Datives in Constructions with Unaccusative Se

Abstract In this talk I argue that in Spanish there is a type construction, involving a verb with the unaccusative seclitic and a dative, where the dative is in fact the external argument, which renders the construction comparable to a transitive structure (in the well-established unaccusative-transitive alternation). A number a tests and criteria are reviewed to support the external argument status of the dative (which in fact can be assimilated to the status of locatives and datives with some impersonal verbs, see Fernández Soriano 1999). These are: unmarked word order, raising and binding. On the other hand, there is also evidence for the non externalized status of other, internal, argument: impossibility of anaphor binding, possibility of being a bare NP, inability to control, among others. I conclude that in this constructions the dative is an instance of quirky case. Key words:Spanish, quirky case, se-unaccusatives, argument structure. Resum. Datius en construccions amb el se inacusatiu En aquesta xerrada defenso que en castellà hi ha un tipus de construcció, que conté un verb amb la marca sed’inacusatiu i un datiu, on el datiu és, de fet, l’argument extern, cosa que fa aquesta construcció comparable a una estructura transitiva (dins l’alternança ben coneguda inacusatiutransitiu). S’hi revisen tots de proves i criteris que donen suport al caràcter d’argument extern del datiu (que, de fet, és assimilable a l’estatus dels locatius i datius amb alguns verbs impersonals, vegeu Fernández Soriano 1999). Com a indicis, tenim: ordre de mots no marcat, elevació i lligam. D’altra banda, hi ha també indicis del caràcter no externalitzat de l’altre argument, l’intern: impossibilitat de lligam d’anàfores, possibilitat de ser un SN nu, incapacitat de controlar, entre d’altres. La conclusió és que en aquestes construccions el datiu és un exemple de cas capriciós (quirky case) Paraules clau:castellà, cas capriciós, inacusatius amb se, estructura argumental. CatWPL 7, 199989-105 Datives in Constructions with Unaccusative Se Olga Fernández Soriano Universidad Autónoma de Madrid [email protected] Received: December 13th 1998 Accepted: March 17th 1999 Table of Contents CatWPL 7 089-105 13/6/00 12:26 Página 89 1. The dative occupies the subject position 2. Internalization of the theme 3. Quirky subjects References In Spanish, as in other Romance languages, most of unaccusative verbs which have a transitive counterpart (in Hale and Keyser’s 1994, 1997 sense) appear with the clitic se, which is also both the reflexive and the impersonal clitic. Some examples are provided in (1): (1) a. Juan rompió los vasos. Juan broke the glasses b. Juan quemó la comida. Juan burned the food c. Los vasos se rompieron. the glasses SE brokePL d. La comida se quemó. the food SE burnedSG In Levin and Rappaport Hovav’s (1995) terms these verbs express an externally caused event and have a «change of state (and position)» meaning. In Hale and Keyser’s approach, the transitive structure (1a) is the result of embedding the «complex dyadic» predicate into a monadic V, which allows for the addition of an external argument in syntax. The monadic V, on the other hand, has been related to a CAUSE predicate (see, for example, Harley, 1996). We will come back to this question, but see the structure below: Mendikoetxea (1998), in the line of Chierchia (1989) and others, proposes that these predicates are very similar to true reflexives, in the sense that the occurrence of se correlates to their dyadic nature as well as to the possibility of transitivization. Unaccusative se predicates select for a PRO in Spec of vP, which establishes a control relation with the theme. Differences in interpretation are to be related to the nature of the CAUSE predicate. In the unaccusative SE predicate, as opposed to the regular reflexive one, the causing factor is understood statively, that is, not as an action performed by an agent but as an event determined by a property of the theme argument (Chierchia, 1987). Mendikoetxea supposes that in (1c) and (1d) the argument which triggers verbal agreement is both the cause and the theme. The fact I would like to focus on is that, together with the transitive/agentive alternative, these constructions, in the unaccusative se version, have the possibility to take a preverbal dative. 90 CatWPL 7, 1999 Olga Fernández Soriano (2) Spec Syntax ………………………… V2 V1l-structure V1Adj CatWPL 7 089-105 13/6/00 12:26 Página 90 The meaning of the resulting sentence is that the person involved (which would appear in the nominative in the transitive counterpart) does not perform a direct action towards the object: the event taking place only involves him/her in a certain way. So together with the ones in (1), we get structures like the ones in (3): (3) a. A Juan se le rompió el coche. to Juan SE CLDT broke the car ‘The car broke on John.’ / ‘John’s car broke (on him).’ b. A Pedro se le ha quemado la comida. to Pedro SE CLDT has burned the food ‘The food has burned on Pedro.’ c. A mí se me ha acabado el dinero. to me SE CLDT has finished the money ‘The money has finished on me.’ / ‘I ran out of money.’ d. A este estudiante se le han olvidado las respuestas. to this student SE CLDT have forgotten the answers ‘This student has forgotten the answers.’ Glosses with benefactive PP’s like on me / him / her are not quite accurate: sentences in (3) do not mean that the argument in the Dative is affected by the event but that the person is responsible for it without directly intervening in the action. I will sometimes translate it as a TO phrase. As the glosses indicate, the dative can be interpreted as the possessor of the theme, although this is not obligatory. I will try to give an analysis of the dative which appears in these constructions. The basic claim I would like to maintain is that this dative is an external argument. That is, I accept Mendikoetxea’s analysis of unaccusative se structures but claim that it changes when the dative is present. In this case, the theme is demoted to internal argument position, although, as can be seen in (3c, d), it triggers verbal agreement. In a structure like (2), then, the higher Spec can bear Nominative or Dative Case, depending on its thematic role. I will therefore claim that the ones in (3) are structures with a quirky subject. In so doing, I will concentrate on three kinds of evidence showing that: a) The dative is the element which satisfies the EPP feature of T. b) The dative is not only the subject but the external argument of the construction, that is, the subject of predication and the closest element to T. c) The theme argument is internal and never «externalizes» when the dative is present, that is, it never raises to Spec TP. d) Verbal agreement with the theme is the result of long distance feature checking with T. It is well known that, within a given language, there are constructions in which the properties usually displayed by subjects seem to be scattered between more than one NP, due to the fact that the nature of those properties is also different Datives in Constructions with Unaccusative Se CatWPL 7, 1999 91 CatWPL 7 089-105 13/6/00 12:26 Página 91 (structural position in which they merge, structural position to which they move, thematic role, Case, agreement, etc.)1. In the cases under consideration, a dative, which does not agree with the verb, is, we claim, the subject of the construction, in the sense that it is the element of which the event is predicated and occupies the external argument (higher) position. Let us now go back to the structure in (2). In recent frameworks dealing with different aspects of grammar it has been proposed, on one hand, that there is an additional node above VP: VoiceP (Kratzer, 1996) , EventP (Harley, 1995), VP (Koizumi, 1993)), TrP (Collins, 1997), (small) v(Larson, 1988, Chomsky, 1995, 1998). It is widely accepted also that this node is directly related to the external argument. On the other hand, the idea that some aspects of argument structure are syntactically definable has been extensively developed and formalized. More specifically, in Hale and Keyser’s (1994, 1997) framework, the nature of argument structure follows from the properties of heads and the structural relations Specifier and Complement. Adjuncts and external arguments are not part of this l-syntax, since they are excluded from the internal structure of verbs. According to this, there are four types of heads. Some heads do not take complements or specifier (usually nouns fall under this category) and some heads take both (as in the case of prepositions, which are ‘dyadic’, in the authors’ terms).2The other two types of (verbal) heads take a complement. If this is nominal in nature, we will have a transitive or an unergative verb (if the noun conflates with the verb, as in run). This type of head, which Hale and Keyser call ‘monadic’, can take an external argument in syntax: the agent of the predicate. There are also heads which take a complement of an adjectival rather than a nominal nature. This case is more complex because the nature of adjectives makes them select for a Spec. In order to achieve this requirement, the adjective will have to be «parasitic» on a verbal head which will provide the structure with a Specifier. This would be a «complex dyadic» structure which gives rise to unaccusative verbs. That the preverbal NP in unaccusative constructions is an internal argument can be seen by the fact that the structure can be further embedded into a simple (monadic) verb (see (2)), that is, transitivized, as in (4): (4) a. The wind [turns [ V [the leaves red]] b. John [clears [the screen A]] So, in Hale and Keyser’s view, external arguments (agents or causers) are introduced in the syntactic structure by a separate verbal head, different from the one which contains the lexical verb and its internal arguments. The presence of an agent is thus structurally derived by projecting an adequate specifier position.3 92 CatWPL 7, 1999 Olga Fernández Soriano 1. See Harley (1995) for a discussion of the properties usually attributed to subjects and how they show in different constituents. 2. See Hale and Keyser (1997) for more details. Here we will be concerned by the other two types. 3. Kratzer (1996) makes a similar proposal. Her claim is also that the external argument is not an argument of the verb and has to be added via a sort of secondary predication. She further proposes CatWPL 7 089-105 13/6/00 12:26 Página 92 On the other hand, Harley (1995), provides the head above VP with content and supposes that it can be of two different types: CAUSE versus BECOME/ HAPPEN. This head, which she calls EventP, appears with all eventive verbs. Harley further claims that only the first type of head (CAUSE) can have a Spec, which will end up being an external argument with an agent/causer thematic role. It should be noted also that this argument has the property of being the initiator of the event. So Harley (1995) states the facts from another perspective: she claims that the separate head that is added to both (di)transitives / unergatives and unaccusatives is an abstract head with semantic content. If it is CAUSE, it will take a Spec (an external argument). If this abstract head is of the type HAPPEN / BECOME it will not merge with an external argument. It thus appears with unaccusatives. In Fernández Soriano (1999) some evidence is provided for the claim that the Spec of EventP can be filled in both cases (CAUSE and BECOME/ HAPPEN). That is, the nature of the abstract head will not determine the emptiness of its Spec but the type of argument that it will take as its Spec. The idea is that the event node, when BECOME / HAPPEN, can have its Spec filled by a locative (or a dative) argument in some cases. More specifically, in impersonal predicates such as the ones with haber «there be», sobrar «to exceed» / «to be extra», faltar «to miss / lack», constar «to state», on one hand, and suceder, ocurrir «to happen» as well as meteorological verbs, on the other, a locative phrase appears as an external argument. The locative argument is in some sense an initiator of the event, given that it always denotes a place where the event or state originates. This fact is related to its ability to appear in subject position.4The evidence for the claim that the locative is a (quirky) subject comes from different grounds. Among other things, the locative can never be a bare NP, in contrast with internal arguments or adjuncts ((5a) vs. (5b)). Second, it is the locative which raises in raising constructions (5c); and third, a quantifier in the locative phrase can bind a pronoun in the theme but not vice versa ((5d) vs. (5e)). Again, this is not the case for adjunct locatives (5f). I provide some examples in (5), but see Fernández Soriano (1999) for details: (5) a *En restaurantes grandes faltan camareros. in big restaurants lack waiters b. En restaurantes grandes no hay que invertir dinero. in big restaurants one should not invest money c. Aquí parece {sobrar / faltar / ocurrir} algo. here seems to be-extra/ lack / happen something (#Algo parece {sobrar / faltar / ocurrir} aquí.) Datives in Constructions with Unaccusative Se CatWPL 7, 1999 93 that there is a VoiceP node above VP, which is a functional category that introduces the external argument and is also responsible for accusative Case assignment. Marantz (1984) also argues for the proposal that external arguments do not appear in the lexical representation of verbs. 4. This is probably a way to restate Grimshaw’s (1990) idea about an «aspectual dimension», which is parallel to the thematic hierarchy. In this author’s framework, elements which participate in the first sub-event of the predicate are realized as external arguments. CatWPL 7 089-105 13/6/00 12:26 Página 93 d. En cada trabajo publicado constará / figurará el nombre de in each published work will be-stated / appear the name of su autor. its author e. *El nombre de cada autor constará en su trabajo. The name of each author will be-stated on his work f. * Contra cada pared debes apoyar su estantería correspondiente.5 against each wall you-must lean its corresponding shelf I would like to extend that analysis to these cases and suppose that a structure like (6) is adequate for unaccusative se structures with a dative. The event node, when CAUSE will take an external argument in the Nominative, and when BECOME / HAPPEN, in the dative case. So what I would like to propose is that the dative in structures like (3) is generated in the Spec of the Ev node above VP, just were the subject of the corresponding transitive sentences would appear. The structure would then be something like the one below (in which we abstract away from the position of the clitic se): As can be expected, if we are correct, the dative structure can never undergo transitivization / causativization, not even appear with analytical causatives. With respect to the possibility to form analytical causatives, dative structures contrast with unaccusative SE structures with no dative, as the following examples show: (7) a. *Juan hizo a Pedro olvidársele su nombre. (cf. ... olvidarse Juan made Pedro forget-SE-CLDT his name forget-SE de su nombre.) of his name b. *El viento hizo hundírseme el barco. (cf. ... hundirse the wind made sink-SE-CLDT the boat (on me) sink-SE el barco.) the boat 94 CatWPL 7, 1999 Olga Fernández Soriano 5. These sentences are good without the quantifier-pronoun relation. (6) Ev DAT Ev’ Ev VP DP V V A CatWPL 7 089-105 13/6/00 12:26 Página 94 c. *Juan (se) olvidó a Pedro el coche Juan (SE) forgot to Pedro the car d. *Tu respuesta hizo disipárseme las dudas. (cf.... disiparse your answer made vanish-SE-CLDT the doubts vanish-SE mis dudas.) my doubts It has been claimed (Chierchia, 1995) that, since these structures involve a propositional function, a process of expletivization is required to provide an external argument for predication. This can be done by externalizing the theme or, we would like to claim, by merging of a dative argument in Spec of EvP. There are other facts which indicate that the above analysis may be on the right track. I will concentrate on them in what follows: 1. The dative occupies the subject position 1.1. The first piece of evidence has to do with constituent order. The order shown in (3) is the unmarked one for these structures: sentences like (8a, b) are odd, with normal intonation, in clear contrast with (8c), with a preposed theme but no dative. (8) a. ?? Se le quemó la comida a Juan./ ??La comida se le SE CLDT burned the food to Juan / the food SE CLDT quemó a Juan. burned to Juan b. ?? Se le hundió el barco a Juan. / ?? El barco se le hundió SE CLDT sank the ship to Juan / the ship SE CLDT sank a Juan. to Juan c. La comida se ha quemado. the food SE has burned d. El barco se ha hundido. the ship SE has sunk In fact, information structure considerations indicate that the dative appears in the position usually occupied by (agentive) subjects. The preferred position for the dative in the cases under study is preverbal, and it is interpreted as unmarked, topic, just like subjects usually are. More specifically, if focus propagates in a direct path from the most embedded element (Zubizarreta, 1998; Cinque, 1993)6it would Datives in Constructions with Unaccusative Se CatWPL 7, 1999 95 6. See also Contreras (1983), Suñer (1982) for a detailed discussion of neutral information structure of sentences in Spanish.. CatWPL 7 089-105 13/6/00 12:26 Página 95 never get to the dative if this had moved from complement or adjunct position. The consequence of this would be that the whole structure containing a preverbal dative could never be interpreted as focus, but this is contrary to fact. A way to test the unmarked status of some particular word order is provided by the possibility of the sentence to be an answer to a question like what happens/ happened?, as proposed, among other, by Contreras (1983). In contrast with elements preposed by topicalization, focalization or left dislocation, the occurrence of a dative in preverbal position with the predicates under study does not affect the «neutral» information structure of the sentence. Sentences in (9a) through (9c) are examples all of which can be appropriate answers to the initial question, whereas (9d), with postverbal dative, cannot. (9e, f) show that this is not the case for postverbal agentive subjects and for other preposed arguments and adjuncts. Specifically the preposed dative in unaccusative se sructures contrasts with left dislocated goal datives, as (9f) shows: (9) ¿Qué ha pasado / pasó? what has happened / happened? a. A Juan se le ha quemado la comida. to Juan SE CLDT has burned the food b. Al niño se le ha perdido el bolígrafo. to he kid SE CLDT has lost the pen c. A tu hermano se le ha roto el coche. to your brother SE CLDT has broken the car d. El coche se le ha roto a tu hermano. e. #Ha roto el vaso el niño. has broken the glass the kid f. #A Juan le han dado el regalo. to Juan CLDT they-have given the present 1.2. Raising. When appearing under raising verbs like parecer «seem» it is the dative which raises to Spec of matrix TP. Raising of the theme is impossible if the dative is present. This is why sentences like (10b) are odd (maybe unless pronounced with marked intonation): (10) a. A Juan parece habérsele roto el coche. to Juan seems to-have-SE-CLDT broken the car b. ?? El coche parece habérsele roto a Juan. the car seems to-have-SE-CLDT broken to Juan c. A María parece perdérsele el niño continuamente. to María seems to-lose-SE-CLDT the kid all-the-time 96 CatWPL 7, 1999 Olga Fernández Soriano CatWPL 7 089-105 13/6/00 12:26 Página 96 d. ?? El niño parece perdérsele a María continuamente. the kid seems to-lose-SE-CLDT to María all-the-time ‘María seems to lose the kid all the time.’ Note that the equivalent structures without the dative are perfect with raising of the theme, so it is the dative which blocks the theme from moving to Spec TP. Although for some speakers the structure with raising of the dative are not perfect, the important fact is that raising of the theme is impossible (for all speakers) if the dative is present, but not in the unaccusative se structure, as can be seen in (11): (11) a. El coche parece haberse roto. the car seems to-have-SE broken b. Todo parece arreglarse sin problema. everything seems to get-fixed without problem c. La comida parece quemarse. the food seems to-burn-SE d. Este niño parece perderse continuamente. this kid seems to-get-lost all-the-time In this respect, this dative contrasts both with goals, which never block raising of the subject, and with experiencers of psych verbs, which are supposed to be in a high position (cf. Belletti and Rizzi, 1987), as (10’) shows. This, we would like to claim, is because the theme in psych verbs is externalized. (10’) a. Juan parece habérselo dicho a Pedro. ‘Juan seems to have told John about it.’ b. Tu respuesta no parece haberle molestado. ‘Your answer does not seem to have bothered him.’ c. La obra parece gustarle al público. ‘The play seems to be pleasant to the audience.’ What this contrast indicates is that the dative argument is closer to matrix T than the theme, a fact which is consistent with the structure proposed in (6). We are facing a special case of quirky dative, similar to the one analyzed for Icelandic (see Chomsky, 1998), but which generates in the embedded clause. 1.3. Binding. In the constructions under study, a quantifier in the dative phrase can bind a pronoun in the theme and the opposite does not hold. This shows again that the dative argument is higher than the theme. See sentences in (12): (12) a. A cada cocinero se le quemó su pescado. to every cook SE CLDT burned his fish ‘Each cook’s fish burned on him.’ Datives in Constructions with Unaccusative Se CatWPL 7, 1999 97 CatWPL 7 089-105 13/6/00 12:26 Página 97 Borer, H. (1997). «Licensing events: the role of locatives». Paper presented at MIT. Bresnan, J. (1990). «Levels of representation in Locative Inversion: A comparison of English and Chichewa». Ms., Stanford University. —(1994). «Locative inversion and the architecture of Universal Grammar». Language 70: 72-131. Bresnan, J. and J.M. Kanerva (1989). «Locative inversion in Chichewa: A case study of factorization in grammar». Linguistic Inquiry 20: 1-50. Chierchia, G. (1996). « Reference to kinds across languages». Ms., University of Milan. Chomsky, N. (1981). Lectures on Government and Binding. Dordrecht: Foris. — (1995). The Minimalist Program. Cambridge: MIT Press. —(1998). «Minimalist inquiries: The framework». MIT Occasional Papers in Linguistics 15. Cinque, G. (1993). «A null theory of phrase and compound stress». Linguistic Inquiry 24: 239-297. Collins, C. (1997). Local Economy. Cambridge: MIT Press. Contreras, H. (1983). El orden de palabras en español. Madrid: Cátedra. Den Dikken, M.; Maers, A. (1993). «Case dependencies: the case of predicate inversion». The Linguistic Review 10. Diesing, M. (1992). Indefinites. Cambridge: MIT Press. Fernández-Soriano, O. (1999). «Two types of impersonal constructions in Spanish: locative and dative subjects». Syntax 2-2. Freeze, R. (1992). «Existentials and other locatives». Language 68-3: 552-595. Grimshaw, J. (1990). Argument Structure. Cambridge: MIT Press. Hale, K.; Keyser, J. (1994). «On argument structure and the lexical expression of syntactic relations». In Hale, K.; Keyser, J. (eds.). The View from Building 20. Cambridge: MIT Press, p. 53-110. — (1997). «The basic elements of Argument Structure». Ms. MIT. Harley, H. (1995). Subjects, Events and Licensing. Doctoral Dissertation, MIT. Harley , H (1996). «Events, agents and the interpretation of VP shells». Ms. MIT. Jonas, D.; Bobaljik, J. (1993). «Specs for subjects: The role of TP in Icelandic». MITWPL 18: 59-98. Kratzer, A. (1996). «Severing the external argument from its verb». In Rooryck, J.; Zaring, L. (eds.). Phrase Structure and the Lexicon. Dordrecht: Kluwer, p. 119137. Levin, B.; Rappaport Hovav, M. (1995). Unaccusativity. At the syntax-Lexical Semantics Interface. Cambridge: MIT Press. Longobardi, G. (1994). «Reference and proper names: A theory of N-movement in Syntax and Logical Form». Linguistic Inquiry 25: 605-665. Luján, M. (1980). «Clitic promotion and mood in Spanish verbal complements». Linguistics 18: 381-394. Marantz, A. (1994). On the Nature of Grammatical Relations. Cambridge: MIT Press. —(1989). «Relations and configuration in Georgian». Ms., Univ. of North Carolina, Chapel Hill. McGinnis, M. (1997). «Case and locality in L-syntax: evidence from Georgian». Ms., MIT. Mendikoetxea, A. (1998). Impersonal Constructions and the Theory of Grammar. Ms., U.A.M. 104 CatWPL 7, 1999 Olga Fernández Soriano CatWPL 7 089-105 13/6/00 12:27 Página 104 Perlmutter, D. (1970). Deep and Surface Structure Constraints in Syntax. New York: Holt. Rigau, G. (1994). «Catalan presentational sentences and the properties of AGR». In Cinque, G. et al. (eds.). Paths Towards Universal Grammar. Studies in Honor of Richard Kayne. Washington: Georgetown University Press, p. 343-360. —(1996). «Relativized impersonality: Deontic sentences in Catalan». Ms., Universitat Autònoma de Barcelona. Rizzi, L. (1978). «A Restructuring Rule in Italian syntax». In Keyser, J. (ed.). Recent Transformational Studies in European Languages. Cambridg: MIT Press. Schütze, C. (1993). «Towards a minimalist account of quirky case and licensing in Icelandic». MITWPL 19: 321-375 Sigurdsson, H. (1996). «Icelandic finite verb agreement». Working Papers in Scandinavian Syntax 57: 1-46. Sobin, N. (1997). «Agreement, default rules and grammatical viruses». Linguistic Inquiry 28-2: 318-343. von Stechow, A. (1995). «Lexical decomposition in Syntax». In Egli et al. (eds.). Lexical Knowledge in the Organization of Language. Amsterdam: John Benjamins. Suñer, M. (1982). The Grammar of Spanish Presentational Sentence Types. Georgetown University Press. Wurmbrand, S. (1997). «Restructuring and Control». To appear in The Interpretive Track, MITWPL 25. Zubizarreta, M.L. (1998). Prosody, Focus and Word Order. Cambridge: MIT Press. Datives in Constructions with Unaccusative Se CatWPL 7, 1999 105 CatWPL 7 089-105 13/6/00 12:27 Página 105