De in Catalan Elliptical Nominals : a Partitive Case Marker

CatWPL 412,1995 243-265 De in Catalan Elliptical Nominals: a Partitive Case Marker* Núria Martí Universitat Autbnoma de Barcelona. Departament de Filologia Catalana 08193 Bellaterra (Barcelona). Catalonia (Spain) [email protected] Received: April 18th 1995 Accepted: Setember 10th 1995 Abstract In Catalan an AP interna1 to an indefinite nominal phrase with an elliptical noun is obligatorily preceded by the prepositional element de: i.e. un de petit (lit. a of smal1,'a small one'). The occurrence of de in elliptical indefinite nominals in Catalan, taken together with its parallel occurrence in dislocated indefinite nominals and with en pronominalization, sheds light on the structure of indefinite nominals. We propose that they are QPs (Quantifier Phrases), where Q selects a Case functional projection (KP) and assigns it partitive Case (cf. Giusti (1992)), and take de to be the head of KP. The clitic en and the element de in dislocated constituents are other morphological manifestations of the partitive Case assigned to KP. Key words: nominal ellipsis, partitive Case, de preposition, quantifier phrase, Case phrase. Resum. De en nominals eldl'ptics en catala: un marcador de Cas partitiu En catala un sintagma adjectiu intern a un nominal indefinit amb nom el.liptic és precedit obligatbriament de l'element prepositiu de: un de petit. L'aparició de de en nominals indefinits el.líptics en catala, en relació amb els fenbmens de la pronominalització de en i de la dislocació, on també apareix de precedint els nominals indefinits dislocats, dóna llum sobre l'estructura dels nominals indefinits. Proposem que són SQ (Sintagmes Quantificador), on Q selecciona una projecció funcional de cas (SK), a la qual assigna partitiu (cf. Giusti (1992)), i considerem que de és el nucli d'SK. El clític en i I'element prepositiu de dels constituents dislocats són igualment manifestacions morfolbgiques del cas partitiu assignat a la projecció SK. Paraules clau: el.lipsi nominal, Cas partitiu, preposició de, sintagma quantificador, sintagma de Cas. * I thank Anna Bartra, Gemma Rigau and an anonymous reviewer for their comments. I am specially indebted to Jaume Sola for his comments, suggestions and invaluable help in the final English version of this paper. I have benefited from a FI (DGR) grant and also grants awarded by the DGICYT and the CIRIT (research projects PB93-0893-C04-01 and GRQ93-2035 respectively). 244 CatWPL 412, 1995 Núria Martí Table of Contents O. Introduction 4. De in front of Dislocated Constituents 1. The Data 5. De as a Realization of KO 2. De Is a Nominal 6. Conditions Regulating the Functional Category Distribution of de 3. A Functional Projection for 7. Conclusion Partitive Case: KP References O. Introduction This paper is concerned with a construction, which is specific to Catalan among Romance languages, where de precedes an AP intemal to an indefinite nominal with a non overt noun. The goals of this paper are to establish the status of the element de, and the conditions that trigger its appearence in such a restricted class of nominals in Catalan. More generally, we want to evaluate the evidence regarding the structure of nominals and APs that this construction offers. We will propose that this construction is related to en pronominalization, and to dislocation of indefinite nominals, which are preceded by the element de too, and we will provide a unitary analysis for all of them. Our proposa1 is a general hypothesis on indefinite nominals. We claim that they are quantifier phrases (QPs), where Q selects a functional projection responsible for Case (KP), whose head K0 is realized as de. Q assigns partitive Case to KP: de is, in fact, a manifestation of partitive Case. The scope of this theory extends to dislocated constructions, where the element de which precedes them is treated as a manifestation of partitive Case as well. We also assume that clitic en is partitive. We concentrate on Catalan, but data from other Romance languages will be taken into account. The paper is structured as follows: in section 1 we introduce the relevant data; in section 2 we argue that de is the head of a nominal functional category; in section 3 we identify this category as KP, a category responsible for Case; in section 4 we bring into consideration dislocated constructions, where de appears too; in section 5 we analyse all the instances of de as a realization of KO; and in section 6 we characterize the general conditions regulating the distribution of de. 1. The Data In Catalan indefinite nominals, the prepositional element de obligatorily appears preceding an AP whenever the noun is elliptical, as shown in (la). In contrast, the presence of de is excluded in indefinite nominals with an overt noun, as in (lb), and also in any definite nominal phrase, whether the noun is overt or not (see (2a,b)). (1) a. un de petit a of small 'a small one' De in Catalan Elliptical Nominals: a Partitive Case Marker CatWPL 412, 1995 245 (1) b. un gat ("de) petit a cat of small 'a small cat' (2) a. el gat (*de) petit the cat of small 'the small cat' b. el (*de) petit the of small 'the small one' Nominals in (1) contain an indefinite article, and those in (2) a definite article. Indefinite quantifiers such as algun ('some'), molt ('much, many'), etc. and nume- . rals behave like indefinite articles in requiring de when the noun is non overt; whereas definite quantifiers such as tot ('all') and demostratives behave like definite articles in excluding it in the same context, as shown in (3) and (4), respectively: (3) a. alguna *( de) buida some of empty 'some empty one' b. molts * ( de) diferents many of different 'many different ones' c. tres * ( de) barats three of cheap 'three cheap ones' (4) a. tots (* de) barats1 all of cheap 'all cheap' b. aquesta (* de) buida that of empty 'that empty' 1. In fact tots barats seems to be grammatical only with the adjective as a secondary predicate (see (i)). This is not the case with the definite article or the demonstratives. Other definite quantifiers such as cada ('each') and ambdós ('both') do not admit adjectives at all in elliptical nominals (see (ii)). (i) Els he comprat tots barats. Acc3pl have-lsg bonght all cheap 'I bought them all for a cheap price.' (ii) *He comprat ambdós barats. have-lsg bought both cheap 'I bought the two cheap ones.' 246 CatWPL 412. 1995 Núria Martí Besides the distributional restrictions depending on the definitness of the nominal, de also shows restrictions on the constituent it precedes: de precedes APs, but not PPs or relative clauses (see (5)), and de is also excluded if there is no modifier inside the nominal (see (6)). (5) a. un (*de) amb taques a of with stains 'one with stains on it' b. uns (* de) que tenen taques some of that have stains 'some ones with stains on it' (6) un (* de) a of ' one ' There are further restrictions on the kind of the adjectives that can appear in these constructions. Prenominal adjectives such as mer ('mere') are excluded, as illustrated in (7); and in the case of adjectives whose interpretation varies significantly depending on whether they precede or follow the noun, the only possible reading in these constructions is the one corresponding to the postnominal position (consider (8) and (9)). (7) a. un mer accident b. * un accident mer 'a mere accident' c. * un de mer 'a mere one' (8) a. una simple novebla 'a mere novel' b. una novel.la simple 'a simple novel' c. una de simple 'a simple one' (9) a. un pobre home 'a pitiable man' b. un home pobre 'an impoverished man' c. un de pobre 'an impoverished one' De in Catalan Elliptical Nominals: a Partitive Case Marker CatWPL 412, 1995 247 None of the other Romance languages considered here shows a similar pattern: only Catalan features de in elliptical nominal con~tructions.~ Compare the Catalan examples in (10) with the corresponding examples in Spanish, Italian and French -(I I), (12) and (13), re~~ectivel~:~ (10) a, un gat petit 'a small cat' b. un de petit 'a small one' (1 1) a. un coche nuevo 'a new car' b. uno nuevo 'a new one' (12) a. un vestito bello 'a beautiful dress' b. uno bello 'a beautiful one' (13) a. un crayon rouge 'a red pencil' b. un rouge 'a red one' (Catalan) (Spanish) (Italian) (French) In sum, de appears in Catalan indefinite nominals when the noun is elliptical and preceding postnominal APs. In the next section we address the question of how to analyse the sequence de AP. 2. De Is a Nominal Functional Category The sequence de AP can in principle be analyzed in two ways: a) de does not form a constituent with the AP, but rather with the whole NP in which N is ellipted; b) de forms an immediate constituent with the AP. 2.. De can also appear in French and Italian elliptical nominals, but the presence of de is more restricted and triggers a special interpretation. We consider it a different construction (see Martí (1994)). Observe the following French examples (from Milner (1976, 1978)): (i) a. J'en ai vu un de bon. I.EN have-lsg seen a of good 'I have seen a good one.' (among them) b. J'en ai vu un bon. 'I have seen a good one.' For Italian examples, see fn. 20. 3. Notice that in Spanish and Italian the indefinite article has a different form depending on wheter the nominal is overt or elliptical: un/uno, respectively. 248 CatWPL 4/2, 1995 Núria Martí There is evidence that approach (a) is to be preferred. If de forms a constituent with the AP, we would expect every AP within the elliptical indefinite nominal to be preceded by de, but this is not the case. Consider the nominals in (14), containing more than one AP, and the ones in (15,16), where an AP co-occurs with a PP: (14) a. Te 'n puc deixar una de sedosallarga. Dat2sg.NE can-lsg lend a of silky long 'I can lend you a silky long om.' b. N'he vista una de musical americana molt divertida. NE.have-lsg seen a of musical american very funny 'I have seen a very funny american musical film.' (15) a. En tinc algun de vermell amb rhdio. NE have-lsg some of red with radio 'I have some red one with radio.' b. En tinc algun amb radio vermell. (16) a. Li n'he regalat una d'elegant de seda. Dat3sg NE.have-lsg given a of.elegant of silk 'I gave him an elegant silky one.' b. Li n'he regalat una de seda elegant. Dat3sg NE.have-lsg given a of silk elegant 'I gave him an elegant silky one.' As examples in (14) clearly illustrate, it is not the case that one de element precedes each AP: de appears only once in each nominal. They also show that de appears in a left-most position right after the determiner. The examples in (15, 16) confirm the independent occurrence of de and AP(s), and the fixed position de occupies. The AP is preceded by de only when it is the fist modifier (see (15a,16a)). When the fist modifier is a PP, de is not realized (see (15b,16b)). The contrast between (17a) and (17b) provides further evidence that each nominal can only contain one realization of de:4 (17) a. N'he comprat de blaus i blancs. NE.have1 sg bought of blue and white 'I have bought blue and white ones.' b. N'he comprat de blaus i de blancs. 'I have bought blue ones and white ones.' 4. In these examples no quantifier is present. As they show the same pattem as the examples with indefinite quantifiers --de appears when the noun is elliptical-, we assume that there is a nul1 qnantifier. De in Catalan Elliptical Nominals: a Partitive Case Marker CatWPL 412, 1995 249 These sentences have a different interpretation: in (17a) the objects purchased are two-colour, while (17b) expresses that we have bought several blue objects and several white ones. The latter reading is determined by the presence of two de, each of which corresponds to a different nominal projection. In (17a), instead, the presence of a single de corresponds to a single nominal projection. In other words, in the first example the two APs are coordinated within one nominal, while in the second one there is coordination of two nominals. In sum, all the evidence presented so far indicates that de occupies a unique structural position within the nominal. From this we conclude that it heads a projection of its own. The next question we address is the nature of this projection. It is reasonable to assume that we are dealing with a functional projection given the lack of semantic content of the element de. In addition, its systematic appearence in a well defined class of nominals (elliptical indefinites) gives plausibility to its functional status. More specifically, we will argue that de heads a functional projection responsible for Ca~e.~ This hypothesis is in the spirit of many traditional proposals which take some apparent prepositions to be Case markers. For instance, it has been argued that the de element preceding genitives nominals appears in order to satisfy Case. Consider (18): (18) la casa *( de) la Maria the house of the Mary One piece of evidence in favor of the Case status of de in elliptical nominal constructions seems to be the fact that it cannot be followed by PPs. This could be explained in the spirit of the Case Resistance Principle, as formulated by Stowell (1981), which states that Case cannot be assigned to a category containing a Case assigning feature, such as prepositions.6 5. See Bernstein (1993a,b) for an alternative analysis which treats de as the realization of the functional category word marker (WM), which she proposes for nominals. For arguments against this analysis, see Martí (1994). 6. This account is not straightforward, since the relevant PPs are not, in our proposal, directly selected by de but are rather a constituent within the nominal. However, it seems that the reluctance of prepositions to be preceded by de is sensible to the word order, in the sense that de and the preposition cannot be adjacent in the sequence. Obsewe the contrast in acceptability between (ia) and (ib), where the adverb absolutament ('absolutely') intervenes between de and the preposition en ('in'): (i) a. *No me n'han donat cap d'en condicions. not Dat 1 sg NE.have-3pl given none of.in conditions 'They have'nt given anyone in conditions to me.' b. ?No me n'han donat cap d'absolutament en condicions. The impossibility of de to precede relative clauses could also be explained as a Case resistance instance in a similar way. The absence of de when there is no material that can follow it, instead, seems to require another type of explanation, perhaps based on phonological reasons: de is a proclitic element and if there is not a potential host, it is not realized. 250 CatWPL 412, 1995 Núria Martí In summary, we claim that de heads a Case functional projection within the nominal. In the following section we dea1 with the characteristics of this Case projection and its position in the nominal structure. 3. A Functional Projection for Partitive Case: KP In the current literature it is generally assumed that every functional morpheme projects its own category. It has been argued for number (Picallo (1991), Ritter (1991), Valois (1991), Cinque (1993), among others) and gender (Picallo (1991)). In this spirit, it is reasonable to assume that Case is also projected as a functional category, which would be morphologically realized in languages with Case endings. We will concentrate on indefinite nominals. We propose that in these nominals there is a Case Phrase (KP), which is the highest functional projection below QP (quantifier phrase), as shown in (19): K NumP (part.) Num ... \ NP According to this analysis, then, indefinite nominals are QPs, whose head is an indefinite quantifier selecting a partitive Case projection. Like indefinite quantifiers, we propose that the indefinite article is generated in Q0 because of its parallel behaviour. The proposa1 of a Case projection in nominals is not new: it has been proposed in Giusti (1992b) for aíl nominals. Giusti argues that all nominals are QPs, headed by Qs which select a Case projection. The distinction between definite and indefinite nominals lies in the type of the quantifier that heads QP: definite quantifiers, such as all, every, each, and indefinite quantifiers, such as some, muny, etc. According to her, they show different selection properties: definite quantifiers transmit the externa1 Case they receive from V or P to their complement, whereas indefinite quantifiers retains it and themselves assign partitive Case to their complement. The nominal structure Giusti proposes is (20), where FP are Case projections (Giusti (1992b: 162)): The highest FP is for the externa1 Case (assigned by a verb or a preposition). The lowest FP is selected by Q: FO is partitive (non overt) when Q is indefinite and FO is realized as an article when Q is univer~al.~ Giusti (1992) proposes this structure in order to account for the parallelism between definite and indefinite quantifiers concerning the possibility to be separated from the rest of the nominal proje~tion:~ (21) a. Els llibres, els he comprat tots. the books Acc3pl have-lsg bought all 'The books, I have bought them all.' b. De llibres, te'n portaré alguns. of books Dat2sg.NE bring-FUT-lsg some 'Books, I will bring you some.' According to Giusti, the pronominalized constituent is the projection selected by Q, and each type of quantifier requires a different clitic: tots ('all') requires the definite clitic els, while alguns ('some') requires the indefinite clitic en. The structure of definite nominals is beyond the scope of the present paper. As for indefinite nominals, the structure we propose in (19) is essentially consistent with Giusti's proposal. 7. Giusti considers the definite article as a Case marker, among other reasons, because it appears as adjective inflection in some languages (old German, Romanian, Greek, etc.) (see Giusti (1993:73)). For her also the indefinite article is generated in the lower FP, in spite of the fact that it is not similarly used as a Case marker in any language. Thls idea does not capture the parallel behaviour of the indefinite article to indefinite quantifiers with respect to de appearence, en cliticization and nominal dislocation (as we will see). 8. The alleged parallelism is not as systematic as Giusti's proposal seems to suggest: while any indefinite can occur in a construction where the noun is dislocated with the clitic ne, the definite quantifiers that can float are very restricted (all and both). 258 CatWPL 412, 1995 Núria Martí Within the NP either the clitic en or an empty category (represented by e)I7 appears. As for the position occupied by the APs, we assume that they are adjoined to the maximal functional projections inside the nominal. The question that we next address is the following: why is de not realized whenever the noun is overt? We must assume that we are dealing with the same structure as in (36), the only difference being that an overt noun is inserted in the NP. We will try to characterize the conditions regulating the distribution of de in the next section. 6. Conditions Regulating the Distribution of de We should first point out that the absence of de in nominals with an overt noun is not systematic. The presence of de is attested for some quantifiers in some dialects, as already pointed out by Fabra (1956: 23): (37) a. molt d'oli a.lot of.oil b. bastant d'aigua quite of.water The existence of such examples is a further argument for the analysis of de as realization of Case. In addition, there is a tendency in some Catalan dialects to have only de in the absence of agreement endings on the quantifier.'8 This suggests an 17. We have only seen examples with en cliticization. In (i) there is an example where NP is generated empty: (i) Tinc un cotxe gros i un de petit. have-lsg a car big and a of small 'I have a big car and a small one.' 18. Catalan dialects show a lot of variation, ranging from dialects where de obligatorily appears without any number or gender restriction (Majorcan and other balearic dialects, see (i)) to dialects which do not admit de after the quantifier (Alguerian, see (ii)). (i) a. molt de pa a.lot of bred b. molta d'aigua a.lot of.water c. molts d'al.lots a.lot of.boys d. moltes d'al.lotes a.lot of.girls (ii) a. tante cavalls so.many horses (Majorcan) (Alguerian) b. assay herbes quite herbs c. poche vi few wine The intermediate situation illustrated in (37) and (38) is attested in some central dialects. There is no such variation when the noun is elliptical: in this case de uniformly appears in all Catalan dialects. De in Catalan Elliptical Nominals: a Partitive Case Marker CatWPL 412, 1995 259 explanation for the systematic presence of de in nominal ellipsis contexts. With an overt noun de appears more easily with the unmarked masculine singular form than with feminine or plural forms, which are marked for gender and number: (38) a. molta (?? de) calor a.lot-fem-sg of heat b. moltes (*de) noies a.lot-fem-pl of girls c. bastants ('de) pisos quite-pl of apartments When the noun is covert, the grammatical features it bears are also covert. In order to make recoverable the information concerning Case, the Case marking element de must be realized (we assume that number and gender features, but not Case features, are manifested by Q). If de is absent, a general condition on ellipsis is violated, namely that only recoverable information can be elided. The ellipsis of the nominal head in Catalan is only possible if the inflectional features corresponding to the noun are realized.19 In other Romance languages, instead, the presence of the quantifier, which is the Case assigner, is sufficient. Romance languages show variation in various degrees as to how partitive Case is morphologically realized. In Spanish, partitive Case has no morphological manifestation: there is no clitic en, and, as we have seen, de is absent in front of both dislocated constituents and APs inside elliptical nominals. Catalan is at the opposite end in this respect: there is a clitic en and de precedes both dislocated constituents and APs inside elliptical nominals (in some dialects de even appears in some non-elliptical nominals). Italian and French show an intermediate position: they share with Catalan the existence of clitic en and the presence of de in front of dislocated constituents, but they do not realize de inside the elliptical nominal projection. Interestingly, with an empty quantifier Italian behaves like Catalan: di is obligatory in nominal ellipsis because in the absence of the quantifier it is the only means to recover partitive Case. 19. Brucart (1987) explains in this way that the definite article obligatorily appears in contexts of nominal ellipsis in Spanish: it becomes the only lexical manifestation of inflectional features of the noun: (i) a. Sobre todo buscava 10s (autógrafos) de 10s escritores. above all looked-3sg-for the autographs of the writers 'He looked for the autographslones of the writers.' b. Sobre todo buscava * (autógrafos) de 10s escritores. above all looked-3sg-for autographs of the writers (Brucart (1987: 326, ex. 54)) 260 CatWPL 412. 1995 Núria Martí (39) a. Ne voglio di belle. b. *Ne voglio belle. NE want-lsg of beautiful 'I want beautiful ones.' (Cordin (1988: 638))20 Let us try now to define the syntactic conditions under which de must appear. We should be able to explain why de is realized with en or an elliptical noun, but not with an overt noun in the general case. We should also account for the limited cases where de co-occurs with an overt noun. We will assume that the key factor lies in N head-movement inside the nominal to check functional features. With Chomsky (1992) we assume that nouns are inserted as already inflected and undergo head-movement to functional heads in order to check morphological features. No reaches K0 after moving through NumO, as you can see in (40). 20. Cordin shows all the possible combinations of ne with an adjective: A. ne + verb + quantifier + di + adjective B. ne + verb + quantifier + adjective C. ne + verb + di + adjective and claims that the combination in (D) is never attested: D. ne + verb + adjective It is not clear to us which is the interpretation assigned to (C). Cordin points out the semantic differences between (A) and (B), conceming the set of elements rhe adjective modifies, but she does not clarify whether the interpretation in (C) corresponds only to (A) or is ambiguous between (A) and (B). The interpretations of (A) and (B) are illustrated in the examples in (ia) and (ib): (i) a. Ne voglio alcune di belle. NE want-lsg some of beautiful 'I want some of beautiful ones.' b. Ne voglio alcune belle. 'I want some beautiful ones.' (Cordin (1988: 638, ex. 33)) In (ia) the adjective refers to the whole set of beautiful objects, of whicb tbe quantifier specifies a subset, while in (ib) the adjective refers to the subset of objects specified by the quantifier. For more information, see Cordin (1988). In Catalan, example (ii) corresponds to (ib) and not to (ia), in spite of the fact that it contains de: (ii) En vull algunes de boniques. NE want-lsg some of beautiful 'I want some beautiful ones.' De in Catalan Elliptical Nominals: a Partitive Case Marker CatWPL 412, 1995 261 un K NumP When the noun is empty, we suppose the syntactic derivation is the same: No head-moves up to K0 in order to check all the functional features. The only difference with the preceding case is that the noun in K0 is not visible. In the case of en the derivation is different. We assume that en is a maximal projection, on the lines of Kayne (1994), which is an NP, and that it moves through specifiers in its way out of the QP.~~ Since, within the QP, en moves as a maximal projection, it will not move through K~.~~ 21. In Kayne's (1994) theory adjunction and specifier positions are undistinguishable. We kept to the more traditional of distinguishing the two in order to have enough structural space: adjoined positions for APs and specifier positions for the movement of en. Otherwise, en could not move in the presence of an AP. 22. We leave open the question how the morphological feature of K0 is checked if No does not move to it, because en moves as a maximal projection. We speculate that the specifier-head configuration is an altemative means for the features to be checked, and that it takes place whenever the unmarked strategy (head-adjunction) does not apply. 262 CatWPL 412. 1995 Núria Martí K' K NumP 1 AP NumP petit '; A A Num' What the derivation with clitic en and with an empty noun have in common is that K0 does not contain phonological material in the syntax. In other words, K0 is not visible. This contrasts with the case where the noun is overt: it moves to K0 and makes it visible. For some reason, in Catalan K0 must always be visible, so that, if no overt noun moves to this position, de must be inserted. In the minimalist program, late insertion is allowed if we are dealing with elements which are not interpretable at LF. Chomsky (1992) proposes such a process of insertion for the element do in English and for the element de of genitives: ccThis formulation allows later insertion of functional items that are vacous for LF interpretation, e.g., the do of do-support or the of of of-insertion>> (Chomsky (1992: fn. 22)). De-insertion in K0 would be another instance this process. It remains to be explained why de sometimes appears with an overt noun. This possibility is lexically restricted. A possible explanation consists in assuming that with certain quantifiers movement of the noun is shorter and only reaches NumO. For this reason the lack of lexical material in K0 would trigger de insertion in this position.23 If the noun reaches KO, de insertion is not activated. This is to say, there would be a double possibility of either short or long No-movement, restricted to some quantifiers. 23. Since de is inserted after spell-out in PF, the problem remains unsolved how the features of K0 are checked in case they are strong. We'leave this question open. De in Catalan Elliptical Nominals: a Partitive Case Marker CatWPL 412, 1995 263 7. Conclusion The analysis proposed in this paper consists in considering the element de that precedes AP inside indefinite elliptical nominals as the head of a Case functional projection, KP. This is the highest nominal functional projection and it is selected by Q, which is projected in QP and assigns it partitive Case. The identification of de with the head of a nominal functional projection explains the fact that it appears only once in each nominal and it occupies a fixed position in the structure, the left-most position right after the determiner (the indefinite article or an indefinite quantifier, in this case). As for the distribution of de, the fact that its appearence is restricted to indefinite nominals derives from the fact that only indefinite quantifiers, projected in QP, select a partitive Case projection. In this way, definite nominals are automatically excluded. In indefinite nominals, de is only realized when the noun is non-overt. De insertion is triggered in order to satisfy the requirement of K0 to be visible in Catalan (i.e., to contain phonological lexical material). It seems that the restrictions on the type of constituent which de precedes derive from its casual nature: de precedes APs,~~ but not PPs or relative clauses because the latter ones reject adjacent Case markers (cf. Case Resistance Principle by Stowell (1981)). The hypothesis of de as a partitive Case marker extends to the dislocated constructions, where de precedes dislocated indefinite nominals. We claim that de is also a partitive marker in these contexts, which is necessary in order to identify the dislocated constituent with its corresponding position in the sentence: (part of) the Case projection selected by the indefinite quantifier. References Argente, J. (1976). <<Un exercici d'anilisi transformacional entorn del pronom en>>. In: V. Sánchez de Zavala (ed.). Estudios de Gramútica C;enerativa. Barcelona: Labor Universitaris. Bartra, A. (1985). <<Qüestions de la sintaxi d'ordre en catalin. Universitat Autbnoma de Barcelona. PhD dissertation. - (1987). <<Encara n'hi ha més (entorn de en i alguns SN genitius),,. Llengua & Literatura 2: 377-427. Belletti, A.; Rizzi, L. (1981). ctThe Syntax of 'ne': Some Theoretical Implications,,. The Linguistic Review 1: 1 17154. Beninci, P. et al. (1988). c<L'ordine degli elementi della frase e le construzioni marcate,,. In: Renzi (ed.) (1988), p. 115-270. 24. The proposed analysis does not explain why pre-nominal APs cannot be preceded by de in elliptical nominals. Probably, they are excluded in these.constructions for the same reasons that account for their restricted distribution: they are excluded (or hardly admitted) in some other contexts, such as dislocated nominals or partitives, where there is not an elliptical noun. We have no explanation of the distribution of pre-nominal adjectives, but it does not seem problematic for the analysis of de AP presented here. 264 CatWPL 412, 1995 Núria Martí Bernstein, J. (1991). <<Nominal Enclitics in Romance,). MIT Working Papers in Linguistics 14: 51-66. - (1993a). <<Topics on the Syntax of Nominal Structure across Romance,,. City University of New York. PhD dissertation. - (1993b). eThe Syntactic Role of Word Markers in Nul1 Nominal Constructions,,. Probus 5: 5-38. Brucart, J.M. (1987a). La elisión sintáctica en español. Bellaterra: Servei de Publicacions de la Universitat Autbnoma de Barcelona. - (1987b). <<La elipsis parcial,,. In: V. Demonte; M. Fernández Lagunilla (eds.). Sintaxis de las lenguas románicas. Madrid: El Arquero, p. 291-328. - (1993). <<Concordancia ad sensum y partitividad en espaiioh. Universitat Autbnoma de Barcelona. Unpublished manuscript. Brucart, J.M.; Gricia, L. (1984) a1 Sintagrni Norninali senza testa: uno studio comparato),. Rivista di Grammatica Generativa 1 1 : 3-31. Cardinaletti, A.; Giusti, G. (1991). <<Partitive ne and the QP-Hypothesis. A Case Study~. CLI-91.1.4. Universitii di Venezia: Working Papers in Linguistics. Published in E. Faba (ed.) (1992). Proceedings of the XVII Meeting of Generative Grammar. Trieste, Febmary 22-24, 1991: Rosenberg & Sellier, p. 121-141. Casadio, C. (1991). <<Ne-cliticization and Partitive Interpretation,,. Proceedings of the XVZZ Meeting of Generative Grarnmar, Trieste, February 22-24, 1991: Rosenberg & Sellier, p. 143-158. Chomsky, N. (1992). (<A Minimalist Program for Linguistic Theory,,. MIT Occasional Papers in Linguistics 1. Cinque, G. (1990). Types of A'-Dependencies. Carnbridge, Massachusetts: MIT Press. - (1993). <<On the Evidence for Partia1 N Movement in the Romance DP,,. Universiti di Venezia. Unpublished manuscript. ' Cordin, P. (1988). <<I1 clitico ne),. In: Renzi (ed.) (1988), p. 633-641. Fabra, P. (1956). Gramhtica catalana. Barcelona: Teide. Giusti, G. (1992a). c<Heads and Modifiers among Determiners Evidence from Romanian and Gennan,). CLI - 92.1.3. Universita di Venezia: Working Papers in Linguistics. - (1992b). La sintassi dei sintagma quantificaat Universita di Venezia. PhD dissertation. Giusti, G. (1993). La sintassi dei determinanti. Padova: Unipress. Monograph of Rivista di Grammatica Generativa. Kayne, R. (1975). French Syntax. The Transformational Cycle. Cambridge, Massachusetts: MIT Press. French translation: Syntaxe du frangais. Le cycle transformationnel. Paris: Editions du Seuil, 1977. Kayne, R. (1994). The Antisymmetry of Syntax. Cambridge, Massachusetts: MIT Press. Longa, V.; Lorenzo, G.; Rigau, G. (1994). dubject Clitics and Clitic Recycling. The Syntax of Locative Sentences in Some Iberian Romance Languages,,. Universidad de Santiago de Compostela/UMass at Amherst; Univ. de OviedoMIT; UABNIT. Unpublished manuscript. Martí, N. (1994). ctAnZilisi de la construcció de + sintagma adjectiu en catalb. Universitat Autbnoma de Barcelona. Master thesis. Picallo, C. (1991). c<Nominals and Nominalizations in Catalan,. Probus 3: 279-316. Renzi, L. (ed.) (1988). Grande grammatica italiana di consultazione. I. La frase. 1 sintagmi nominale e preposizionale. Bologna: I1 Mulino. De in Catalan Elliptical Nominals: a Partitive Case Marker CatWPL 412, 1995 265 Ritter, E. (1991). eTwo Functional Categories in Noun Phrases: Evidence from Modem Hebrew,,. In: S. Rothstein (ed.). Syntax and Semantics. Perspectives on Phrase Structure: Heads and Licensing. London: Academic Press. Stowell, T. (1981). <<Origins of Phrase Structure,,. MIT. PhD dissertation. Valois, D. (1991). c<The Interna1 Syntax of DPn. University of California Los Angeles. PhD dissertation.