clac 9/2002



Marta Luján


The University of Texas at Austin



1    Introduction

This paper challenges the notion that the structure of the Determiner Phrase (DP) contains a virtual category which answers for DPs’ referential and anaphoric uses. Some current DP proposals, for instance, assume a null pronoun to account for the nominalized Adj(ective) in Romance, while others in Germanic postulate an abstract operator (or demonstrative) to derive the specificity of definite DPs. Such accounts strip the D(eterminer) of denotative function, turning it into a mere locus for Agreement and Case features. Moreover, they run afoul of an important notion which laid the foundation for the DP analysis. This is the hypothesis that D is a variant form of P(ronoun). For instance, Postal (1969) equates the definite D with the third-person P, as do earlier grammarians like Bello (1847) and Jespersen (1924), who further identify it as a Dem(onstrative). If these authors are on the right track, it would appear that the definite D already has the attributes that the null categories in the current DP proposals are trying to capture.

Following the lead of the earlier authors, I argue that the reference of various uses and interpretations of nominals may be straightforwardly derived, if Ds are viewed as modified Ps. In the DP analysis, D is a functional head modified by a clause or predicate, while P is a D with no complement. On such a structural basis, the reference of definite vs. indefinite DPs may be traced to the different discourse functions of two types of D, by assuming that the definite D is a form of the third-person P, and the indefinite D is a variant of the numeral one (Perlmutter 1970). In contrast to competing analyses, I locate DP’s referential functions directly on its D. On such a basis, Ds may be semantically described as discourse-linking functions. Adapting from Enç’s (1991) theory of specificity, they relate DP denotations to discourse antecedents by the inclusion relation, a subset of which is the identity relation, as commonly found with bare Ps.

I develop a binding approach to DPs’ referential properties which gives a novel view of the nominalized Adj in Romance and elucidates the antecedent-pronoun relation, the discourse functions of different Ds, and the disjoint and bound interpretations of DPs with common Ns. I show that the D type plus its restrictive vs. nonrestrictive modification determine whether the binding relation denotes identity of entity or of class, and why obviation results with common Ns, but not with the nominalized Adj. The analysis of Ds as modified Ps assimilates the disjointness of identical clausemate common noun DPs to the obviation of clausemate bare Ps. Moreover, it treats the specificity of DPs with common Ns as a binding fact, parallel to the contextually nominalized Adj in Romance. The definite Ds picking out a salient or a reconstructible discourse antecedent follows since in this account Ds has the same binding function as bare Ps.


2    D-headed structures

In the latest DP analysis of nominals, D is a functional head modified by a clause or predicate (Kayne 1994):


(1)        [DP  D  -  CP  ]    (where CP may be a Small Clause)


In this structure CP is a relative clause, complement to D. Such a clause may lack Tense, as a defective or Small Clause (SC). The common N and the Adj proceed from the predicate nominal of a SC (Holmberg 1993, Campbell 1996).

The structure in (1) readily fits restrictive relatives that freely occur with Ds in Spanish, in alternation with the strong forms corresponding to lexical Ps, which co-occur with appositive relatives, as in:


(2)        a. lo /*ello   que  agrada  seduce          ‘that which pleases seduces’

                 it             that  pleases seduces

            b. ello /*lo, que  agrada,  seduce         ‘that, which pleases, seduces’

                 it             that  pleases  seduces

(3)        a. las /*ellas   que  te  conocen            ‘the (fem) ones who know you’

                 the-fem/pl that you know

            b. ellas /*las, que  te  conocen,            ‘they (fem), who know you,’

                 they-fem   that you know


Although, Ds and lexical Ps look very much alike in Spanish, they are usually seen as separate categories. In contrast, Bello (1847), who points out the above facts, argues for the identity of the definite D and the third-person P, and for the analysis of such nominals as P-headed structures, where D=P.1 In addition, he claims a related structure for the so-called ‘nominalized’ Adj, as in:


(4)        a. lo   bueno                                         ‘that which is good’

                 it   good

            b. las            más  bonitas                    ‘the (fem) prettier (ones)’

                 they-fem more pretty-fem/pl

            c. los              buenos                           ‘the (masc) good (ones)’

                 they-masc good-masc/pl


Restrictive relatives with Ds or personal Ps are also known to occur in other languages in a variety of degrees. For instance, Hestvik (1992) reports that Norwegian Ps take restrictive modifiers, as shown in (5), and cites similar, though more constrained, data from English by Postal (1969), as reproduced in (6):


(5)        a. han med rød hatt                              ‘the (masc) one with red hat’

                 he  with red hat

            b. han uten      hår  på  hodet               ‘the (masc) one without hair’

                 he  without hair on  the-head

c. han some går     der                         ‘the (masc) one who walks there’

     he  who  walks there

(6)        a. you men who wish to escape

            b. we who are opposing Fascism

            c. You who wish to survive had better shape up.


According to the DP analysis, a P must be a bare D, i. e. one with no complement. Radford (1997: 153) makes this point, arguing that Ds are much like Ps with modifiers on the basis of the data reproduced in (7).2 He also claims that bare nominals, as in (7c), have an abstract they heading their phrase (cf. §4.2):


(7)        a. We syntacticians take ourselves/*yourselves/*themselves too

                 seriously, don’t we/*you/*they?

            b. You syntacticians take yourselves/*ourselves/*themselves too

                 seriously, don’t you/*we/*they?

c. Syntacticians take themselves/*ourselves/*yourselves too seriously,

     don’t they/*we/*you?


Noticeably, however, in contrast to Ps and Ds, the demonstratives that and those lend themselves more readily to restrictive modification by relatives:


(8)               a. that which pleases

            b. those who work hard

            c. those we bought


The sequences in (8) may be filling the gap for the paucity of D with restrictive relatives. Quite likely, the lack of inflection on D disallows such DPs in English. However, since the may be considered a reduced form of that (Jespersen 1924), and this may be related to it, a form of the third-person P, it becomes possible to class them together. The analysis in this paper then extends to English and languages with similar restrictions on D/P, as becomes evident later on.


3    Structural hypothesis

On the basis of the structure in (1), and adapting from Campbell’s (1996) account of DP with common Ns, I assume the structure for DP with Adj or N shown in (9). As indicated in (a), the lexical D (=P) ellas originates in the Spec of a SC which contains a predicate nominal (Adj or N), and it raises to an empty D-head position, as shown in (b): (I give the full P-form with the underlined portion in bold type to indicate the corresponding clitic D form)


(9)        a. [DP    D         [SC ellas  {pequeñas /  niñas } ]]   las pequeñas/ las niñas

            b. [DP    ellas     [SC     t    {pequeñas /  niñas } ]]    ‘the small ones/ the girls’

                        they-fem            small-fem/pl  girls


This movement creates an operator-variable structure, and provides the trace needed for the interpretation of D/P as a variable.3 DP’s internal predication is expressed by the SC with the predicate nominal.4 The structure for DP with a full relative is similar, but with D moving through Spec of C, as indicated by the trace t’ in (10): (See Appendix for the derivation of other types of DP.)


(10)      a. [DP --  [D’    ello        [CP t’  que  tú   sabes  t   ]]      lo que tú sabes

                                 it                     that  you know              ‘that which you know’

            b. [DP --  [D’    ellos      [CP t’  que  tenemos  t   ]]        los que tenemos

                                 they-masc       that  we-have                ‘the ones we have’


In these structures, the lexical P in Spec of SC or CP does not lose its referential and binding functions once it moves to D, where it cliticizes. Rather, D keeps the same denotative capacity as P. This notion is contrary to the view of D as a purely functional head, or mere repository of DP’s Agreement and Case features, as commonly found in current DP analyses. For instance, Campbell (1996), adapting from Holmberg (1993), postulates an abstract specificity operator in Spec of DP, which binds a variable in Spec of SC, as in:


(11)      [DP  Opi  [D’  the  [SC ei  book   ]]]      (where Op= Demonstrative)


Obviously, such specificity operator, which Holmberg equates with a null Dem, derives a DP’s specificity of reference.5 Thus, the determiner the lacks such referential function in the structure in (11). It has, instead, an exclusively formal role; namely, to license by formal feature checking the null specificity operator in its Spec. However, it is unclear what features may be involved, since the is an uninflected, invariant form. Moreover, if it is a deixis feature of Dem which the also has, then there is little reason for the presence of the special Op in (11).

A similar notion about the presence of a virtual category prevails in DP analyses formulated to account for the nominalized Adj in Romance, as in the ambiguous Spanish nominals in (12). I combine two common proposals in (13):6


(12)      los ricos  =  los {hombres/ objetos} ricos     ‘the rich {men/ objects} ones’

            el azul     =  el {color/ objeto} azul               ‘the blue {color/ object} one’

            la bella    =  la {mujer/ cosa} bella               ‘the beautiful {female/ thing}’


(13)      [DP   la     [NP  { pro / ØNpequeña  ]]]       la pequeña

                  the-fem/sg                     small                 ‘the small (one)/ the little girl’


The virtual N and pro would account for the multiple interpretations that attach to the nominalized Adjs in (12), while the definite D is, again, responsible only for licensing the null element in the structure in (13). However, since Adjs and Ns show the same inflections as Ds in Spanish, it is unclear why the predicate Adj in SC does not suffice to legitimize the virtual category in its Spec.7

By contrast, our structures in (9)-(10) present none of the problems that relate to assuming a separate virtual category. In them, the definite D, being a P, is referential, and may denote specificity of reference. The only virtual category in DP is D’s variable-trace in Spec of SC/CP, which is licensed by its movement into D. Thus, DP’s specificity and its ambiguity with the bare Adj in Romance are predictably expected from the meaning of D as a form of the third-person P.


4    Anaphoric and nonanaphoric uses of Ds

We now account for the so-called ‘contextually nominalized’ Adj in Romance in terms of the antecedent-pronoun relation, consistent with the structural hypothesis in §3, and maintaining the definite D’s denotative capacity as a P. The approach leads to aligning the D+Adj nominals in Romance with the anaphoric (or bound) uses of definite DPs, and opposite to the obviation of identical clausemate DPs and bare Ps. It will be seen that the contextually nominalized Adj in Romance and their unified treatment with data previously thought unrelated, considerably extend the area of binding phenomena, while bringing the antecedent-pronoun relation into a different, new light. In addition, the unbound uses of DPs in general statements may be explained without resorting to genericity operators or predicates, and simply on the basis of the meaning of the personal P (=Dem), and the numeral one which underlie the definite and the indefinite Ds, respectively.


4.1    Nominalized Adjs and the specificity property

Our structural hypothesis readily accommodates the contextually nominalized Adj in Romance, which involves relating a D+Adj phrase to a contextual antecedent. To illustrate, the sentence in (14) allows the two distinct readings, synonymous with the paraphrases in (a)-(b), since the underlined DP with the Adj is expectedly ambiguous. It may be interpreted as bound to an agreeing antecedent DP, as in the reading in (14a), or it may be unbound, as in the interpretation in (14b):


(14)      Este libro   y    el  vecino       se  contradicen.

            this  book and the neighboring Refl-contradict

            (a) Este libro y el libro vecino se contradicen.

                  ‘This book and the neighboring book contradict each other.’

            (b) Este libro y el hombre vecino se contradicen.

                  ‘This book and the male neighbor contradict each other.’


On the premise that the definite D is identified as the third-person P, these data must fall under binding. Thus, the denotation libro ‘book’ of el vecino ‘the neighboring (one)’ in the interpretation (14a) is easily explained as deriving from the D (=P) el being bound by este libro ‘this book’. By contrast, in the reading (14b) el is interpreted as a free or unbound P, and denotes a male neighbor. The latter reading is that of the third-person masc/sg P as restrictively modified by vecino, as in el (que es) vecino ‘he (who is a) neighbor’; hence, the presence of a virtual N (e. g. hombre) or pro is redundant and inconsistent with our analysis.

Noticeably, the antecedent-pronoun relation of Ds with attributive Adjs differs from the usual binding of bare Ps in one important respect: it does not involve identity of individual referent, but only of class. Observe that in (14) the modified P (=D) refers to the same class of objects denoted by the antecedent, but not to the same entity, a fact which is expected, since the Adj vecino modifying el adds restriction to the class denoted by the antecedent este libro ‘this book’.

Consequently, the analysis of the nominalized Adj in our terms leads to broadening the antecedent-pronoun (or binding) relation so as to allow it to denote identity of class rather than of individual referent, as so far assumed. Accordingly, binding may relate DPs that denote distinct referents but that belong to (different subsets of) the same class, as with the contextually nominalized Adj. This result is very interesting, and it has important implications for understanding the function of binding in discourse. In addition, it points to the relevance of restrictive vs. nonrestrictive modification for determining the outcome of the binding function of Ds. Evidently, the contextually nominalized Adj under the terms of our account has immediate consequences for analyzing other types of bound DPs.

The account immediately extends to the common anaphoric use of definite DPs found across languages, which is described as ‘specificity’ by some authors, or as ‘uniqueness’ by those using the Russellian terms.8 I illustrate with the parallel examples from Spanish and English in (15). Such a referential property derives in our account from the binding capacity of the definite D as a modified P:


(15)      a. Un hombre apareció en la entrada. El hombre parecía agobiado.

            b. A man appeared at the entrance. The man seemed overburdened.


In this area of overlap between the two languages, the predicate nominal modifying the anaphoric definite D in (15) works as a nonrestrictive modifier because it is identical to the predicate nominal in the antecedent DP. And given that it does not add restriction to the class denoted by the antecedent, the binding relation gives not only identity of class, but also identity of individual entities. In examples of this sort, therefore, the anaphoric DPs are equivalent to the corresponding unmodified Ds (=Ps) pro/ he.9 The argument is of course valid for English, if the determiner the is a reduced form of the Dem that (Jespersen 1924).

A similar account may be given to other common uses of the definite D which are univocally interpreted in relation to a given domain of discourse. Some illustrative examples from Spanish are given in (16) with the corresponding glosses exemplifying the parallel data in English:10


(16)      a. No hables con Juan, pues el tonto no puede guardar un secreto.

               ‘Don’t talk with Juan, as the fool cannot keep a secret.’

            b. Su novela es entretenida, aunque el argumento es trillado.

               His novel is entertaining, though the plot is commonplace.’

            c. No aceptaron el ensayo que escribió Ernesto.

               ‘They didn’t accept the essay that Ernesto wrote.’


In these sentences the anaphoric or bound DPs select a referent that must be understood in relation to either a previously mentioned entity, or one that may be deduced from an already established discourse referent. Thus, in (16a) the epithetic Adjs tonto/ fool work as nonrestrictive modifiers to el/ the (=that), respectively. Since the Adjs do not restrict the bound denotation of the Ds, these may select the previously mentioned Juan as their referent. On the other hand, in (16b) and (16c) the same Ds respectively denote un argumento/ a plot, which the novel talked about in (16b) must have, and un ensayo/ an essay which Ernesto is said to have written in the restrictive relative in (16c).

It is worthwhile to notice that an elliptical interpretation based on the presence of a null N or pro in DP structure, as is commonly assumed for the D+Adj nominals in Romance, does not allow their unified treatment with the anaphorically interpreted DPs found in most languages. Such an analysis simply does not extend to the anaphoric uses of DPs in (15)-(16), which fall instead under studies of specificity. Ellipsis affords a much narrower perspective, one in which these sets of data are completely unrelated. In addition, ellipsis is not relevant for explaining the disjoint interpretation of identical clausemate DPs and bare Ps. By contrast, under our binding perspective such obviation data find a common ground, and stand opposite to the bound uses of DPs, as we see next.


4.2    Obviation of clausemate DPs and bare Ps

Our unified binding account of the nominalized Adj and the anaphoric definite DPs puts together the typical obviation of identical clausemate DPs, which has so far defied explanation, with the obviation of agreeing bare Ps thought to be the result of the binding Principle B. This dictates that a P be free in its minimal governing category, thus forbidding it to be bound to a clausemate antecedent.11 In our perspective such DPs, as those illustrated and shown in bold type in (17), must be interpreted with disjoint reference for exactly the same reason that excludes binding of identical clausemate Ps, as in (18):


(17)      a. Los niños atacaron a los niños.                   Los niños ≠ los niños

            b. The boys attacked the boys.                        The boys ≠ the boys

(18)      a. Ellos los atacaron a ellos.                             Ellos ≠ ellos

            b. They attacked them.                                    They ≠ them


Clearly, whether bare or modified, the definite D of the identical DPs in (17) would select the same referent as the antecedent subject were it not for a clausemate restriction such as that in Principle B. By contrast, the binding relation is possible with the contextually nominalized Adj in a simple clause, where it does not lead to identity of individual referent, e. g.


(19)      a. Esta parte precede a la anterior.                              la = la parte

                 ‘This part precedes the previous one.’

            b. Los tonos claros sustituyen a los oscuros.               los = los tonos

                 ‘(The) light tones substitute (the) dark ones.’


Nevertheless, precisely because binding is possible with the Adjs in (19), Principle B cannot hold; its implied clausemate prohibition runs contrary to our revised binding notions. Our earlier conclusion about binding not requiring identity of individual referents also conflicts with Principle B. In fact, once the requirement of individual coreference and the restriction about clausematehood disappear, Principle B is untenable. Consequently, the obviation of identical clausemate Ps and common noun DPs must have some other explanation.

However, no new principle should be needed to account for the data in question. In fact, we claim that the obviation of identical clausemate Ps must be the result of restricting individual coreference to clausemate Ps that are reflexive in form. In other words, the interpretation of identical clausemate Ds (=Ps) that are nonreflexive must be the effect of a requirement such as that in Principle A, which dictates that reflexive Ps must be bound in their minimal governing category. Consequently, clausemate Ds and Ps which do not have the appropriate reflexive form, such as those in (17)-(18), cannot be interpreted as coreferential.

In conclusion, these data corroborate our interpretation of the binding function of the modified definite D (=P), while they also lead to a reassessment of the antecedent-pronoun relation and the binding theory, in particular, Principle B. This principle conflicts with our analysis, since it arbitrarily excludes binding of identical clausemate Ps and common noun DPs, but not of Ds with modifying Adjs. This consequence is not surprising: our interpretation of the facts in this section shows that the antecedent-pronoun relation does not exclusively denote identity of individual referent (or individual coreference). Accordingly, the binding relation must be more broadly defined, in terms of discourse structure or discourse coherence, and as denoting merely class inclusion, a subset of which is the identity relation. Structurally, the relation minimally requires Person and Number (and sometimes Gender) agreement of the DPs bound by it.


4.3    Generic nominals

In our perspective, the generic sense of certain DPs must be a semantic property of their D, much as the specificity of reference is a property of the anaphoric D. If the definite D preserves its full denotation as a form of the personal P and Dem, then there is no need for abstract ‘genericity’ operators or predicates.12 Such categories are redundant, much as the specificity operator previously dismissed. Our analysis of Ds as modified Ps, is consistent with Radford’s (1997) hypothesis that English bare plural DPs are headed by an abstract they. Interestingly, such nominals correspond to Spanish generic DPs with the definite D in its plural form:


(20)      a. Syntacticians take themselves/*ourselves/*yourselves too seriously,

     don’t they/*we/*you?

b. Los sintaxistas se toman muy en serio, no?


Our structural hypothesis favors the occurrence of they in the structure of bare plural DPs, and the use of the definite D in comparable Spanish generic plural DPs, as in (20b), would corroborate the hypothesis. In addition, other uses of the plural third-person P in both languages support the idea that ‘genericity’ derives from the meaning of this P. For instance, in both languages the weak form (in Spanish a null pro) may be used in sentences that are traditionally labelled ‘impersonal’, for they do not refer to any specific person, e. g.13


(21)      a. They show foreign movies in that theater.

b. pro Pasan películas extranjeras en ese cine.


It is precisely because they lack specificity of reference that bare plurals may occur in existential sentences with there, along with the Dem and other forms of the personal P used in a general sense, as exemplified in (22):


(22)      a. There are syntacticians who take themselves too seriously.

            b. There are those who would believe anything.

            c. There’s the perfect solution for every problem.

            d. There’s always him/ her at the lead.


However, generic sentences may be expressed with singular DPs as well, and with either the definite and the indefinite Ds, as shown in (23). This is to be expected in our analysis, since such sentences are also possible with the numeral one, as well as with the 3rd- and 2nd-person Ps or the Dem, all of which may be bare or modified, as in the English examples in (24), and the Spanish ones in (25):


(23)      a. A beaver builds dams.

b. The lion is carnivorous.

(24)      a. One is always bound to make mistakes.

b. He who works hard is likely to succeed.

c. { Those/ You } who wish to survive had better shape up.

(25)      a. Uno siempre se equivoca.     ‘One always makes mistakes’

                 one   always  Refl-errs

b. Los   que  esperan  desesperan.      ‘Those who wait despair’

     they  that  wait      despair

c. Abriga el pellejo, si  quieres    llegar   a  viejo.     ‘Cover your skin, if you

     cover   the skin,   if  you-want  to-get-to old          want to reach old age.’


Then, the generic interpretation arises because the personal P and the numeral one may refer to all the elements in a class, or to a prototype, a sample, or any free-choice member therein, in order to denote its ‘kind’. 14 It is no accident that it is usually the subject of generic statements that is responsible for their sense, since the subject is either a bare P or a DP with a modified P. If the personal P and the numeral one may denote ‘kinds’, and Ds are but modified Ps, then it is unnecessary or redundant to appeal to specialized operators or predicates to derive the property of genericity.14


5    Binding as discourse linking

The analysis of Ds as modified Ps leads to a new version of the binding relation that is broader in its denotation, but minimally constrained by agreement. As the locus of DPs’ referential function, Ds may be seen as discourse-linking functions. They relate DP denotations to discourse antecedents by the inclusion relation, a subset of which is the identity relation. The analysis extends to other Ds, in particular, the indefinite D, seen as the numeral one, and to all other quantifier Ds. Although sharing in spirit with Enç (1991), our approach shows her partitive structures to be superfluous to account for the meaning of linked Ds.


5.1    Ds as discourse-linking functions

The arguments previously made are universally valid, even when the contextually nominalized Adj is not universally found across languages. However, it seems undeniable that the interpretation of the Adj in relation to a contextual antecedent is universal, regardless of the structural details of the DP containing the Adj. This is already illustrated for English with the glosses of the examples cited above. In addition, it may be seen that the contextually interpreted Adj is not restricted to the definite D, and neither is the antecedent-pronoun relation. The anaphoric relation is indeed possible with the indefinite bare P, as shown in (26), as well as with the indefinite P modified by a restrictive Adj, as in (27):


(26)      a. Vestí una falda larga y Eva también vistió una.         una=una falda larga

            b. I wore a long skirt, and Eva wore one, too.               one=a long skirt

(27)      a. Liz tenía dos rosas rojas y Bill le dio una amarilla.            una=una rosa

            b. Liz had two red roses, and Bill gave her a yellow one.       a one=a rose


Undoubtedly, in these examples the indefinite D is in a binding relation with the underlined DPs in the first conjunct clause, its various forms being interpreted in relation to such agreeing antecedents. In (26) the bare one/ una selects the same class and subclass as denoted by the restrictively modified skirt/ falda in the antecedent. But in (27) the restrictively modified D a/ una, as the numeral one/ un- (Perlmutter 1970), selects the same general class, but not the subclass red/ rojas, denoted by the antecedent. Rather, it sets up a contrast in relation to that subclass on the basis of its restrictive modifier yellow/ amarilla.

However, in the two sets of examples, although one/una is bound, and selects the same class as that of an antecedent, it must refer to a new entity, distinct from the antecedent. This follows from the meaning of the indefinite D as a numeral, and is consistent with the denotation of binding as class inclusion, rather than individual coreference. The function of a numeral is to ‘enumerate’, as if specifying the elements in a list, where each must be distinct from the others. Consequently, the discourse-linking function of this D is that of introducing new referents in the discourse domain, in contrast to the function of the definite D. Accordingly, the indefinite D must be nondemonstrative, opposite to the definite D, which is ‘demonstrative’. Likewise, the definite D, as a personal P, also derives its discourse-linking function from its meaning. Since it is a Dem, its function is to point to a contextual antecedent, signalling a familiar entity or class as an already established referent in the domain of discourse.15

As previously noted, the presence and type of modification together with the type of D decides the outcome of the linking relation, that is, whether it results in an identical referent, as with the anaphoric definite D in the earlier examples (15)-(16), in contrast to the examples with the bound indefinite D in (26)-(27), which involve identity of class but not of individual referent. Furthermore, complementation and D type also determine the disjoint reference of the clause-mate DPs and bare Ds in the examples of (17)-(18), but not of the DPs with the contextually interpreted Adjs in (19) and in (26)-(27). The reflexive vs irreflexive form of the personal P distinguishes individual coreference vs. class inclusion, and gives rise to the clausemateness condition in many languages.

The distinct discourse-linking functions of the two kinds of D may be seen at work in the contrasting examples of (28) and (29). In the former (=15), but not the latter, a linked DP that is equivalent to a bare P can refer to an individual or entity previously introduced in the domain of discourse. Apparently, whether Ds may be linked requires no restriction, but rather depends on general discourse coherence. For instance, linking, but not otherwise, turns the consecutive clauses in (28) into a well-formed discourse. However, in (29) no coherent discourse may result, as linking fails with the indefinite D modified by the same common N as in the antecedent. The reason is that since no class distinction is conveyed by the N, this form of the numeral D does not suffice for identifying a newly introduced referent. The more explicit variant otro/ another is called for in this context:


(28)      a. Un hombre apareció en la entrada. El hombre (=pro) parecía agobiado.

            b. A man appeared at the entrance. The man (=He) seemed overburdened.

(29)      a. Un hombre apareció en la entrada. ?Un hombre parecía agobiado.

            b. A man appeared at the entrance. ?A man seemed overburdened.


Accordingly, a D may be bound or free with no restrictions other than those arising from its form, its own complement and that in the antecedent. Thus, (30a) may have the bound and free readings of el inglés, with the favored reading depending on contextual factors. On the other hand, the DPs in (30b) must be disjoint, since linking el to este libro, agreement permitting, is disqualified by the predicate profesor, which automatically puts the P’s referent in a disjoint class:


(30)      a. Este libro y el inglés se contradicen.

                 ‘This book and the English {man/ one} contradict each other.’


            b. Este libro y el profesor se contradicen.

                 ‘This book and the professor contradict each other.’


Our analysis extends easily to other nondemonstrative Ds, as illustrated in (31)-(32), where the quantifier Ds in the second conjunct must be contextually interpreted. This fact supports the notion of Ds as discourse-linking functions. Here again, their bound and free readings must be freely available, as they give the expected distinction between a coherent discourse vs. a sentence list:


(31)      a. Several girls attended the play. Three left before the end.

b. Varias niñas asistieron a la obra. Tres salieron antes del final.

(32)      a. All of them attended the play. A few left before the end.

b. Todos ellos asistieron a la obra. Unos pocos salieron antes del final.


Observe, moreover, that since the bound reading of these quantifiers must denote operations on the antecedent sets, it is unnecessary to assume virtual partitive structures, as those proposed by Enç (1991), to explain their meaning.


5.2    Agreement

Explicit agreement in Person and Number (and sometimes Gender) is what is minimally needed for DPs to hold the binding relation. Overt inflection is what allows Spanish Ds to occur with Adjs in DPs that are contextually interpreted, as in the (a) examples below. By contrast, lack of inflection on English Ds bars them from appearing with such Adjs. In fact, English Ds with attributive Adjs cannot be bound to an agreeing antecedent, unless their DP indicates explicit Number agreement by means of the form one/s, as indicated in the (b) examples:


(33)      a. Empaqué la televisión grande, pero dejé la pequeña sin empacar.

            b. I packed the big television, but I left the small *(one) unpacked.

(34)      a. Empaqué una televisión grande, pero dejé una pequeña sin empacar.

            b. I packed a big television, but I left a small *(one) unpacked.


As the only condition on linking, agreement must be strictly exceptionless. This may be seen in the contrasting behavior of the Spanish Masc forms un/ uno, in the context of an Adj like pequeñ- ‘small’. The reduced un lacks Number and Gender inflection, hence, it cannot link to an antecedent, unlike the unreduced uno, which has the appropriate ending to function in a bound DP. The difference in acceptability shown in (35), then, arises from using one or the other form:


(35)      Empaqué un reloj grande, pero dejé  { uno/*un } pequeño   en la  mesa.

            I packed   a  clock big       but   I-left    one     a      small-masc on the desk

            ‘I packed a big clock, but I left {a small one/*a small boy } on the desk’


The bound uno gives an appropiate discourse, since uno pequeño ‘a small one’, in the context of un reloj grande ‘a big clock’, denotes un reloj pequeño ‘a small clock’. Instead, the undifferentiated D in un pequeño ‘a small boy’ gives an unbound DP, as well as an incoherent discourse.

On the other hand, in (36) there are two discourses, but only the one with uno has a bound DP, referring to un empleado viejo ‘an old employee’ from the set of los empleados ‘the employees’:


 (36)     Los empleados piden  la huelga.{ Uno/ Un } viejo  habla acaloradamente.

            the employees ask-for the strike    one   an  old-masc speaks heatedly

            ‘The employees ask for the strike. An old { one/ man } speaks heatedly.’


However, how agreement works needs to be further investigated since it is obvious from examples like (36) that the Number category contributes in crucial ways in the operation performed by a quantifier D that links to an antecedent.


6    Conclusion

The analysis of Ds as modified Ps allows a binding analysis for the contextually interpreted DPs with bare Adjs in Romance, which have traditionally been treated as elliptical structures. By so doing it extends the area of binding phenomena, and integrates such nominals with various anaphoric and nonanaphoric uses of DPs. In addition, it dispels received notions about the special status of the (in)definite Ds, and of individual coreference in anaphora, among others. The perspective has many interesting consequences and opens new paths that merit full investigation.



The following are sample DP derivations for Spanish and English that implement the D-CP structure in Kayne (1994):


(37)      los libros que leemos    ‘the books that we read’

            a. [DP    ellos   [CP   que  leemos  [DP wh- [SC t  libros ]]]]

                        they-masc  that  we-read                      books

            b. [DP    ellos   [CP wh-libros que  leemos  t  ]]                    (WH-Criterion)

            c. [DP    ellos-libros     [CP t’  que  leemos  ]]                    (Predicate-raising)

            d. [DP    los-libros        [CP  que  leemos   ]]                         (Enclisis)

(38)      las razones en las que se basan     ‘the reasons on which they are based’

            a. [DP  ellas     [CP que  se basan  en  [DP  ellas  [SC t  razones ]]]]

                       they-fem   that Refl-base on         they-fem    reasons

            b. [DP  ellas     [CP en ellas-razones [C’ que se basan t ]]]       (WH-Criterion)

            c. [DP  ellas-razones [CP en ellas t  [C’ que se basan t ]]]        (Pred-raising)

            d. [DP  las razones  [CP en las  [C’ que se basan  ]]]                (Enclisis)

(39)      the { book / one } that we read

a. [DP  that      [CP that we read wh-{ book / one } ]]

b. [DP  that      [CP wh-{ book / one } that we read  t ]]         (WH-Criterion)

c. [DP  that { book / one }  [CP wh-t  that we read  t ]]          (Pred-raising)

d. [DP  the { book / one }  [CP  that we read  t  ]]                  (Enclisis)

(40)      She gave us one that we read.

a. She gaves us  [DP  D   [CP that we read wh-one ]]

b. She gaves us  [DP  D   [CP wh-one  that we read  t ]]         (WH-Criterion)

c. She gaves us  [DP one [CP t’ that we read  t ]]                    (Pred-raising)



1. Cf. Also Luján (1972, 1980, 1999, and 2000), Lázaro-Carreter (1980), Ojeda (1983), Bosque y Moreno (1990), and Garrido (1991), among others.

2. Radford also cites supportive data from infants’ speech, such as it ladder.

3. Cf. Luján (2000) for the operator-variable structure for DP and for D as a covert relative Op.

4. Bach (1968) argues for the internal predication in NP. Holmberg (1993) and Campbell (1996) are recent DP implementations of this notion.

5. Holmberg (1993) has a virtual Dem in his own DP hypothesis:

[DP Dem  [D’ the  [NP  book  ]]].

6. Cf. Brucart (1987) for ellipsis, and Contreras (1989), Eguren (1989), Bernstein (1993), among others, for a null N or pro in DP structure.

7. Cf. Luján (2000) for a full critique of DP formulations such as (11)-(12).

8. Cf. Enç (1991), Campbell (1996), Garrido (1991), and Leonetti (1999), among others.

9. Here pro indicates the null pronoun form of Spanish.

10. Garrido (1991) and Leonetti (1999) describe numerous such uses of the definite Determiner.

11. Principle B resumes a similar statement which is found in most traditional grammars.

12. Cf. Schubert and Pelletier (1987) and Carlson and Pelletier (1995), for comprehensive surveys on generics and the problems of their analysis.

13. It is well-known that personal Ps may have generic antecedents: e.g.

(i) Snakes are reptiles. They creep.

(ii) The director makes the final decision. He or she bears the responsibility.

14. Here I am referring to the personal P (or Dem) as a unitary category, whose various forms are the realizations of its component features for Person, Number and Gender.

15. The discourse-linking function of the demonstrative vs. nondemonstrative Ds would replace the Familiarity and Novelty Conditions (cf. Karttunen 1968, and Heim 1982, among others).



Bach, Emmon. 1968. Nouns and Noun Phrases. Universals in Linguistic Theory, ed. by Emmon Bach and Robert Harms, 90-122. New York: Holt, Rinehart and Winston.


Bello, Andrés (y Rufino Cuervo). 1970. Gramática de la Lengua Castellana. Edición de Niceto Alcalá-Zamora y Torres, con Notas de Rufino Cuervo. Buenos Aires: Sopena.


Bosque, Ignacio y Juan Carlos Moreno. 1990. Las construcciones con lo y la denotación del neutro. Lingüística 2/1, 5-50.


Bernstein, Judy. 1993. Topics in the Syntax of Nominal Structure across Romance. Doctoral dissertation. City University of New York.


Brucart, José M. 1987. La Elisión Sintáctica en Español. Barcelona: Publicaciones de la Universidad Autónoma de Barcelona.


Campbell, Richard. 1996. Specificity operators in SpecDP. Studia Linguistica 50/2: 161-188.


Carlson, Gregory N. and Francis Jeffry Pelletier, eds. 1995. The Generic Book. Chicago: The University of Chicago Press.


Contreras, Heles. 1989. On Spanish empty N’ and N*. Studies in Romance Linguistics, ed. by Carl Kirschner and Janet Decesaris, 83-95. Amsterdam: John Benjamins.


Eguren, Luis. 1989. Datos del español en favor de la hipótesis de la frase determinante. Revista Argentina de Lingüística 5/1-2: 163-203.


Enç, Mürvet. 1991. The semantics of specificity. Linguistic Inquiry 22/1: 1-25.


Garrido, Joaquín. 1991. Elementos de Análisis Lingüístico. Madrid: Fundamentos.


Heim, Irene. 1982. The Semantics of Definite and Indefinite Noun Phrases. Doctoral dissertation, University of Massachusetts, Amherst.


Hestvik, Arild. 1992. LF movement of pronouns and antisubject orientation. Linguistic Inquiry 23/4: 557-594.


Holmberg, Anders. 1993. On the structure of predicate NP. Studia Linguistica 47/2: 126-138.


Jespersen, Otto. 1924. The Philosophy of Grammar. London: George Allen & Unwin.


Kayne, Richard S. 1994. The Antisymmetry of Syntax. Cambridge, Mass.: MIT Press.


Karttunen, Lauri. 1968. What do referential indices refer to? RAND Corporation Report No. P38543, Santa Monica, California.


Lázaro-Carreter, Fernando. 1980. El problema del artículo en español. Estudios de Lingüística, Fernando Lázaro Carreter, 27-59. Barcelona: Editorial Crítica.


Leonetti, Manuel. 1999. El Artículo. Gramática Descriptiva de la Lengua Española, ed. de Ignacio Bosque y Violeta Demonte, 787-890. Real Academia Española. Madrid: Espasa-Calpe.


Luján, Marta. 1972. On the so-called neuter article in Spanish. Generative Studies in the Romance Languages, ed. by Casagrande, Jean and B. Saciuk, 162-174. Rowley, Mass.: Newbury House.


Luján, Marta. 1980. Sintaxis y Semántica del Adjetivo. Madrid: Cátedra.


Luján, Marta. 1999. Minimalist Bello: basic categories in Bello’s Grammar. Advances in Hispanic Linguistics, ed. by Javier Gutiérrez-Rexach and Fernando Martínez-Gil, 428-446. Somerville: Cascadilla Press.


Luján, Marta. 2000. Sobre el sintagma determinante definido. To appear: Describing, Theorizing and Teaching Grammar: Spanish towards the 21st. Century, Papers from the International Symposium, Ohio State University, 1999.


Ojeda, Almerindo. 1993. Linguistic Individuals. CSLI Vol 31, Center for the Study of Language and Information, Stanford University.


Perlmutter, David 1970. On the article in English. Progress in Linguistics, ed. by Manfred Bierwisch and Karl. E. Heidolph, 233-248. The Hague: Mouton.


Postal, Paul M. 1969. On the so-called “pronouns” in English. Modern Studies in English, ed. by David A. Reibel and Sanford A. Schane, 201-224. New Jersey: Prentice-Hall.


Radford, Andrew. 1997. Syntactic Theory and the Structure of English, a Minimalist Approach. Cambridge: Cambridge University Press.


Schubert, Lenhart K. and Francis Jeffry Pelletier. 1987. Problems in the representation of the logical form of generics, plurals and mass nouns. New Directions in Semantics, ed. by Ernest Lepore, 385-451. London: Academic Press.




© Marta Luján. Circle of Linguistics Applied to Communication/ Círculo de Lingüística Aplicada a la Comunicación 9, February 2002. ISSN 1576-4737. Delivered at the (Preferably) Non-lexical Semantics Conference, Université de Paris VII, Paris, May 2000. Published in: Chicago Linguistic Society 36, Vol. 1: The Main Session, ed. by Okrent, Arika and John Boyle. Chicago Linguistic Society, University of Chicago. 259-273. 2000


clac 9