1. Introduction

This article develops a new analysis of bare plurals in Spanish that attempts to present a unified and coherent view of the apparently diverse conditions on their distribution and interpretation. We will demonstrate that a theory based on prosodic, morphological and semantic constraints can account for the derivations that converge at both interfaces, Phonological Form(PF) and Logical Form (LF). The interaction of these constraints explains why Spanish bare plurals behave so differently in comparison to the rest of the Romance languages, a fact that has been largely unnoticed. The new data presented are discussed within Chomsky's "Minimalist Program" (Chomsky, 1993), and in particular, the development of his ideas in Bare Phrase Structure (Chomsky, 1994).

2. Spanish Bare Plurals: Basic Distributional Facts

As has been noted extensively in the literature, the distribution of bare plurals (henceforth BPs) in Spanish under normal conditions of stress and intonation follows a pattern. On the one hand, they are allowed in the following positions: as objects of verbs, as in (1a) and (1b);objects of prepositions, as in (1c-d); objects of nouns (1e) and (1f); inverted subjects, (2a) and (2b); topics, as in (3b) and focused subjects as in (3a).

(l) a. Quiero tortillas (Contreras, 1986)

(I) want tortillas

b. Los esclavos construyeron pirámides

The slaves built pyramids

'Slaves built the pyramids'

c. Ayer salí con amigos

Yesterday went-I with friends

'Yesterday, I went out with friends'

d. Va a fiestas todos los sabados

goes-he A parties every Saturday

e. Cosas de niños

things of children

f. Varias clases de emociones

several classes of emotions

(2) a. Llegaron muebles

Arrived furniture

The furniture arrived'

b. En este árbol anidan cigüeñas (Torrego, 1989)

In this tree, storks shelter

(3) a. ESCLAVOS construyeron las pirámides

SLAVES built the pyramids

b. Dinero, no tengo

Money, NEG have-lstsg

'Money, I don't have'

c. Sonrieron NIÑOS

smiled CHILDREN

On the other hand, it has been claimed that BPs cannot occur as preverbal subjects in both main and subordinate clauses, as in (4a to c) and as indirect objects, as in (5a-c):

(4) a. "Esclavos construyeron las pirámides (Contreras, 1986)

Slaves built the pyramids

b. *Muebles llegaron

Furniture arrived-3rdpl

'There arrived furniture'

c. *Niños fueron asesinados

Children were killed

d. *Me dijeron que bombas explotarían

me told-3rdpl. that bombs would explode

'They told me that bombs would explode'

(5) a. *El director ha devuelto los documentos a empleados (Brugè & Brugger, 1993)

The director has returned the documents to the employees

b. *Adriana no le habla a lingüistas (Masullo, 1992)

Adriana NEG CL speak A linguists

'Adriana doesn't speak to linguists'

c. *Yo siempre le juego partidos de tenis a principiantes (Masullo, 1992)

I always CL play games of tennis A beginners

'I always play tennis games with beginners'

3. Previous Approaches

The proposals that have been made in the literature of the last decade can be grouped in three types of approaches:

3.1. The Naked Noun Constraint.

In her study of Spanish presentational sentences, Suñer (1982) includes a chapter on the distribution of Spanish bare nouns, which is the most exhaustive account of the distribution of Spanish bare plurals in subject position (Masullo, 1992). She proposes the following constraint on their distribution: The Naked Noun Constraint (NNC). According to the NNC "an unmodified common noun in preverbal position cannot be the surface subject of a sentence under conditions of normal stress and intonation (p.209) ".

The NNC would give an explanation of the contrasts in grammaticality between the sentences in (3) and (4), since it would predict that the sentences in (3) are grammatical because they do not have "normal stress and intonation". Suñer's account has been criticized on the grounds that it proposes a filter with little syntactic import (Lois, 1989; Masullo, 1992).Nevertheless her study contains important insights that have not been explored nor developed so far. As can be inferred from our conclusions, the NNC is in reality a partially valid generalization in the end. It is true, however, that it is conceived to apply only to subjects, losing the perspective of a wider generalization.

3.2. The ECP approach.

Contreras proposes a structure for bare nouns that treats them as elements headed by an empty quantifier. Since this quantifier is an empty category, it must satisfy the Empty Category Principle (ECP, Chomsky, 1981). Occurrences of bare plurals in the subject position, as in (4),or in other ungoverned positions -typically the indirect object position- would violate the ECP, since the empty quantifier would not be governed. In other words, the distribution of bare plurals would follow from the theory of government.

Longobardi (1994) and Delfitto & Schroten (1993) treat bare plurals as "defective" Determiner Phrases (Abney, 1987). In these DPs, the D0 position is empty and the DP requires lexical government. Longobardi proposes also that when the D0 position is empty, the DP receives an existential interpretation by default, following Diesing (1992).

3.3. Incorporation-based approaches

Lois (1989: 152) claims that NPs without a determiner have to incorporate into an X0 =V0, P0 to be licensed. This constraint would predict the grammaticality of the examples in (1), since the bare plural would incorporate into the verb or the preposition. No case assignment is necessary in order for the NP to be licensed. The NP is visible at LF after incorporation.

Masullo (1992: 263) follows Contreras (1986) in treating bare plurals as defective nominal projections or projections that do not reach their maximal level of expansion within an" extended projections" view of syntactic constituents (Grimshaw, 1991). Hence, bare nouns require a "stronger" means of identification, specifically, they have to receive inherent case (Belletti, 1988) and they have to incorporate into another head.

Our approach is inspired by the feature-based theory in Chomsky (1994) and will integrate the prosodic, morphological and lexical aspects of the distribution (spell-out) of Spanish bare plurals. In the next paragraph, we present our basic theoretical assumptions.

4. Structures and Operations

One of the more relevant features of the Minimalist Program developed by Chomsky(1993, 1994) is the elimination of D-Structure and S-Structure as necessary levels of grammatical representation. Crucially, conditions operate only at the interface levels Phonetic Form (PF) and Logical Form (LF). As a consequence, there are no conditions operating at SS or DS, or conditions that mention exclusively X-bar structure. Conditions on government such as the ECP are no longer considered a conceptual necessity, since they mention X-bar configurations and relations within X-bar trees in a non-trivial way. In addition, there is no licensing between a Head and its complements (i.e. out of the checking domain).

Grammatical processes are conceived as a series of ordered computational operations (computational steps). Grammatical derivations are a set of computational steps. Every step is an operation on features. There are two fundamental operations on features: MERGE and MOVE.

The evaluation of competing derivations follows the Principle of Economy. Economy is broken down into Shortest Movement, Greed and Procrastination (Marantz, 1994). The Shortest Movement Constraint eliminates Minimality, Subjacency and the ECP. Greed allows us to eliminate derivations that are well-formed but unmotivated, and Procrastinate delays the visibility of a feature or set of features until the LF-interface (i.e. beyond the Spell-Out). The Spell-Out is the splitting point to PF and LF. When a feature is added to the computation, it triggers the derivation (FEATURE INSERTION). The Minimalist Program faces two challenges: (i) to describe the mechanisms that create computations and (ii) to explain how derivations are compared. The most economical derivation depends on the "feature carriers" or lexical items involved in it, which are language-specific. The derivation proceeds and expands by projection in the sense of the bare phrase structure theory.

The principle of GREED dictates that a lexical element moves only to check its own features: prosodically or morphologically realized (relevant or "visible" at PF), or semantic (relevant at LF). A lexical item is (from the computational point of view) a set of two sets of features: ¹-features (PF features) and l-features (LF features). In every computational step (application of an operation) a feature or sets of features is checked. Therefore, two or more features can be checked at the same time in a single computational step.

Operations have a target. Only equidistant targets are valid extensions of the domain of the computational step (operation). MERGE is a local operation by definition, since it is an operation that forms larger units out of those already constructed (Chomsky, 1994: 10). MOVE is obligatorily subject to the Minimal Link Condition (Chomsky, 1993) and implies feature insertion.

There are two sorts of features: strong and weak. Strong features are visible at PF and LF, whereas weak features are visible (interpretable) at LF. In other words, weak and strong features are interpretable at LF, but strong features are also interpretable at PF. A computation converges when both interfaces satisfy the Full Interpretation principle (FI), i.e., all the features present in the computation have been checked.

One of the more surpnsing consequences of the theory we have presented so far is that it is no longer necessary to limit maximal projections to at most one specifier. We have the following three assumptions: (i) computation creates structure, (ii) every operation involves all the features relevant to it, and (iii) features have to be checked in a specifier (in a Spec-Headrelation; Sportiche, 1993b). It follows that there will be as many specifiers as there are computational steps necessary for the satisfaction of FI at the interface levels.

A derivation starts when we select a number of lexical resources (feature carriers) and compare competing derivations. Every feature or set of features added (inserted) affects the domain of the computation creating different competing derivations.

5. Prosodic and Morphological Focus

It has been recently claimed by Masullo (1992) and Brugè & Brugger (1993) that BPs cannot occur as indirect objects because this position is either ungoverned or receives structural case (6a-c):

(6) a. *El director ha devuelto los documentos a empleados (Brugè & Brugger, 1993)

the director has returned the documents to employees

b. *Le molesta el humo a señores

CL bothers the smoke A men

'Smoke bothers some men'

c. *Diste tu dinero a políticos

gave-2nd sg. your money A politicians

'You voted for politicians'

Nevertheless, the sentences above become fully grammatical when the BP in indirect object position is focused or when what we will call a "depreciative" affix is attached to it:

(7) a. El director ha devuelto los documentos a empleadillos/EMPLEADOS

the director has returned the documents to employees-deprec. / EMPLOYEES

b. Diste tu dinero a politicuchos / POLÍTICOS

gave-2ndsg. your money A politicians-depreciative / POLITICIANS

c. Le aposté dinero anoche a primerizos / PRINCIPIANTES

CL bet-lstsg. money last-night A beginners-deprec. / BEGINNERS

'I made a bet with some beginners last night'

The contrast that arises between the examples in (6) and (7) casts serious doubts on the ECP-based hypotheses proposed by Contreras, Longobardi and Brugè & Brugge. Consider, for example (6a) and (7a). The BPs empleados and empleadillos are both DPs (or QPs) headed by an empty determiner or quantifier. Since this empty category is not properly governed, an ECP violation should arise in both sentences. This is clearly not the case. It could be argued that the BPs in (7) have incorporated into the depreciative affix. But if we assume, following Chomsky (1993), that lexical items formed by affixation are already present in the lexicon, then empleados and empleadillos would enter the computation in an identical stage of syntactic complexity and would be structurally similar. In other words, in the two cases there should be an empty element heading the relevant functional projection DP or QP and the ECP violation would be unavoidable. An alternative explanation in terms of the Adjacency Condition on case assignment, along the lines proposed by Torrego (1984) or Belletti (1988), would not predict the contrast either because, as mentioned before, the structural configurations are identical.

We see that either Government or Case should work uniformly for both groups of examples. Similar considerations apply to Incorporation-based explanations, like the ones in Lois(1989), Masullo (1992) and Delfitto & Schroten (1991). According to these proposals, if a is treated as a preposition, then we explain the grammaticality of the sentences in (7) but the ones in (6) turn out to be a problem because they should be grammatical (the BP would be able to incorporate into the preposition a). Since this is not the case, we can claim that a is not a preposition but rather a case marker. Now, the ungrammaticality of the examples in (6) is predicted and those in (7) cannot be accounted for. Finally, we see that Masullo's theory rules out BPs in indirect object position in general. This follows from the claim that, for instance, politicuchos in (7b) would receive structural Dative Case, thus remaining unidentified at LF. The sentence is again predicted to be ungrammatical, contrary to facts.

In what follows, we are going to propose a feature-based theory in order to account for the contrast. We claim that depreciative affixes such as -ucho in politicucho , -illo in empleadillo and -izo in primenzo carry a functional feature that must be checked under the same projection as the focused EMPLEADOS, POLÍTICOS and PRINCIPIANTES in (7). Affixation mechanisms creating lexically depreciative elements are quite common in Spanish. All the lexical items in (8) carry depreciative affixes, and turn sentences similar to the ones in (6) grammatical when substituted for the non-depreciative bare plural.

(8) ricacho 'money-bag', vivales 'crafty-devil', esquiroles 'scabs',

señoritingos 'rich little daddy's boy', querindongas 'lover-fem-deprec.

The minimal hypothesis would be that the [+depreciative] lexical feature and [+focus] are checked under the same specifier, in other words, they are indistinguishable from a computational point of view. The prosodically or semantically focused BPs in (7) would move to the target specifier where [+depreciative] is checked in order to check a prosodic or semantic feature [+focus]. The feature [+depreciative] carried by the affix -ucho must be considered "strong" (Chomsky, 1993) and checked before LF, since it forces the BP to move to a projection higher than its natural target position, probably NumP. When this feature is carried by non affixated lexical items such as avaros 'tightwads', fascistas 'fascists' or pipiolos 'freshmen-deprec.' in (9), the feature would "procrastinate" and would be checked at LF. This is confirmed by the grammaticality of the following sentences:

(9) a. He prestado demasiado dinero a avaros

(I) have lent too much money to tightwads

b. Nunca confio mis hijos a fascistas

(I) would never leave my sons in the confrdence of fascists

c. Le hicieron bromas a pipiolos

CL made-3rdpl. jokes A freshmen-deprec.

'They made fun of freshmen'

In support of the relevance of the [+/- depreciative] distinction in the computation of the syntactically sensitive features of Spanish feature carriers, it should be noted that animate exocentric compounds in Spanish are made up of [V + BP] configurations and most of them are depreciatives. Some examples of this generalized mechanism are given in (10a). Therefore, t seems a very plausible hypothesis to claim that MOVE is sensitive to [+depreciative] in Spanish. Additional support for this apparently perplexing characteristic comes from the fact that "appreciative" lexical items cannot normally fit as indirect object BPs, as can be seen in (10b-c).

(10) a. sacamuelas ' dentist-deprec. ', matasanos ' doctor-deprec. ',

picapleitos 'lawyer-deprec. ', rompetechos 'short-deprec.', tragaldabas 'glutton', chupatintas 'clerk- deprec.

b. *Requisó los coches a generosos

Impounded-3rdpl. the cars A generous-people

c. *Diste tu dinero a politicazos

You gave your money to politicians-appreciative

Following Chomsky (1994), we adopt the "Multiple Specifiers" hypothesis, according to which a functional category has as many specifiers as features to be checked under restricted conditions. We claim that Spanish DPs have at least two specifier slots, as shown in (11): one for the specificity feature (the Spec2), and the other for the focus feature (Specl).

(11) Dmax

spec n D

Spec2 D

[+specificity]

Specl D

[+focus]

Dmin Nummax

Under our perspective, focus can be seen as a condition on the interfaces, namely, as a condition on Feature Interpretation:

- At LF the feature [+focus] is interpreted as [+depreciative], like in avaros and politicuchos.

- At PF, [+focus] can be either (morphologically) realized in a suffix like -ucho or (prosodically) realized in a specific intonational contour or pitch accent, as in poLItico.

BPs with depreciative affixes can show up also in preverbal position, a fact which was explicitly not accounted for or was even explicitly ignored for all kinds of BPs by previous approaches. In (12a) and (12c) políticos and mujeres are ruled out as expected. The sentences in (12b) and (12d) show the alternation between a prosodic and a morphological triggering of the derivation: POLITICOS/politicuchos in (12b) and MUJERES/mujerzuelas in (12d).

(12) a. *Políticos coparon todos los puestos del partido

Politicians won all the party positions

b. POLÍTICOS / Politicuchos coparon todos los puestos del partido

POLITICIANS / politicians- deprec. won all the party positions

c. "Mujeres te han puesto en aprietos

women CL have put in a-bind

Women have put you in a bind'

d. MUJERES / mujerzuelas te han puesto en aprietos

WOMEN / women-depr. CL have put (you) in a-bind

Sentences (12c) and (12d) differ minimally in the fact that in the latter the BP carries an additional feature [+focus] that has to be checked at both interfaces, either prosodically or morphologically at PF and semantically at LF. Another type of similar examples that remains unexplained under previous approaches is constituted by recursive and 'heavy' prefixes like minirrobots, requeteacaparadores, and posposimpresionistas in (13):

(13) a. Minirrobots / *robots hacen el trabajo con igual calidad

Small-robots / *robots do the job with the same quality

b. Requeteacaparadores / *acaparadores han acabado con las existencias

Intens-hoarders ! *hoarders have finished-off with the merchandise

'Horders finished-off all the merchandise'

c. Posposimpresionistas / *impresionistas publicaron sus obras

Post-postimpressionists / *impressionists published their works

d. MINIrobots es lo que tú necesitas y no ese trasto

TINYrobots is LO that you need and not that piece-of-junk

'TINY robots is what you need and not that piece of junk'

e. MiniROBOTS y por qué no minipersonas

'Tiny ROBOTS and why not tiny people'

Our explanation of (13 a-c) is that the prefixed BPs have a morphological feature related to contrastive focus that must be checked prior to Spell-Out. The prefixes contribute to the "heaviness" of the BPs. In this respect, BPs are p(rosodic)-constituents in the sense of Zec & Inkelas (1990), namely, branched constituents. An additional piece of evidence is shown in (13d-e), where the root or the affix in MINIrrobots and MiniROBOTS can receive narrow focus. This demonstrates that mini and robots are prosodical constituents that can enter in the domain of the operation that assigns focus at PF and, as a consequence, the prosodic node dominating them branches. Therefore, we can conclude that Spanish BPs are prosodically licensed if they are "heavy" or, in more technical words, if they are p-constituents.

The facts presented so far show that government plays no major role in a significant part of the syntax of bare plurals: its occurrence in subject and indirect object position is a source of counterexamples for all the previous the ones based on government. Similar conclusions apply naturally against those who propose a Case-based approach, because subjects and indirect objects would occupy structural positions and should receive structural case.

6. Multiple Specifiers in the DP

Several arguments can be presented in favor of the "Multiple Specifiers" Hypothesis presented in Chomsky (1994). The first one we consider is clitic doubling. In Suñer (1988), it is argued that clitics are manifestations of object agreement obeying the Matching Principle, namely, agreement between the clitic and the NP (Suñer, 1988: 393).

(14) a. le / *les diste tu dinero a politicuchos / POLITICOS

CL.sg / *CL.pl. gave-lstsg. your money A politicians-depr. / POLITICIANS

b. "le / les diste tu dinero a los políticos

*CL.sg / CL.pl. gave-lstsg. your money A the politicians

c. "le ! 'les diste tu dinero a políticos

In (14b) the clitic les and the DP los políticos match, since they exhibit overt number agreement. The Matching Principle is violated when we substitute the singular clitic le for the plural les. The example (14c) shows that the BP políticos does not satisfy the Matching Principle by mere number agreement neither with the plural clitic les nor with the singular le. Finally, in (14a) we see that in the grammatical version le diste tu dinero a politicuchos, the clitic le and the BP politicuchos/POLITICOS can "coexist". This has to be because their projections match in complexity at some point of the derivation.

It is a traditional observation about Romance clitics that they are specific (Givon, 1976),a fact that constitutes a widely accepted view in current work in linguistics (Dobrovie-Sorin,1985; Unagereka, 1992; Sportiche, 1993, among many others). Clitics have to be considered specific determiners and their behaviour follows from the syntax and semantics associated with specificity (Enç, 1991). Although the parallelism between direct and indirect objects is not complete, Sportiche (1993) has proposed that the instances with indirect objects display an AGR similar to AGR-O and AGR-S. Suñer (1988) and Lyons (1990) identify clitic and agreement systems entirely. If the picture sketched above is correct, then there are should not be instances of unspecific clitics. The examples in (15) illustrate this point:

(15) a. Vi a uno / alguno / alguien

saw-lst sg. A one / someone / someone

b. *Lo vi a uno / alguno / alguien

CL saw-lstsg. A one / someone / someone

If clitics check [+specificity] and politicuchos check [+focus] in (14a), as claimed before, then there should be a violation of the Matching principle and le and políticuchos should not coexist. The contradiction vanishes if we recall that one of the consequences of the minimalist assumptions is that two features can be checked at the same time (by a single application of an operation). The fact that the matching Principle is not violated in (14a) shows precisely that the features [+specificity] and [+focus] have to be placed within the same computational domain. As a matter of fact, the same lexical item can carry the two of them. Sentence (14a) has an interpretation in which politicuchos, besides being focused, is interpreted as specific, i.e., the speaker has in mind a specific group of politicians. Notice also that in (14b) both the determiner los of los políticos and the clitic les of les diste tu dinero have to match and check its specificity feature against the higher specifier (Spec2). If los políticos in (14b) were focused, it would follow trivially that both Spec1 and Spec2 would be filled/checked.

7. DP Extractions

Extractions of a DP out of another DP constitute a second powerful argument in favor of the application of the "Multiple Specifiers" Hypothesis to DPs. DP extractions are a well studied phenomenon in Spanish, including Torrego (1984), Demonte (1987), Mallén (1991),Martín (1993) and many others. The paradigm in (16a-c) illustrates possible options for extraction:

(16) a. De María he visto *los juguetes copy

b. De Mana he visto Ø juguetes copy

C. De María he visto Ø JUGUETES copy

Of Mary have-lstsg. seen *the toys ! Ø toys / Ø TOYS

A well-known factor that intervenes in extractions from Spanish DPs is the so-called "Specificity effect". Effects of this sort appear to block extractions in wh-questions and restrictive relative clauses. In (16), the topicalized de María exhibits the same effects. Sentences (16 b-c) show instances of extractions through the BP juguetes and through the focused BP JUGUETES. Following the standard assumptions in the literature, we can claim that De María is moved through a specifier of the DP (our Spec2), the position against which specificity is checked. Therefore, the ungrammaticality of (16a) becomes transparent: both the trace of De María and the determiner los compete for the same position. They both have to check their respective [ + specificity] to avoid crash but only one target is available. Therefore, the derivation "crashes" at LF. Let us now consider (17):

(17) a. DE MARIA he visto los juguetes

b. DE MARIA he visto Ø juguetes

c. *DE MARIA he visto Ø JUGUETES

Following the line of reasoning that we have been pursuing and comparing (16a) and (17a) it is not difficult to conclude that the movement of DE MARIA in (17a) is not made through the specifier of los juguetes, but through the [+focus] specifier. In (16a), the position against which specificity is checked is filled by los and is not available for the movement of the non focused possessive De María. This is not the case in (17a), where the focused DE MARIA can escape through Spec1, regardless of whether the second specifier is filled by los juguetes. On the other hand, JUGUETES in (17c) blocks the movement of the focused DE MARIA, since it needs to check its [+focus] feature and the two focused elements are competing for the same specifier (the [+focus] one). If we take a look at (16c) what we see is quite revealing. The focus position in which that particular feature [+focus] is checked does not block the movement of De María. Let us evaluate the two options that we naturally arrive at in (18):

(18) A. Checking specificity and focus together within the same computational cycle

and

B. Checking the two features in different computational cycles,

Alternative (18A) is, first, more local and, second, more economic. In other words, the shortest movement is no movement. Thus, this kind of checking that we will call JOINT CHECKING saves steps in the computation so that GREED is not violated. Since the motivation for triggering the extension of the domain, namely, the strong feature to be checked is fulfilled, there is no reason not to get joint checking in its natural locus, the Multiple Spec cycle , in the way shown in (11).

8. LF Interpretation and Same Domain Effects

If [specificity] and [focus] share the Dmax domain, we should expect some specific LF effects involving both of them. Consider the examples in (19):

(19) a. Un montón de los invitados, de los que dos hablaban de lingüística

group of the guests, of the that two were-talking about linguistics

'a group of guests, two of which were talking about linguistics'

b. Un montón de Ø invitados, de los que dos hablaban de lingüística

a group of guests, ....

c. Un montón de Ø INVITADOS, de los que dos hablaban de lingüística

In (19a), the DP un montón de los invitados de los que dos hablaban de lingüística is ambiguous between two different interpretations. Under one interpretation, the two individuals who talk about linguistics, besides being guests must also be members of the group referred to. Under the second interpretation, the two individuals who talk about linguistics are only required to be guests, but not members of the relevant group. In other words, the difference resides in whether the ones who talk about linguistics are or are not part of the group (montón) of guests. In (19b), the ones who talk about linguistics have to be members of the specific group of guests, not just simply guests. There is no ambiguity at all. Not surprisingly, if we consider (19c) we see that both readings are recovered: to be or not to be group members, as happened in (19a). Here our predictions seem to be confirmed. In (19a), los invitados checks the [+specifity] feature, since it denotes a particular set of guests. The embedded relative can be coindexed with los invitados or with un montón, since they both represent computationally equivalent projections (they are specific). The bare plural invitados in (19b) is not [+specific]. Therefore, it is not available for coindexation. By the same reasoning, the focused INVITADOS in (19c) must have checked the [+specificity] feature since the possibility of coindexation to it is recovered. This shows that [+focus] and [+specificity] must be features in the domain of the same computational cycle (the DP).

One prediction we can make is that the number of features checked all at once -joint checking- varies cross-linguistically as a function of the feature carriers. Not all features are compiled in the same item the same way. For instance, the "depreciative effect" displayed by -ucho in Spanish doesn't appear in other Romance languages such as Italian (20a-c), Catalan (20d,e), French (21a-c) and Romanian (21 d,e):

(20) a. Il direttore ha restituto i documenti *a / agli / a degli impiegati-deprec.

The director has returned the documents *A / A-the / A-the-part. employees-deprec.

b. Non voto mai per *Ø / i / dei fascisti

Not vote never for *Ø / the / the-part. fascists

'I never vote for fascists'

c. Ho prestato troppi soldi (?? a) / a dei tirchi

'I have lent too much money to tightwads'

d. El director ha donat els documents a *(uns) empleadots (=20a)

e. He deixat massa diners a *(uns) rics

'I have lent too much money to rich people'

(21) a. Le directeur a rendu les documents *à / aux gratte-papiers (= 20a)

b. J'ai donné mon vote *à / aux politiciens-deprec.

I have given my vote A / A-the politicians-deprec.

'I have voted for politicians-deprec.

c. J'ai prête trop d'argent à *(des) avares (=20c)

d. Directorul a înapoiat documentele *(unor) functionarási / functionarásilor ( =20a)

e. Am imprumutat prea multi bani *(unor) avari / avarilor (=20c)

At the same time, Siswati, a Bantu language, is going to provide a good example (22 and 23) of both morphological and prosodic strategies to reach the PF interface by a derivation triggered by the appreciative/depreciative feature.

(22)Diminutive a. umutfw-ana 'a little human/child'

b. kwe-mutfu 'a little human'

c. kwe-mutfw-ana 'a little human' (depreciative)

(23)Augmentative a. umfati-kati 'a big woman'

b. umfati-kaÿti 'a big woman' (appreciative) (ÿ=LH tone)

Siswati has two diminutive affixes: the prefix kwe- and the affix -ana. There are two possibilities for their combination with a noun. The combinations kwe-noun (22b) or noun-ana (22a) yield a neutral diminutive meaning 'little-noun'. In other words, when they appear in isolation combined with a noun they have the same meaning (diminutive). The second combinatory possibility is that they appear together, forming the complex kwe-noun-ana. There is a significant difference in this second case: when kwe- and -ana appear together they acquire a depreciative lexical content, as in (22c). Therefore, the [+depreciative] feature is checked by a morphological operation of word formation. This operational behaviour cannot be transferred to augmentatives, since there is only an augmentative suffix and no augmentative prefix. Therefore, the feature [+appreciative] has to be checked phonologically for the derivation to converge. In Siswati, a L(ow) H(igh) tone has to be linked to the augmentative affix kati in order to check the feature at PF and get the appreciative meaning, i.e., check [+appreciative] at LF..

9. Embedded subjects

Suñer (1982: 217) claims that the same contrasts in grammaticality that are found in matrix contexts are also found in embedded clauses. Therefore, bare plurals cannot occur as pre-verbal subjects of subordinate clauses.

(24) a. *Me dijeron que bombas explotarían

me told-3rd.pl. that bombs would explode

'They told me that bombs would explode'

b. "La alarma sonará si ladrones entran

the alarm will go off if thieves enter

She concludes that the Naked Noun Constraint operates also in embedded contexts, as the well-formed examples below show. In (25a) ladrones sin experiencia is a modified bare plural, and bombas in (25b) is not preverbal:

(25) a. La alarma sonará si ladrones sin experiencia entran

the alarm will go off if thieves without experience enter

'The alarm will go off if unexperienced thieves enter'

b. Me dijeron que explotarían bombas

me told-3rd.pl. that would-explode bombs

Nevertheless her predictions are not born out, since it is very easy to find unmodified preverbal subjects in embedded clauses (26 b,c), whereas the corresponding matrix sentence is ungrammatical (26a):

(26) a. *Submarinistas encontraron el tesoro

Scuba divers found the treasure

b. Dice que submarinistas encontraron el tesoro

He says that scuba divers found the treasure

c. Si submarinistas encuentran el tesoro, el gobierno estará contento

If scuba divers find the treasure, the government will be happy

The same contrasts between root and embedded clauses seem to be absent in other Romance languages like Italian, Catalan, French and Romanian, as illustrated in(27).

(27) a. Ha detto que "(dei) subaquei hanno scoperto il tesoro (=26b)

b. Diu que *(uns) submarinistes van trovar el tresor (=26b)

b. Il a dit que *(de) plongeurs ont découvert le trésor (=26b)

c. A spus ca° scafandri*(i) an descopent comoara (=26b)

The existence of this asymmetry in Spanish sheds new light on the structure of the CP and the nature of the complementizers in this language. Let us assume that the multiple specifiers constituent-structure proposed in Chomsky (1994) and already presented here for DPs carries over to CPs, so that we have the following CP-structure:

(28) Cmax

Spec n C

spec2 C

[+wh]

specl C

[+focus]

Cmin IP

We are not claiming that the features checked in the CP are exactly the same as those checked in the DP. For instance, it seems that [specificity] is not going to be checked at the CP level in the same way as in DPs, whereas [wh] is a typical CP-feature. What we are trying to explore here is the interaction between the morphological nature of the CP head and the mechanisms and possibilities involved in the checking of the [focus] feature in the CP computational domain. Recall that we have claimed before that the [+focus] feature can be checked within the DP. Our additional hypothesis here is that it can be also checked in the CP. There is actually a wide variety of CP-types in which bare plurals in ungoverned position can occur. For example, in (29) we have two direct questions (29a,b) and two indirect questions (29c,d):

(29) a. ¿Ladrones robaron la oficina?

Thieves robbed the office?

Did thieves rob the office?

b. ¿Por qué ladrones robaron la oficina?

Why did thieves rob the office?

c. Preguntó por qué ladrones robaron la oficina

He asked why thieves robbed the office

d. Preguntó si ladrones robaron la oficina

He asked whether thieves robbed the office

The syntactic and semantic difference between direct and indirect questions is not mirrored at PF. The embedded yes/no question in (29d) has the same intonational contour of a declarative sentence, whereas the sentences in (29a-c) share many prosodic features. Let us briefly consider some of them. Following the terminology in Pierrehumbert (1980) and Pierrehumbert & Hirschberg (1990), we can observe that por qué attracts nuclear (focal) stress both in (29b and c) and is linked to a High tone. This is what characterizes prosodically the [+wh] feature at PF. Direct and indirect questions differ at PF in that the former have a High final boundary tone (29a,b) whereas the latter are characterized by a Low final boundary tone (29c,d).

(30) a. Por qué ladrones robaron la oficina ?]]

| |

H* H%

b. Preguntó por qué ladrones robaron la oficina ]]

| |

H* L%

c. Preguntó si ladrones robaron la oficina ]]

|

L%

The fact that in these sentences the feature [+wh] is associated with a particular intonational contour and stress pattern (namely, obligatory nuclear stress over the wh-word) shows that the feature is undoubtedly "strong" in these cases, since it is checked both at PF and at LF. The feature [+wh] must be considered "weak" in (30c), since there is no prosodic clue identifying the sentence as a question. Therefore, the feature must be checked at LF in order to get a convergent derivation. The computation of the [+wh] feature at LF attracts the abstract focus feature of the bare plural in the three cases triggering an absorption-like mechanism (Higginbotham & May, 1981).

Longobardi (1994) observes that bare plurals are licensed when they are vocatives, as shown in the contrast in (31):

(31) Espectadores, aplaudan / *Espectadores aplauden

Audience, applaud / *Audience applaud

Since Spanish does not have an overt morphological vocative affix, the abstract vocative feature of the bare plural in a command-denoting sentence is checked off at PF by inserting a phonological phrase boundary after the bare plural and aligning it to a distinctive boundary tone at PF ( typically a Low tone). Otherwise the derivation crashes, since the vocative feature is not checked off and remains at the PF-interface level, thus violating the Principle of Full Interpretation that holds at both interfaces. We have to stress here again that the fact that there is no morphological affix realizing the feature in the overt syntax does not necessarily mean that the feature is weak. As we have shown in the case of interrogatives and imperatives, the strength of a feature can well be prosodic.

(32)[[Espectadores] [aplaudan ]]

| |

L% L%

A final illustration of this point are the sentences in (33):

(33) a. Lo que hay que ver: Monjas vociferando!

Imagine that. Nuns shouting!

b. Es asombroso ver cómo computadoras hacen el trabajo

it is amazing to see how computers do the work

It is incredible how computers work'

Bare plurals are also strongly licensed when associated with an exclamative intonational contour at PF (33a). The prosodic pattern of embedded exclamatives is again different from matrix ones: the qu-word attracts nuclear stress and the final boundary tone tends to be Low (33b). As seen in the case of interrogatives, in both types of exclamatives the feature of the bare plural is checked off by absorption.

10. Feature checking in CPs

Let us turn back now to the issue of how the focus feature of bare plurals is checked in very different environments. Iatridou (1992) and Iatridou and Embick (1994) suggest that the difference between because- and since-clauses is that the latter are used when the proposition expressed by the clause is presupposed or discourse-old. Since old information cannot be focused, the distinction accounts for the fact that because-clauses can be clefted (34), uttered as answers to a question (35), or be within the scope of a focus operator (36), whereas since- clauses cannot.

(34) a. It is because he was poor that he had to leave home

b. "It is since he was poor that he had to go home

(35) A: Why did John leave?

B: a. Because he wasn't feeling well

b. *Since he wasn't feeling well (Iatridou & Embick, 1994: 199)

(36) a. He had to leave home only because he was poor

b. *He had to leave home only since he was poor

Spanish porque-clauses can be considered non-presuppositional, since they can be clefted, modified by sólo or uttered as answers to a question. On the other hand, ya que-clauses fail these tests so they have to be treated as presuppositional.

(37) a. Es porque era pobre por lo que tuvo que dejar su casa

b. "Es ya que era pobre por lo que tuvo que dejar su casa

c. Tuvo que dejar su casa sólo porque era pobre

d. Tuvo que dejar su casa sólo ya que era pobre

e. A: Por qué se fue Juan?

B: i. Porque no se sentía bien

ii. *Ya que no se sentía bien

We can also infer from sentences (38a,b) that bare plurals cannot occur as preverbal subjects of weak ya que-clauses, but they do occur in porque-clauses.

(38) a. Se enfadó porque hijos suyos fueron al cine

He got upset because his sons went to the movies

b. *Se enfadó ya que hijos suyos fueron al cine.

He got upset since his sons went to the movies

If we assume, following Iatridou, that porque-clauses can be focused whereas ya que-clauses cannot, due to the fact that they constitute old information, we can give an account of the contrast within the theory that we have developed so far. Recall that it has been shown previously that in some specific circumstances a bare plural needs to check off the feature [+focus] in order to be licensed. It is not possible to claim with respect to a sentence like (38a) that the prosodic feature has been checked since the construction is correct even if no focus accent is linked to the bare plural hijos suyos. There are two alternatives left. The first one is that, as we have seen before, the bare plural checks the LF feature [+focus] after an application of the operation MOVE that targets the lowest specifier of the DP. The second one is that the feature is not checked there. The first option has to be discarded on semantic grounds because in a sentence like (38a) what seems to be focused is not the bare plural alone but the whole porque-clause, as shown by the question/answer pair in (37e).

According to the multiple specifiers theory, for every maximal projection there will be as many specifiers as operations necessary to check the features in the lexical item. Hence, we can extend our analysis of the DP to the CP. If a CP is focused, the feature [+focus] has to be checked at LF, and in parallel to the DP case, the feature will be checked in the lowest specifier of the CP. In (37e) and (38a) the bare plural moves at LF to the specifier of the CP and checks its [+focus] feature. As a consequence, the bare plural is licensed and the whole CP will be interpreted as focused (new information). That the [+focus] feature has to be checked can be seen as a morphological requirement of the complementizer porque. As we have observed, ya que does not have that requirement, since it heads a presuppositional CP. Therefore, even though a bare plural like hijos suyos in (38b) needs to move to check its [+focus] feature, there is no target available in the CP. The only alternative for the derivation to converge would be that the bare plural move to its own specifier position to check the [+focus] feature. Then, the bare plural would be prosodically focused, and as a result at LF only the bare plural would be focused and we should get a narrow focus interpretation like the one in (39):

(39) Se enfadó ya que HIJOS SUYOS fueron al cine

He got upset since HIS SONS went to the movies

Another illustration of this last resort mechanism is provided in (40):

(40) Esta molesto ya que *niños / niñatos / NIÑOS entraron en su casa

He is upset since "kids / kids-depr. / KIDS got into his house

If niños is not prosodically focused in (40), then the feature [+focus] will not be checked at PF, violating the principle of Full Interpretation. As a consequence, the derivation crashes. Only the prosodically focused NIÑOS and the morphologically focused niñatos 'kids- deprec.' are able to check the [+focus] feature at their respective [Spec,DP] target, thus satisfying FI.

A second possibility is shown in (41):

(41) Ya que submarinistas encontraron el tesoro ] , para qué ir a buscarlo

|

H%

Since scubadivers found the treasure, why go looking for it?

Here the bate plural is not focused itself, but rather the whole clause is focused as the associated obligatory intonational contour clearly shows: in this type of sentences the ya que-clause final boundary tone is H. The derivation converges here again, since the checking of the features at PF forces them to be checked at LF as well.

Up to this point, a fairly consistent picture seems to emerge. There are two ways of licensing a feature: (i) in the computational cycle (projection) where it is supposed to be checked by lexical requirements , and (ii) in the next computational cycle providing an adequate feature-target, if there is no target available in the natural computational cycle of the feature carrier. In this case, the feature will be attracted by the next feature-target available in the derivation. We will call this second mechanism FEATURE INCORPORATION, in order to distinguish it from the standard checking mechanism in (i) above.

The analysis of rationale clauses that we have sketched above also gives us a clue for a more general explanation of the convergence of PF and LF derivations in which there is a bare plural in the subject position of a CP. The FEATURE INCORPORATION mechanism implies that the DP cycle and the CP cycle are connected, in other words, the bare DP incorporates into the C. We can hypothesize that the bare plural incorporates into the CP by an application of a special sort of the STRICT MERGE operation, namely understanding here merging as feature union (union restricted to the attracted features or UNITIVE MERGE) and not mere feature projection, which is the only possibility presented in Chomsky (1994: 10). A unified account of the behaviour of bare plurals in embedded or adjoined that-clauses, if/when-clauses, etc., thus emerges.

(42) a. Dice que estudiantes tomaron el rectorado de la universidad

He says that students took over the chancellor's building of the

university

b. Dice que ESTUDIANTES tomaron el rectorado de la universidad

He says that STUDENTS took over the chancellor's building of the

university

In (42a) the [+focus] feature of the bare plural is attracted to the lowest specifier of the CP and the feature is computed at LF as CP-focus. As a consequence, the activated focus operator has wide scope over the whole CP. In (42b), this possibility is not at hand, since ESTUDIANTES checks the focus feature at its own specifier, as argued before. Thus, the focus target in the CP is not activated and no wide-focus interpretation is available.

The systematic incorporation of the bare plural into the complementizer at PF, understanding again incorporation as UNITIVE MERGE, is interpreted as absorption of a CP operator and the bare plural at LF. These sort of interpreted logical forms are to some extent equivalent to the Lewis-Kratzer style LFs which use unselective binding.

(43) Cuando estudiantes aprueban una asignatura, se van a tomar copas.

When students pass a course, they go partying

Incorporation is interpreted at PF as membership to the same phonological phrase. It follows that no prosodic boundary (pause) can be inserted between the complementizer and the bare plural.

(44) a. *Cuando ][ estudiantes aprueban una asignatura, se van a tomar copas

When students pass a course, they go partying

b. Cuando ][ los estudiantes aprueban una asignatura, se van a tomar copas

When the students pass a course, they go partying

A different version of the same idea would be Delfitto & Schroten's use of X-1 level categories. It is also worth noticing that the incorporation-based treatment of bare plurals in embedded clauses that we are defending allows for an unification of these cases with the standard patterns V + BP, P + BP that we find in Spanish (examples in (1), (2)), and in most Romance languages. As seen in §3.3, the arguments in favor of treating the latter cases as incorporation seem quite compelling, but needed for a refined notion of incorporation in order to constitute a convincing hypothesis for CP incorporation. The feature attraction mechanisms and operations available (or obligatory) in the CP computational domain are quite different in their interpretative LF effects from those involved in the VP or the PP, as we have extensively shown. Nevertheless, the syntactic operation of feature incorporation applies similarly in the three domains and consequently bare plurals are licensed in the three environments.

11. Adverbs and bare plurals

It is well known that focus adverbs like sólo, hasta, incluso always license bare plural subjects, as observed by Suñer (1982) and Masullo (1992):

(45) a. Sólo adultos entendieron la película

Only adults understood the movie

b. Hasta musulmanes comerian de esta carne

Even muslims would eat (of) this meat

c. Incluso millonarios sufrirán las consecuencias de la crisis

Even millionaires will suffer the consequences of the crisis

The explanation of the examples above follows from our hypotheses. Focus adverbs attract the feature [+focus] of the elements they take scope over acting as targets for the MOVE operation. Bare plurals check the focus feature there satisfying the principle of GREED.

As we have seen, full DPs and constituents in which the focus feature has been checked represent different levels of computation, since in the former the specificity feature has also been computed. As expected, they cannot be coordinated.

(46) a. ?Los niños y SACERDOTES acudieron a la fiesta

The kids and PRIESTS came to the party

b. *Los niños y solo sacerdotes acudieron a la fiesta

The kids and only priests came to the party

Similarly, when a focus adverb takes scope over coordinated bare plurals, all the members of the coordination check their focus feature.

(47) a. [ Sólo [niños y sacerdotes]] acudieron a la fiesta

Only kids and priests came to the party

b. *[Sólo niños y sacerdotes acudieron a la fiesta

Only kids and priests came to the party

c. [Sólo niños] y SACERDOTES acudieron a la fiesta

Only kids and PRIESTS came to the party

The sentence in (47c) provides further evidence for our idea that the checking of a feature at PF implies that it is also being checked at LF. If SACERDOTES in Sólo niños y SACERDOTES acudieron a la fiesta were only checked at PF, it could not be coordinated with sólo niños since they would represent different computational stages. The fact that weak and strongly focused constituents are coordinated or, in other words, computed at the same time, confirms also the idea of JOINT CHECKING: partially different features are computed in a single step.

We finish this article by suggesting that focus adverbs are not the only ones that license bare plurals. Speaker-oriented adverbs -frecuentemente in (48a,b)- and adverbs of quantification -a todas horas, siempre, a menudo in (48c)- are also able to do so:

(48) a. Frecuentemente niños pasean por el parque.

Frequently kids walk in the park

b. Niños pasean por el parque frecuentemente.

Kids walk in the park frequently

c. Clientes vienen por aquí a todas horas / siempre / a menudo.

Clients come by here all the time / always /frequently

Manner adverbs license bare plurals when the adverb is focused, being then able to attract the same feature of the bare noun:

(49) a. Subrepticiamente ladrones se introdujeron en la casa.

Surreptitiously thieves broke into the house

b. ladrones se introdujeron en la casa SUBREPTICIAMENTE /

*subrepticiamente.

Thieves broke into the house SURREPTITIOUSLY /

surreptitiously

12. Conclusions

Some of the conclusions we have reached in the paper are the following:

1. A feature-based approach to the behaviour of bare plurals is superior to a X-bar theoretic one, since we account for the occurrence of bare plurals in ungoverned positions.

2. There are at least three types of grammatical conditions on features: prosodic, morphological and semantic. Among the prosodic features we can cite the focus or nuclear stress and special intonational contours. Depreciative affixes and exocentric compounds are instances of a morphologically realized feature. Finally, depreciative lexical items carry this feature as lexical.

3. Feature checking takes place in a refined constituent structure that allows multiple specifiers. This idea suggested in Chomsky's Bare Phrase Structure applies successfully in both the domain of the DP and the domain of the CP.

References

ABNEY, S. (1987), The English Noun Phrase in its Sentential Aspect, Ph.D. diss., MIT.

BAKER, M. (1988), Incorporation: A Theory of Grammatical Function Changing, The University of Chicago Press, Chicago.

BELLEITI, A. (1988), "The Case of Unacussatives", Linguistic Inquiry, 19, 1-34.

BOSQUE, I. (ed.), (1995), Presencia y ausencia del artículo en español, Madnd, Visor.

BRUGE, L. & G. BRUGGER, (1993), "On the Accusative 'A' in Spanish", ms., Univ. di Venecia.

CHOMSKY, N. (1993), "A Minimalist Program for Linguistic Theory" in K.Hale & S.J. Keyser (eds.), The View from Building 20: Essays in Linguistics in Honor of Sylvain Bromberger, MIT Press, Cambridge, 1-52.

CHOMSKY, N., (1994), "Bare Phrase Structure", MIT Occasional Papers in Linguistics, 5.

CONTRERAS, H. (1986), "Spanish Bare NPs and the ECP", in I.Bordelois, H.Contreras, K.Zagona (eds.), Generative Studies in Spanish Syntax, Foris, Dordrecht.

DELFITTO, D. (1993), "A propos du statut lexical de l'article partitif en français. Quelques hypotheses sur l'interaction entre morphologie et forme logique", in A.Hulk, F.Melka, J.Schroten (eds.), Du lexique à la morphologie: du côté de chez Zwaan, Rodopi, Amsterdam, 103-116.

DELFITTO, D. & J.SCHROTEN (1991), "Bare Plurals and the Number Affix in DP", Probus, 3.2, 155-185.

DEMONTE, V. (1987), "Rección y minimidad en el sintagma nominal", in V.Demonte, M.Fernández Lagunilla (eds.), Sintaxis de las lenguas románicas, Ediciones El Arquero, Madnd, 252-290.

DIESING, M. (1992), Indefinites, MIT Press, Cambridge.

DOBROVIE-SORIN, C. (1985), "Redoublement clitique, relatives et interrogatives en roumain et espagnol", Lingvisticae Investigationes, 9, 269-306.

EMONDS, J. (1985), A Unified Theory of Syntactic Categories, Fons, Dordrecht.

FRANCO, J. (1993), On Object Agreement in Spanish, Ph.D. diss., USC.

GIVON, T. (1976), Subject and Topic, Chicago University Press, Chicago.

GRIMSHAW, J. (1991), Extended Projections, ms., Brandeis University.

HIGGINBOTHAM,J. & R.MAY (1981), "Questions, Quantifiers and Crossing", The Linguistic Review, 1, 41-80.

HOEKSTRA, T. & R.MULDER, (1990), "Unergatives as Copular Verbs; Locational and Existential Predication", The Linguistic review, 7, 1-79.

HORVATH, J. (1994), "Structural Focus, Structural Case, and the Notion of Feature-Assignment", in K. E. Kiss (ed.), Configurational Languages, Oxford University Press.

IATRIDOU, S. (1992), Topics in Conditionals, Ph.D. diss, MIT.

IATRIDOU,S. & R.EMBICK, (1994), "Conditional Inversion", NELS, 24, University of Massachusetts, Amherst.

KOIZUMI, M. (1994), "Layered Specifiers", NELS, 24, 255-269.

LOIS, X. (1989), Aspects de la Syntaxe de I'Espagnol et Theon'e de la Grammaire, Ph.D. diss., Université de Pans VII.

LONGOBARDI, G. (1994), "Reference and Proper Names: A Theory of N-Movement in Syntax and Logical Form", Linguistic Inquiry, 25, 609-665.

LYONS, C. (1990), "An Agreement Approach to Clitic Doubling", Transactions of the Philological Society, 88.1, 1-57.

MALLEN, E. (1991), "Noun Phrase Structure, Clitic Doubling and Extraction", Linguistiche Benchte, 134, 276-309.

MARANTZ, A. (1994), ms., MIT.

MARTIN, J. (1993), "On Extraction from NPs in Spanish", in W.Ashby, M.Mithun, G.Perissinotto, E.Raposo (eds.), Linguistic Perspectives on the Romance Languages, Selected papers from the LSLR 21, John Benjamins, Amsterdam/Philadelphia, 303-313.

MASULLO, P. (1992), Incorporation and Case Theory in Spanish. A Crosslinguistic Perspective, Ph.D. diss., University of Washington.

PIERREHUMBERT,J. (1980), The Phonology and Phonetics of English Intonation, Ph.D. diss, MIT.

PIERREHUMBERT,J. & P.HIRSCHBERG (1990), "The Meaning of Intonational Contours in Discourse", in P.R.Cohen, J.Morgan, M.Pollack (eds.), Intentions in Communication, MIT Press, Cambridge, 271-312

RAINER, F. (1986), "Recursiveness in Word-Formation with Special Regard to Spanish", Acta Linguistica Academiae Scientianun Hungarícae, 197-209.

SPORTICHE, D. (1993), "Clitic Constructions", ms., UCLA

SPORTICHE, D. (1993b), "A Reductionistic Program", ms., UCLA

SUÑER, M. (1982), Syntax and Semantics of Spanish Presentational sentence-Types, Georgetown University Press, Washington D.C.

SUÑER, M. (1988), "The Role of Agreement in Clitic Doubled Constructions", Natural Language and Linguistic Theory, 6, 391-434.

TORREGO, E. (1984), "Determinerless NPs", ms., University of Massachusetts, Boston.

URIAGEREKA, J. (1992), "Extraction Parameters: A Case Study on Underspecification" , ms., University of Maryland.

ZEC, D. & S. INKELAS (1990), "Prosodically Constrained Syntax", in S. Inkelas, D. Zec (eds.), The Phonology-Syntax Connection, Chicago University Press, Chicago, 365-378.