Definite Articles As Relative Operators
∗
Karlos Arregi
MIT, 14 December 2000
1. Introduction
It is often the case that relative operators have other functions. Typically, a
relative operator in a given language can also be either a question wh-word,
as in English (cf. 1), or a definite article, as in German (cf. 2, taken from
Wiltschko 1998). In Spanish, there are relative operators of both kinds, as
shown in (3).
(1) a. the man who Bill saw
b. Who did Bill see?
(2) Peter
Peter
bewundert
admires
den
the
Mann,
man
der
the
das
the
Bier
beer
erfand.
invented
Peter admires the man who invented beer.
(3) a. el
the
hombre
man
con
with
el
the
que
that
estuvimos
we-were
hablando
talking
ayer
yesterday
the man we talked to yesterday
b. ¿Con
with
qui´en
who
estuvisteis
you-were
hablando
talking
ayer?
yesterday
Who did you talk to yesterday?
c. el
the
hombre
man
con
with
quien
who
estuvimos
that
hablando
we-were
ayer
talking yesterday
the man we talked to yesterday
Given that this phenomenon does not seem to be accidental, an analysis of these
relative operators in which they are seen as homophonous with definite articles
or question wh-words cannot be considered satisfactory. Rather, the strategy
should be to assume that relative operators in these languages are what they
appear to be (definite articles or wh-words) and find an explanation for the
∗
The research reported in this paper has benefited greatly from discussion with Paul
Elbourne, Sabine Iatridou and David Pesetsky. I would also like to thank Calixto Ag¨uero-
Bautista and Irene Heim for useful comments on several topics discussed here. Needless to
say, all errors are mines.
1
fact that they, as opposed to any other type of word, can function as relative
operators.
In this paper, I concentrate on definite articles that can function as relative
operators, drawing mainly on Spanish data. Following the strategy suggested
in the previous paragraph, I show that if we assume that these items are indeed
definite articles, we can find a very natural explanation for their use as relative
operators, and for why other determiners cannot have this use. Specifically, I
argue that these facts follow directly from the copy theory of movement (see
Chomsky 1993 and §3.1 below), and from certain independently motivated
assumptions about the interpretation of lower copies of moved items. Thus, the
analysis defended in this paper not only provides an explanation for the use of
definite articles as relative operators, but it also provides strong support for the
copy theory of movement.
As I show in more detail in §4, the structure I assume for relative clauses is
the one represented in (4). In this structure, the phrase containing the definite
article undergoes
A-movement to [Spec,CP].
(4) [
DP
DNP[
CP
[ ...el...] [
C
...[ ...el...] ...] ] ]
✻
The only (semantic) purpose that this movement has is to make the sister of the
upper copy (i.e.
C) a predicate. Furthermore, the relative clause (i.e. CP) must
be interpreted as a predicate. Thus, in order to be able to interpret the CP as
a relative clause, the phrase in [Spec,CP] must be deleted (i.e. ‘reconstructed’).
The result is that only the lower copy is interpreted. Since, within the copy
theory of movement, lower copies of moved items must be interpreted as definite
descriptions (see Fox 2000b, Elbourne 2000, and §3.1 below), the result is that
the movement in (4) is only possible if the moved phrase is headed by a definite
article. As shown in §4, phrases headed by other determiners, such as quantifiers,
are not possible in this structure, due to an independently motivated identity
condition on the deletion of higher copies of moved items (see §3.2).
The paper is organized as follows. The analysis of Spanish relative clauses,
which was summarized above, is developed in §4. The preceding sections, §2
and §3, discuss the Spanish data and the theoretical assumptions needed for the
analysis, respectively. Finally, §5 discusses other types of relative pronouns, and
offers some suggestions as to how the analysis defended in §4 can be extended
to them.
2. Spanish Relative Clauses
As examplified in (3a) above, definite articles can be used as relative operators
in Spanish. However, there are other types of relative operators in Spanish, as
shown in (5).
2
(5) a. Definite Article
la
the
mesa
table
sobre
on
la
the
que
that
est´a
is
el
the
libro
book
the table the book is on
b. Definite Article + Cual ‘which’
la
the
mesa
table
sobre
on
la
the
cual
which
est´a
is
el
the
libro
book
the table the book is on
c. The Wh-Word Quien ‘Who’
el
the
hombre
man
con
with
quien
who
estuvimos
that
hablando
we-were
ayer
talking yesterday
the man we talked to yesterday
d. The Wh-Word Donde ‘Where’
la
the
casa
house
donde
where
naci´o
was-born
Cervantes
Cervantes
the house were Cervantes was born
e. The Relative Pronoun Cuyo ‘Whose’
el
the
libro
book
cuyo
whose
autor
author
naci´o
was-born
en
in
esa
that
casa
house
the book whose author was born in that house
In this paper, I concentrate only on those which are or contain a definite article,
i.e. those exemplified in (5a, b). As for the ones which are wh-words and
can be used in questions (cf. 5c, d), no complete analysis is offered, but some
suggestions are made in §5. Finally, cuyo ‘whose’ (cf. 5e) is, for the purposes
of this paper, the least interesting of all Spanish relative pronouns, since it does
not have any other use. In particular, even though it is derived from a Latin
word (cuius) which could be used both as a relative pronoun and a question
wh-word, in Spanish it does not have this latter use. Thus, I will have nothing
to say about it in this paper.
It is important to note that, as in other languages, there are certain cases
in which no overt relative operator is used. This occurs in subject and object
relative clauses, as shown in (6, 7).
(6) a. el
the
hombre
man
que
that
vino
came
ayer
yesterday
b. * el
the
hombre
man
el
the
que
that
/el
the
cual
which
/ quien
who
vino
came
ayer
yesterday
(7) a. el
the
libro
book
que
that
le´ı
I-read
ayer
yesterday
the book I read yesterday
3
b. * el
the
libro
book
el
the
que
that
/el
the
cual
which
le´ı
I-read
ayer
yesterday
the book I read yesterday
For the purposes of this paper, I adopt the standard assumption that this type
of relative clause in fact does involve relative operators (i.e. the ones examplified
in 5). In certain cases, this relative operator is deleted at PF (i.e. in 6 and 7,
but not in 5).
1
The analysis of Spanish relative clauses developed in the next
two sections assumes that a definite article can always be used as a relative
operator in Spanish, and that in some cases, it is deleted at PF.
3. The Syntax and Semantics of Movement
In this section, I present the basic assumptions on which the analysis of relative
clauses in §4 is based. They have to do with the mechanisms by which move-
ment chains are created in the syntax and intepreted by the semantics. The
first two of these assumptions, which have been justified in the literature inde-
pendently of relative clauses, are the following: (i) movement involves copying
(cf. Chomsky 1993), and (ii) the lower copy of a moved item is interpreted as
a bound definite description (cf. Fox 2000b, Elbourne 2000). The third as-
sumption, which could be considered more controversial, is that scope (‘total’)
reconstruction can be achieved by deleting the higher copy of a moved item (see
Chomsky 1993, Fox 2000b).
In the following subsections, I present these assumptions, and the arguments
that have been offered in the literature in favor of them. I will only discuss these
topics to the extent that they are helpful for the analysis of relative clauses
developed in §4. §3.1 concentrates on the copy theory of movement (i and ii
above), and §3.2 is a discusssion on total reconstruction.
3.1. The Copy Theory of Movement
Within the framework of the Minimalist Program, Chomsky (1993) proposes
that movement does not leave traces. Rather, movement consists in making a
copy of a constituent of the tree and merging it in some other position. The
result is that, after movement, there are two copies of the same constituent
in two different positions in the tree, and, typically, only one of the copies is
pronounced at PF. Chomsky (1993) argues that this theory provides a natural
explanation for certain cases of reconstruction. In particular, this analysis of
movement can account for cases in which
A-movement cannot circumvent a
violation of Condition C of Binding Theory, as exemplified in (8).
(8) * Which claim that John
was asleep was he
willing to discuss.
1
See Kayne (1976), Chomsky (1977), Chomsky and Lasnik (1977), Chomsky (1980),
Cinque (1981, 1982) and Pesetsky (1996). For a specific implementation of this analysis for
Spanish, see Arregi (1998), which is based on suggestions made in Brucart (1992).
4
Under the copy theory of movement, the wh-phrase in (8) leaves a copy in its
base position, as represented in (9). This copy contains a name c-commanded
by a co-referring pronoun, which causes a Condition C violation.
(9) [ Which claim that John
was asleep ]
was he
willing to discuss [ which claim that John
was asleep ]
✻
Although the copy theory of movement accounts for certain cases of recon-
struction in an elegant way, it also raises certain questions about the interpre-
tation of lower copies of moved items. Consider, for instance, the lower copies
left by wh-movement and QR in (10, 11) respectively.
(10) a. Which book did Mary read?
b. [ Which book ] did Mary read [ which book ]
✻
c. For which x, x a book, Mary read x.
(11) a. Mary read every book.
b. [ every book ] Mary read [ every book ]
✻
c. For all x, x a book, Mary read x.
Intuitively, the sentences in (10a, 11a) should have the meanings in (10c, 11c),
i.e. the lower copy of the moved item in (10b, 11b) should be interpreted as
a variable bound by the upper copy, not as a quantifier or wh-phrase. Thus,
we need a mechanism to interpret these lower copies as bound variables, or as
expressions containing a bound variable. In Fox (2000b), it is suggested that
these lower copies are in fact interpreted as bound definite descriptions. Thus,
the interpretation of (10, 11) above would be (12a, 12b), rather than (10c, 11c).
(12) a. For which x, x a book, Mary read the book x.
b. For all x, x a book, Mary read the book x.
In Elbourne (2000), a formal mechanism is given to achieve this result. This
analysis of the interpretation of lower copies of movement is couched within a
more general theory of definite articles and their relation to pronouns. He pro-
poses that the definite article takes two arguments, not just one as is standardly
assumed. Thus, a definite DP such as the book has the structure in (13).
(13)
the
1
D
NP
book
DP
5
In fact, Elbourne proposes that the lexical entry for the definite article is the
substructure headed by D in (13), which contains the and an index. The takes
two arguments, the first of which is that index. Informally speaking, this index
introduces a variable which can be bound from outside. Thus, the meaning of
a definite article under this view is that shown in (14).
(14) For any [ the i ] and assignment g,
[[ the i]] =
λf : f ∈ D
e,t
& there is exactly one x such that x = g(i)&f (x)=1.
the unique x such that x = g(i)&f (x)=1
(14) is basically the Fregean definite article, with the addition of a variable (g(i)
in 14) which can be bound from outside. The sister of this D, NP in (13), is the
outer argument of the definite article. The resulting DP is a definite description
containing a variable, i.e. the unique y such that y = g(1) and y is a book in
the example above.
Apart from allowing definite DPs to be bound (see Elbourne 2000 for de-
tails), this approach to definite articles allows us to formalize in a simple way
the mechanism suggested in Fox (2000b) for the interpretation of lower copies of
movement. Recall that what is needed is to interpret the lower copy of a moved
DP as a bound definite description. Thus, Elbourne proposes that, within the
lower copy of the DP, D is replaced with [ the i ] , whose meaning is (14), and
where i is the same index as the one on the upper copy.
2
Following Fox (2000a),
let us call this procedure Trace Conversion. Thus, after Trace Conversion, the
sentences in (10a, 11a) above, repeated here as (15a, 16a) are interpreted as in
(15c, 16c), respectively, which is the correct result.
(15) a. Which book did Mary read?
b. [ Which book ] did Mary read [ which book ]
✻
c. For which x, x a book, Mary read the unique y such that y = x &
y is a book.
(16) a. Mary read every book.
b. [ every book ] Mary read [ every book ]
✻
c. For all x, x a book, Mary read the unique y such that y = x & y is
abook.
As I show in §4 the copy theory of movement, together with the procedure
presented above for the interpretation of lower copies, provides a straightforward
analysis of relative clauses in Spanish. In particular, the fact that a lower
copy has to be interpreted as a bound definite description goes a long way in
2
More precisely, i has to be bound by the lambda abstract created by the upper copy after
movement.
6
explaining why definite articles can be used as relative operators. However,
before we proceed with the analysis of relative clauses in §4, we must first
consider the properties of total reconstruction in §3.2, which will also be an
important ingredient in the analysis.
3.2. Total Reconstruction
In the previous subsection, we discussed certain cases of movement in which
reconstruction was necessary in order to account for Condition C violations. As
shown there, this type of reconstruction is explained once we assume the copy
theory of movement. There are, however, other cases of reconstruction which
cannot be handled in the same way. Consider the example in (17).
(17) [An Austrian]
is likely to t
win the Gold medal.
In one of the possible readings of (17), the quantifier phrase an Austrian has
scope under likely. One way to obtain the correct LF for this interpretation is
to ‘reconstruct’ the whole phrase to its base position in the embedded clause.
This is what in the literature has been called ‘total’ or ‘scope’ reconstruction
(see May 1977, Fox 2000b, Sauerland and Elbourne 2000 and references cited
there).
Note that this case cannot be handled in the same way that the examples in
the previous subsection were. In this case, it is not enough to leave the upper
copy intact and apply Trace Conversion to the lower copy. That would still
assign the QP scope over likely. The upper copy has to disappear completely,
so that the QP is interpreted only in the base position. Following Fox (2000b)
(and others), I assume that this can be achieved by simply deleting the higher
copy at LF. Thus, (17) has the structure in (18b) at LF.
(18) a. [An Austrian] is likely to [an Austrian] win the Gold medal.
✻
b. is likely to [an Austrian] win the Gold medal.
However, this view of total reconstruction is too strong as it stands. In
particular, some conditions under which deletion of a copy is allowed must be
added. One such condition is that the two copies be identical. This might seem
like a painfully obvious observation, but is nevertheless necessary, especially
if one assumes operations like Trace Conversion, which alters the structure of
lower copies. Consider, for instance, the sentence in (19).
(19) John read every book.
If deletion of higher copies were not restricted in some way, we would predict
that (19) could mean John read the book, which, of course, is not the case.
Specifically, after applying QR to every book and applying Trace Conversion to
the lower copy, deletion of the upper copy would yield this wrong interpretation.
If, on the other hand, we add an identity condition, deletion in this case is not
7
possible, and the wrong reading is not predicted.
3
I conclude, then, that the
identity condition is necessary for deletion of upper copies of movement.
As we will see in §4, total reconstruction, together with the identity condi-
tion proposed above, are an important part of the analysis of relative clauses
defended here, in allowing us to explain why determiners like quantifiers cannot
be relative operators.
4. The Syntax and Semantics of Relative Clauses
As promised in the introduction, in this section I show that the independently
motivated assumptions about movement that were presented in §3 are all that
is needed for the correct analysis of Spanish relative clauses. Specifically, the
assumption that lower copies of movement are interpreted as bound definite
descriptions allows us to explain in an elegant way why definite articles can be
used as relative operators. More importantly, it allows us to explain why other
determiners, such as quantifiers, cannot have that use. Before I develop the
analysis in §4.2, in §4.1 I present some evidence that relative clauses in Spanish
have to be analyzed using the ‘head-raising’ analysis. Although this is not an
important part of the account offered in §4.2, there is nevertheless important
evidence for it in the literature, and I will therefore adopt it.
4.1. A Head-Raising Analysis of Relative Clauses
In a head-raising analysis, the nominal head of the DP containing the relative
clause is generated inside the relative clause, from where it moves to its external
surface position. Several arguments have been offered in the literature in favor of
this analysis (see, among others, Schachter 1973, Vergnaud 1974, Kayne 1994,
Sauerland 1998, Bhatt 1999, Hackl and Nissenbaum 1999). The arguments I
present below are adapted mainly from Bhatt (1999).
All the arguments have to do with cases in which the NP head needs to be
interpreted inside the relative clause. All these cases receive a natural explana-
tion in the head-internal analysis, but not in accounts in which the NP head is
not generated inside the relative clause.
The first argument derives from Condition A of binding theory. As shown
in (20), the reflexive pronoun s´ımismoinside the head NP can be bound by a
constituent inside the relative clause.
(20) el
the
retrato
portrait
de
of
s´ı mismo
himself
del
of-the
que
that
me
me
habl´o
talked
Juan
Juan
the portrait of himself Juan talked to me about
The possibility of scope reconstruction for numerals contained inside the
head NP provide a further argument. Consider (21).
3
Fox (2000b, p. 190) reaches the same conclusion by looking at the interaction between
movement and late insertion of adjuncts.
8
(21) Me
me
preocupa
worries
las
the
25
25
personas
persons
con
with
las
the
que
that
es
is
probable
probable
que
that
hablemos
we-talk
hoy.
today
I am worried about the 25 people it is probable we will talk to today.
A possible reading of (21) can be paraphrased as ‘I am worried about the fact
that it is probable that we will talk to 25 people today’. In this reading, the
numeral has scope under the modal probable.
4
Finally, another argument comes from possile readings that adjectives have
when they are inside the head NP of a relative clause. (22) contains two relevant
examples.
(22) a. la
the
primera
first
persona
person
con
with
la
the
que
that
Pedro
Pedro
dijo
said
que
that
hab´ıa
had
estado
been
hablando
talking
Juan.
Juan
the first person Pedro said Juan had been talking to
b. la
the
´unica
only
persona
person
con
with
la
the
que
that
Pedro
Pedro
dijo
said
que
that
hab´ıa
had
estado
been
hablando
talking
Juan.
Juan
the only person Pedro said Juan had been talking to
In both these sentences, the adjectives primera ‘first’ and ´unica ‘only’ can have
either a ‘high’ reading or a ‘low’ reading. Under a high reading, the DP in (22a)
refers to the first person mentioned by Pedro. Under a low reading, this DP
refers to the first person that, according to Pedro, Juan talked to. Similarly,
under the high reading, the DP in (22b) refers to the only person mentioned by
Pedro; under a low reading, it refers to the only person that, according to Juan,
Pedro talked to. The existence of the low reading provides evidence for the
head-internal analysis. The low reading under this analysis can be generated by
interpreting the head NP (including the adjective) inside the embedded sentence
in the relative clause, as shown in (23) (for details, see Bhatt 1999).
(23) the [
CP
Pedro said [
CP
Juan talked to first/only person ] ]
These three arguments argue for a head-raising analysis of Spanish relative
clauses. Note, in particular, that all the examples in (20-22) contain relative
clauses where the operator is the definite article. Therefore, I will adopt it in
the analysis presented in the next section. Note that there are several possible
analyses compatible with these arguments. What these arguments show is that
4
As pointed out by Bhatt (1999) it is not clear exactly how it is that in (21) the worry
is about the fact that we will talk to 25 people, not about the people themselves. Probably
related to this is the fact that, as shown in (21), the verb preocupar ‘worry’ has third person
singular agreement, rather than the expected third person plural.
9
the head NP is generated in an internal position. Under standard assumptions,
this NP, together with the relative operator, moves to [Spec,CP], as illustrated
in (24).
(24) [
DP
...[
CP
[ ... NP ...] [
C
...[ ... NP ...] ...] ] ]
✻
A different question is how the head NP gets to its CP-external position (if,
indeed, it does). In the following section, I concentrate mainly on the CP-
internal structure, and nothing hinges on this detail, so I will therefore remain
agnostic (but see Kayne 1994, Sauerland 1998, and Bhatt 1999 for recent
alternatives).
4.2. The Role of the Definite Article
Given that all the necessary assumptions presented in the previous sections have
been made explicit, we are now in a position to offer an explanation for the fact
that definite articles can be used as relative operators. Consider the relative
clause in (3a), repeated here as (25), and whose structure after movement to
[Spec,CP] is that shown in (26).
(25) el
the
hombre
man
con
with
el
the
que
that
estuvimos
we-were
hablando
talking
ayer
yesterday
the man we talked to yesterday
(26)
PP
con el hombre
hablando
VP
I
estuvimos
IP
pro
C λx.wetalkedtothemanx
que
CP
PP
con el hombre
✻
el hombre
DP
I adopt the fairly standard assumption that movement of an XP makes the sister
of the higher copy a predicate (see Heim and Kratzer 1998, Nissenbaum 1998).
Thus, movement of the PP to [Spec,CP] makes its sister, i.e.
C, a predicate,
10
i.e. λx.wetalkedtothemanx. Furthermore, this is precisely the interpre-
tation that is needed for the CP. As is standardly assumed, relative clauses are
interpreted as predicates. However, if the PP remains in that position at LF,
that would not be the denotation of the CP. Rather, it would have the meaning
of a proposition, which is not the right type for a relative clause. Therefore,
the upper copy of the PP must be deleted at LF. The only (semantic) role that
the movement has in relative clauses is to create the predicate, as is standardly
assumed (see Heim and Kratzer 1998).
Note that this deletion essentially results in total reconstruction of the moved
phrase (cf. §3.2). The only difference with the cases of total reconstruction we
saw in §3.2 is that, crucially, in relative clauses the predicate created after the
movement remains a predicate after deletion of the upper copy.
5, 6
The fact that in relative clauses reconstruction is obligatory explains why
definite articles, as opposed to other determiners, can be used as relative op-
erators. Note that in (26), it is not necessary to apply Trace Conversion (cf.
§3.1), since the moved phrase is already a definite description, i.e. the man x.
If, on the other hand, some other determiner, such as quantifiers like each or
a, were to be used, Trace Conversion would be necessary. Trace Conversion in
this case would in effect destroy the chain created by movement, since the two
copies would now be different. As shown in §3.2, total reconstruction is blocked
in this context, due to the identity condition. Since, as shown above, total re-
construction is obligatory in relative clauses, we obtain the desired result that
quantifiers cannot be used as relative operators.
7
One possible objection to this analysis might go as follows. In the example
above, the lower copy must be interpreted as a definite description. Since it is
the internal argument of hablar ‘talk’, it must be of type e. However, consider
the case in (27), in which the quantifier each is used as a relative operator, and
it is heading a DP in subject position.
8
(27) [
CP
[eachboy]λx[
C
that [ each boy ] read his
x
copy of El Quijote]]
✻
Since the lower copy is in subject position and QPs can be interpreted in this
position, one might argue that Trace Conversion in this case is not necessary in
this case. Thus, deletion of the upper copy would not be blocked by the identity
condition on total reconstruction. Furthermore, in this specific example, the
lambda abstract created by the movement is binding a variable, i.e. the one
5
On this difference between standard cases of total reconstruction and the one argued for
here, see §4.3
6
This means that whatever operation makes C a predicate must take place before deletion.
A straightforward way to implement this is to assume, following Heim and Kratzer (1998),
that, as part of the movement operation itself, an index is adjoined to
C which turns it into
apredicate.
7
This raises the obvious question of how we can account for other kinds of relative operators,
such as wh-words (cf. §1and§2). Although, at the moment, I do not have a full answer to
this question, see §5 for some suggestions.
8
For the purposes of the argument, I am ignoring the movement of the subject from the
VP-internal position.
11
introduced by the pronoun his, so we cannot appeal to some kind of prohibition
on vacuous binding. Thus, the argument would go, we seem to be predicting
that, at least in subject relativization, it is possible to use each as a relative
pronoun in certain cases.
The answer to this objection is obvious: Trace Conversion is always oblig-
atory (although vacuous in the case of definite descriptions). The trick that
apparently allowed to derive the previous example as a relative clause was to
not interpret the lower copy of the moved phrase as bound by the lambda ab-
stract created by the movement, and to let a pronoun be bound by the lambda
abstract. That this is indeed a trick is shown by the simple fact that movement
must always result in binding of the lower copy by the upper copy. This can be
seen in very simple examples of movement, as in (28).
(28) a. Each boy seems to have read his copy of El Quijote.
b. [ Each boy ] λx [ seems [ each boy ] to have read his
x
copy of EQ].
✻
c. For all x, x a boy, it seems that for all y, y aboy,y read x’s copy of
El Quijote.
In this example, each boy raises from the subject position of the embedded clause
to the subject position of the main clause. Trace Conversion does not apply,
so the lower copy is not interpreted as bound by the upper copy. Furthermore,
the lambda abstract created by the movement binds the pronoun his in the
embedded clause. This derivation for (28a) is represented in (28b), and would
yield the meaning in (28c), which is paraphrasable by every boy seems to have
read every boy’s copy of El Quijote. This is obviously the wrong result. The
only way to block this derivation is to apply Trace Conversion obligatorily, so
that the lower copy is interpreted as bound by the upper copy. Therefore, the
possible objection mentioned above rests on obviously false assumptions.
It is important to stress that the crucial difference between the definite ar-
ticle and other determiners is precisely its definitiness. In relative clauses, only
the lower copy is interpreted, and lower copies must be interpreted as defi-
nite descriptions. Note also that all the independently motivated assumptions
presented in §3 are crucial in the argument. In particular, the copy theory of
movement forces us to have the rule of Trace Conversion, which, due to the iden-
tity condition on deletion, results in the impossibility of total reconstruction if a
definite article is not used. Therefore, these assumptions find additional support
by the facts studied here, since they allow us to explain the facts without added
stipulations.
In the following subsection, I discuss certain problems that the analysis of
total reconstruction assumed here might have. Specifically, I review the argu-
ments found in Sauerland and Elbourne (2000) in favor of accounting for total
reconstruction phenomena in terms of PF-movement. As I show below, these
arguments do not apply for total reconstruction in relative clauses, which in fact
makes the claims made in Sauerland and Elbourne (2000) perfectly compatible
with the analysis defended here.
12
4.3. On Total Reconstruction in Relative Clauses
As discussed in §3.2, I assume that total reconstruction can be achieved by
deleting the higher copy of a moved item. Furthermore, as shown in §4.2,
this assumption is crucial in the analysis of relative clauses presented there.
However, Sauerland and Elbourne (2000) (henceforth S&E), drawing on total
reconstruction phenomena found in different cases of A-movement, argue that
total reconstruction should be viewed as the result of PF-movement.
In order to exemplify their analysis, consider example (17) from §3.2, re-
peated below as (29).
(29) [An Austrian]
is likely to t
win the Gold medal.
As was noted in §3.2, (29) has a reading in which the raised subject has scope
under likely. This reading is possible if the raised subject is interpreted only in
its base position, as represented in (30b).
(30) a. [An Austrian] is likely to [an Austrian] win the Gold medal.
✻
b. is likely to [an Austrian] win the Gold medal.
S&E argue that (30b) is the LF structure of (29) because the movement occurs
at PF and therefore does not feed semantic interpretation.
S&E give convincing arguments that this is the right approach to total recon-
struction phenomena in A-movement.
9
Unfortunately, the empirical arguments
presented there are not applicable to relative clauses. One of the main argu-
ments that they give is that the PF-movement theory of total reconstruction
derives Barss’ Generalization (see Barss 1986), according to which total recon-
struction of a QP to a position X is blocked when the QP does not c-command
X in the overt form. In the structures relevant to this generalization, a phrase
containing the lower copy of the raised QP moves to a position above the QP.
In these cases, total reconstruction is not possible. Obviously, Barss’ general-
ization is not testable in relative clauses, since it would involve extraction of the
consituent containing the lower copy of the relativized XP outside the relative
clause island.
10
S&E also give a conceptual argument against the LF-deletion analysis of
total reconstruction. They observe that the assumption that movement makes
the sister of the higher copy a predicate poses a problem for the LF-deletion
theory. This is because deletion of the higher copy of a moved item would
leave the lambda abstract intact, which would make the wrong prediction that,
9
Specifically, they offer three arguments, based on the interaction between raising and wh-
movement in English, facts from agreement with group terms in British English, and multiple
scrambling in Japanese
10
The other empirical arguments presented by S&E are also irrelevant for relative clauses.
One of these arguments involves verbal agreement in number with the raised XP in subject
position. The other argument involves Japanese sentences in which two instances of scrambling
target multiple specifiers of the same head. As far as I can tell, these arguments are not testable
in (Spanish) relative clauses.
13
for instance, the sentence in (29) above could be interpreted as a predicate,
rather than a proposition. Although this argument is relevant for other cases
of movement, it is clearly not so for relative clauses. As we saw in §4.2, in
relative clauses, total reconstruction must in fact leave the predicate created
by movement intact, since relative clauses are interpreted as predicates, not
propositions. Furthermore, this kind of reconstruction cannot be achieved via
PF-movement, since this kind of movement has no effect on interpretation. This
is the crucial factor that distinguishes total reconstruction in relative clauses
from other cases of total reconstruction.
To conclude, even if we accept the PF-movement analysis for the cases of
total reconstruction examined by S&E, the arguments presented by them are
not relevant in relative clauses, which makes their proposal compatible with the
analysis of relative clauses defended here.
5. Conclusion
In the previous sections, I have offered an explanation for the fact that definite
articles can be used as relative operators. Given that the analysis explains
the facts, it is to be preferred to alternative accounts in which they are due
to accidental homophony. Furthermore, the analysis derives the facts directly
from the copy theory of movement, together with certain assumptions that are
necessary within this theory. Thus, the analysis defended here can be seen as
offering further support for the copy theory of movement.
Although the analysis explains why definite articles can be used as relative
operators in Spanish (and other languages, like German), there are several ques-
tions that still remain open. As we saw in §2, definite articles are not the only
type of relative operators in Spanish. These other relative operators are some
wh-words, the definite article followed by cual ‘which’, and cuyo ‘whose’, which
can only function as a relative operator. As for el cual, the fact that it contains
a definite article is, of course, predicted by the analysis defended here. The fact
that cual can also be used as a question wh-word does not seem to be particu-
larly problematic, since these two cual’s have different syntactic properties: in
relative clauses, it seems to behave like a noun, since it is preceded by a definite
article; in questions, it is not preceded by an article, and seems to behave like a
determiner. Thus, I tentatively assume that cual in relative clauses is simply a
semantically empty noun.
The other types of relative operators seem more problematic. The analysis
presented here seems to make the strong prediction that only definite articles
should be used as relative operators. However, this is not the case. What the
analysis predicts is that, if the relative operator has other uses, the DP headed
by the relative operator can be interpreted as a (bound) definite description in
those other uses. Thus, the analysis is not contradicted by operators like cuyo
‘whose’, which can only be used as a relative operator, since there is no reason
to suppose that a DP headed by it is not definite. As far as I can tell, this
argument also applies to relative operators in other languages which have no
14
other use.
This brings us to the more serious problem of explaining the use of wh-words
as relative operators, especially given the fact that their use as relative operators
is also possible in other languages like, for instance, English. Under standard as-
sumptions, wh-words are interpreted as quantifiers, which the analysis defended
here predicts should not be able to function as relative operators. One possible
solution is to assume, contrary to the standard assumption, that wh-words, or,
at least the ones that can be used as relative operators, are definite determiners.
In fact, this is precisely what is suggested for wh-words like which in Rullmann
and Beck (1998). Although this seems like a promising approach, more work is
needed in this area, and I leave it as a question for further research.
Finally, I would like to mention languages that use personal pronouns as rel-
ative operators. One such language is Modern Hebrew. Probably, the solution
to this problem will come from a better understanding of the relation between
personal pronouns and definite articles. It has often been proposed in the litera-
ture that personal pronouns and definite articles are in fact the same lexical item
(or, at least, different ‘varieties’ of the same lexical item; see Elbourne 2000
and references cited there). It seems that the right path to follow in this case is
to investigate the differences in distribution of personal pronouns and definite
articles in both Spanish and Hebrew, and use this to find an explanation for the
difference between these two languages in their choice of relative operators.
15
References
Arregi, K. (1998) “Spanish el que Relative Clauses and the Doubly Filled COMP
Filter,” ms., Massachusetts Institute of Technology.
Barss, A. (1986) Chains and Anaphoric Dependence. On Reconstruction and Its
Implications, Doctoral dissertation, Massachusetts Institute of Technolog.
Bhatt, R. (1999) Covert Modality in Non-finite Contexts, Doctoral dissertation,
University of Pennsylvania.
Brucart, J. (1992) “Some Asymmetries in the Functioning of Relative Pronouns
in Spanish,” Catalan Working Papers in Linguistics 1992, 113–143.
Chomsky, N. (1977) “On Wh-movement,” in P. Culicover, T. Wasaw, and A.
Akmajian, eds., Formal Syntax, Academic Press, New York, 71–132.
Chomsky, N. (1980) “On Binding,” Linguistic Inquiry 11:1, 1–46.
Chomsky, N. (1993) “A Minimalist Program for Linguistic Theory,” in K. Hale
and S. J. Keyser, eds., The View from Building 20: Essays in Honor of Sylvain
Bromberger, MIT Press, Cambridge, MA, 1–52.
Chomsky, N., and H. Lasnik (1977) “Filters and Control,” Linguistic Inquiry
8:3, 425–504.
Cinque, G. (1981) “On Keenan and Comrie’s Primary Relativization Con-
straint,” Linguistic Inquiry 12:2, 293–308.
Cinque, G. (1982) “On the Theory of Relative Clauses and Markedness,” The
Linguistic Review 1:3, 247–294.
Elbourne, P. (2000) “Pronouns and Definite Articles,” ms., Massachusetts In-
stitute of Technology.
Fox, D. (2000a) “Antecedent Contained Deletion and the Copy Theory of Move-
ment,” ms., Harvard University.
Fox, D. (2000b) Economy and Semantic Interpretation, MIT Press and MIT
Working Papers in Linguistics, Cambridge, MA.
Hackl, M., and J. Nissenbaum (1999) “Variable Modal Force in For-infinitival
Relative Clauses,” in A. Giorgi, J. Higginbotham, and F. Pianesi, eds., The
Proceedings of the Bergamo Conference on the Syntax and Semantics of Tense
and Mood, Oxford, Oxoford University Press.
Heim, I., and A. Kratzer (1998) Semantics in Generative Grammar, Blackwell,
Oxford.
Kayne, R. S. (1976) “French Relative Que
,” in M. Luj´an and F. Hensey, eds.,
Current Studies in Romance Linguistics, Georgetown University Press, Wash-
ington D.C., 255–299.
16
Kayne, R. S. (1994) The Antisymmetry of Syntax, MIT Press, Cambridge, MA.
May, R. (1977) The Grammar of Quantification, Doctoral dissertation, Mas-
sachusetts Institute of Technology.
Nissenbaum, J. (1998) “Movement and Derived Predicates: Evidence From
Parasitic Gaps,” in U. Sauerland and O. Percus, eds., The Interpretive Tract,
volume 25 of MIT Working Papers in Linguistics, MITWPL, Cambridge, MA,
147–295.
Pesetsky, D. (1996) “Some Optimality Principles of Sentence Pronunciation,”
in P. Barbosa, D. Fox, P. Hagstrom, M. McGinnis, and D. Pesetsky, eds., Is
the Best Good Enough?, MIT Press and MIT Working Papers in Linsuistics,
Cambridge, MA, 337–383.
Rullmann, H., and S. Beck (1998) “Reconstruction and the Interpretation of
Which-Phrases,” in G. Katz, S.-S. Kim, and H. Winhart, eds., Reconstruc-
tion: Proceedings of the 1997 T¨ubingen Workshop, Universit¨at Stutgart and
Universit¨at T¨ubingen, 223–256.
Sauerland, U. (1998) The Meaning of Chains, Doctoral dissertation, Mas-
sachusetts Institute of Technology.
Sauerland, U., and P. Elbourne (2000) “Total Reconstruction, PF-Movement,
and Derivational Order,” ms., T¨ubingen University and Massachusetts Insti-
tute of Technology. To appear in Linguistic Inquiry.
Schachter, P. (1973) “Focus and Relativization,” Language 49:1, 19–46.
Vergnaud, J.-R. (1974) French Relative Clauses, Doctoral dissertation, Mas-
sachusetts Institute of Technology.
Wiltschko, M. (1998) “On the Syntax and Semantics of (Relative) Pronouns
and Determiners,” Journal of Comparative Germanic Linguistics 2, 143–181.
17
