Empty Objects in Catalan

EMPTY OBJECTS IN CATALAN Anna Gavarr6 Universitat de Barcelona It is the purpose of this paper to consider the characteristics of Catalan null objects in the context of the theory of empty categories in general. The licensing of an empty category and recovery of its content are made fully independent and thus it is postulated that the difference between subject and object pro-drop is not one of licensing but rather should be placed in the way in which their features are identified. In this way, I propose a unified treatment of empty subject and object licensing. 1. The Identification Hypothesis The main reason to postulate a phonetically null argument lies in its being implicated in the interpretation of a sentence, especially in relation to control and binding; to the extent that control and binding are represented syntactically in the principies-and-parameters framework, the nul1 element must appear in the syntactic representation as well. The licensing and recovery of the content of empty categories (ECS hereafter) has been the object of a considerable literature. For a long time the possibility of having an EC has been correlated with the richness of agreement: (1) In many languages the distinction between the three persons is found not only in pronouns, but in verbs as well, thus in Latin (amo, arnas, amat) ... In such languages many sentences have no explicit indication of the subject, and ego amo, tu amas is at first said only when it is necessary or desirable to lay special stress on the idea "I, thou" (Jespersen ( 1%5[1924]:213)). Catalan Working Papers in Linguistics (CWPL) 1992: 145-161 Universitat Autbnoma de Barcelona The fact that agreement correlates with the possibility of dropping a nominal category has more recently been known as Taraldsen's Generalisation (cf. Taraldsen (1978)). In the work of Jaeggli (1982), the same idea is expressed in the "identification hypothesis", which states that agreement is the means of recovery of the content of an EC, and inextriqbly associates the licensing by agreement of an EC with the recovery of its content. Although this phenomenon has often been illustrated with examples from Latin and the Romance languages, it is challenged in a broader survey. Thus standard Finnish allows empty possessors, as in (24, at the same time that it exhibits agreement on the noun; from these agreement markers the number and person of the dropped nominal are recoverable. And in colloquial Finnish a new inflectional paradigm, (2b), has arisen; the noun is invariant and the possessors are obligatory. (2) a. (Minun) taloni b. Minun talo 'My house' (Sinun) talosi Sinun talo 'Y our house' (Hiinen) talonsa Hien talo 'Hislher house' (Meidiin) taiommme Meidiin tai0 'Our house' (Teidiin) ialonne Teidiin talo 'Y our house' (Heidiin) talonsa Heidiin talo Their house' This much is in accordance with the identification hypothesis. However, a "rich" system of agreement is not sufficient in itself to allow empty categories. Thus standard and colloquial Finnish both present verb agreement, but while the former permits empty non third person subjects, the later allows none: (3) a. (Mina)olen. b. Maoon. 'I amo (Sina) olet. Sa oot. 'You are' Hi on. Se on. 'Slhe is' (Me) olemme. Me ollan. 'We are' (3) a. (Te) olette. He ovat. b. Te ootte. 'YOU aret Ne on. T hey are' Moreover, it has been pointed out that an agreement system is not even a necessary condition for the occurrence of empty categories -empty categories and absence of agreement coexist, for instance, in Japanese and Chinese (cf. Huang (1984)). 2. Empty Objects In the absence of object agreement, the data on empty objects is in conflict with the identification hypothesis. In the following I shadow Rizzi's (1986) arguments for a syntactic analysis of empty objects with Catalan examples that closely parallel those of Italian. Thus wrresponding to Rizzi's Italian example (4) is the Catalan example (5). (4) Questo conduce a concludere quanto segue. this leads to conclude-INF what follows This leads one to conclude what follows.' (5) Aixb porta a concloure el següent. this leads to conclude-INF the following This leads one to conclude what follows.' To motivate an empty category, note first that there must be some nul1 element to act as wntroller in e.g. (6). (6) Un general pot obligar [ec] a obeir les seves ordres. a general can force-INF to obey-INF the his orders 'A general can force one to obey his orders.' Second, the EC can act as the antecedent of an anaphor as in (7), where amb un mateix 'with oneself needs to be bound. (7) La bona mdsica reconcilia amb un mateix. the good music rewnciles with one self 'Good music reconciles one with oneself.' The interpretation of the empty category is that of an arbitrary, generic, human, so that e.g. (4) can be paraphrased as in (8): (8) Questo wnduce la gente a concludere quanto segue. this leads the people to conclude-INF what follows This leads people to conclude what follows.' Rizzi (1986512) characterises a generic NP like the one in (4) by the following collection of feaiures : That Catalan empty objects are also [+human] is illustrated in (10), where it is shown that a verb selecting for [-human] objects does not allow for an EC as complement. (10) *Aquesta cafetera aboca [ec] bé. this wffee-maker purs well This coffee-maker purs well.' While Cataian empty objects share with Italian their specification for human and generic, they are [-plural]. The contrast is shown in (1 1) from Italian (Rizzi (1986:504)) and (12) from Catalan: (1 1) La buona musica riconcilia con se stessi. the good music reconciles with themselves 'Good music reconciles one with oneself.' (12) *La bona música reconcilia amb ells mateixos. the good music reconciles with themselves 'Good music reconciles one with oneself.' The value for the feature [*plural] must thus be determined language-specifically. Because the EC is [+plural] in Italian it can bind a reciproca1 (cf. Rizzi (1986), who took this example from Belletti (1982)) .The [-plural] EC of Catalan means that this type of sentence is ruled out due to a mismatch in agreement: (13) Un buon pranzo pub riconciliare [ec] gli uni con gli altri. a good meal can reconcile-INF the ones with the others 'A good meal can reconcile with one another.' (14) *Un bon Apat reconcilia [ec] els uns amb els altres. A good meal reconciles the ones with the others 'A good meal reconciles people with each other.' Argument small clauses also allow empty objects, as in (IS), and display agreement of the empty category with the adjective.1 (15) Aquesta música fa [[ec] felid*felip]. this music makes happy-SGlhappy-PL This music makes one happy.' Adjunct small clauses can marginally contain an empty object in Catalan, while they are supposed to be perfect in Italiaa. Authier (1989) points out that French prohibits arbitrary nul1 objects in adjunct small clauses altogether. In relation to person, Catalan allows for the expression of a generic by second person pronouns (in much the same way that English does), as in (16ii), which 1s a paraphrase of (7): (16) La bona mdsica et reconcilia amb tu mateix. the good music you reconciles with yourself (i) 'Good music reconciles you with yourself.' (ii) 'Good music reconciles one with oneself.' The instantiation of the second person generic NP must be overt; there is no empty counterpart to i t: (17) *La bona mdsica reconcilia amb tu mateix. the good music reconciles with yourself 'Good music reconciles one with oneself.' Hence, since a generic NP is not necessarily third person, an empty NP must be specified as such in Catalan; we shall assume the same for Italian. The syntactic agreement behaviour encountered with respect to person and number corroborates the motivation for an empty category originating in binding and control; the EC is the triggerer of agreement. And since the missing object is interpreted as generic and human with all verbs, it is natural to assume that its interpretation arises from syntactic representation, rather than from the lexical properties of verbs with which it occurs. Assuming that we have established the existence of this empty category, we tum now to consider its nature. 3. Empty Objects as Pronominals. When trying to define the EC at issue with respect to the features [+anaphoric] and [rtpronominal], we can rule out it being [+anaphoric, +pronominal], i.e. PRO, since the EC is governed by the verb, while PRO is ungoverned; equally, we can mle out it being [+anaphoric, -pronominal], ¡.e. NP-trace, since there is no (overt) antecedent for it. The two remaining kinds of analysis have both been proposed for empty objects: the one which considers them to be [-anaphoric, -pronominal], i.e. variables, and the other that treats them as [-anaphoric, +pronominal], i.e. empty pronominals. The variable account was first put forward by Huang for Chinese (cf. Huang (1984)), and presupposes that an empty object is a variable bound by an empty operator in topic position, and as such subject to Principle C of the binding theory. This kind of analysis has been applied to German (Huang (1984)), Japanese (Hasegawa (1984)), American Sign Language (Lillo-Martin (1986)), KiNande (Authier (1988)), and Swedish (E. Engdahl, cf. Huang (1989)), as well as Spanish (Campos (1986)), European Portuguese (Raposo (1986)), Quiteño Spanish (Suñer and YCpez (1988)), and French (Authier (1989)). Huang (1987, 1991) revises somewhat his earlier position. Let us simply note here that Rizzi (1986) excludes a variable analysis by ruling out an implicit operator in the following manner. If we assume that an empty operator is to be found in Italian, as in European Portuguese (cf. Raposo (1986)), in the same position as a wh-operator, it would be predicted that such an EC cannot occur in a clause with a wh-operator; this prediction is incorrect, however, for Italian and Catalan: (18) a. Quina mdsica reconcilia [ec] amb un mateix? which music reconciles with oneself 'Which music reconciles one with oneself?' b. Les seves paraules, no SC qu2 poden portar [ec] a PRO concloure [t]. hislher words, not know what can lead-INF to conclude-INF 'I don't know what hislher words can lead one to conclude.' If the EC is pro, no such erroneous prediction is made; Rizzi concludes that the Italian EC at hand is a pure pronominal. Treatments in terms of pro have also been proposed for empty objects of Chamorro (Chung (1984)), French (Zribi-Hertz (1984)), Imbabura Quechua, Korean and Thai (Cole (1987)), Hungarian (Farkas (1987)), Nonvegian (Afarii and Creider (1987)), and Brazilian Portuguese (Farrell(1990)). In Gavarr6 (1991) I considered somepros occumng, under quite specific conditions, in the context of the clitic system of colloquial Catalan. 3.1. Licensing versus Recoverability To address the problem of identification of this EC, Rizzi distinguishes the licensing of ECS and their identification, i.e. the recovery of their features, as is standardly assumed for all other ECS. In this way, the licensing of pro in object position can be made analogous with that of the licensing of that EC in subject position in the Romance languages with nul1 subjects, while the recovery of its content is, in the normal cases, different, insofar as there is subject-verb agreement to make subject drop transparent, but there isn't object-verb agreement. As concerns the licensing of pro, Rizzi puts fonvard the following principle: (19) pro is governed by Xo. X stands for syntactic categories that may vary from language to language. Briefly, (19) is to be understood as parametrically defined, according to a multivalued parameter. That English doesn't allow any instance of pro would be captured by having no instance of category X; languages allowing subject pro-drop have Infl as a governor licensing pro, and Italian, allowing pro in object position, has V as another licenser of this EC. Thepro licensed by (19) is made recoverable according to the following convention: (20) Let X be the licensing head of an occurrence of pro: then pro has the grammatical specification of the features on X coindexed with it. In the case of pro-drop with agreement, the licensing head presents the +features that make the identification of pro possible. Agreement lacking in object predrop in Italian, Rizzi puts to use the rule of arbitrary interpretation assignment (21). (21) Assign arb to the direct 0-role. Two further refinements are indispensable for this theory of pro. First, Rizzi notes that arbitrary nul1 objects appear unavailable to Move a, so that sentences where pro has been raised to subject position are ungrammatical, as (22) illustrates. (22) * [pro,b] 6s matat [t] per la milfia. is killed by the Mafia 'One is killed by the Mafia.' The proposal that Move a is inapplicable is derived by appealing to Case theory, by requiring thatpro be Case-marked by its licenser, V, instead of Infl (as it would be in subject position). This triggers the substitution of the stronger principle (23) for (19). (23) pro is Case-marked by Xo. Second, most theta roles can be satisfied by pro in object position, as shown by (24) and (25), meant to illustrate EXPERIENCER and BENEFACTIVEpros respectively. (24) A vegades el Mtirius impressiona [pro]. sometimes the Mtirius impresses 'Sometimes Mhius impresses one.' (25) A vegades un director promou [pro] sense ra6. sometimes a director promotes without reason 'Sometimes a director promotes one without good reason.' Gavarr6, A. (1991) 'A Note on Catalan Clitics', Catalan Working Papers in Liriguistics 1991, UAB, 65-73. Grimshaw, J. (1991) Lectures at the Linguistic Institute of the LSA, UCSC. Hasegawa, N. (1984) 'On the So-called 'Zero Pronouns' of Japanese', The Linguistic Review, 4, 289-341. Huang, C.-T. J. (1984) 'On the Distribution and Reference of Empty Pronouns', Linguistic Inquiry, 15, 53 1-574. Huang, C.-T. J. (1987) 'Remarks on Empty Categories in Chinese', Linguistic Inquiry, 18, 321-337. Huang, C.-T. J. (1989) 'Pro-drop in Chinese: A Generalised Control Theory' in O. Jaeggli and K. Safir (eds.) The Null Subject Parameter, Foris, Dordrecht. Huang, C.-T. J. (1991) 'Remarks on the Status of the Null Object' in R. Freidin (ed.) Principles and Parameters in Comparative Grammar, The MIT Press, Cambridge, Massachusetts. Jaeggli, 0. (1982) Topics in Romance Syntax, Foris, Dordrecht. Jespersen, 0. (1%5[1924l) The Philosophy of Gramrnar, W.W. Norton, New York. Koopman, H., and D. Sportiche (1988) 'Subjects', ms., UCLA. Lillo-Martin, D. (1986) Two Kinds of Null Arguments in American Sign Language', Natural Language and Linguistic Theory, 4,415-444. Pollock, J.-Y. (1989) 'Verb Movement, UG and the Structure of IP', Linguistic Inquiry, 20, 365-424. Raposo, E. (1986) 'On the Null Object in European Portuguese' in O. Jaeggli and C. SilvaCorvalhn (eds.) Studies in Romance Linguistics, Foris, Dordrecht. Rizzi, L. (1986) 'Null Objects in Italian and the Theory of pro', Linguistic Inquiry, 17,501557. Suñer, M., and M. YCpez (1988) 'Null Definite Objects in Quiteño', Linguistic Inquiry, 19, 51 1-519. Taraldsen, K. T. (1978) 'On the NIC, Vacuous Application and the that-trace Filter', mimeographed, MIT. Distributed by the Indiana University Linguistics Club, Bloomington, Indiana. Zribi-Hertz, A. (1984) 'PrCpositions orphClines et pronoms nuls', Recherches Linguistiques, 12, 46-91. Logofidia, Facultat de Psicologia Universitat de Barcelona C/ Adolf Florensa, s/n 08028 Barcelona