Argument
As a preliminary to the first argument, let us briefly sum up Emonds' treatment of the copula be as in sentences like 1)
John is a farmer.
2)
John is very skillful.
3)
John is in Denver.
4)
John is the mayor in Denver.
5)
John is happy in Denver.
Emonds terms NPs like a farmer· and APs like
very skillful
as
"predicate attributes" and gives a number of semantic and syntactic arguments for wanting to distinguish them structurally from NPs and APs in such constructions as 6)
John saw a farmer.
7)
John works skillfully.
He makes this distinction by marking such phrases as a farmer in 1) and very skillful
in 2) with the feature +PRED.
more, in contrast to Chomsky
Further-
(1965), Emonds enters the copula
be into the lexicon as a verb like eat and subcategorizes it such that it can take a predicate attribute as in 1) and 2), or a PP as in 3), or both as in 4), 5).
Actually, the system
of subcategorization just described adheres to Emonds' 1969 discussion of tAere-insertion; we will discuss his later version of subcategorization in 6.4. below. The first argument Emonds gives against the standard formulation of tftere-insertion concerns the generation of sentences like 8)
There is a God.
9)
There was a fire.
If we are to generate sentences 8), 9) by standard
there-inser-
tion, then it is necessary to postulate the underlying forms 10) and 11) 10)
*A God is.
116 11)
*A fire was.
which are then converted into the surface forms 8) and 9) by making tftere-insertion obligatory for these cases.
Normally, though,
be has a deep structure complement and usually tfrere-insertion is optional; compare 12), 13) 12)
A man is in the garden.
13)
There is a man in the garden.
Furthermore, we must restrict the underlying abstract intransitive be structures to having indefinite subjects to avoid generating 15) from 14) 14)
The God is.
15) *There is the God. However, this last restriction is somewhat questionable since, as we pointed out in 3.4., there seem to be cases with a different semantic interpretation which require definites 16)
We have the angel, the devil...and then there's the God.
However, observe that in Emonds' structure-preserving framework, the correct set of sentences is generated without any special statements. That is, underlying 8), 9) we have (more precisely, +PRED) NP
where
17)
AGodis
18)
A fire was
Np
[
N p
[ Np
]Np [
)Np
] N p is an empty NP node generated in deep structure
which satisfies the subcategorization condition on the verb be; viz., that be co-occurs with a predicate attribute NP, empty in this case.
Now i^ere-insertion will apply to 17) and 18),
moving the NPs a God., a five onto the empty NP node (and thus satisfying the structure-preserving constraint) to yield 8) and 9).
Theve-insertion
must
apply in these cases or else 17) and
18) will be ruled out by the convention mentioned above that at some stage of the derivation, either in deep structure or after certain transformations have applied, an empty node must be fil-
117 led by a terminal element.
Thus in the case of 15) , which has
the underlying structure 19)
The God is
Np[
]Np
t^ere-insertion will not apply, since the subject is not "definite and the empty node will not be filled and will hence be ruled out by the convention above.
Note also that in 17), 18)
above, the vacated subject node provides an NP for the insertion of there, thus avoiding the problem of derived constituent structure of there discussed in 3.3.
6.3.
The Semi-modal Argument
The second argument Emonds presents in favor of his structurepreserving formulation of ifeere-insertion involves the semi-modals be going
to and be to.
As he points out, tÄere-insertion
normally postposes the subject NP after the first be, if there are two be s in the string
(see also 4.2.)
1)
Some new houses are being built.
2)
There are some new houses being built.
3) *There are being some new houses built. 4)
Only a few students were being obnoxious.
5)
There were only a few students being obnoxious.
6) »There were being only a few students obnoxious. Emonds notes that there are two exceptions to this generalization, involving be going
to and be to
7)
A demonstration is going to be in the gym.
8)
There is going to be a demonstration in the gym.
9) *There is a demonstration going to be in the gym. 10)
Three senators are to be at the conference.
11)
There are to be three senators at the conference.
12) *There are three senators to be at the conference. In 9) and 12) it is impermissible to place the subject NP after the second be in contradiction to the generalization just stated.
Thus a special ad hoc restriction must be placed on
insertion for the case of be going
there-
to and be to (see also 4.3.).
However, in Emonds' structure-preserving framework the prob-
118 lem does not arise. treating be going
He points out that there is motivation for
to and be to as modals.
Thus neither occurs
in infinitives or in gerunds 13)
*John seems to be going to come.
14)
*John is likely to be to be home by six.
15)
*John's being going to come surprised us.
They do not occur after other modals 16)
*John can be going to come.
17)
*John will be to come by seven.
But these properties are shared with the modals will, mustj
can,
etc. 18)
*John seems to will come.
19)
*John will can come.
Thus, argues Emonds, we may generate be going the position of modal (S
NP-TENSE-M-VP).
to and be to in
But since there is
no NP generated in the position of M(odal), there will be no empty NP node on which the subject NP may be postposed by insertion.
Thus 9) and 12) are ruled out.
there-
In the next section
we will see that in the case of 8) and 11) with the sequence NP-PP, there is an empty NP node provided by the phrase structure rules and hence both sentences are acceptable.
6.4.
The Predicate Restriction
A third argument against standard t^ere-insertion and for the structure-preserving hypothesis is provided by examples like 1) 2) 3) 4)
Some graduate students are union members. *There are some graduate students union members. Few taxi drivers are too rich. *There are few taxi drivers too rich.
In the standard account of there-insertion,
it would be neces-
sary to put a special condition on the rule to prevent deriving 2) from 1) (see 4.4.) or 4) from 3). restrict there-insertion predicate attribute.
That is, we would have to
from applying if be is followed by a
119 But in the structure-preserving framework where 2) would have to have the underlying structure (more precisely, +PRED) NP 5)
Some graduate students are
we note that there is no empty NP,
N p [union
Np[
^P'
members] N p for
students to move onto in order to generate 2).
sorr e
>
graduate
Or in other
words, be is subcategorized to precede NP but not NP NP.
Sim-
ilarly, in the case of 4) be is subcategorized to allow a following AP, as in 3), but not for a following NP AP, which would be required to provide an empty NP node for few taxi drivers in 3) to move onto to generate 4).
Thus no new statements are need-
ed in the structure-preserving framework to rule out 2) and 4). Observe, furthermore, that it follows in Emonds' framework (Emonds1
that we optionally may have there-insertion
indefinite
subject movement) with be plus prepositional phrase as in 6)
A man is in the garden.
7)
There is a man in the garden.
Thus 6) shows us that we must subcategorize be to allow a following PP.
If this PP is filled in deep structure, as in 6),
then there-insertion
may not apply.
However, since we also have 8)
John is a mayor in Denver.
we must also subcategorize be to allow a following predicate attribute (here a predicate nominal)-PP sequence.
If the pred-
icate attribute position is filled in deep structure as in 8) and in 9)
Some people are workers in Denver.
then tfcere-insertion may not apply. However, suppose the predicate attribute position is empty as in 10)
A man is
Np[
lNp
pptin
the garden]
Then tfcere-insertion may apply, moving a man onto the empty NP spot, yielding 11)
There is a man in the garden.
(=7)
120 In the above it has been assumed
(in accordance with Emonds
(1969)) that empty deep structure nodes are subject to subcategorization conditions.
Emonds (1970) discusses the possibili-
ty that empty nodes in deep structure need not fulfill sub-categorization conditions. there-insertion
However, as far as I can determine, the
facts on the basis of which Emonds proposes
this innovation do not support any such modification of the grammar.
Thus Emonds poses the question of how to account for the
facts if it is argued that be may have as complements either a predicate attribute is in Denver), 12)
(as in John is a mayor)
but not both.
or a PP (as in John
In this view then, in
John is a mayor in Denver.
(=8)
be is subcategorized only for the deep structure complement predicate attribute NP, and the PP in Denver
is outside the verb
phrase be a mayor. Now suppose the VP expansion rule is roughly as follows, as Emonds suggests 13)
VP
V-(
NP
I
)-(PP)
+PRED
That is, we may have a V followed optionally by a direct object NP or a predicate attribute followed optionally by a PP.
Thus
the sequence V-NP-PP yields 14)
I gave a book to Mary.
The sequence V-+PFED yields NP 15)
Peter is a mayor.
Now suppose the phrase structure rules generate the sequence 16)
V-+PRED-PP NP
Under the assumption just made above that be subcategorizes only for an NP or a PP but not both, we would not be able to insert be into the structure 16). and we have
But suppose the +PRED is empty NP
121 17)
V-+pRED[ NP
]-PP
If we relax the condition that empty deep structure nodes are subject to subcategorization conditions, then we could insert be into such a structure as 17) yielding, say 18)
Some d o c t o r s - b e - + p R E D [
] - p p [ i n Denver]
NP Thus, in effect, the empty NP node is "invisible" for purposes of subcategorization and since be is subcategorized for PP, it may be inserted into the structure 18). by there-insertion some doctors 19)
(=indefinite subject
Then we would get 19) movement),
which moves
onto the empty NP node
There are some doctors in Denver.
Emonds points out that allowing empty nodes to be ignored for purposes of subcategorization rules out the possibility of subcategorizing find in sentences like I found boring
for NP-PRED
in I consider
the movies
(or presumably of subcategorizing
John a fool as NP-PRED).
quite
consider
For otherwise we would
be able to insert be into such structures as 20)
Some graduate students-be~ N p [
]Np-PRED
since be is subcategorized for PRED and since we may ignore empty nodes for purposes of subcategorization.
Then we Would not
be able to exclude 21) as before.
*There are some graduate students union members. Thus, argues Emonds, the complements to find
presumably consider)
(=2)
(and
must be generated by fce-deletion from full
sentences and the problem will not arise. However, we note that the whole idea of allowing empty nodes not to be subject to subcategorization conditions with its attendant consequences all rest on the assumption that 22)
There are some doctors in Denver.
is derived from 18) above.
(=19)
Note, however, that if we were to
still insist that empty nodes above be subject to subcategorization conditions
(and hence ruling out 18) as the deep struc-
122 ture for 22)), we still have a possible source for 22); viz., 23)
Some d o c t o r s ~ V p [ b e - + p R E D [
] ] v p - p p [ i n Denver]
NP where be, we recall, is subcategorized for NP or PP, but not both.
The PP in Denver
is here outside the VP in accordance with
the assumption made by Emonds, which led to the whole preceding discussion.
Thus whatever motivation there is for assuming that
empty nodes in deep structure need not fulfill subcategorization conditions
(see Emonds
(1970) for other discussion), the facts
about the existential there do not seem to provide any such support for this modification of the grammar, as Emonds claims; rather one can continue to make the assumption that empty nodes are subject to subcategorization and all the facts can still be accomodated. Recall that the facts illustrated in 24) discussed by Milsark (see our discussion in 5.6.) provide an additional argument against the structure-preserving analysis 24)
There were some people sick. There were some people hungry.
Emonds notes such facts, but offers no explanation for them. As Milsark
(1974) points out, these examples have no syntactic
source under the structure-preserving analysis. no reduced relatives some people 25)
Thus there are
sick, etc. of the required form
*Some people sick are here.
Thus we have a gap which is unexplained in terms of the structure-preserving analysis, but explained in terms of the phrase structure analysis
6.5.
(see 5.6.).
Criticism of the Structure-preserving Formulation
The fact that the output of tfcere-insertion is always a structure which is independently generated by the phrase structure rules is accounted for by Emonds 1 sis, as we have seen.
structure-preserving hypothe-
However, observe that in any case where
the output of a transformation is a structure which is independently generated by phrase structure rules there is an additional possibility; viz., that the structure is directly generated in
123 the base by the phrase structure rules.
Furthermore, everything
else being equal, the phrase structure alternative is to be preferred if it involves rules which are independently motivated for other constructions.
The transformational alternative, on
the other hand, involves the postulation of a new rule in the transformational component. Thus in the phrase structure analysis we do not need to add any rules at all to the grammar to handle such
there-construc-
tions as there are ghosts; these are generated by phrase structure rules which are needed in any case to generate such sentences as John is a soldier.
The structure-preserving formula-
tion of there-insertion involves the postulation of a new transformation in the grammar; viz., there-insertion
(=indefinite
subject movement) in addition to the phrase structure rules which are needed for independent reasons.
Even when the phrase struc-
ture analysis is extended to accomodate more complex constructions such as there is a man standing on the corner, there is a man being beaten up outside, etc., we are able to generate these constructions with rules (such as Cleft Reduction) which are independently required for the description of cleft constructions in general.
Thus, given that the phrase structure analysis also
explains all the facts that the structure-preserving analysis does, it is to be preferred since it is in effect the null hypothesis, involving no special tftere-insertion transformation. However, the structure-preserving formulation of there-insertion fails in other ways.
Thus there is no way to prevent the
generation of sentences like 1) without recourse to ad hoc conditions like we discussed above 1)
*There are being some new houses built.
In the phrase structure analysis this construction is ruled out by the same principle that rules out * there is being a God, which is needed in the grammar in any case.
But the structure-preserv-
ing analysis of t/zere-insertion. must derive 1) from 2)
Some new houses are being built.
where the first occurrence of be is the passive be and the second occurrence of be is the progressive be.
But since Emonds
124 (1970) explicitly allows empty NP nodes to be generated in deep structure after the passive and progressive be s (see below) which he regards as verbs in deep structure, there is no way to prevent the generation of 1) without some ad hoc condition like we discussed in connection with standard t/zere-insertion to the effect that if there are two be s in a string satisfying the structural description of the rule, the first be must be chosen 4.2.).
(see
Thus with this ad hoc condition on t^ere-insertion we
get 3)
There are some new houses being built.
However, in the alternative phrase structure analysis, the fact that 3) is well-formed while 1) is not is explained without recourse to any ad hoc conditions in the grammar. There is a formulation of there-insertion
which could avoid
the ad hoc condition needed to exclude 1) in Emonds 1
analysis.
We could assume that tTzere-insertion applied to deep structures of the form 4)
NP Aux be
Νρ[Δ]Νρ
where Δ is a dummy element.
Thus, in essence, anything that can
be generated as an NP in subject position, including reduced relatives, may follow there
(after there-insertion).
In this
analysis 1) can be ruled out by the same condition discussed in connection with the phrase structure analysis.
However, this
leaves the parallels between existentials and clefts unexplained. For example, we have seen that by analyzing the constituent string following there as only an NP or reduced relative we were unable to accomodate such facts as the intonation contour of the existential there construction, which, however, were explainable by deriving there constructions from clefts. Furthermore, "subject-of" could be defined the same way for both a God is good and there is a God by deriving the latter from a deep structure
Np[a
G o d ] N p Aux be NP'-^-'NP·
However,
this is a weak argument for such an analysis of tftere-insertion, given the internal syntactic evidence in English just mentioned which argues for generating there in deep structure. Furthermore, the English order dummy
element-V(existence)-NP
125 has many cross-language parallels.
Thus compare German
(and
similarly many other languages) 5)
Es gibt einen Gott. "There is a God".
Notice that German lacks even the superficial syntactic evidence that einen Gott is the subject of geben.
Thus einen Gott
in 5) is in the accusative case. Moreover, there is no superficial subject-verb agreement; i.e., there is only 6)
Es gibt Götter. "There are Gods".
but no 7)
*Es geben
(pl.) Götter.
Thus the fact to be explained is why existential constructions pattern typically as dummy element-V(existence)-NP in deep structure . Thus the parallels between clefts and existentials on the one hand and the remarks about the null hypothesis above suffice to reject such a structure-preserving
solution.
Observe that Emonds' formulation of the structure-preserving analysis requires him to somewhat weaken the force of the structure-preserving hypothesis in general.
Recall that the main idea
behind the definition of a structure-preserving
transformation
(for movement transformations) was "the notion of moving a constituent labeled X into a position where a node X is already provided for by the phrase structure rules".
But if this is true,
consider what this means for the progressive and passive be s as in 8)
John is eating dinner.
9)
Two windows were broken.
(John-T(ie+ing)-eat dinner) (Two windows-T(be+en)-breakby Δ )
Emonds argues that the structure-preserving formulation of thereinsertion,
if correct, would have interesting implications for
deep structure constituency, in particular for the progressive and passive morphemes be.
Thus consider the progressive in
126
10)
Some children may be riding horses.
11)
There may be some children riding horses.
If 11) is related to 10) by tTzere-insertion, then the NP some children in 11) must have been moved onto an empty NP spot ] N p to satisfy the structure-preserving constraint. That Np[ is, be takes NP complements in underlying structure and thus 11) has the following deep structure
13)
N p[some
children] Np ~ M [may] M " vp [ v [be] v - Np [
]NP"VP
[riding horses] V p ] v p Similarly, 10) has a deep structure like 12) except that the empty NP node is missing. In the same way we can represent 14)
There were many windows broken,
with the following (derived) structure 15)
Np [many
windows]Np-vp[be]v~Np[
] Ν ρ ~ ν ρ [broken] v p ] y p
Thus in order to maintain the "structure-preserving" hypothesis, Emonds postulates empty NP nodes after the progressive and passive morphemes be in 13) and 15) in order that tftere-insertion can move the subject onto these nodes. Thus, in effect, be becomes a V(erb) which allows (empty) NP complements as in 12). But notice that there is no independent motivation for the generation of an NP node in the complement of the passive and progressive bes. However, this is precisely what is required by
127 Eitionds' definition of structure-preserving operation. As Emonds himself points out, given no constraints on the phrase structure rules, it is always possible to formulate non-structure-preserving transformations to be structure-preserving. But the crucial empirical claim of the structure-preserving hypothesis is that many logically possible rules are excluded from natural languages. Thus in Emonds' framework it is no accident that passive places noun phrases in positions independently generated by the phrase structure rules, but does not, say, place the subject in the auxiliary or does not, as under the hypothetical language "A" discussed by Emonds (1970), place the agent phrase directly before the verb. However, given Emonds' analysis of the passive (and progressive) be, where an NP node is generated in the complement of the passive, it would be presumably quite possible for the passive subject postposing rule to postpose the subject after the auxiliary be (exactly as is the case with indefinite subject movement)
as in
16)
Russia
defeated
Germany
It should be emphasized that Emonds' move does not weaken the structure-preserving hypothesis to the extent that empty nodes are allowed to be generated anywhere. Indefinite Subject Movement does actually move the subject to an NP position which is provided for by the phrase structure rule VP-*-V-NP. However, in the case of the auxiliary be this NP position is never filled except by indefinite subject movement. This is different than the case in 6.4. for the derivation of such sentences as 17)
There is a mayor in Denver.
128
from 18)
A mayor is
Np[
] N p PP
In fact, be occurs in deep structure with full noun phrases 19)
John is a mayor.
What is lost in Emonds' account is the fact that NPs do not typically appear in auxiliary position in English 20)
*John is Peter eating lunch. *John was Bill seen by Mary. *John will Mary be here by six.
And in fact, NPs seem to end up in what appears to be auxiliary position only by the application of root transformations, like Subject-aux Inversion as in 21), but not through the application of structure-preserving operations, like Passive, as we noted above 21)
Is Peter eating lunch? Was Bill seen by Mary? Will Mary be here by 6:00?
Thus we have one more reason for rejecting the structure-preserving account of tTzere-insertion; viz., that it slightly weakens the explanatory power of the theory. However, the examples discussed above presented no problem for the phrase structure analysis. As we have shown (Chapter 2), the underlined be morpheme in 22) is in both cases the copula be generated in deep structure rather than the progressive and passive be 22)
There may be some children riding horses. There were some windows broken.
We furthermore conclude that no inferences can be drawn from the "structure-preserving" hypothesis in order to substantiate the claim which has appeared in much of the literature of transformational grammar that the auxiliary be (and have) is generated by the same set of rules which generate other English V(erbs) and their complements (see Jenkins (1972)). Thus the structure-preserving formulation of tTzere-insertion is to be rejected on the grounds that it involves the addition of an otherwise unnecessary transformational rule to the grammar, on the grounds that the statement of the rule involves the
129
addition of an ad hoc condition to the grammar to exclude 1), and on the grounds that it weakens the claim of the structurepreserving hypothesis. Thus, since no (structure-preserving) rule of tfcere-insertion exists, such a rule cannot provide evidence in favor of Emonds1 structure-preserving hypothesis, although it should be noted that the correctness of the phrase structure analysis does not call into question the validity of the structure-preserving hypothesis. On the contrary, the problem for the structure-preserving hypothesis discussed immediately above dissolves if the phrase structure analysis is correct.
7.
THERE-INSERTION IN GENERATIVE SEMANTICS
7.1.
Introduction
In what follows we will consider a proposal for a rule of tftere-insertion proposed by McCawley (1970, 1972).
It is im-
portant to point out that, although McCawley's presentation of there-insertion within the framework of generative semantics is different than standard accounts in that the rule operates on semantic representations containing existential quantifiers as predicates, in what follows we will be concerned primarily with those features of his analysis which can be readily stated in the standard theory;
e.g., the fact that the rule applies to
verb initial structures.
Thus, abstracting the variables and
existential quantifiers from McCawley's underlying semantic representations (see McCawley (1970)), which, we argue, do not have any bearing on what follows.
McCawley's rule of there-
insertion will apply in the generation of the structure underlying 1) from the structure 2) 1) 2)
There is a unicorn in the garden. s [y[is]y~ N p[a
unicorn] Np -[in the garden]]g
This takes place in two steps.
First there-insertion applies
to 2) to yield 3)
sCVtis]V~NP[thereINP~NP[a
unicorn
lNP~tin
the
garden]]g
and secondly, a rule of Verb-second placement (the inverse rule to Subject-aux Inversion in standard analyses) applies, yielding the correct surface structure 1) . In what follows we will show that McCawley's analysis of there-insertion is subject to many of the same defects that are inherent in standard accounts of there-insertion.
But before
doing so, let us consider what advantages McCawley claims for
131 his tfrere-insertion analysis.
In McCawley
the discussion of there-insertion tions.
(1972) he takes up
along with other transforma-
In the section entitled "On the non-equivalence of some
'notational variants'" McCawley states that "one of the goals of linguistics is to provide as tight as possible a characterization of what a 'possible grammar 1 is" and that "the proposal of verb-first order allows much tighter constraints to be imposed on the class of 'possible transformations' than does the proposal of verb-second order".
Thus McCawley maintains that giv-
en the underlying structure 2), his rule of tftere-insertion need perform only one elementary operation; viz., the insertion of the morpheme there, whereas in standard analyses, the
there-in-
sertion transformation must perform two operations, insertion of there and permutation of the subject with be. Although, as we have insisted throughout this study, we agree with McCawley that an important goal of linguistic theory is the characterization of "possible transformations"; viz., the constraints on the operations that such rules can perform, it is clear that we can't learn much about such constraints on transformations, unless the individual transformational rules like tfcere-insertion are well-motivated syntactically.
For instance,
if it is the case, as we claim, that there is no rule inserting there in English at all, then no conclusion can be drawn about the underlying order of subject and verb and the facts about there are irrelevant to the issue about what operations transformations can perform. A similar argument applies to McCawley's argument that the rule of Raising provides support to underlying verb first order. As Chomsky
(1971) points out, if there is no syntactic rule of
Raising in English, no conclusions can be drawn about underlying deep structure order.
Note also that standard
there-inser-
tion can be made to apply in two parts, as in Milsark's formulation; see 5.1. McCawley also hopes to show that the semantic deep structures that the linguist postulates converge with one or another logical structure postulated by logicians for other reasons—compare his discussion of the logical systems of Lukasiewicz and Peano
132 in the reference above.
Thus he argues that
iftere-insertion
is more easily stated with underlying verb first order, corresponding to Lukasiewicz's
system.
Again this conclusion is only as strong as the arguments are for the transformations on which it is based.
For some comments
on "logical" underlying orders and for many other arguments against the verb-initial hypothesis, see Berman
7.2.
(1974).
Criticism of McCawley's Analysis
Turning now to McCawley's analysis, consider the derivation of sentences like 1) There
There has been a man in the garden.
is inserted by tftere-insertion into the configuration 2)
[Tense][have][be] N p[a m a n ] N p [ i n the garden]
and moved up the tree by the obligatory rule Raising.
A post-
cyclic rule applies, McCawley's Verb-second rule, inverting the main verb and subject to yield the correct surface
structure.
Thus, schematically the derivation of 1) proceeds as follows 3)
[Tense] - [have]
4)
[Tense] v ~[have] v ~[be] v ~there-[a m a n ] N p [ i n the garden]
5)
[Tense] v -[have] v ~there-[be] v ~[a m a n ] N p [ i n the garden]
6)
[ T e n s e ] v ~ t h e r e - [ h a v e ] [ b e ] v ~ [ a m a n ] N p [ i n the garden]
7)
[be] - [a m a n ] N p [ i n the garden]
there-Tense-have-be-a man in the garden
Raising has applied in 5), 6) and Verb-second has applied in 7). In examining McCawley's analysis of existential there,
we
are assuming for purposes of argument Ross's analysis of auxiliary verbs as main verbs, an assumption which Postal insists on
(1971).
However, for counterarguments to the auxiliary as
main verb proposal, see Jenkins
(1972).
of such sentences as a man has been
Thus in the derivation
in the garden,
a man
orig-
inates in the lowest S and is moved upwards through the sentences containing the main verbs have,
be, and
(according to McCawley)
Tense by a rule called Raising, which applies obligatorily
for
133 auxiliary verbs (see Postal, 1971).
Thus in the derivation of
such sentences as there is someone
working
were some people
there is going
(being) arrested,
on presentences,
there
to be a scan-
dal, and in the derivation of all of the sentences to be discussed, we will be assuming underlying phrase structure trees such as found in Ross (19 72).
We overlook refinements by McCawley
which derive the perfect have from past tense. Returning to our main point, the problems which are involved can be observed in the generation of the sentence 8)
There are some new houses being built.
This will have roughly the following underlying structure
(after
passive) 9)
Tense-be(prog)-be(pass)-build-some new houses-(by s.o.)
(Be(prog) means the progressive be and be(pass) means the passive be.) 10)
After Raising on the first cycle, we get Tense-be(prog)-be(pass)-some new houses-build
Since we have the context be
NP, we may optionally apply
there-
Insertion , yielding 11)
Tense-be(prog)-be(pass)-there-some new houses-build
which, however, by successive Raising, which is obligatory with auxiliaries, and Verb-second will give 12)
*There are being some new houses built.
However, suppose that we do not apply tTzere-insertion, which is optional, until the next cycle, applying instead Raising to some new houses, 13)
yielding
Tense-be(prog)-some new houses-be(pass)-build
Now we may apply tftere-insertion, giving 14)
Tense-be(prog)-there-some new houses-be(pass)-build
which by successive Raising of there and Verb-second will yield 15)
There are some new houses being built.
However, the only way we can generate the correct 15) and exclude the ungrammatical 12) is by applying there-insertion
only
when it reaches the second be; i.e., the topmost be in the tree.
134 Thus we must add the following ad hoc condition to McCawley's t^ere-insertion rule 16)
Do not apply tftere-insertion in its context be unless be is the topmost be.
NP
More exactly, the topmost be in a sequence of auxiliaries
(and
copula) is meant here.
There may, of course, be several such
sequences in the tree, each containing several bes.
Condition
16) applies to each such sequence. Notice, however, that this ad hoc condition is the same restriction which we need in the standard formulation of insertion
(see 4.2.).
there-
Thus in standard tfrere-insertion analyses,
in order to generate the correct 15) and exclude the ungrammatical 12), we needed a condition, roughly as follows 17)
Do not apply tTzere-insertion in its contexts unless the be mentioned in the structural condition is the first (i.e., leftmost) be.
Thus McCawley's formulation of tftere-insertion offers no advantages here over the standard formulation of
there-insertion,
but rather both formulations have exactly the same defect. ever, in the phrase structure analysis of there
How-
(see Chapter
2), exactly the right set of sentences is generated. The grammatical sentence 15) is generated from a reduced cleft (or a reduced relative) 18)
There-Tense-be~ N p [some new houses] s [being built]
while the ungrammatical sentence 12) is excluded by an independently needed constraint which rules out sentences like 19)
There are being Gods.
20)
There are being ghosts in the garden.
A second major defect in McCawley's formulation is seen by considering examples like 21)
There is going to be a demonstration in the gym.
with roughly the following underlying structure 22)
Tense-be-go-be-a demonstration in the gym
Here, we cannot apply there-insertion
since the be in question
135 is n o t t h e t o p m o s t be t i o n s of R a i s i n g , w e 23)
By two successive
Tense-be-a demonstration-go-be-in
We can now apply be,
in the tree.
applica-
get the
gym
tfcere-insertion in the c o n t e x t of t h e
topmost
giving 24)
Applying
Tense-be-there-a demonstration-go-be-in R a i s i n g to there
25)
and Verb-second we
applying as
gym
get
» T h e r e i s a d e m o n s t r a t i o n g o i n g t o b e i n the
In f a c t , t h e o n l y w a y to g e n e r a t e t h e g r a m m a t i c a l be's
the
there-insertion
21)
gym. is b y
i n t h e c o n t e x t of t h e l o w e r of t h e
two
in
26)
Tense-be-go-be-there-a
demonstration
in the
gym
which, after successive Raising and Verb-second yields the
de-
s i r e d 21) . B u t n o t e t h a t i n o r d e r t o g e n e r a t e t h e c o r r e c t s e t of t e n c e s , w e a r e v i o l a t i n g C o n d i t i o n 16)
above; viz., the
sencondi-
t i o n t h a t ifcere-insertion a p p l y o n l y i n t h e c o n t e x t of t h e m o s t be.
T h u s w e m u s t a d d a s u b c a s e to t h i s a d h o c
condition;
v i z . , w e m u s t say t h a t in t h e c a s e o f t h e r e - i n s e r t i o n , w e p i c k t h e t o p m o s t be, w e m u s t p i c k the a completely generated.
be.
where
I t is r e a d i l y e v i d e n t t h a t t h i s
exactly the same ad h o c
Thus once again McCawley's
l y s i s o f there
However,
thereexhibit
the phrase structure
can again generate the proper sentences,
use of independently needed constraints
to as
ana-
making
(4.3.).
A s i m i l a r example, s h o w i n g the same p o i n t as w a s m a d e in n e c t i o n w i t h be going
now
formulation
over the standard account, b u t
the same defects.
is
subcondition
to is r e q u i r e d i n s t a n d a r d f o r m u l a t i o n s of
(see 4 . 3 . ) .
offers us no advantages precisely
to
must
ad hoc move, although the proper sentences are Furthermore,
f o r be going insertion
e x c e p t i n t h e c a s e of be going
lowest
top-
to c a n b e g i v e n u s i n g t h e s e m i - m o d a l
conbe
in 27)
*There are three senators to be at the
conference.
28)
T h e r e a r e to b e t h r e e s e n a t o r s a t t h e
conference.
136 The final argument to be made is in connection with predicate nomináis.
Notice that still another ad hoc condition must be
added to McCawley's description of tftere-insertion to prevent the derivation of 29) from 30) 29) 30)
*There are some graduate students union members. Some graduate students are union members.
Thus we must in effect require the following condition on thereinsertion 31)
there-insertion may not apply in the context be NP-Predicate Nominal
Again this is the same restriction as is required in standard i/zere-insertion (see 4.4.).
But in the phrase structure analy-
sis it is impossible to generate 29) since the rules of English grammar do not generate the phrase some graduate students union members as either a noun phrase or as a reduced cleft 32)
* N p [Some graduate students union members] N p
33)
»There are
N p [some
graduate students]g[union members)
Thus in McCawley's there-insertion
analysis (as well as in stan-
dard tAere-insertion) the grammar must be complicated by the addition of a condition like 31) to the statement of
there—in-
sertion, in order to rule out 29), but in the simpler phrase structure analysis nothing new need be stated at all to generate the correct forms. Another argument in this connection against McCawley's formulation of iftere-insertion is based on the contrast between 34)
*There are people tall.
35)
There are people sick.
See the discussion in 5.6. and 6.4. As we have seen, McCawley's formulation of iftere-insertion, when spelled out in detail, has many of the same defects of the standard formulation, defects which can, as far as I can see, only be overcome within the phrase structure analysis
(Chapter
2) where it is assumed that there is generated in deep structure.
Furthermore, McCawley presents no other syntactic argu-
ments internal to English which argue for his particular form-
137 ulation of there-insertion lations of tftere-insertion.
as against the more standard formuThus no conclusions about the "log-
ical" order of constituents in deep structure can be drawn from the data on there.
See also McCawley (1970) for other arguments
for underlying verb second order, as well as criticism by Chomsky (1971), Berman (1974).
8.
LOCATIVE ANALYSES OF
8.1.
Introduction
MEM-INSERTION
In this chapter we will consider very briefly three other analyses of the English existential there, Kimball, and Fillmore.
the analyses of Kuno,
What they all have in common is that
a locative constituent plays a crucial role in underlying ture.
Thus we may term them "locative" analyses.
struc-
In the case
of Kuno's and Fillmore's analyses the existential morpheme is derived from an underlying locative phrase. Kimball's analysis, there is a verb locate form of the existential
8.2.
there
In the case of
in the underlying
construction.
The Locative Analysis of Kuno
In Kuno
(1971) it is proposed that the underlying
of 1) is 2) 1)
structure
(which in turn is possibly derived from 3) )
There is a man in the garden.
2)
In the garden is a man.
3)
A man is in the garden.
Thus the existential morpheme there locative constituent.
derives from an underlying
Let us ask what evidence Kuno has.
One piece of evidence, Kuno claims, is that we can explain the superficial form of the morpheme there, from a locative.
since it derives
However, as Kuno himself notes, the plausib-
ility of this explanation is seriously diminished by the fact that there is no obvious underlying locative constituent the full range of existential constructions
(unless one postu-
lates an ad hoc underlying LOC marker in each case) as in 4)
There are people willing to help.
5)
People are willing to help.
for
139 Furthermore, in the phrase structure analysis too we were able to provide a partial explanation for the existential there morpheme, by relating it to the there morpheme of cleft sentences.
However, none of the parallels between existentials and
clefts captured by the phrase structure analysis are captured in Kuno's analysis. The only other principal argument that Kuno advances is an account of various existential constructions containing quantifiers.
Thus compare such sentences as
6)
Many people are in all classes.
7)
There are many people in all classes.
Thus while 6) is ambiguous in interpretation between many ple attend
all classes
and all classes
7) means only the latter.
have many people
peo-
in them,
Now the explanation that Kuno suggests
involves constraints that extend the expressive power of the theory.
Milsark
(1974)
shows that Kuno's facts can be described
within the traditional theory of scope of quantifiers without such undesirable extensions of the theory.
Note that the dif-
ference of interpretation in 6) and 7) can be related to the "definiteness restriction" discussed in 5.5.
(See Milsark
(1974)
for a much more detailed criticism of the locative theory of Kuno.) Thus the data discussed by Kuno does not require us to relax the null hypothesis for the English existential. 8.3.
The Locative Analysis of Kimball
Kimball (1973) suggests that the locative sentences in 1) and 2) both contain the item locate in their underlying form 1)
A bird is in the tree.
2)
There is corn in the manger.
Thus the underlying form of 1) is 3) 3)
I/one locate(s) a bird in the tree.
After applying Passive, we get 4)
A bird is located in the tree.
Then locate is dropped, giving us 1).
140 Similarly, the underlying form for 2) is 5)
I/one can locate corn in the manger,
and the underlying form for 6) is 7) 6)
There is a God.
7)
I/one can locate a God.
Kimball hopes this analysis will serve to explain "the close connection between existence and possibility" as well as provide a source for such sentences as 8)
There is no even prime greater than two.
which would have the underlying form 9)
I/one cannot locate an even prime greater than two.
Kimball gives no further syntactic or other justification of this analysis. Notice that when we consider the meaning of constructions like 8) and 9) more carefully, it is apparent that they are not semantic paraphrases. Thus compare 10) 11)
There is no four-leaf clover in this field. I/one can't locate a four-leaf clover in this field.
12) 13)
There is no solution to Fermât's last theorem. I/one can't locate a solution to Fermât's last theorem.
It is obvious that neither 10) and 11) nor 12) and 13) are paraphrases, as is entailed by Kimball's analysis. If they were, then 14) should be contradictory, like its source 15), which is not the case 14)
15)
There's a solution to Fermât's last theorem, but I/one can't locate a solution to Fermât's last . theorem. I/one can locate a solution to Fermât's last theorem, but I/one can't locate a solution to Fermât's last theorem.
Thus the analysis collapses on semantic grounds and in the absence of any other evidence must be rejected.
141 8.4.
The Case Grammar Analysis
In the case grammar analysis of Fillmore (1968) it is proposed that 1), 2) , 3) have identical underlying structures 1) 2) 3)
Many toys are in the box. There are many toys in the box. The box has many toys in it.
However, Fillmore offers no specific arguments for the derivations he proposes, referring the reader in a footnote to a forthcoming work on existential sentences. In the lack of any more detailed evidence in favor of this analysis, it can be pointed out once more that, as in the case of Kuno's analysis, such an analysis holds little promise of being able to be extended to the full class of non-locative existential sentences, nor of explaining the varied syntactic parallels between existentials and clefts that have formed the subject of this study. Nor does the particular case grammar framework seem to offer any advantages from the semantic point of view. As has been pointed out in Chomsky (1972), the semantic descriptions given in case relation terms can be mapped into the (extended) standard theory that we are presupposing here as a notational variant. Furthermore, other semantic frameworks—such as the thematic relation framework of Jackendoff (1972)—seem to me to offer more promise of explaining the kind of facts that case theory was devised for. In any case there is no syntactic or semantic justification available from the case grammar analysis which would require us to modify (and hence undesirably extend) our minimal phrase structure analysis.
9.
CONCLUSION
To sum up, then, we have presented and defended a hypothesis which we have entitled the phrase English existential there
struoture
analysis
of the
against a variety of competing hypoth-
eses which we referred to as varieties of the transformational analysis.
The main characteristic of the phrase structure ana-
lysis is that it is assumed that the English existential construction is base-generated with there (perhaps as a determiner).
in deep subject position
That is, we have argued for the null
hypothesis for the existential; viz., that the deep structure and surface structure of these constructions is essentially the same.
Thus there is no special transformation of tftere-inser-
tion that relates deep and surface structure, but only such transformations as are independently needed in the grammar rules of relative and cleft formation, Agreement, It was furthermore shown that the existential exhibited the properties of such base-generated
(like the
etc.). constructions
copula-predicate
constructions as are independently generated in the g r a m m a r — locatives, relatives, and cleft.
This is, as we pointed out,
a necessary consequence of the phrase structure analysis, while under the transformational analysis it is an accident. One important theoretical consequence of this study is that it has been shown to be possible to describe the English existential construction without introducing transformations which are rich in expressive power, as apparently the tftere-insertion transformation would be, if it existed.
For example, it would require
the statement of quantifiers in the structural condition of the transformation.
This is an important result as it allows us
to retain a highly-restricted theory of transformations.
And
such very restricted theories of language are what we are aiming for in order to explain the mysteries of language
learning.
lo.
BIBLIOGRAPHY
Akmajian, Adrian. 1970. Aspects of the grammar of focus in English. M.I.T. Dissertation. Cambridge/Mass. Akmajian, Α., F.W. Heny. In press. Introduction to the principles of transformational syntax. Cambridge/Mass., MIT Press. Allan, Keith. 1971. A note on the source of there in existential sentences. FL 7. 1-18. Bach, Emmon. 1974. Syntactic theory. and Winston, Inc.
New York, Holt, Rinehart
Baker, C.L. 1973. Definiteness and indefiniteness in English. Bloomington/Ind., Indiana University Linguistics Club (mimeographed) . Berman, Arlene.
1974.
On the VSO hypothesis.
LI 5.1
1-37.
Borkin, Ann, David Peterson (eds.). 1972. Where the rules fail: a student's guide. Bloomington/Ind., Indiana University Linguistics Club (mimeographed). Bresnan, Joan. 1970. An argument against pronominalization. LI 1.1 122-23. 1971. Sentence stress and syntactic transformations. Lg 47. 257-81. Burt, Marina K. 1971. From deep to surface structure. York, Harper and Row.
New
Chomsky, Noam. 1957. Syntactic structures. The Hague, Mouton . 1965. Aspects of the theory of syntax. Cambridge/Mass., MIT Press. 1970a. Remarks on nominalization. Readings in English transformational grammar, Roderick A. Jacobs, Peter S. Rosenbaum (eds.), 184-221. Waltham/Mass., Ginn and Co. 19 70b. Deep structure, surface structure, and semantic interpretation. Studies in general and oriental linguistics presented to Shiro Hattori on the occasion of his sixtieth birthday, Roman Jakobson, Shigeo Kawamoto (eds.). Tokyo, TEC Co. Ltd. 1971. Conditions on transformations. Bloomington/Ind., Indiana University Linguistics Club (mimeographed). Also in A festschrift for Morris Halle, Stephen R. Anderson, Paul
144 Kiparsky (eds.), 232-86. New York, Holt, Rinehart and Winston, Inc. 1972. Empirical issues in the theory of transformational grammar. Studies on semantics in generative grammar, 120202. The Hague, Mouton. 1975. Questions of form and interpretation. Linguistic Analysis 1.1 Forthcoming. Lectures on grammar. Emonds, Joseph. 1969. A structure-preserving constraint on NP movement transformations. Papers from the fifth regional meeting Chicago linguistic society, Robert I. Binnick et al. (eds.), 60-5. Chicago, Chicago Linguistic Society. 1970. Root and structure-preserving transformations. M.I.T. Dissertation. Bloomington/Ind., Indiana University Linguistics Club (mimeographed). Erteschik, Nomi S. 1973. On the nature of island constraints. M.I.T. Dissertation. Cambridge/Mass. Fiengo, Robert W. ture. M.I.T.
1974. Semantic conditions on surface strucDissertation. Cambridge/Mass.
Fillmore, Charles J. 1968. The case for case. Universals in linguistic theory, Emmon Bach, Robert Harms (eds.), 1-90. New York, Holt, Rinehart and Winston. Grinder, John. 19 72. On the cycle in syntax. Syntax and semantics, Vol. 1, John P. Kimball (ed.), 81-111. New York, Seminar Press. Grinder, John T., Suzette H. Elgin. 1973. Guide to transformational grammar. New York, Holt, Rinehart and Winston, Inc. Hellberg, Staffan. 1970. On existential sentences. Gothenburg papers in theoretical linguistics. Göteborg (mimeographed) . Higgins, Francis R. 1973. The pseudo-cleft construction in English. M.I.T. Dissertation. Cambridge/Mass. Horn, Laurence. 19 72. On the semantic properties of logical operators in English. U.C.L.A. Dissertation. Los Angeles/Cal. Jackendoff, Ray S. 19 72. Semantic interpretation in generative grammar. Cambridge/Mass., MIT Press. Jenkins, Lyle. 1972. Modality in English syntax. M.I.T. Dissertation. Bloomington/Ind., Indiana University Linguistics Club (mimeographed). 1974a. Cleft reduction. Actes du colloque franco-allemand de grammaire transformationnelle, Vol. 1, Christian Rohrer, Nicolas Ruwet (eds.),182-91. Linguistische Arbeiten. Tübingen, Max Niemeyér Verlag. 1974b. Tfcere-insertion in generative semantics. Löwen und Sprachtiger, Referate des VIII. Linguistischen Kolloquiums Löwen. Tübingen, Rotsch Verlag.
145 Jespersen, Otto. 1940. A modern English grammar on historical principles. Heidelberg, Carl Winter. Kimball, John P. 19 72. Cyclic and linear grammars. Syntax and semantics, Vol. 1, John P. Kimball (ed.), 63-80. New York, Seminar Press. 1973. The grammar of existence. Papers from the ninth regional meeting Chicago linguistic society, Claudia Corvini et al. (eds.), 262-70. Chicago, Chicago Linguistic Society. Kirsner, Robert S. 1973. Natural focus and agentive interpretation: on the semantics of Dutch expletive er. Los Angeles, U.C.L.A. (mimeographed). Kuno, Susumu. 1971. The position of locatives in existential sentences. LI 2.3 333-78. Lyons, John. 19 67. A note on possessive, existential and locative sentences. FL 3. 390-9 6. 1968. Existence, location, possession and transitivity. Logic, methodology and philosophy of science III, B. Van Rootselaar, J.F. Staal (eds.), 495-504. Amsterdam, NorthHolland. 19 73. Deixis as the source of reference. Trier, Linguistic Agency University Trier (mimeographed). Milsark, Gary L. 1974. Existential sentences in English. M.I.T. Dissertation. Cambridge/Mass. McCawley, James. 19 70. English as a VSO language. Lg 4 6 286-99. 19 72. A program for logic. Semantics of natural language, Donald Davidson, Gilbert Harman (eds.), 498-544. Dordrecht, D. Reidel. Perlmutter, David M. 1970. The two verbs begin. Readings in English transformational grammar, Roderick A. Jacobs, Peter S. Rosenbaum (eds.), 107-119. Waltham/Mass., Ginn and Co. 1970b. On the article in English. Progress in Linguistics, Manfred Bierwisch, Karl Erich Heidolph (eds.), 233-248. The Hague, Mouton. 1971. There and Equi. Papers in Linguistics 4.2 389. Postal, Paul. 1970. On so-called pronouns in English. Readings in English transformational grammar, Roderick A. Jacobs, Peter S. Rosenbaum (eds.), 56-82. Waltham/Mass., Ginn and Co. 1971. A remark on the verb-initial hypothesis. Papers in Linguistics 5.1 124-137. Poutsma, H. 1916. A grammar of late modern English. en, P. Noordhoff.
Groning-
146 Roberts, Paul. 1964. English syntax. Brace and World, Inc.
New York, Harcourt,
Rosenbaum Peter S. 1967. The grammar of English predicate complement constructions. Cambridge/Mass., MIT Press. Ross, John R. 1967. Constraints on variables in syntax. M.I.T. Dissertation. Bloomington/Ind., Indiana University Linguistics Club (mimeographed). 1969. Guess who. Papers from the fifth regional meeting Chicago linguistic society, Robert I. Binnick et al. (eds.), 252-86. Chicago, Department of Linguistics. 1972. Doubl-ing. LI 3.1 61-86. Vendler, Zeno. 1964. Nominalizations. Transformations and discourse analysis papers, No. 55. Philadelphia, University of Pennsylvania. Wasow, Thomas. 1972. Anaphoric relations in English. Dissertation. Cambridge/Mass. Williams, Edwin S. 1971. Small clauses in English. Mass., M.I.T. (mimeographed).
M.I.T. Cambridge/