Asymmetries in Spanish Relative Pronouns: A Study on Que, Quien, and El Cual, Apuntes de Catalán

¡Descarga Asymmetries in Spanish Relative Pronouns: A Study on Que, Quien, and El Cual y más Apuntes en PDF de Catalán solo en Docsity! SOME ASYMMETRIES IN THE FUNCTIONING OF RELATIVE PRONOUNS IN SPANISH Jos6 M. Brucart Universitat Autbnoma de Barcelona The goal of this paper is to outline an explanatory theory of the syniactic behaviour of relative pronouns in Spanish. The requirement that such a theory must fulfill is one of globality: i.e., it must include both types of relative clauses: restrictives and appositives. This differentiates our proposal from some others (such as Rivero (1982) and Schroten (1984), for instance), which focus primarily on restrictives1. For purely operational (not theoretical) reasons. we will exclude the relative adjective cuyo from our study, not because of its decreasing presence in speech, but rather because its syntactic behaviour is perfectly regular and does not show the asymmetries that characterize the other relative pronouns in Spanish. Our expository strategy will be as follows: in section 1 we will study the data and the proposals previously made and in section 2 we will present our analysis. 1. The data and their account Grammars of Spanish usually describe the asymmehies that affect the syntactic functioning of the relative pronouns que (with its variants el que and Prep + (el) que), quien and el cual. Bello (1847, § 303-332 and 1073-1085), Fernández Ramírez (1951, ch. X), Alcina & Blecua (1915, ch. 8) , RAE (1973, 3.20) and Martinez (1989, ch. IV) refer to the different behaviour of these units. First of all, we will examine the main characteristics of each of them: The classic works on this topic are Kayne (1976) on French, and Cinque (1978; 1982) on Italian. Brito (1991) and Ojea (1992) are two recent studies dealing with both classes of relative clauses in Portuguese and Spanish, respectively. Caialan Working Papers in Linguistics (CWPL) 1992: 113-143. Univasitai Autbnoma de Barcelona 1 .l. Relative que Que is the unmarked relative form, since it is the only one that can occur in both restrictive and appositive clauses and develop any syntactic function (when necessary, preceded by the corresponding preposition and by the definite article).* The examples in (1-2) show the complex syntactic paradigm of que in restrictives and appositives, respectively: (1) a. El ministro (*el) que pronunció el discurso ... The-masc.,sg. minister (*the-masc.,sg.) that gave-3sg the speech The minister that gave the speech ...' b. El discurso (*el) que el ministro no pudo terminar ... The-masc.,sg. speech (*the-masc.,sg.) that the minister not could3sg finish The speech that the minister could not finish ...' c. La actriz (*la) que vimos anoche ... The-fem.,sg. actress (*the-fem.,sg.) that saw-lpl last night The actress that we saw last night ...' d. La actriz a *(la) que vimos anoche ... The-fem.,sg. actress to-acc. *(the-fem.,sg.) that saw-lpl last night The actress that we saw last night ...' e. La actriz a *(la) que entregaron el premio anoche ... The-fem.,sg. actress to-dat. *(the-fem.,sg.) that gave3pl the prize last night The actress to whom they gave the prize last night ...' f. El bolígrafo con (el) que escribo todas mis cartas.. . The-masc.,sg. pen with (the-masc.,sg.) that write-lsg. all my letters The pen with which 1 write al1 my letters ...' As will be argued later. we do not consider queand el que as two different relative forms, as do Ojea (1992) and other linguists. (3) a. [ ~ p El ministro, [ ~ p que pronunci6 el discurso I] b. [ ~ p El ministro, [Dp el pro [cp que pronunci6 el discurso I]] (4) a. [Dp El discurso, [cp que el ministro no pudo terminar I] b. [ ~ p El discurso, [Dp el pro [cp que el ministro no pudo terminar I] Some authors (such as d11ntrono (1979, ch. 15) and Martínez (1989, $ 4.8.2)) have proposed considering the type exemplified in (3-4 b) as a third class of relatives ("restrictive appositives", following dfIntronofs terminology). In our account there is no need to distinguish this set from the one of restrictives, given that the apposition in these cases is not exhaustively represented by the relative clause, but by a DP, whether we follow Bello or Rivero. There are some empirical proofs that el que does not constitute a unitary constituent here. One is provided by the fact that some units such as mismo or Único can intervene between the article and the complementizer in examples like (3-4 b). The possibility of commuting the definite article by a demonstrative such as aquel or by an indefinite such as uno suggests the same. Note that all these changes are not possible when the article is inside the relative clause, with the relative operator functioning as its complement, as in (1-2 d-h). Ojea (1992, $ 4.5) studies these constructions carefully and, after concluding that they do not forn a class different from restrictive relatives, assumes that they are connected with free relatives (i.e., relatives without an explicit antecedent). Following her view, the structure of (4b) would be like (9, where pro is the empty antecedent of the free relative clause: (5) El discurso, [Dppro el que el ministro no pudo terminar], ... As can be seen, unlike the analysis in (4b), in (5) it is assumed that el que functions as a relative pronoun whose grammatical features can identify the empty category preceding it. The prediction that this account seems to make is that all relative pronouns introducing a f r e relative clause can also occur in a "restrictive appositive". However, we think that this assumption does not work with quien. According to Ojea (1992, p. 122), (6) is ambiguous between a pure appositive reading (quienes = those present) and a restrictive appositive one (quienes = some of those present), as (7 a,b) exemplify, respectively: (6) Los asistentes, quienes querían participar en el coloquio, esperaron hasta el final. The people present, who wanted to take part in the colloquium, waited until the end' (7) a. Los asistentes, que querían participar en el coloquio, esperaron hasta el final. The people present, who wanted to take part in the colloquium, waited until the end' b. Los asistentes, lm que querían participar en el coloquio, esperaron hasta el final. 'Of the people present, those who wanted to take part in the colloquium, waited until the end' In our idiolect, however, (6) has only the interpretation of (7a), not the one corresponding to (7b). If our judgement of grarnmaticality is correct, quienes cannot occur in a "restrictive appositive" clause and, therefore, the parallelism between these constructions and free relatives is not so close as Ojea (1992) assumes. Therefore, we think that the appropriate structure of clauses including el que is that of (3-4 b), not that of (5). 1.2. Que preceded by preposition As for the obligatory intervention of the definite article intervening between que and the majority of prepositions in (1-2 d,e,g,h) as opposed to its optionality in (1 f), there is no satisfactory explanation in the literature at the moment. It is probable that a combination of factors are involved in this case, as has been often suggested. First, the syllabic nature of the preposition: bisyllabic prepositions, as well as prepositional expressions (locuciones preposicionales) reject the absence of the article before que, whereas the monosyllabic ones tend to accept it, aithough its presence is always more frequent. Allusions have also been made to the homonimic effects that can result from the lack of the article. Thus, sequences of preposition + que such as para que, a que, hasta que, por que could be confused with the corresponding subordinating conjunctions. It is probable that there are also syntactic factors conditioning these cases. In fact, Spanish does not always accept the lack of the determiner in an NP complement of a preposition: (8) a. Escribe con (el) bolígrafo.lE1 bolígmfo con (el) que escri be... Writes with (the) pen./The pen with (the) that writes '(S)he writes with (the) pen.'/The pen with which (s)he writes ...' b. Trabaja con *(su) amigo./El amigo con *(el) que trabaja ... Works with *(herlhis) friend.lThe friend with *(the) that works '(S)he works with *(her/his) friend.'/The friend with whom (s)he works ...I (9) a. Dispongo de(1) dinero./El dinero de(1) que dispongo ... Have- lsg of (the) money .lThe money of (the) that dispose- 1 sg 'I have of somelthe moneytlThe money of which I have ...I b. Te quejas de *(la) contaminaci6n.lia contaminaci6n de *(la) que te quej as... You-dat wmplain about *(he) pollution./The pollution about *@e) that you-dat complain 'You complain about the pollutionllThe pollution about which you complain ...I At first sight, some parallelism seems to exist between both types of construction, although a very careful scrutiny of these facts is still necessary to reach reliable conclusions. It must also be noted that the sequence preposition + que in relatives can only be used when the antecedent is definite, as shown in (10a,b): (10) a. El dinero de(1) que disponia ... b. Un dinero dell*de que disponía ... Moreover, when the article does not occur before the relative operator, minimality effects are manifested, as is shown by the presence of a negation intervening between the relative operator and its trace: (13) e. El autor, contra quien se ha querellado el alcalde, ... The author, against whom the mayor made a complaint, ...I As can be seen in (12-13), the only syntactic asymmetries that affect quien are the contrast of grammaticality between (12a) and (13a), and the logical lack of an appositive counterpart for (12 f-h). In our analysis, both phenomena have the same origin. The structure of free relatives poses interesting problems. The most important one is determining to which category they belong. Plann (1980) and Suñer (1984) claim that in these cases the relative pronoun has an empty antecedent pro, which permits proposing a structure for these constructions parallel to the other restrictive relative clauses. Notwithstanding, in a work about Catalan, Bartra (1990) argues that in examples like (12 h) the relative pronoun, which has a non-referential reading and acts as a logical variable with existential import, does not have an empty antecedent and that consequently the maximal projection of the overall sequence is CP. Note that in these examples it is not possible to commute quien to el que. This effect can be derived from the different category in which they are projected, following Bartra's analysis: CP and DP, respectively. Thus, the indefinite semantic value of the variable occurring in wnstructions like (12 h) would be incompatible with the intrinsic definite nature of el que: (14) a. No encontrk quienl*el que me atendiera. 'I could not find anyone to attend me' b. Hay quienlYel que te desea mal. There is someone that wishes you harm' (examples from Plam (1980), p. 113) 1.4. The relative el cual Finally, the paradigm of the relative pronoun el cua1 is characterized by having less referential autonomy than quien. Therefore, it must always have an explicit antecedent, which implies that this form cannot occur in free relatives. Ojea (1992, p. 99) associates this weakness to the speciai structure of this form, which contains an empty nominal head.3 According her, the definite article, having to identify pro in (15), cannot at the same time identify the otherpro that functions as antecedent in a free relative. El cua1 reproduces the same asymmetries between restrictive and appositive clauses that quien displayed: (16) a. *El bolígrafo el cual no escribe bien.. . The pen which does not write well ...I b. *El bolígrafo el cual te compr6 para tu cumpleaños ... The pen which 1 bought you for your birthday ...I c . El bolígrafo con el cual escri bo... The pen with which I write ...I d. *El autor el cuai escribi6 la obra enfermo ... The author who wrote the play while ill ...' e. El autor al cua1 entregamos el primer premio ... The author to whom we gave the first prize ...I (17) a. ?El bolígrafo, el cual no escribe bien, The pen, which does not write well,' b. ?El bolígrafo, el cual te compr6 para tu cumpleaiios The pen, which I bought you for your birthday,' 3 .In favour of this structure, Ojea (1992) adduces the cases in which this relative pronoun includes a lexical noun. They are examples such as the one that Cuervo (1893, s.v. cual) extracts from El Quijote: "A grandes voces llam6 a Sancho que viniese a darle la celada, el cual Sancho, oy6ndose llamar, dej6 a 10s pastores". (17) c. El,bolígrafo, con el cual escribo veinte cartas al dia, The pen, with which I write twenty letters a day' d. ?El autor, el cual escribi6 la obra enfermo, The author, who write the play ill' e. El autor, al cua1 entregamos el primer premio, The author, to whom we gave the first prize, The low degree of acceptability in (17 a,b,d) derives from some additional conditions affecting the use of el cual. Bello (1847, 1077) states that el cua1 can substitute que in appositive clauses when they are quite long and there is a perceptible pause at the beginning of the clause. 2. The analysis It seems natural to think that at least some of the asymmetries between restrictive and appositive clauses are derived from structural differences between the two. However, the same structure, that represented in (18), is usually assigned to both types, : Other proposals for distinguishing between both structures are not so influential at the present. In general, the linguists that defend this option (such as Smits (1989), Brito (1991) and Ojea (1992)) posit the structure of (19) for restrictive relatives, and tend to keep (18) for appositive relatives: The main argument for p i t i n g an identical structure for both types of relative clauses relies on the fact that in (19) the antecedent of the relative operator is N', not NP. The coindexing between the relative pronoun and the antecedent does not seem possible in this structure, given (21) b. Luis no sabe [que María se ha ido]. 'Luis does not know that Mary left' c. Luis no sabe [quiCn 10 hizo]. 'Luis does not know who made it' d. Luis no sabe [quC hacer]. 'Luis does not know what to do' (22) a. Luis preparó la cena. 'Luis cooked the dinner' b. iQu6 preparó Luis? 'What did Luis cook?' C. i Qut fn'o hace! What cold makes 'How cold it is!' d . Que tengas suerte. That have-2sg-subjunctive luck 'I wish you luck' In (21) we have four examples containing a complement clause. The first one is the only one in which the subordinate is not introduced by a [+QUI unit. Note that in this case the explicit mark of subordination is not placed in CP, but in IP (it is expressed by means of a defective verbal ternporality). On the contrary, in (21 b,c), the existence of an independent temporality in the subordinate clause is counteracted by the presence of a [+QUI unit that functions as a subordination rnarker in CP. From this point of view, we can characterize (21 d) as containing a complement clause with two marks of dependency: the [+QUI unit relates these constructions to (21 b,c). Furthemore, the lack of an independent temporality permits the association of these clauses with the special modal (epistemic) value that the literature often refers to. On the other hand, the sentences in (22) are syntactically independent. But only the first forms an independent proposition.5 Sentences in (22 b,c) each coníain an operator binding a variable. They both behave as "open sentences", with a marked modality (interrogative and exclarnative, respectively). The presence of a modal operator [+QUI in (22 d) implies that the corresponding sentence does not have propositional independence. Therefore, we can relate the [+QUI feature to an extended concept of "subordination" including marked modalities. Relative clauses are always introduced by a [+QUI element expressing the subordinate nature of these constructions. It is interesting to note that this mark is independent of the occurrence of a finite verb, as (23 c) shows: (23) a. Busca un libro que vio la semana pasada. Looks for a book that saw3sg last week '(S)he is looking for a book that (s)he saw last week' b. Busca un libro que trate de 16gica. h k s for a book that deais-SUBJ with logic '(S)he is looking for a book about logic' c. Busca un libro con que pasar el rato. h k s for a book with that to pass the time '@)he is looking for a book with which to pass the time' Moreover, we will suppose that the head of the CP of a relative clause is marked with the feature [+Rel], which must be adequately saturated in the course of derivation. The values of [+QUI and [+Rel] are different. [+QUI síands for the subordinate nature of the clause, as we have argued. As for [+Rel], it expresses the relative nature of the construction. We will see later If it is assumed that (22 a) constitutes a projection of IP (not of CP), this sentence would not be marked with respect to [iQu. What is important here is the contrast between this example and the rest of the sentences in (22). which present a [+QUI mark in spite of being "independent sentences". that in Spanish [+Rel] units exist that do not saturate a [+QUI feature. Therefore, both features must be considered separately.6 We will assume that the features [+QU, +Rel] must be adequately saturated in order to license a relative construction. Moreover, we will suppose, as usual in the generative grammar from its very beginning, that the relative pronoun is generated in the position that corresponds to its syntactic function in the subordinate clause and that it is moved to Spec,Comp in the syntax, a particular case of the general phenomenon of movement of XP to Spec. Therefore, the S- structure representation of a DP containing a relative clause like La persona a quien tu' admirar ... is reflected in (24): (24) [ ~ p La personai [cp [sp, a quieni ] [C <+QU> <+Rebi] [rp tú admiras ti]]] The person whom you admire ...' Suppose that the index of a relative clause is fixed in the head of CP and that it is licensed by the presence of a relative operator in Spec,CP. This relation is an instance of the general mechanism of spec-head agreement. Therefore, the coindexing relations shown in (24) are due to the interaction of three different mechanisms: principles of trace theory, which coindex quien with its trace; spec-head agreement, which relates quien with the [+Rel] feature in CO; and principles of predication stablishing the relation between the subordinate clause and its antecedent. At LF, the relative pronoun in (Spec,CP) is interpreted as an operator that binds the variable t. Being a true relative pronoun, guien is lexically rnarked [+QU, +Rel] and, therefore, it saturates the identical features placed in the head of CP. It is likely that [+Rel] results from the combination of some more basic features. As a matter of fact, it is equivalent to the cluster [+wh, +pred(icative)] proposed by Rizzi (1991) as a characterization of relative clauses. However, in that which follows we will continue using the feature [+Rel] for ease of exposition. As has been said in 9 1.1 and 1.3, a very important difference between Spanish and the rest of the languages for which an empty operator has been proposed is that que can occur as object of preposition, often preceded by the definite article. Thus, besides constructions as (25) -identical to the corresponding ones in English, French, Italian, Catalan or Portuguese-, Spanish also has the cases in (26), which are only possible in the other languages with the occurrence of a true relative pronoun: 8 (26) a. El niño que jugaba en el parque ... The boy that played in the park ...' b. El niño que vimos en el parque ... The boy that we saw in the park ...' (27) a. El niño al que viste ... The boy to-the that saw-2s The boy whom you saw ...I * In Catalan, the occwence of a tonic forn homonymous with the mstressed que is possible when preceded by a preposition, although other pronouns (such el qual) are preferred: (i) El llibre que vam llegir l'any passat ... The bwk that we r a d last year..! (ii) *El llibre deo) que et vaig parlar ahir... (iii) El llibre de qub et vaig parlar ahir... The book about which I spoke to you yesterday ...I (iv) *El llibre del qu2 et vaig par1 ar... (v) El llibre del qual et vaig parlar ahir... The book about which I spoke to you yesterday ...' The contrast in tonicity hetween que and 9 3 seems to reflect the different nature of these forns: the first is the complemetizer, the second behaves as a relative pronom. As the other true relative pronouns, qu2 cannot occw in a restrictive clause with an overt antecedent if it is not preceded by a preposition. Another significant difference with respect to Spanish is the non-availability of an intervening definite article between the preposition and qu& as is shown in (iv). (27) b. El niño de que te hablt? ... The boy of that to-you spoke-1s The boy about whom I spoke to you ...I If we accept the presence of an empty relative operator in (26), we must explain why the insertion of the complementizer is obligatory in these cases. Our idea is that the empty operator cannot saturate the [+QUI feature, by virtue of its lack of phonetic content. Therefore, it is necessary to insert a complementizer to show the subordinate nature of the construction.9 The analysis of the examples in (27) is a bit more problematic. The minimal hypothesis consists in supposing that Spanish has recourse to the empty operator strategy in these cases also, with the subsequent insertion of the complementizer at SS. The main advantage of this view is that it gives a unified account of the relative que, whether it is preceded by a preposition or not. However, such an analysis poses an important problem, since it makes it possible for an empty L. . unit to act as the objes of a preposition in Spanish, an option clearly not attested in any other case. However, we do not think that this problem about restrictivity is insoluble, because relative constructions possess a cluster of peculiar characteristics that can license some mechanisms that are not accessible to other constructions (in particular, the existence of a local antecedent of the relativized element, as well as the presence of the [+Rel] feature in the operator that binds the variable). Notice that English does not accept the identification of nul1 subjects and objects in general. Nevertheless, that relatives are possible in both cases. The fact that the complementizer can be absent in English (as in (i)) does not necessarily imply that the empty relative operator saturates the [+QUI feature, since its absence is also possible in complement clauses such as (ii). For an account of the ungrammaticality of (iii). cf. Rizzi (1990. 8 2.7): (i) The man (that) I spoke to yesterday was a famous actor. (ii) I think he is wrong. (ui) The man *(that) spoke to you yesterday was a famous actor. In order to avoid the restrictivity problem, almost all studies about Spanish relatives propose differentiating between three types of "relative que": the complementizer (that occurs in absence of a preposition); the true relative pronoun que, and the compound relative pronoun el que (these latter two forms being in complementary distribution with the former and occumng as objects of a preposition). If this analysis is accepted, the [+Rel] feature must be attributed to the second and to the third units. However, we think that it is possible to posit a unified analysis for all these forms; one that does not compel us to consider their similarity as a mere case of homonim y. First we will examine the contrast between the two patterns represented in (27). As has been suggested in 9 1.3, there are several factors that seem to intervene in the choice between preposition + que and prep + article + que. In our opinion, the increasing preference of Spanish for the second pattem is due to the fact that the definite article reproduces the grammatical features of the antecedent and therefore permits the identification of the empty relative operator more easily.10An empirical argument for the analysis we are presenting is provided by the superlative constructions of Canariense Spanish studied in Bosque & Brucart (1991). In this dialect, (28) is grammatical, just like the standard forms El amigo con el que tengo más conjianza and El amigo con el que más conjianza tengo The friend in whom I have the most confidence" (28) El amigo con el más confianza que tengo ... In (28), the phrase containing the superlative operator has raised to a position between the article and que. Consequently, the sequence con el que cannot form a unitary constituent, as is l0 It is interesting to note that the possibility of the occurrence of the definite article before que is cmelated to the fact that this unit has retained sorne deictic force in Spanish, in contrast with other Romance languages (cf. Brucart & m c i a 1987). However, this principle of vacuousness cannot adequately explain the asymmetry in (30) by itself, since direct objects are also affected by the phenomenon, as (30d) shows. As a matter of fact, any grammatical function that can be performed without a preposition manifests the same contrast. For instance, locative and temporal NPs: (33) a. Un sAbado que I *el cua1 viajt? a Tenerife ... 'A Saturday that I traveled to Tenerife ...I b. La hora que I *la cual pasé en la parada del autobús ... The hour that I spent at the bus stop ...I A slightly diferent, perhaps more promising way of explaining the contrasts in (30) consists in relying on the idea that transformationai processes are sensitive to ecomomy principles, as has been proposed in Chomsky (1991). Both quien and el cua1 share the characteristic of including grammatical features that permit the identification of their antecedent. Quien has number inflection in present-day Spanish and can head a relative without an overt antecedent. For its part, el cua1 includes the definite article, whose features identify the antecedent (by means of gender and number inflection). When the relative operator is not preceded by a preposition, the grammatical information camed out by these units is redundant with the one obtained through the coindexing between the head of CP and the antecedent. We can suppose that one of the contexts in which the application of a transformational rule is vacuous is found when a relative pronoun reproduces the grammatical features of the antecedent. This situation takes place when a relative pronoun is projected as an NP, but not when it is inside a PP, because the lexical and structural information carried out by the preposition is not contained in the antecedent. If there is an absolute identity, Spanish adopts the strategy of using the empty relative operator and inserting the cornplementizer. This mechanism seems to be governed by economy criteria similar to the ones that have been proposed in the "Avoid Pronoun Principle", exemplified in (34): (34) a. 61 lleg6 tarde. I Lleg6 tarde. He came3s late. I Came3s late. 'He came late' b. Luis puso la mesa y (*C1) cen6. 'Luis laid the table and (he) had dinner' In (34), the pronoun tends to be omitted -except if it includes some emphatic operatorll-, given that the verbal morphology guarantees the recoverability of the information it cames out. In a similar way, in (30b,d) the relative reiterates the grammatical content of the antecedent. In such a context, the recourse to the empty operator implies "least effort". Finally, it must be assured that this analysis N ~ S out sequences like (35), in which the relative pronoun has remained in situ and the complementizer has been inserted:l2 (35) *El libro que leímos el cu al... The book that we read which ...I It is not difficult to determine the illformedness of this sequence: as a true relative pronoun, el cua1 possesses the features [+QUI and [+Rel]. In order to license the relative pronoun, its features must be rnatched with the selectional features in COMP. But in (35) such a matching is impossible for various reasons. On one hand, the insertion of the complementizer saturates the [+QUI feature in COMP. On the other hand, the relative in situ cannot enter into an agreement relation with the head of CP. In the event that an empty relative operator is inserted in order to saturate the [+Rel] feature in COMP (as is presumably the case in examples of relative clause Cf. Rigau (1986) on the emphatic nature of subject pronouns in pro-drop languages. l2 For a different account of the ungrammaticality produced by the presence of two operators, cf. Bok-Bennema (1990). with a resumptive clitic as in (25)), the problern lies in the fact that the relative in situ is not licensed. since its features cannot be matched. 2.4. The analysis of appositive relatives The last asyrnmetry is that which sets up the ungrammatical status of (30b,d) against the grammaticality of its corresponding explicative counterparts in (31). Acwrding to our account, what (31) shows is that the reasons that precluded raising of the relative pronoun to COMP in (30b,d) have disappeared. This fact intuitively has some correlation with the higher degree of syntactic independence that appositive relatives have with respect to the matrix clause. In effect: whereas the restrictive clause acts as an intensional modifier of the antecedent and contributes to the very reference of the NP or DP to which it is subordinated, the appositive carries an independent secondary predication that does not affect the intensional wntent of the NP or DP that includes it. This explains why personal pronouns or proper nouns can have appositive relatives, but not restrictive ones. In the literature there are severa1 proposals that stablish this essential difference between both types of relatives. In our analysis, we will take the proposa1 presented in Safir (1986), that is independently motivated by reasons related to binding theory. Safir's proposa1 consists in considering that the grammatical level where correference between the relative and the antecedent is fixed is different in restrictive and appositive relatives. Whereas the coindexing between the relative and the antecedent in restrictive sentences applies at SS or LF.13 the same relation in appositive clauses is not stablished unti1 a later derivational stage, namely LF' (as proposed in Chomsky (1986b)). If we accept this suggestion, the reasons that inhibited the raising of some pronouns in restrictive relatives when they were not preceded by a preposition are not operating anymore. Thus, we can predict that relative pronouns will not be blocked in l3 In order for our amunt to work, we must assume that the coindexing in restrictive relative clauses takes place at SS. Perhaps &is point is subject to some parametric variation. For instance, the nonexistence of extraposed relatives in Spanish could be derived from the different value taken by Spanish and English, if we accept that the coindexing in an extraposed relative implies a previous operation of rmnstruction in LF. Browning, M. (1987), Nul1 Operator Constructions, Unpiblished PhD, Cambridge (Mass.), MIT. Brucart, J.M. & L. GrAcia (1987), "I SN senza testa: uno studio comparato", Rivista di Grammatica Generativa (Padova), 1 1: 1-32. Chomsky, N. (1977), "On Wh- Movement", in P.W. Culicover, T. Wasow & A. Akmajian (eds.) (1977), Formal Syntax, New York, Academic Press, 71-132. Comsky, N. (1982), Some Concepts and Consequences of tlie Tlleory of Government and Binding, Cambridge (Mass.), MIT Press. Chomsky, N. (1986a), Knowledge of Language, New York, Praeger. Chomsky, N. (1986b), Barriers, Cambridge (Mas.), MIT Press. Chomsky, N. (1991), "Some Notes on Economy of Derivations and Representation", in R. Freidin (ed.) (1991), Principies and Parameters in Comparative Grammar, Cambridge (Mass.), MIT Press, 417-454. Cinque, G. (1978), "La siniassi dei pronomi relativi cuie quale nell'italiano moderno", Rivista di Grammatica Generativa (Padova), 3: 3 1-126. Cinque, G. (1982), "On the Theory of Relative Clauses and Markedness", The Linguistic Review, 1: 247-294. Cinque, G. (1990), Types of A'-Dependencies, Cambridge (Mas.), MIT Press. Cuewo, R. J. (1893), Diccionario de construcción y régimen de la lengua castellana, Bogoki, Instituto Caro y Cuervo, 1952. D'Introno, F. (1W9), Sintaxis trarlsformacional del español, Madrid, CBtedra. FernBndez Ramírez, S. (1951), Gramútica española, Madrid, Revista de Occidente. Pnlarged and critica1 ed. by J. Polo and I. Bosque, 5 volms., Madrid, Arcollibros, 198519861 FernBndez Soriano, 0 . (1992), "Pronombres reasuntivos y doblado de clíticos", paper presented to the Segundo Coloquio de Granuitica Generativa, Vitoria, april 1992. Kayne, R. (1976), "French Relative <<que>>", in M. LujBn & F. Hensey (eds.) (1976), Current Studies in Romance Linguistics, Washington, Georgetown University Press, 255-299. Martínez, J.A. (1989), El pronombre. Numerales, indeJnidos y relativos, Madrid, Arcollibros. Ojea, A. (1992), Los sintagmas relativos en inglés y en español, Oviedo, Universidad de Oviedo. Plann, S. (1980), Relative Clauses in Spanish without Overt Antecedents and Related Constructions, Berkeley, University of Califomia Press. RAE (1973), Esbozo de una nueva gramática de la lengua española, Madrid, Espasa-Calpe. Rigau, G. (1986), "Some Remarks on the Nature of Strong Pronouns in Null-subject Languages", in 1. Bordelois, H. Contreras & K. Zagona (eds.) (1986), Generative Studies in Spanish Syntax, Dordrecht, Foris, 143-163. Rivero, M.L. (1980), "That-Relatives and Deletion in COMP in Spanish", Cahiers Linguistiques d'ottawa, 9: 383-399. Rivero, M.L. (1982). "Las relativas restrictivas con que", NRFH, 3 1: 195-234. [Reproduced in M.L. Rivero (1991), 35-77] Rivero, M.L. (1991), Lar construcciones de relativo, Madrid, Taurus. Rizzi, L. (1990), "Speculations on Verb Second", in J. Mascar6 & M. Nespor (eds.) (1990), Grammar in Progress, Dordrecht, Foris, 357-386. Rizzi, L. (1991), Relativized Minimality, Cambridge (Mass.), MIT Press. Safir, K. (1986), "Relative Clauses in a Theory of Binding and Levels", LI, 17: 663-689. Schroten, J. (1984), "Two Approaches to the Distribution of Spanish Relative Pronouns", Estudis Grarnaticais (UAB Working Papers in linguistics), 1: 295-327. Smits, R.J.C (1989), Eurogrammar. Zhe Relative and Cleft Constructions of the Germanic and Romance Languages, Dordrecht, Fons. Suñer, M. (1984), "Free Relatives and the Matching Parameter", The Linguistic Review, 3: 363-387. Departament de Filologia Espanyola Facultat de Lietres EdiJci B E-O8193 Bellateva