The Galician Inflected Infinitive and the Theory of UG

The Galician Inflected Infinitive and the Theory of UG Víctor Manuel Longa Universidade de Santiago de Compostela and Universitat Authnoma de Barcelona In this paper, I will focus on the properties of the Galician Inflected Infinitive (heuceforth. GII). I argue, against Raposo (1987), that the behaviour of the Iníiected Infinitive in Portuguese and Galician is not the same. I will contend that all the GI1 structures are full CPs. Another important property of Galician, closely related to the existence of the Inílected Infinitive, is that agreement with non-finite forms is not restricted to the infinitivd forn: the gerund can also bear agreement. The analysis of Iníiected Infinitive constructions as CPs will be extended to the iníiected gerund. 1. Introduction Although the linguistic tradition is full of references to the Galician-Portuguese domain, the study of its linguistic properties has been actually focused mostly on Portuguese. Thus, the study of Galician has been always subsidiary to the study of Portuguese. The same applies to the study of the Inflected Infinitive in the referred area: although this phenomenon exists in both languages, the Galician Inflected Infinitive has been described as showing the same pnnciples governing the Portuguese Inflected Infinitive (henceforth, PII). Within the GB framework, the situation has not changed at all: for insíance, Raposo (198792 fn 9) claims that the properties of the Inflected Infinitive in both languages are presumably the same. In this paper, I shall focus on (some of) the main properties of the GII, with special reference to the characteristics in which both languages differ. Although the properties of the Inflected Infinitive in both languages are not expected to be extremely different, GI1 shows some properties that differ subsíantially from PII. Catalun Working papers in Linguistics (CatWPL) 4.1 (1994): 2344 Universitat Autbnoma de Barcelona This paper is organized as follows: in section 2 general data of GI1 are presented, showing the main differences observed in relation with the Portuguese phenomenon. In section 3, Raposo's analysis will be briefly presented. Sections 4 and 5 will provide an analysis of the GI1 appearing as subject clause and as subcategorized complement, respectively. In section 6 the structure of the Inflected Infinitive introduced by a preposition will be analyzed, and some aspects related to clitics will be briefly discussed. Section 7 presents an analysis of the Galician Inflected Gerund, and finally, in section 8 some general conclusions will be pointed out. 2. The GI1 Data Both in Galician and in Portuguese, the Infl(ection) element of infinitives is not specified for Tense distinctions, but it may be specified for Agr(eement) distinctions.1 The Agr-markers in Galician are esentially the sarne than the ones in Portuguese, abstracting from some phonetical differences: (1) 1 eu ter+@ n6s ter+mos I have-INF we have-INF+ lpl 2 ti ter+es v6s ter+des you have-INF+2sg you have-INF+2pl 3 el ter+a2 eles ter+en slhe have-INF they have-INF+3pl 1 Usually, an infinitival forn is neither specified for tense distinctions nor for Agr distincticns. Thus, as Raposo (1987:93-4) points out, the Inflected Infinitive is an extremely rare combination with respect to the options that Universal Grammar offers. 2 As Gondar (1978:26) points out, in some diaiects of Galician first and third person singular end im ..e. 24 2.1. Some general restrictions on the ocurrence of the GI1 are not different to the ones that apply in Portuguese according to Raposo (1987:87): GI1 cannot appear as an independent clause nora matrix clause. (2) a. *Eles arranxaren, o muiño. they arrange-INF-3pl the mill b. *Eles admitiren chegar onte. they admite-INF-3pl anive-INF yesterday c. É doado supoñeren as cousas. is easy suppose-INF-3pl the things 'It is easy that they suppose the things.' d . *E doado que supoñeren as cousas. is easy that suppose-INF3pl the things Thus, the GI1 structures are only possible as embedded clauses, but without a complementizer, as the contrast (2c,d) shows. 2.2. More specifically, the contexts in which GI1 can appear are very much the same as PII, but with some differences in certain contexts that we are going to examine in next sections. 2.2.1. Subject Clauses. In this context, a preverbal subject is not possible in Galician, in clear contrast with Portuguese? where the preverbal position for the subject is the canonical choice: (Portuguese) (Galician) 2.2.2. Complements Subcategorized by Certain Predicates. PI1 is allowed with matrix epistemic, declarative, and factive verbs, but GI1 can only appear in the subcategorization context of declmtive verbs.6 This is shown below: 2.2.2.1. Epistemic Verbs: (4) a. EU penso terem os deputados trabalhado pouco. I think-lsg have-INF-3p1 the deputies worked little (Portuguese) (3) a. Non esta claro aprobármo-10 exame.4 not is clear pass-INF-lpl-the exam b. Será difícil eles aprovarem a proposta.5 be-FUT-3sg difficult they approve-INF-3pl the proposal c. *Non esta claro n6s aprobármo-la proposta. not is clear we approve-INF-lpl-the proposal In wntrast with Portuguese, the subject clause may be at the beginning of the sentence in Galician: (i) Aprobánno-10 exame non está claro. pass-INF-lpl-the exam not is dear On the other hand, as regards (3) and similar examples, it must be observed that in Galician the definite article cliticizes onto the verb, depending on some phonetical conditions (i.e., when the verbal forn finishes in -s or -r). For this reason, this cliticization should not be wnfunded with the infinitival Agr affix. The Portuguese data have been taken from Raposo (1987). The use of GI1 in this kind of constructions is much more restricted than in Portuguese. (4) b. *Xoan pensa xantaren os pais m~ito.~ Xoan thinks eat-INF-lpl the parents a-lot (Galician) 2.2.2.2. Factive Verbs: (5) a. EU lamento os deputados terem trabalhado pouco. (Portuguese) I regret-lsg the deputies have INF-3pl worked little b. *Lamentei traballaren os meus amigos. (Galician) regreted-lsg work-INF-3pl the my friends 2.2.2.3. Declarative Verbs: (6) a. EU afirmo terem os deputados trabalhado pouco. (Portuguese) I claim-lsg have-INF-3pl the deputies worked little b. O mestre afirmou fdrmo-las cousas. (Galician) the teacher claimed3sg make-INF-lpl-the things 2.2.2.4. Portuguese epistemic and declarative verbs do not allow the subject of the Inflected Infinitive to appear in preverbal position, while factive matrix verbs allow both preverbal and postverbal position. In Galician, the Inflected Infinitive appearing in subcategorized complements of dedarative verbs does not allow preverbal subjects neither, as exemplified in (7): (7) a. *O mestre afirmou os nenos faceren as cousas. the teacher claimed3sg the boys make-INF-3pl the things b. O mestre afirmou faceren os nenos as cousas. A few speakers of Galician may accept this kind of construction, but this choice is clearly marginal and it must be related to contact between Galician and Portuguese. 2.2.3. Adjunct and Predicative Clauses: (8) a. Fixérono para traballaren ledos. made-3pl-3sgAcc for work-INF-3pl happy b. Isto non C para te recolleres. this not is for yourself retire-INF-2sg Subjects of infinitival adjuncts and predicative clauses appear usually in postverbal position, but they can also appear in preverbal position: (9) Pra ti sanares téíienche que levar de rneia noite.8 for you cure-INF-2sg have-INF-2sgAcc that carry-INF by mid night 'If you want to cure you have to be canied in the midnight.' As we will see below, this is possible because of the prepositional status of the construction. 2.2.4. Infinitives Subcategorized by N or A: (10) a. Admitiu o feito de faceren a tarefa. admite-PAST3sg the fact of make-INF-3pl the task b. Estades desexosos de rematárde-10 traballo. are anxious about finish-INF-2pl-the job In the examples of (10), a dummy preposition must be introduced in order to license these constructions because N and A cannot assign structural Case to their complement.9 ~aken from Gondar (1978:64). Infinitives do not need Case in any language, but in section 3 we will see that, in Raposo's framework. Case assignment to the infinitive is crucial to license constructions with infiected infinitives. 2.3. As for the position of clitics, some important differences can be found between Galician and Portuguese, as we can see in these examples: (1 1) a. de nos entenderdes I * de entend6rdesnos (Portuguese) if lplAcc understand-INF-2pl if understand-INF-2pllplAcc b. de nos entenderdes l de entendbrdesnos (Galician) Enclisis is not possible in this context in Portuguese (cf. Benucci (1992)), but both enclisis and proclisis are allowed in Galician. 3. Raposo's Analysis 3.1. I will adopt the main points of the analysis offered by Raposo (1987). In Raposo's view, the existence of the Inflected Infinitive is due to the interaction of two parameters: the Infl pararneter and the nul1 subject parameter. 3.1.1. The Znfl Parameter. This author states that, in many languages, if Infl is finite, then it is specified for Agr, and that if Infl is specified for Agr, then it is finite. However, this does not hold for Galician nor Portuguese, because in both languages an infinitival Infl may be specified for overt Agr distinctions. Thus, in many languages, if Agr is overtly specified, the choice of [&Tensel is predicted not to be free, but in Galician and Portuguese the choice is actually free. 3.1.2. The Nul1 Subject Parameter. According to Chomsky (1982), Agr may be specified for Case in the pro-drop languages. The central hypothesis of Raposo derives from this assertion: "In the absence of [+Tensel, Infl (or Agr in Infl) is capable of assigning nominative Case to a lexical subject only if it is itself specified for Case." (Raposo (198792)) 3.2. Raposo analyzes subject clauses and the complements of factive verbs, with subject-verb order, as bare IPS, without a CP level: the matrix Infl governs and assigns Case to the embedded Infl. For factive with verb-subject order, epistemic, and declarative constructions he proposes a CP structure: the Infl element of the matrix verb cannot govern the embedded Infl, and, consequently, cannot assign Case to it, because CP constitutes a barrier. However, following Belletti and Rizzi (1981), Raposo assumes that a maximal projection is not an absohte barrier in the sense that an element outside lt can govern its specifier and head positions. In the case of CP structures, V governs the head of CP; therefore, if the embedded Infl raises to C, it will be governed and assigned Case features. Thus, Infl-to-Comp raising is the crucial proposal of Raposo's analysis.10 3.3. As I have already mentioned, I will adopt the essentials of this analysis, but I will assume that ai1 GI1 structures are full CPs. Each case will be briefly anaiyzed in the next sections. 10 There are, however. some problems concerning the siatus of the subcategorized structures in Raposo's framework. Galves (1991) points out some of these problems, but she does no! address all of hem. In my view, the main problem of Raposo's conception is that his analysis depends excessively on the position of the subject. The double option IPlCP is unelegant in order to explain the double possibility allowed by factive verbs. For a similar conception to the one offered in this paper, see Hye-suk Yoon and Bonet-Farriln (1988). wheae all the non-nominal infinitival constructions are analyzed as CPs. 30 4. The Structure of GI1 as Subject Clauses Consider (3a), repeated here as (12): (12) Non está claro aprobármo-10 exame. not is clear pass-INFlpl-the exam (13) is part of the structural representation of (12): (13) non está claro [cp [C [C aprobarmosl] [rp pro [I t'l] [vp tl o exame]1]11 In this S-structure representation, V, generated in the VP node, has raised to the head position of IP, picking up the features of Infl in this way. But this movement is not enough, because the embedded Infl is not govemed in and it can not receive Case in that position. For this reason, [V+I] must raise further to the head of CP. In this position, the embedded Infl will be governed and assigned Case features by the matrix Infl (cf. Belletti and Rizzi (1981), Rizzi (1982), Chomsky (1986), Raposo (1987)). As a consequence, the embedded Infl will be able to assign Case to its lexical subject (if present). If V does not raise to the C position, the embedded Infl will not be govemed nor assigned Case, and therefore, will not be able to assign Case to its lexical subject. Of course, the embedded Infl could be govemed by V if an IP is postulated, but the structure with a bare IP would not explain the verb-subject order.12 It must be noted that the analysis presented here does not disagree with the one supported by Raposo, who suggests that the CP projection causes the verb-subject order. However, . Raposo's analysis is empirically incomplete in the view of the Galician data, and, therefore, not really valid for the GII. The analysis presented here applies only to GII, not to PII. For this reason, here we do not pursue a unified approach to the subject PI1 (subject-verb order) and to 11 As was said before, preverbai subjects are not ailowed in this context (cf. (3c)). l2 I assume that the subject appears in the [SpecJP] position 6.6. As was said in section 2, the clitics appeanng in the GII constructions show much more mobility than in the Portuguese equivalent constructions. In Galician the canonical position for clitics in these structures is the enclitic position: V + clitic, unlike Portuguese. Consider (22): (22) a. de nos encontramos17 (Portuguese) of lplAcc meet-INFlpl a'. *de encontrármonos b. de nos atoparmos (Galician) of lplAcc meet-INF-lpl b ' . de atopámonos c. de o faceren (Galician) of 3sgAcc make-INF3pl Therefore, the double possibility, enclisis and proclisis, is allowed in Galician. These data are relevant to the question of the status of prepositions and its relation with cliticization, as 1 will show. In Benucci's view, full prepositions may be assimilated to the que ('that') complementizer, but this is not possible in Galician. If strong prepositions were really complementizers, we would expect proclisis, not enclisis. For reasons that we will see directly, clitics offer evidence that P does not occupy the first position of the clause because the canonical position is enclisis, not proclisis. (23) shows the inexistence of parallelisms between P and the complementizer: (23) a. Penso que Xoan o magullou (vs. *magullouno). think-lsg that Xoan 3sgAcc scratch-PAST3sg scratch-PAST-3sg-3sgAcc b. Pensamos para facérmolo. thinklpl for make-INF-lpl3sgAcc 17 The Portuguese data have been taken from Benucci (1992). 38 If P and the complementizer were really equivalent, we would not expect enclisis in (23b). However, if we assume that P is generated outside the CP projection, and if, according to Benucci, clitics have to occupy the second position inside CP, then we can account for the enclisis facts: (24) a. ... [cp que o fagan 1st. 2nd b. ... para [cp fackren-o 18 1st. 2nd 7. The Galician Inflected Gerund 7.1. The last characteristic of Galician I would like to refer to in this paper is that the possibility of having [+Agr] with [-Tensel is not restricted to the infinitive. The gerund may also have full agreement with its lexical subject. This agreement is phonetically realized in less forns of the verbal paradigm than it is in the infinitival forn. The paradigm is offered in (25): (25) 1 eucantandw0 nós cantando+mos I singing we singing+lpl 2 ti cantandM vós cantándo+des you singing you singing+2pl 3 el cantand4 eles cantando+@ s/he singing they singing l8 Benucci presents empirical evidence on the double possibility in Portuguese, enclisis and proclisis, but only when an adverb is placed between both elements, clitic and P. He thinks the adverb is transparent, and for this reason, proclisis is allowed; In Galician, as we have seen, we can have proclisis quite independently of the presence of any intervening element. In proclisis, the clitic would climb to the left of V element, being adjoined to it. On the other hand, this hypothesis cannot explain examples like (22b). At the present stage, I have no consistent explanation for this example. Consider these two examples: (26) a. Cantandodes así, gañaréde-10 premio. singing-2pl in-this-way win-FUT-2sg-the prize b. TCndomos paciencia pasará todo. having-lpl patience finish-FUT-3sg everything In (26a,b) there is no lexical subject. The following examples show that preverbal subjects are not possible: 19 l9 A preverbal subject is possible in certain cases, like the following one offered to me by C. Folgar: (i) Nós traballando e vós cautando. we working and you singing The use of the Galician Inflected Gerund (henceforth, GIG) is not possible in these sentences (cf. Longa (1993)). as this example shows: (ii) *Nós traballándomos e vós cantándodes. Fnrthermore, structures like (i) can not be inserted in a matrix senteuce, as opposed to example (28). Iu the sentence considered in this footnote, the Aux element has been deleted, but the crucial point is not deletion in itself but the fact that such a deletion is incomplete. In spite of the absence of the Aux element, some of its features remain inert, mainly agreement. We can explain in this way the impossibility of the presence of the Iníiected Gerund. The structure (ii) is made ungrammatical by the presence of the agreement feature because it violates a basic condition of the Galician and Portuguese non-personal inflected forns, which detennines the ungrammaticality of a structure like (iii): (ui) *... Z...X-Agr Y -Agr...W... where Z and W are iu the place of variables. X is a finite verb and Y is a non personal forn, as in (iv): (iv) *Queremos cantarmos. wantlpl sing-INF-lpl With a postverbal subject these restrictions disappear because the Aux element has not been generated. 40 (27) a. *Vós cantándodes así, gañarkde-10 premio. you singing-2pl in-this-way win-FüT-2sg-the prize b. *Nós téndomos paciencia pasar6 todo. we having-lpl patience finish-FüT-3sg everything Only subjects in postverbal position are ailowed: (28) a. Cantándodes v6s así, gaiiarkde-10 premio. singing-2pl you in-this-way win-FüT-2sg-the prize b. Tkndomos n6s paciencia pasará todo. having-lpl we patience finish-FUT3sg everything The explanation for the exarnples of (28) is quite similar to the one proposed for the case of the Inflected Infinitive with verb-subject order. The structure of (28a) is represented in (29): (29) [CP [ct [c cantAndodesi1 [rp vós [It [r t'il [VP t illlll . .. As in the case of infinitives, the proposa1 of an IP projection would ailow Infl to be governed, according to Raposo's Case theory, but it would not give an account for the verb-subject order. So, V must raise to Comp after passing through I. In its final placement, V is governed by the matrix Infl. This element assigns Case features to the verb in such a way that Agr will be able to assign Case to the lexical subject. The Infl-to-Comp approach is once more the most plausible analysis. 8. Conclusions The analysis o€ the daia offered in this paper support the claim that, although GI1 and PI1 are not extremely different phenomena, they differ in some crucial properties like the basic order (subject-verb vs. verb-subject), the position of clitics, the levels of projection in the structures, and so on. Moreover, it has been shown that given the existence of enclisis and the impossibility of contraction with the article, Galician prepositions seem to be true prepositions. These facts show that Galician prepositions do not occupy the [Spec,C] position, against Kayne's analysis for some Romance prepositions.20 It has also been shown that agreement can manifest overtly not only in the infinitive but also with gerund forms. One of the main objectives of this paper is to defend the idea that all structures with non finite inflected forms should be analyzed as CPs, not as IPS. So, we have found further evidence for Kayne's (1991) proposal about the CP nature of infinitival complements of a V, and we have unified the structures of infinitives and gerunds. On the one hand, Raposo establishes an automatic relation between Infl-to-Comp raising and the existence of a CP structure, that result in the verb-subject order, and, on the other hand, between the lack of Infl-to-Comp raising and the absence of a CP projection in the cases of subject-verb order. This kind of automatic relation is not plausible for Galician: although there is no Infl-to-Comp movement in the cases with an Inflected Infinitive introduced by P and with subject-verb order, the relevant structure has been shown not to be an IP but a CP. Summing up, Galician grammar chooses inherently a CP structure in all the cases with non finite inflected forms, independently of the subjectverblverb-subject order. The hypothesis developed in this paper should be applied also to the PII. However, as it has been mentioned before, to undertake this task is beyond the scope of these pages. 20 As we have seen, there are some cases in which contraction is possible, but in those cases P occupies a position outside CP in such a .way that another preposition can stay in [Spec,C]. Therefore, when only one preposition is present it must occupy the position of a true P. 42 -- This is an extended version of a paper presented at the I1 Workshop on the Syntax of Central Romance Languages, Barcelona, May, 1993. The beginning of this research was supported by a grant of the Xunta de Galicia that made possible my stay as visiting scholar in the Universitat Autbnoma de Barcelona. I would like to thank the comments on an earlier version of this paper offered by F. D'Introno (University of Massachusetts at Amherst), M.L. Hernanz (UAB), T. JimCnez JuliB (Universidade de Santiago), G. Rigau (UAB), and a CatWPL anonymous reviewer. I am specially indebted to G. Lorenzo (Universidad de Oviedo and MIT) for his comments and his help. References Baker, M. (1988) Incorporation. A Theory of Grammatical Function Changing , Chicago University Press, Chicago. Belletti, A. and L. Rizzi (1981) 'The syntax of ne: Some Theoretical Implications', The Linguistic Review 1, 1 17154. Benucci, F, (1992) 'Prepositional Particles and the Portuguese Personal Infinitive', WPL University of Venice, 1-20. Chomsky, N. (1982) Some Concepts and Consequences of the Theory of Government and Binding, MIT Press, Cambridge, Mass. Chomsky, N. (1986) Barriers, MIT Press, Cambridge, Mass. Galves, C. (1991) 'Inflected Infinitive and Agr Licensing', ms. Universidade de Campinas. Gondar, F. G. (1978) O infinitivo conxugado en galego, Universidade de Santiago, Santiago de Compostela. Hye-suk Yoon, J. and N. Bonet-FarrBn (1988) 'The Ambivalent Nature of Spanish Infinitives', ms. University of Chicago. Kayne, R. S. (1991) 'Romance Clitics, Verb Movement, and PRO', Linguistic Inquiry 22, 647-686. Longa, V. M. (1993) 'Gerundios galleges, posición de sujetos y concordancia', paper presented at the XXIII Symposium of the Sociedad Española de Lingüística, Lleida, december 1993. Raposo, E. (1987) 'Case Theory and Infl-to-Comp: The Inflected Infinitive in European Portuguese', Linguistic Inquiry 18, 85-109. Rigau, G. (1992) 'Propiedades de Flex en las construcciones temporales de infinitivo: la legitimación del sujeto', ms. Universitat Autbnoma de Barcelona. Rizzi, L. (1982) Issues in Ztalian Syntan, Foris, Dordrecht. Rizzi, L. (1990) Spiegazione e Teoria Grarnmaticale, Unipress, Padova. Praza Isabel a Catdlica, 2, 2"E E-36204 Vigo