Preliminary Remarks on Old Spanish Auxiliares : haber, ser and estar

PRELIMINARY REMARKS ON OLD SPANISH AUXILIARIES: HABER, SER, AND ESTAR Montse Batllori Universitat de Girona The main purpose of this paper is to provide the reader with a preliminary description of the Old Spanish auxiliary system. As has been established in the literature, the future and conditional auxiliary is functional and the auxiliaries that constitute infinitival periphrases, either with temporal or modal meaning, and perfect compound tenses are lexical (cf. Rivero (in press)). From this standpoint, ow goal is to show that: (a) Old Spanish temporal and modal periphrases can hardly conform to an analysis in which a VP subcategorizes for another VP; (b) the non-finite verb of the periphrastic future and conditional constructions undergoes excorporation in its movement to cO; (c) the so-called perfect auxiliaries are lexical verbs which subcategorize for a small clause. Although ow proposal is not totally free of the problems associated to the nature of participle preposing in compound tenses, it allows us to relate this movement to more general linguistic phenomena such as object preposing and topicalization in embedded sentences. Following Martínez-Gil (1989), we will assume that Old Spanish is a VO language. Section 1 will be devoted to infinitival periphrases. For the time being, we will limit this paper to the description of empirical data. In section 2, we will present an analysis of periphrastic futures and conditionals following Cardinaletti & Roberts (1991). Our purpose is to provide empirical support for the view that the non-finite verb, which moves to CO in order to satisfy Tobler-Mussafia's Law, excorporates from the auxiliary. Catalan Workhg Papers in Linguistics (CWPL) 1992: 87-1 12. Universitat Autbnoma de Barcelona Section 3 will focus on the nature of participle-preposing constructions and the so-called 'perfect auxiliaries'. Since we postulate that they are lexical verbs, we should reverse the selection frame and instead of talking about 'auxiliary selection' we should say that: (a) haber subcategorizes for a small clause the predicate of which is the participle of a transitive or an active intransitive verb; (b) if the main verb is ser, the predicate of the small clause is unaccusative (the participle is either ergative, reflexive, deponent or passive). Tremblay's (1991) proposal about the differences and similarities between avoir (haber) 'have' and Etre (ser) 'be' concerning Case Theory gives an account of the differences in subcategorization. As for the structure of the Old Spanish constructions under discussion in this section, we will follow Cardinaletti & Roberts (1991), Kayne (1985) and Suñer (1990). In accounting for the syntactic change that has taken place from Old to Modern Spanish, we will follow the theory of Language Change revised and improved in Roberts (1992). We will assume Lois' (1989) analysis of compound perfect tenses in Modern Spanish and will make use of Lema's (1991) hypothesis concerning ASPP. 1. Infinitival periphrases According to Yllera (1973), Old Spanish periphrastic constructions in which haber, ser and estar play a central role belong to three main different sets: modal periphrases, temporal periphrases, and 'haber + que + injinitive'. For modal periphrases which express obligation (root meaning), Picallo (1991) postulates an analysis for catalan according to which a VP selects another VP as a complement: Although we will assume this analysis at this time, it should be noticed that Old Spanish data do not fit in the analysis completely because, in many cases, there are not enough stmctural sites to accommodate all the elements which appear in the attested sentences. As shown in (2), the subject usually precedes the auxiliary:' (2) Por esto nós avemos de trabajar mucho por aver ... (Cal.: 128) for this we have-AUX to work-INF much to have ... This is the reason why we have to work so much to have ...' However, we find two examples in which the subject follows the infinitive: (3) desi a pocos dias ovo de venir el religioso a aquella +dat. (Cal.:3 19) thus in few days had-AUX to come-INF the religious to that city 'and in a few days the religious had to come to that city.' (4) et estés en el camino por do ha de par el venador. (Cal.:222) and stay-SUB in the path by where have-AUX to pass-INF the deer-hunter 'and stay in the path by which the deer-hunter must go.' Notice, though, that in both cases the infinitive is an ergative verb, which could account for the postposition of the subject.2 The insertion of an object between the infinitive and the subject is exceptional. We find a single example (with epistemic meaning): Tiuoughout the paper Old Spanish examples are labelled according to source. Cal. stands for Libro de Caiila e Dimna. Faz. stauds for La Fazienda de Ultra Mar and Buen Amor for Libro del Buen Amor (se references). We thank Dr. J. M. Brucart for this observation. 89 (5) Ovo de cuer entre ellos contienda sobre las wsas deste mundo. (Cal.:325) had-AüX to fall-INF among them wntest about the things of-this world They happened to discuss about the things of this world.' Likewise, there are only two attested examples in which the subject intervenes between the 'auxiliary' and the preposition: (6) et avrk ni de creer 10 que te dixeren de mi. (Cal.:3 14) and have-AüX-FUT you to believe-INF it that to-youcl tell-FUT.SUB of me 'and you will have to believe what they will tell you about me.' (7) E ovo esto a seer. (Faz.:55) and had-AUX this to be-INF. 'And this had to be (this way).' In contrast, the insertion of clitic object pronouns is fairly frequent, due to the positional restrictions of Tobler-Mussafia's Law: (8) Puesto qu'el le6n non me pensase mal (...), avríalos de escuchar. (Cal.: 162) given that-the lion not of-meu think-SUB badly, have-AüX-COND-to-themu to listen-INF To prevent the lion from thinking wrong of me, I would have to listen to them.' As for the occurrence of adverbs or adverbial phrases in this position, only one example was attested: (9) et por la tribulaqion que avia en parte de aver (Cal.: 157) and for the tribulation that I had-AUX in part to have-INF 'and for the trouble I was to endure in some degree' As said and illustrated above, the interposition of such constituents between the 'auxiliary' and the infinitive could represent a problem for Picallo's (1991) proposal because it is not only the clitic but also the subject and some adverbial phrases that may appear in this position. Moreover, as the following example shows, there may be a negative marker between the preposition and the infinitive of the periphrasis, which suggests that this constmctions should be analyzed as a VP which selects a CP: (10) et yo sd tenudo de te non enbiar sin respuesta de tu ruego. (Cal.:245) and I arn-AUX obliged to to-you~~ not send-INF without answer to your plea 'and I must not send you back without an answer to your plea.' Notice that along with the passive form of 'tener de + infinitive', which according to Yllera (1973) means 'to be obliged', focusing on the person who must do the action, the negative marker appears between the clitic that follows the preposition and the infinitive. If we assume that negation is generated under NEGP, the projection subcategorized must be a CP. It could be argued that the negative marker could have inwrporated into the verb owing to its clitic nature. However, our first impression is that non is not unstressed in Old Spanish because clitic pronouns can occur after the negative marker, at the beginning of a sentence, and after a pause without bringing about a violation of Tobler-Mussafia's Law: (1 1) ca el que se atreve a fiarse en alguno , non lo aviendo provado ... (Cal.:317) for the who SECL dares to rely-SECL on someone, not to-himCL having-AUX proved 'for the one who dares to rely on someone without having proved his worth ...' Finally, we will devote the last part of this section to illustrate clitic climbing in Old Spanish and, in particular, the sites in which unstressed object pronouns may appear. Three positions for clitics are attested: preceding the infinitive (12), following the 'auxiliary' (13), and preceding the 'auxiliary' (14). (12) oviCsemos de nos enemistar de la natura. (Cal.:206) have-AUX-PAST.SUI3 to USCL become-enemies-INF with the nature 'we had to become enemies of nature.' (13) avríalos de escuchar. (Cal.: 162) have-AUX-COND-bthema to listen 'I would have to listen to them.' (14) et yo te he de dezir muchas estorias et fazañas. (Cal.:209) and I to-youa have-AUX to say-INF many stories and deeds 'and I have to tell you many stories and deeds.' 2. Periphrastic futures and conditionals In this section we will examine periphrastic future and conditional constructions. Many of the data show that the infinitive moves to @to prevent the clitic pronoun from occupying the first position in the sentence. Whenever there is no unstressed pronoun, the infinitive undergoes incorporation into the auxiliary. Compare the two forns in italics in (15): (1.5) Quando ovieses muerto estos, apoderarse an de ti et avrh tu reino asi wmmo 10 ante avían. (Cal.:286) when have-AUX-PAST.SUB killed these, seize-INF-SEcL have-AUX of you and have-FUT your kingdom so as itCL before had. 'After having killed them, they will seize you and will have your kingdom as they did before.' Sometimes a stressed element precedes the periphrastic future (or conditionai). This might be regarded as a problem because, aithough this element already occupies the first position in the sentence, it does not preclude the movement of the non-finite verb to the left of the clitic: (16) et yo tajarte he esta red un nudo en pos otro. (Ca1.:271) and I cut-INF-to-youa have-AUX this net one knot in-after other 'and I will cut you this net knot by knot.' Actually, what is happening is that the initial constituent is topicalized and occupies a position which is external to CP. Its presence does not affect Tobler-Mussafia's Law, which, as suggested by Lema and Rivero (cf. Lema (1991), Rivero (1988, 1989, 1990, 1991, in press), and Lema & Rivero (1989, 1991, 1992)), applies within CP. It is well known that Old Spanish discourse linkers such as et, mas, and desído not count as initial elements because they do not appear inside CP. As for the structure of Old Spanish periphrastic futures and conditionals, we could adopt either the analysis proposed by Lema and Rivero or the one postulated in Cardinaletti & Roberts (1991). We prefer the latter because its explanatory adequacy is higher (crosslinguistically as well). It must be pointed out, though, that this analysis will be applied here only to Old Spanish. Cardinaletti & Roberts (1991) propose that V2 languages and all languages in which clitics must satisfy either Tobler-Mussafia's or Wackernagel's Law have a maximal projection -AGRlPbetween CP and the traditional AGRP, which they call AGR2P. This proposal provides us with a position for clitics in languages which show clitic second effects (or for the verb in V2 languages) -AGRlOand a landing site to place topicalized elements in embedded sentences -SPEC of AGRlP. Thus, it can be claimed that Old Romance languages have two TOP positions: one external to CP and the other one within CP. Furthemore, whenever the subject follows the future or conditional periphrasis, it is located in SPEC of AGR2P, where it receives nominative Case under government by the auxiliary haber (which occupies the head of AGRlP). Examples (17), (18), and (19) illustrate subject postposition in periphrastic futures and (20) is the analysis given to these constructions: et dezirte ha él que ante fue 61 all6 que iú. (Cal.:342) and tell-INF-to-youCL have-AUX he that before went he there than you 'and he will tell you that he went there before you did.' e bendezirte a la mi alma antes que muera. (Faz.:47) and bless-INF-to-youcL have-AUX the my sou1 before that die-PRES.SUB 'and my sou1 will bless you before I die.' Al tercer0 dia sacarte a Pharaon de la carcel. (Faz.:53) at the third day remove-INF-to-youCL have-AUX Pharaoh from the prison 'On the third day the Pharaoh will release you from prison.' u( = [tv ti] Government of the infinitive trace is realized in tems of 'extended chain coindexation', as proposed by Lema and Rivero, so that there is no violation of the Empty Category Principle (ECP). In our opinion, the non-finite verb incorporates into the auxiliary under lQ and later on, under AGRlo, it excorporates and moves to CO . Thus, the Head Movement Constraint (HMC) is obeyed both in periphrastic and synthetic future and conditional constructions. Cardinaletti & Roberts (1991) suggest two possibilities: either Long Head Movement of the infinitive to C? or excorporation. Although they prefer the former option, we will adopt the latter. They argue that if excorporation is assumed, we should find examples in which the whole complex moves into CO. In fact, there is evidence of this type in La Fazienda de Ultra Mar and therefore we adopt the latter option. Notice that in these examples the synthetic future moves to CO preventing the clitic from occupying the first position within CP: (21) e combredeslo ayna; esta es la pascua al Semor. (Faz.:70) and eat-INF-have-AUX-itCL in-haste; this is the passover to the Lord. 'and you will eat it in haste; it is the Lord's passover.' (22) epornedesh sobre vuestras fijas. (Faz.:63) and put-INF-have-AUX-themCL upon your daughters 'and you will put them upon your daughters.' (23) e toldredeslm a 10s Egiptos. (Faz.:63) and remove-INF-have-AUX-themCL to the Egiptians 'and you will take them from the Egiptians.' As shown in these exarnples, the non-finite verb has not excorporated from the auxiliary: combr-edes, porn-edes, toldr-edes. Furthemore, the analysis given in (20) provides the correct account for embedded clauses such as (24), as illustrated in (25): (24) que en esto saber és¿ sodes barruntes (Faz.:55) that in this know-INF have-AUX if are spies 'that thus I will know if you are spies' Furthermore, this hypothesis is supported by the position of floating quantifiers: (50) su cuerpo estava todo bañado en sangre. (Cal.:280) his body was all bathed in blood 'his body was all covered with blood.' Therefore, following Cardinaletti & Roberts (1991), Kayne (1985), and Brucart (IBI), and assuming Suñer's (1990) analysis of small clauses, we obiain the structure shown in (51): (51) Old Spanish su cuerpok I I-- I 4 en sangre We assume that haber, ser and estar subcategorize for a small clause in Old Spanish. Neither of them assign a 8-role to the external argument and only haber assigns Case (cf. Tremblay (1991)). We believe, haber has lost its ability to assign Case to its internal argument and this led to a change in subcategorization. In Modern Spanish it selects a VP (cf. Lois (1989)). Furthermore, following Lema (1991), we assume that in Modem Spanish haber and estar are base-generated under a functional category ASPP (VPasp): (52) Modem Spanish To summarize, the major syntactic change from Old Spanish haber, ser, and estar to the Modem Spanish corresponding auxiliaries is due to a change in subcategorization: a lexical category VP has been reanalyzed as a functional category ASPP (represented here by means of VPasp to avoid confusions with ASPP in Old Spanish small clauses). There is one exception, though. Ser has kept its original subcategorization frame in passive constructions. In addition to the loss of its ability to assign Case, haber ceased to mean tener 'have', which led to its gradual intrusion into the syntactic functions of ser. Consequently, the former started to be used in constructions where the latter had been used before. In the 13th century, reflexives, deponents, and intransitives with haber are quite rare. From the 14th century onwards, estar specializes in the expression of aspectual features and at the end of the 16th century it ousts ser from the aspectual compound tenses. In contrast, ser remains in passive constructions, which is expected if we take into account that it has an active counterpart in the paradigm of syntactic functions. Notice that it is the only construction in which participle agreement has survived in Modem Spanish. Acwrding to Yllera (1973), at the end of the 15th century participle agreement disappears completely (except for passives). Antonio de Torquemada, in the Manual de Escribientes, published at the beginning of the 16th century, regards periphrastic futures and conditionals as archaic constructions. It follows from this that Tobler-Mussafia's Law was no longer applicable at that time. Although we have not been able to confirm this, we guess .that at that time pamciple preposing was considered to be archaic as well. Cardinaletti & Roberts (1991) suggest that all languages in which Tobler-Mussafia's Law or Wackemagel's Law apply have a maximal projection AGRlP. They add that the loss of cliticsecond effects in Old Italian and the loss of V2 in Old French are to be related to the loss of this projection. Following their analysis, we propose that the syntactic change that carne to an end at the tum of the 16th century in Spanish is due to three basic aspects: (53) a. Grammaticalization of several lexical elements which originated the gradual loss of their semantic content. b. Reanalysis that led to changes of subcategorization in the verbal system and to the change of some lexical categories into functional categories (some of the VPs which subcategorized for a small clause were reanalyzed as ASPPs which subcategorize for a VP). Haber and estar ousted ser, except for passive constructions, because their syntactic functions took over those of ser. (53) c. AGRlP was lost leading to the loss of Tobler-Mussafia effects, participle agreement, and participle preposing. The existence of AGRlP provides an explanation for the position of unstressed object pronouns in the following examples: (54) que te yo he fecho mal. (Cal.:314) that to-youcI have done harm 'that I have hurt you.' (55) Si me yo oviese llegado al le6n. (Cal.: 129) if myselfCL I have-PAST.SUB anived at-the lion 'If I had gotten to know the lion.' The position of clitics in these sentences is commonplace in Calila e Dimna. Taking the pronoun to be in AGRlo, we could argue that in these examples nominative Case assignment takes place under 'spec-head' agreement within AGR2P, which is the only one available in Modern Spanish. The fact that this order is not possible in Modem Spanish supports the hypothesis about the loss of AGRlP. Taking this into account, we will focus on participle preposing. Lema and Rivero, in the different works cited above, examine this construction and argue that it is neither VP preposing nor Long Head Movement. Despite sharing some properties with these operations, participle preposing does not display all of their properties. González-O116 (1983) documents the existence of participle fronting in the second half of the 10th century and, since there are about ten examples in Men6ndez-Pidal's (1%5) edition of the Glosas Silenses (56), we will assume that participle preposing is productive in early Spanish Romance. (56) betatu 10 ayat. (Glosas Silensex295) vetoed it~~ have-PRES.SUB '(that he) have forbidden it.' Notice that in the 13th century the participle could be fronted together with and adverb modifying it: (57) que ya tornados serfan. (Cal.: 144) that aiready wme-back-PART be-COND 'that they would have already come back.' (58) Ya entendido he agora esto. (Cd.:305) already understood have now this 'I have already understood this now.' (59) Ya llegado ha tu fazienda a tal lugar. (Cal.: 188) aiready come has your deed to such place 'Your deed has already reached such a point.' (60) fecho as td otro tal a 10s otros. (Cal.:301) done have you other such to the others 'you have done the same to the others.' (61) Benedicto sea Abraam de Dyos. (Faz.:44) blessed be-SUB Abraham by God 'Let Abraham be blessed by God.' In (57) the non-finite verb ser has incorporated to the auxiliary (giving the conditional as a result) and the complex 'adverbparticiple' has been fronted. This suggests that participle preposing is an Xmax movement.4 It can be claimed that the fronted element is the predicate of However. it is difficult to sustain this because esto 'this' in (58) and a tal lugar 'such a point' in (59) are objects of their respective participles. Notice that (60) and (61) present the same problem. The adverb modifying the small clause, an AP headed by the participle. Since participle preposing is possible in embedded clauses such as (57), we must assume that the predicate of the small clause is situated in SPEC of AGRlP and the main verb in AGRlO. In general, the subject occurs after the auxiliary. In (59), (60), and (61), nominative Case is assigned to the subject in SPEC of AGR2P under government from AGR10. Whenever a clitic pronoun follows the Wciple, it is adjoined to the main verb in AGRl? (63) Engaííados nos ha este cuervo. (Cal.:240) deceived us^^ has this raven This raven has deceived us.' the participle can be something other than ya 'already' (for instance muy or mucho 'very, much'). According to A. Suiíer @.c.) these facts are the same for Modem Spaoish small clauses. Example (64) could be interpreted as a possible counterexample to our proposal. It can be argued, however, that the object is occupying a TOP position externa1 to CP, a view supported by the presence of the resumptive pronoun, and the participle is in AGR10: (64) Esto oido 10 he. (Cal.: 122) this listened bitCL have 'I have listened to this.' Negation also supports the treatment of participle preposing as an Xmm movement because it does not prevent the movement from taking place: (65) si provados non fueren (Cal.: 133) if proved not were 'if they were not proved' (66) 10 que en muchos días acabado non as (Buen Amor:579) it that in many days finished not have 'what you have not finished in many days' As Cardinaletti and Roberts (1991) show, participle preposing does not only occur in Old Spanish. Icelandic stylistic fronting allows participle preposing, too. Compare the following sentences from Cardinaletti & Roberts (1991: 13): (67) a. arna er konan sem kosin var forsetí. there is woman that elected was president There is a woman that was elected president.' b. %rna er konan sem varkosin forsetí. there is woman that was elected president There is a woman that was elected president.' Adjectives can be fronted in the same way participles are in Old Spanish. Notice that there are examples attested in which the whole AP is fronted: (68) Vivo es e sano. (Faz.56) alive is and sound 'He is alive and well.' (69) Plus bellos son sos ojos. (Faz.59) more beautiful are her eyes 'Her eyes are more beautiful.' (70) Non podian bever de las aguas (...) que arnargas eran. (Faz.:72) not could drink of the waters that sour were They could not drink that water because it was sour.' Thus, it must be claimed that adjective and participle fronting correspond to one and the same strategy: topicalization of the small clause predicate. As (71) shows, the participle may be fronted together with its complement ((72) is the underlying structure for (71)): (71) maguer abonida cosa sea (Cal.: 185) though loathed thing be-SUB 'though (death) is a loathed thing.' (72) sea [[abonida cosa] [pro]] Notice that what has been fronted is the small clause that constitutes the predicate of another small clause.5 Finally, we find some examples in which the whole small clause has been fronted: - - Thanks to N. Martí for this observation (73) et desque él muerto fuere, ligera cosa es de matar a ella (Cal.:345) and since he dead be-FUT.SUB, easy thing is to kill her 'and when he will be dead, it will be easy to kill her.' Taking into account that the small clause is analyzed as an ASPP, the subject of the small clause moves to SPEC of ASPP and the predicate to ASP (spec-head agreemet takes place). Then the whole ASPP is fronted. Considering that the small clause is the interna1 argument offuere 'be', this movement can be related to object preposing. 4. Conclusion To summarize, infinitival periphrases do not seem to conform to an analysis in which a VP selects another VP. The non-finite verb excorporates from the auxiliary in AGRlO and moves to CO preventing the clitic from being sentence initial in periphrastic futures and conditionals. Haber, ser and estar are lexical verbs which subcategorize for a small clause the predicate of which is a participle. Participle preposing can be related to the fronting of small clause predicates and to object preposing. The main syntactic changes from Old to Modem Spanish can be explained according to the following facts: (a) the grarnmaticalization of severa1 lexical elements originated their gradual loss of semantic content; (b) reanalysis led to changes of subcategorization in the verbal system and to the change of some lexical categories into functional categories; (c) AGRlP was lost leading to the loss of Tobler-Mussafia effects, participle agreement, and participle preposing. References Almerich, Arcidiano de Antiochia (12th century) La Fazienda de Ultra Mar: Biblia Romar~eada et Itinéraire Biblique en Prose Castillane du XIle s&b, edited by M. Lazar (1%5) Acta Salmanticensia: Filosofia y ietras, 18,2. Brucart, J. M. (1991) 'Pasividad y atribuci6n en español: un análisis generativo' in V. Demonte and B. Gurza (eds) Estudios de Lingüística de España y México, UNAM El Colegio de Mkxico, Mkxico. Cardinaletti, A. and 1.Roberts (1991) 'Clause Structure and X-second', Ms. UniversiiA di Venezia and University of Wales. Gondlez 0116, F. (1983) 'Enclisis pronominal en el participi0 de las perífrasis verbales', Revista de Filologia Espaiiola, 63, 1-32. Kayne, R. (1985) 'L'accord du participe passk en frangais et italien', Modkles Linguistiques, 7, 73-91. Lema, J. (1991) 'Tiempo y Aspecto, correlatos sinkícticos y semánticos' , Ms., Universidad Autbnoma Metropolitana, Iztapalapa Lema, J. and M.L. Rivero (1989) 'Long Head Movement: ECP vs. HMC', NELS, 20,333347. Lema, J. and M.L. Rivero (1991) Types of verbal movement in Old Spanish: Modals, Futures and Perfects', Probus, 3, 1-42. Lema, J. and M.L. Rivero (1992) 'Inverted Conjugations and Verb-second Effects in Romance' in C. Laeufer and T.A. Morgan, (eds) Theoretical Analyses in Romance Linguistics, John Benjamins, Amsterdam. Libro de Calila e Dimna (13th century) edited by J. M. Cacho Blecua and M. J. Lacarra (1987). Clásicos Castalia, Madrid. Lois, X. (1989) Aspects de la syntaxe de l'espagnol et theorie de la grammaire, Universitk de Paris VI11 Ph.D. dissertation. Martínez-Gil, F. (1989) 'Las inversiones del orden de palabras en el Romancero', Hispania, 72, 895-908. Menkndez Pidal, R. (1%5) Crestomatia del espaiiol medieval (vol. 1, 4-9), Gredm, Madrid. Picallo, C. (1991) Funcions dobles: tres estudis de sintaxi catalana, Universitat Autbnoma de Barcelona Ph.D. dissertation. Rivero, M.L. (1988) The Structure of IP and V-movement in the Languages of the Balkans', Ms. University of Ottawa.