Expressing Modality by Recycling Clitics

CatWPL 511, 1996 67-79 Expressing Modality by Recycling Clitics* Víctor M. ~on~a', Guillermo ~orenzo~ and Gemma ~igau~ 1. Universidade de Santiago de Compostela. Departamento de Filoloxia Española, Teoria de Literatura e Lingüística Xeral. Plaza Isabel a Católica, 2, 2" E. 36204 Vigo. Galicia (Spain) e-mail: [email protected] 2. Universidad de Oviedo. Departamento de Filologia Española. Campus de Humanidades <<El Milán,,. C/ Teniente A. Martínez s/n. 3301 1 Oviedo. Principado de Asturias (Spain) e-mail: [email protected] 3. Universitat Autbnoma de Barcelona. Departament de Filologia Catalana. 08193 Bellaterra (Barcelona). Catalonia (Spain) e-mail: [email protected] Received: March 3rd 1996 Accepted: June 1 lth 1996 Abstract This paper tries to interpret the role of certain clitics attached to functional categories located over AgrS. They seem to be in charge of the expression of the modal orientation of the sentence. We show that while certain languages have elements specifically designed for this purpose, other languages resort to a recycling strategy which enables morphologically accusative clitics to do the job. In clear contrast with all these languages, Catalan is an example of a language in which modal clitics seem to be completely absent from the clitic paradigm. Key words: clitics, modality, functional heads. Resum. L'expressió de la modalitat a través del reciclatge de clitics Aquest article vol oferir una explicació del comportament de determinats clitics relacionats amb categories funcionals superiors a SConc i encarregats de l'orientació modal de la frase. Mentre algunes llengües compten amb elements especificament dissenyats per a aquesta funció, altres llengües recorren a l'estratkgia del reciclatge, la qual habilita els clítics d'acusatiu per a l'expressió de la modalitat. En contrast amb totes aquestes llengües, el catala és un exemple de llengua que no compta amb clitics de modalitat. Paraules clau: clitics, modalitat, categories funcionals. * We thank Anna Bartra, Ken Hale, M. Llui'sa Hernanz, Jay Keyser, Carme Picallo, Jaume Soli, and an anonymous reviewer for their cornments and support. This paper has benefited from the following grants: DGICYT PB93-0893-C04-01 and Generalitat de Catalunya 1995SGR00486 (Gemma Rigau), and DGICYT PB93-0887-C03-02 (Guillermo Lorenzo). 68 CatWPL 511, 1996 Víctor M. Longa, Guillermo Lorenzo and Gemma Rigau Table of Contents 1. Galician and Asturian Modal Clitics 4. Conclusions in the Context of Romance Clitization References 2. Clitic Recycling: An Overview 3. An Analysis of the Modal Clitic Constmction 1. Galician and Asturian Modal Clitics in the Context of Romance Cliticization 1.1. Modal Clitics in Galician Galician exhibits the property of using morphol?gically accusative clitics as subject clitics under certain conditions, as in (1) (see Alvarez, Regueira and Monteagudo (1986: 173)):' (1) a. ¡Aquí as ve fien elas! here c1.-3p.-pl.-fem.-ac. come-3p.-pl. they-fem. 'Here they come!' b. ¡Ai os van os veciños! there c1.-3p.-pl.-masc.-ac. go-3p.-pl. the neighbors 'Here come the neighbors!' As can be noted from the examples, the verbal inflection, the lexical subject (which is optional, as usual), and the accusative clitic all share their f-features: third person and plural, in the case of (1). The lexical subject and the clitic also share the gender feature: feminine in (la) and masculine in (Ib). The conditions that make this construction possible are the following: (a) It is only possible if the individuals denoted by the subject are present in the utterance situation. Thus this construction can only be used in present tense, as a means of expressing a ((deixis ad oculos>), using Bühler's (1934) tem. Moreover, it has an appellative function and therefore an exclamatory meaning. (b) The use of the accusative clitics observed in (1) is only possible in sentences with a presentational meaning. The verbs which can express this meaning are those belonging to stative (estar 'to be, to appear') and movement classes (venir 'to come', ir 'to go', chegar 'to arrive, to come'). A similar phenomenon is found in some Northern Italian dialects, such as 1. Sentences of this kind are also used in Galician Spanish, as can be observed in the traslation of (1): (i) ;Aquí las vienen ellas! (ii) ;Ah; 10s van 10s vecinos! As they are identical in their properties, we are going to illustrate the phenomenon by using only Galician data. Expressing Modality by Recycling Clitics CatWPL 511, 1996 69 those studied by Poletto (1992) and Brandi and Cordin (1989). In Veneto, for instance, we can find sentences very similar to (1): (2) No la vien. neg. c1.-3p.-sing. come-3p.-sing. 'She doesn't come.' (Poletto 1992: ex. (I)) Nevertheless, the differences between (1) and (2) are more notable than their similarities. Actually, this kind of clitic in Veneto is neither restricted to sentences with a deictic use --cf. property (a)- nor with a presentational verb --cf. property (b)-. This is the reason of the contrast between (3a), a correct sentence from Veneto, and (3b), an incorrect one from Galician: (3) a. La magna pomi. c1.-3p.-sing. eats apples 'She eats apples.' b. *A xanta mazás. c1.-3p.-sing. eats apples From this, we can conclude that Veneto has 'pure subject clitics', which is not the case of Galician. Galician clitics can also be related to a different kind of subject clitics in other Northern Italian dialects. As discussed by Poletto (1992: 66-103), certain subject clitics of these dialects should be expressing a head position higher than AgrS but lower than Comp. These clitics always appear to the left of negation, and sometimes clitizice over Comp. The first property is illustrated in (4a), a sentence from Polesano, and the second one in (4b), taken from the Polesano dialect of Adria: (4) a. A no vegno. cl. neg. come-1p.-sing. 'I'm not coming.' (Poletto 1992: ex. (10a)) b. A no so sa vegnaró. cl neg know-1p.sing. whether-cl come-1p.sing.-fut. 'I don't know whether I will come.' (Poletto 1992: ex. (23b)) Poletto's proposal is that this kind of subject clitics is located in a functional head related to the modal orientation of the sentence. Therefore, it is clearly related to the clitic topic marker of Fiorentino, as illustrated by the examples given by Brandi and Cordin (1989): (5) Te, e tu parli troppo. you cl. you talk-2.p.-sing. too much 'As for you, you talk too much.' (Brandi and Cordin 1989: ex. (9a)) The similarity to Galician is found in that subject clitics in this language are only permitted if the sentence has a special modal orientation: appellative, in the case 70 CatWPL 511, 1996 Víctor M. Longa, Guillenno Lorenzo and Gemma Rigau of Galician, and emphatic, in the case of Fiorentino. But again, important differences are found in this case. First of all, Galician clitics show agreement markers with the subject, as shown in (6a); in the Italian cases, the form of the clitic is invariable, as shown in (6b) (cf. (5)): (6) a. ¡Aquí o ven el! here c1.-3p.-sing.-masc. come-3p.-sing. he-masc. 'Here he comes!' b. La Maria, e la parla troppo. the Maria cl. she talk-3p.-sing. too much 'As for Mary, she talks too much.' (Brandi & Cordin 1989: ex. (1 14)) Secondly, in Galician the clitic is in complementary distribution with negation, as shown in (7), which is not the case in the Italian dialects: (7) *i Aquí non as veñen elas! here not cl. come they-fem. Finally, this kind of clitic is not restricted to appearing in the context of verbs of movement in the Italian dialects, as can be seen in (5) and (6b). Surnming up, what is found in Galician can be described as a compound of the pure subject and the modal subject clitics of the Northern Italian dialects described by Poletto, though much more restricted in use. As an expression of a modality content, the Galician clitics could relate with two different functional positions: (a) the Comp head, and (b) a lower head specifically designed for modal purposes. Our proposa1 is that modal clitics found in Galician are to be attributed the last type of head position, a head in which the relation between the speaker or the sentential subject and the utterance situation is coded. Actually, this is basically Uriagereka's (1995) proposa1 about an FP projection in Northwestern Iberian dialects. However, the projection we are looking for looks very much like the CP proposed by Laka (1990); i.e., a projection in which sentential negation and certain emphatic features are complementarily generated. This fits well with the fact pointed out in (7). In the next section, devoted to Asturian, we offer an argument in favor of this analysis for modal clitics in the Northwestern Iberian languages. 1.2. Modal Clitics in Asturian The Galician sentences examined so far lose part of their exceptional flavor if we look at another construction found in Asturian. This language offers a structure in which a pronoun corresponding with the subject of a related presentational sentence cliticizes onto a locative wh-phrase. In these sentences, the original verbal head deletes (see D'Andrés (1993: 14-16)). This ltind of construction is mainly used as a request for locative information missed from a previous utterance. The pheno- Expressing Modality by Recycling Clitics CatWPL 511, 1996 71 menon is illustrated in (8b) and (8c), with (8a) offering the corresponding sentence with a verbal head and a non-clitic subject: (8) a. ta Xuan / Maria? where be-3p.-sing John / Mary 'Where is John / Mary?' b. ¿Úlu? where-c1.-3p.-sing-masc. 'He's where?' c. ¿Úla? where-c1.-3p.-sing-fem. 'She's where?' The most important restriction that affects this structure is that cliticization is only possible over the wh-word ú, a very special one in the Asturian ~h-~aradigm.~ U is restricted to matrix questions, as in (8); it cannot be used either in indirect questions or as a relative pronoun, as shown in (9): (9) a. *Nun sé u' ta Xuan. neg. know-1p.-sing. where is-3p.-sing. John b. *La casa u' Xuan vive. the house where John lives The sentences in (9) become correct by substituting ónde for ú. Ónde can also be used in matrix questions, but it does not allow cliticization: A wh-word almost equivalent to the Asturian ú is found again in the Italian dialects studied by Poletto. She points out that Friulan has a locative wh-word, do, with exactly the same restrictions observed in relation with Ú: do is an alternate of dula (the more extensive one, that is, the equivalent of Asturian ónde) in matrix questions. What is somehow unexpected is that do, contrary to dula, is incompatible with the modal subject clitics of Friulan, as shown in (1 1): 2. Galician has expressions similar to those in (8b) and (8c). The main difference is that in the case of Galician the special wh-word used in these cases cannot be used if the verb remains in the sentence. This means that Galician does not exhibit sentences equivalent to Asturian (8a) (see Álvarez et alii 1986: 431). A similar construction was also observed by Menéndez Pidal(1906) in Old Castilian, where the locative wh-word do admitted cliticization of an morphologically accusative clitic in sentences such as the following: (i) Do 10s mis amores, dolos? Where the my loves where-cl-3p-plu-masc 'Where are my loves, where are they?' 72 CatWPL 511, 1996 Víctor M. Longa, Guillermo Lorenzo and Gemma Rigau (1 1) a. *Do a comprial i1 pan? where cl. buys the bread (Poletto 1992: ex. (19b)) b. Do comprial i1 pan? where buys the bread 'Where does he buy the bread?' (Poletto 1992: ex. (19a)) Poletto, following Beninca (1986), argues in favor of the XO character of do. According to her, do is a head wh-word located in compO, as opposed to dula, a XP wh-word located in [Spec, CP]. Poletto's proposa1 about the impossibility of (1 la) is that do already includes the modal clitic, after incorporation from its modal head position. We can extend some of these observations to Asturian, and assume the head character of ú, as distinct to the XP nature of ónde. We can further postulate that the Asturian ú subcategorizes for an emphatic CP as a part of its properties as a head. Thus, the opposition between an ú matrix question and an ónde matrix question is based on the emphatic character of the former, in contrast with the neutral character of the latter. The intuition of Asturian speakers about the use of each form seems to confirm this. Asturian li does not include however a subject clitic, as Friulan do does, but it can host an accusative clitic used as an emphatic subject clitic (see (8b) and (8c)), which is only possible in a presentational context. Therefore, what we have found in Asturian is a construction similar to the Galician one introduced in the previous section in that it makes use of a morphologically accusative clitic. This clitic exhibits the features of the subject, and it clearly relates with the emphatic feature which we are positing within the C functional head; therefore, we consider it an emphatic subject clitic. Moreover, the construction is restricted to presentational contexts in both languages. An explanation is then needed for the following facts: (a) the foundations of the lexical restriction observed both in the Galician and Asturian constructions, which confines the phenomenon we are considering to presentational contexts, and (b) the nature of the connection established by the clitics between the features of C and AgrS within these constructions. But an even more basic question still remains to be explained: (c) why do we find morphologically accusative clitics in these contexts. 2. Clitic Recycling: An Overview We have observed that, Galician and Asturian not having clitics specifically designed for emphatic purposes -as opposed, for instance, to Polesano and Fiorentino-, they use the same clitics that are used for other purposes in the appropriate contexts. What we suggest is that in the constructions considered in this paper, Galician and Asturian 'recycle' the accusative clitic for an emphatic nominative-like use. The 'recycling mechanism' that we postulate is a very productive one in the Northwestern Iberian languages. We have defined it in previous work as a kind of last resort strategy which applies in the lexicon and extends the use of an Expressina Modalitv bv Recvcling Clitics CatWPL 511, 1996 73 element of the clitic paradigm in order to fill in a gap. It can be formulated as follows: (12) Clitic Recycling Strategy: 'Use the (most) unmarked clitic to fill in gaps in the paradigm.' (Longa, Lorenzo and Rigau (1995: ex. (40)) Asturian and Galician, which lack both partitive and locative clitics, make use of the items from the accusative inventory in recycling operations. Some illustrations are given bellow: (13) 'Accusative + ~artitive':~ a. Asturian - ~Habia bruxes? had witches 'Were there witches?' - No, nun les habia. no not c1.-acc-pl-fem. had 'No, there weren't any.' b. Galician - ~Habia meigas? had witches 'Were there witches?' - No, non as habia. no not c1.-acc-pl-fem. had No, there weren't any.' 3. The recycling strategy is starting to be used in the context of the existencial verb haber of Spanish dialects other than Northwestern (see Suñer (1982: 58-61)). An illustration is given bellow: (i) Aviones, 10s hay todos 10s días Planes, cl. has everyday 'There are flights everyday.' Nevertheless, the use of recycled accusatives is by far more general in Northwestern than in Castilian Spanish, as can be observed in the contrasts shown in (ii.) and (iii.): (ii) Fiebre, no la tengo / Cast. * Fiebre, no la tengo (Northwestern Spanish) fever not cl. have fever not cl. have 'I do not have any fever.' (iii) Dolor, no 10 siento I Cast. * Dolor, no 10 siento (Northwestern Spanish) Pain, not cl. feel Pain not cl. feel 'I do not feel any pain.' In cases in which a clitic can be interpreted both as accusative and as partitive, both interpretations are perrnitted because there is no an alternative for any of the competing constructions. An example is given in (i.), in which la may stand both for a partitive object (as fiebre 'fever') or for an accusative object (as la llave 'the key'): (iv) la tengo (Standard Spanish) cl. have 'I have any / it.' 74 CatWPL 511, 1996 Víctor M. Longa, Guillermo Lorenzo and Gemma Rigau (14) 'Accusative + Locative': a. Asturian Les muyeres taben xunto la creizal, pero 10s paisanos nun the women were besidethe cherry tree but the rnen not 10 taben cl-acc-sing-masc were 'The women were beside the cherry tree, but the rnen were not there.' b. Galician As mulleres estaban xunto á cerdeira, pero os homes non the women were beside the cherry tree but the rnen not o estaban. cl-acc-sing-masc were 'The women were beside the cherry tree, but the rnen were not there.' The Asturian and Galician sentences are to be compared with their equivalents in Catalan, whose clitic paradigm has both partitive and locative clitics: (15) a. -Hi havia bruixes? cl-loc. had witches 'Were there witches?' - No, no n hi havia. no not cl-part.-cl-loc. had 'No, there weren't any.' b. Les dones són prop del cirerer, per6 els homes no hi són. The women are beside the cherry tree but the rnen not cl-loc. are 'The women are beside the cheny tree, but the rnen are not there.' Some version of the Elsewhere Conditon must be adopted here in order to prevent that unmarked elements as Asturian la and 10 appear in other contexts as dative. Our idea is that such a condition avoids an unmarked clitic to appear in contexts for which specific elements exist in the paradigm. 3. An Analysis for the Modal Clitic Construction Going back to the Galician cases introduced in section 1.1 ., it is important to note that the appellative character of the construction does not lie so much on the locative element introducing the sentences as on the deictic content of the utterance as a whole. Actually, aquí 'here' is not a locative argument in these presentational sentences, as proven by the fact that there is no paraphrase of (16a) parallel to that of (17a) found in (17b): (16) a. Aquí as veñenelas! # b. "Aquí C onde as veñen elas! here cl. come they-fem. here is where cl. come they-fem. 'Here they come.' 'Here is where they come.' Expressing Modality by Recycling Clitics CatWPL 511, 1996 75 (17) a. Aquí xantan elas. = b. Aquí é onde xantan elas. here eat they-fem. here is where eat they-fem. 'They eat here.' 'Here is where they eat.' This means that aquí-or a similar locative-type element, such as ai 'there'- is in this context nothing more than a marker of the emphatic-deictic meaning of (16a) as a whole. We thus postulate for this element a specifier position within the functional projection in which the modal character of the sentence is expressed, i.e., EP. Therefore, this element is not related to the lexical structure of the verb, since it is not one of its arguments, nor does it modify the relation between them in any relevant sense. We thus conclude that it does not move from the VP to the functional projections; instead, we are going to assume that an element like aquí in (16) is base-generated within the functional structure of the sentence. It is not to be considered as an emphasized argument, but rather as a 'emphasizing operator' generated externally to the verbal lexical projection. For instance, ir 'to go' implies the movement to a place neccessarily different from the position of the speaker. This aspect of its meaning must be reflected in the lexical structure projected by the verb. Ir has thus a dynamic meaning which cannot be modified by a word with the stative meaning of aquí--cf. va desde aquí 'ske goes from here' vs. * va aquí 'ske goes here'-. Aquí can only be used in the context of ir reduced to a very basic deictic meaning understood independently of the lexical properties of the verb. Another emphatic element that can introduce this sentences is the non-aspectual xq 'already', which should be translated in these cases as 'really':4 (18) ¡Xa as chegan elas! already cl. come they-fem. 'Here they come!' As for the recycled clitic, we hypothesize that it is inserted under the AgrS node, and thus compelled to show agreement with the subject. From this position, the clitic reaches the C position, in which the emphatic features are posited and reinforced by the emphasizing operator. The emphatic character of the sentence as a whole obtains after the verb moves overtly to AgrS, and covertly to E.5 The DP sententia1 subject -elas in (16a)- reaches the [Spec, AgrSP] position also at LF. All these movements are represented in (19): 4. One of the authors had the privilege of listening the following sentence in a recent visit to Galicia: (i) ¡Mira cómo lo viene, el avión! (Galician Spanish) Look-imperative how cl. comes the airplane 'Look, here comes the airplane!' In this sentence, it is the imperative form of mirar 'look' the element that acts as the appellative introducer. See the Catalan example in (22.b.). 5. Beninca (1983) also analyses the Paduan subject clitic a, which appears within exclamatory sentences, as a modal clitic. This can be seen as another instance in which a subject clitic assumes the function of a modal marker. The difference is that in the case of Paduan the clitic has lost its person features.