Prospective Readings Revisited

Abstract In this paper I analyse the Prospective readings of sentences like El Ton cantava ahir, però li van ajornar la funció (‘Ton was to sing yesterday, but they postponed the performance’). I follow Delfitto’s (1998) proposal for generics and try to extend it to prospectives. I will point out that in prospective readings the event asserted is not the one expressed by the verb, but a complex event created from the material in the VP plus an operator binding the event variable of the verb: a modal operator with a lexical meaning paraphrasable as was to + Infinitive. This idea allows me to make a distinction between the temporal modifiers that are associated with the Event Time of the event of the verb and those that affect the temporal reference of the sentence. I will suggest that in prospective readings, the temporal modifiers associated with the Event Time of the event expressed by the verb modify this event as event-type, not as a concrete token: they behave as circumstantials. Key words: prospective readings, complex event, event-type, circumstantials. Resum. Una revisió de les lectures prospectives En aquest article analitzo les lectures prospectives d’oracions com El Ton cantava ahir, però li van ajornar la funció. Segueixo la proposta de Delfitto (1998) pels genèrics i miro d’estendrela als prospectius, que són una altra lectura de la morfologia imperfectiva. Proposaré que en les lectures prospectives l’esdeveniment respecte al qual es fa una asserció no és l’expressat pel verb, sinó un esdeveniment complex creat a partir del material que hi ha dins de l’SV més un operador que lliga la variable esdeveniment del verb: un operador modal amb una semàntica lèxica parafrasejable com was to + Infinitiu. Aquesta idea em permet fer una distinció entre els modificadors temporals que van associats amb el Temps de l’Esdeveniment de l’esdeveniment expressat pel verb i els que afecten la referència temporal de l’oració. Suggeriré que en les lectures prospectives, els modificadors temporals associats amb el Temps de l’Esdeveniment de l’esdeveniment expressat pel verb modifiquen aquest esdeveniment en tant que esdeveniment-tipus, no com una realització concreta: es comporten com circumstancials. Paraules clau: lectures prospectives, esdeveniment complex, esdeveniment-tipus, circumstancials. * For writing this paper I have benefited from the Proyecto DGS PB96-1199-C04-02 of the Ministerio de Educación y Cultura and from a FI scolarship granted by CIRIT 1997 SGR 00033 and 1999SGR 00113 of the Generalitat de Catalunya. CatWPL 8, 2000 23-46 Prospective Readings Revisited* Zulema Borràs Barber Universitat Autònoma de Barcelona. Departament de Filologia Catalana 08193 Bellaterra (Barcelona). Spain [email protected].es Received: April 3rd 2000 Accepted: October 27th 2000 CatWPL 8 001-137 29/5/2001 17:27 Página 23 1. Introduction The aim of this paper is to present an analysis of prospective readings associated with imperfective morphology (the reading of, for example, En Joan cantava demà, però s’ha ajornat l’actuació ‘Joan was to sing tomorrow, but the performance was postponed’). I adopt Delfitto’s (1998) idea that imperfective morphology encodes the instruction that the VP must be predicational and not propositional. The author proposes an analysis for habituals and progressives based on this conception of imperfectivity. In this paper I point out that this idea applies also to prospective readings of imperfective morphology, and I develop an analysis of prospective readings in these terms. If this analysis of imperfective morphology is on the right track, then we will be able to defend a uniform conception of imperfectivity and, thus, a unique lexical entry for imperfective morphemes. I will point out that, as it happens with habitual readings, in prospective readings the event1asserted is not the one expressed by the verb, but a complex event created from the material in the VP plus an operator binding the event variable of the verb. In prospective readings this operator is a modal operator with a lexical meaning paraphrasable as was to + Infinitive, which is characteristic of prospective readings. This idea that in prospective readings the event expressed by the verb is not asserted allows me to make a distinction between the temporal modifiers that are associated with the Event Time of this event and those that affect the temporal reference of the sentence. I will suggest that in prospective readings, the temporal/aspectual modifiers associated with the Event Time of the event expressed by the verb modify this event as event-type, not as a concrete token of this event-type. If I am right, in these cases the temporal/aspectual modifiers behave as circumstantials, that is, they contribute circumstances of the predication and enter the derivation in the thematic phase (the phase headed by the category v , according to Chomsky (1998)). Those modifiers that contribute to determine the temporal reference of the sentence are associated with the interval of truth of this sentence2, which in turn is included in the Event Time of the event which is asserted. 24 CatWPL 8, 2000 Zulema Borràs Barber 1. I use the term event in a broad sense, including events in a strict sense and states. From here on, if I do not state the contrary explicitly, I use it in this broad sense. 2. The interval of truth of a sentence is the interval with respect to which you must check the truth conditions of the sentence as a whole and with respect to which the speaker makes an assertion when stating the sentence. Table of Contents 1. Introduction 2. A unique analysis for imperfective morphology 3. The prospectives: an analysis 4. The role of temporal/aspectual modifiers 5. Concluding remarks References CatWPL 8 001-137 29/5/2001 17:27 Página 24 2. A unique analysis for imperfective morphology In this section I will present Delfitto’s (1998) analysis for generics (2.1). This author associates genericity with imperfective morphology, and points out without going into detail that the same analysis should extend to other readings of imperfective morphology, like the progressive one. In subsection 2.2 I present an analysis for the progressive reading of imperfective morphology, based on Delfitto’s (1998) comments. This notwithstanding, in section 2.3 I will present another group of readings, the prospective readings, which are the central subject matter of this paper. 2.1. Generics Delfitto (1998) deals with the relation between genericity on the one hand and the absence of any aspectual encoding or the presence of default aspectual morphology (which crosslinguistically, the author claims, is imperfective morphology)3 on the other hand. He puts forward a non quantificational analysis of genericity, in which generic sentences like the example in (1) (Delfitto’s (1)) have a predicational logical form: they involve the ascription of a property to an intensionally defined individual. (1) John is intelligent. Predicates associated with imperfective morphology involve the realization of one of their arguments as external to the maximal projection of the head. This argument occupies the specifier position of a functional projection PredP, and the head Pred performs a type shifting operation. As a result, the VP is interpreted as a property (type <e,t>), and the argument in [Spec,PredP] is interpreted intensionally, as a property set. Therefore, from Delfitto’s (1998) point of view, generic sentences are not based on episodic sentences, by inserting an implicit operator that quantifies over time variables or situation variables as in other approaches (see Delfitto (1998:1-2) for some references). The intensional/nomic meaning of generic sentences depends on the intensional/modal interpretation of predicational structures. The same analysis applies to cases like (2) (Delfitto’s (5)), with an eventive predicate: the verbal predicate has a PredP in its extended projection, and its external argument occupies the [Spec,PredP]. (2) Abigail fetches my newspaper. Eventive predicates like the one in (2) have an extra event argument, realized as a Davidsonian event variable associated with the verb head of VP. This requires the presence of a Q-adverb that binds this event variable, and yields a logical representation like (3) (Delfitto’s (15)) Prospective Readings Revisited CatWPL 8, 2000 25 3. From now on I will refer to both imperfective morphology and absence of aspectual marking as imperfective morphology. CatWPL 8 001-137 29/5/2001 17:27 Página 25 (3) λx [Qe [C(e,x)] [fetch my newspaper (e,x)]] (Abigail), which means literally: «the property of x such that for n events einvolving x, x fetches the newspaper in e, is a property of Abigail». Thus, with eventive predicates, it is a complex property which is ascribed to the individual associated with the argument in [Spec,PredP]. This complex property is created by means of an implicit Q-adverb quantifying over the event variable of the verb. Therefore, according to Delfitto (1998), both eventive and non-eventive generic sentences share the same syntactic format (predication in the sense explained above), and involve the ascription of a property to an individual. Then, generic sentences are not based on episodics, and the nomic/intensional reading of generics depends on their predicational structure. With respect to imperfective morphology, it encodes that the maximal projection of the predicate is predicative (i.e., the VP does not constitute a proposition, but a property, which is predicated of the argument in [Spec,PredP]); that is, one of the arguments of the verb occupies a position outside the VP. 2.2. The progressive reading Delfitto (1998) points out that, in order to defend a unitary interpretation of the imperfective morphology, we should be able to analyse it in terms of the predicational analysis when presenting the progressive reading as well (see (4)). (4) Ara el Joan canta.4 now the Joan sings ‘Now Joan is singing.’ However, Delfitto (1998) does not explicitly present an analysis of the progressive reading of imperfective morphology. Here I will try to construct an analysis for this case, taking as a point of departure Delfitto’s comments scattered throughout his paper. According to this author, when presenting the progressive reading, the verb with imperfective morphology enters the derivation endowed with an extra argument of spatio-temporal location5. This spatio-temporal argument can be syntactically realized as a phonetically null argument when it is interpreted as a speaker-orientated indexical, as corresponds to structures interpreted as observational reports on the passing scene, for instance progressive sentences. To Delfitto, the progressive reading obligatorily presupposes a given/familiar time at which the ongoing event takes place. This given/familiar time would correspond to the speaker-orientated null spatio-temporal argument. The presence of this indexical spatio-temporal loca26 CatWPL 8, 2000 Zulema Borràs Barber 4. Unless I say the contrary, examples which are not English are Catalan sentences. 5. This argument of spatio-temporal location would be the Kratzerian spatio-temporal argument of stage-level predicates. According to Delfitto, this argument is optional, and it is not present in the syntactic derivation with all stage-level predicates, but only when these predicates somehow select it, according to the construction in which they appear. See footnote 7 for more comments on this issue. CatWPL 8 001-137 29/5/2001 17:27 Página 26 tion of the event triggers an operation of existential closure at the VP level and, consequently, of existential quantification over the verb’s event variable (the other possible semantic trigger for this operation being, according to Delfitto, perfective morphology). This explains the fact that structures associated with imperfective morphology which present a progressive reading are episodic, despite their predicational character. In accordance with the analysis presented in Delfitto (1998), an operation of lambda-abstraction applies to the trace of the verb’s argument displaced to [Spec,PredP]. This is the type-shifting operation performed by Pred, which makes the VP be interpreted as a property, as a lambda-abstract predicated of the argument displaced to [Spec,PredP] (see, in subsection 2.1, the discussion regarding example (2)). Because of this predicational relation between the argument in [Spec,PredP] and its VP-internal trace, the displaced argument cannot undergo reconstruction to its original position. This ensures the intensional interpretation of this argument. When imperfective morphology is associated with a progressive reading, the subject of the predicate (i.e., «el Joan» in (2)) receives a non-intensional, existential reading. This could be accounted for straightforwardly if this argument was reconstructed to its VP-internal position (because of the operation of existential closure operating at the VP level, triggered by the presence of the indexical spatiotemporal location mentioned earlier). However, this is impossible if this argument has previously been displaced to [Spec,PredP]. The idea suggested explicitly by Delfitto is that, in structures with a progressive reading, it is the spatio-temporal argument of the predicate that is displaced to [Spec,PredP] in order to satisfy the requirement encoded in the imperfective morphology that the VP be non-propositional. The indexical character of this external argument would prevent the application of the type-shifting operation, because indexicals are not sensitive to intensional operators. To summarize, in the progressive reading of imperfective morphology, the argument that goes to [Spec,PredP] is a spatio-temporal null argument, which is an indexical. Moreover, there is existential quantification over the event variable of the verb, and consequently the VP is propositional. 2.3. The prospective reading In this section I describe the data that present what I call the prospective reading. Consider the data in (5) and (6): (5) El Ton cantava ahir, però li van ajornar la the Ton sang-imperf. yesterday but to-him they-postponed the funció. performance ‘Ton was to sing yesterday, but they postponed the performance.’ Prospective Readings Revisited CatWPL 8, 2000 27 CatWPL 8 001-137 29/5/2001 17:27 Página 27 (6) El Ton cantava demà, però li van ajornar la the Ton sang-imperf. tomorrow but to-him they-postponed the funció. performance ‘Ton was to sing tomorrow, but they postponed the performance.’ The examples in (5) and (6) have prospective readings, which, as these sentences show, are stative readings6. However, this kind of reading only appears with predicates that are endowed with a Davidsonian event variable7(see (7)). (7) a. #El Joan estava malalt ahir. the Joan was-imperf. ill yesterday ‘Joan was ill yesterday.’ b. #Antonio era futbolista a esta hora. (Spanish) Antonio was-imperf. footballer at this time ‘Antonio was a footballer at this time.’ 28 CatWPL 8, 2000 Zulema Borràs Barber 6. In the sentences in (5) and (6), the predicate associated with imperfective morphology presents a counterfactual reading. Here I will leave this issue aside. Note, however, that this counterfactuality of prospective readings with past imperfective morphology is a conversational implicature, not an assertion. In fact, there are sentences like (i) that have a prospective reading which is not counterfactual: (i) a. El Joan venia, a sopar, aquesta nit, no? the Joan came-imperf. for dinner this night not ‘Joan was to come, for dinner, tonight, didn’t he?’ b. Sí, amb ell som set. yes, with him we-are seven ‘Yes, counting him, we are seven.’ On the other hand, even in the case of sentences that can have a counterfactual prospective reading, the counterfactual implicature can be cancelled. See (ii): (ii) A: Ah, per cert, tu quin dia cantaves? particle by certain you what day sang-imperf. ‘Oh, by the way, what day were you going to sing?’ B: Jo cantava ahir. I sang-imperf. yesterday ‘I was to sing yesterday.’ A: Ai, ho sento, jo volia venir a sentir-te! particle it I-feel I wanted-imperf. come-infinitive to hear-infinitive you ‘Oh, I’m sorry, I wanted to come to hear you!’ 7. I follow Delfitto (1998) in the distinction that he establishes between the Davidsonian event variable and the Kratzerian spatio-temporal argument. The obligatorieness/non-obligatorieness of the Davidsonian event variable permits Delfitto to distinguish individual-level from stage-level predicates. According to this author, individual-level predicates are those that can receive an eventive or a non-eventive interpretation, in his frame those that can enter the syntactic derivation with or CatWPL 8 001-137 29/5/2001 17:27 Página 28 In (7a), with a predicate like to be ill, we only obtain the prototypical stative reading, and we cannot create the complex ‘predicate’ that would give rise to the prospective reading. In (7b), with the modifier at this time indicating the Utterance Time of the sentence8, we can only obtain a habitual reading. In this reading of (7b), the predicate to be a footballer presents an eventive reading that can be paraphrased, for example, as ‘behaved/acted as a footballer’. Delfitto (1998:15) proposes to consider i-level (individual-level) predicates as those for which the presence of a Davidsonian event variable and the consequent eventive reading are optional. This definition seems to apply to the predicate in (7b), which in the derivation of this sentence takes an event argument. If the habitual reading of (7b) shows that the predicate to be a footballer can take an event variable, even though we cannot obtain a prospective reading for this sentence, it seems that the necessity of an event variable in the derivation is not the restriction that accounts for the presence/absence of a prospective reading in (7). It seems then that the presence of a Davidsonian event variable is a necessary but not sufficient condition for prospective readings. There is another kind of evidence in the same direction. It might be difficult to obtain the prospective reading with certain activities (see (8)), despite the fact that they are clearly endowed with an event variable: (8) a. ?Ando mañana.9(Spanish) I-walk tomorrow ‘I walk tomorrow.’ b. ?Parlava avui. s/he-talked-imperf. today ‘S/he was to talk today.’ I believe that the restrictions illustrated in (8) are pragmatic in nature, since it might be hard to obtain the prospective reading in (8a), but it is much easier, I believe, with a verb like to run, for example. Thus, the Spanish sentence Corro mañana (‘I run tomorrow’) can be paraphrased by ‘I take part in the competition tomorrow’ more easily, for example, because the verb correr (‘to run’) is very usual with this kind of meaning. As regards (8b), it is quite common in the jargon of linguists, in which to talk is very usual with the meaning ‘to give a talk’, ‘to give a lecture’. In short, it is difficult to obtain the prospective reading Prospective Readings Revisited CatWPL 8, 2000 29 without a Davidsonian event variable associated with them. On the other hand, stage-level predicates are those that always enter the syntactic derivation with a Davidsonian event argument associated with them. In contrast, according to Delfitto (1998), the Kratzerian spatio-temporal argument is optionally present with stage-level predicates, and it can undergo syntactic realisation as a null category depending on the syntactic configuration or on specific selectional requirements of the predicate. 8. With the term Utterance Time I refer to the time at which the speaker utters the sentence. This term corresponds, then, to the Speech Time of Reichenbach (1947). 9. I thank Myriam Uribe-Etxebarria for calling my attention to these restrictions. CatWPL 8 001-137 29/5/2001 17:27 Página 29 with activities that do not easily take an accomplishment-like reading, a bounded reading. This leads to the hypothesis that it is the boundedness or the ability to be bounded of the predicate that is required to obtain prospective readings10, and not just the presence of an event variable (which is independently required in order to be bound by the modal operator). This also explains the absence of prospective readings in the examples in (7): it seems that the predicates to be ill and to be a footballer can hardly take a bounded reading, they codify an unbounded reading. Then both prospectives and habituals require the predicate to take a Davidsonian event variable, but prospectives also require it to take a bounded reading. On the other hand, prospectives and habituals differ from each other in that the relation between the event expressed by the verb and the complex predicate that results from an operator (a modal operator and a Q-adverb respectively) binding the event variable of the verb in each case is different. In a sentence like (9), (9) John smokes. the complex property ascribed to John is literally, «the property that John has in virtue of the relation between (...) the events in which John is involved that naturally favor smoking (...) [and] the events in which John actually smokes» (see Delfitto (1998:2)). This relation depends on the lexical meaning of the overt/covert Q-adverb that quantifies over the event variable of the verb. In the case of prospective readings, I will propose that the relation between the event expressed by the verb and the complex event which is the one asserted in the sentence depends on the lexical meaning of a modal operator that binds the event variable of the verb. The meaning of this complex event roughly corresponds to the meaning of the periphrasis was to + Infinitive combined with the meaning of the VP (that is, the meaning of the verbal predicate plus its arguments and modifiers). Although this periphrasis only combines with past tense, and the prospective readings appear both with present and past imperfective forms, the periphrasis with was to + Infinitive is useful to see the contribution of the operator that binds the event variable of the verb to the prospective reading. Both in habitual and prospective sentences, the event is not the one encoded in the lexical meaning of the verbal predicate, but a complex one that results from an operator binding the event variable of the verb. However, the event expressed by the verbal predicate can have temporal modifiers that restrict or specify the Event Time that corresponds to this event (see (10) and (11))11. 30 CatWPL 8, 2000 Zulema Borràs Barber 10. I mean predicates that are or can be interpreted as having a bounded Event Time without requiring the presence of perfective morphology nor the presence of a modifier or of a sentence that introduces another Event Time (something like En Joan estava malalt, però es va prendre la medicina i de seguida es va posar bé, ‘Joan was ill, but he took the medicine and he recovered at once’, where the event to recover interrupts the event to be ill). 11. With the term Event Time I refer to the total interval in which the event or the state of the sentence holds. CatWPL 8 001-137 29/5/2001 17:27 Página 30 (10) El Lluís corria la marató demà, però l’ han the Lluís ran-imperf. the marathon tomorrow but it they-have ajornada. postponed. ‘Lluís was to run the marathon tomorrow, but it was postponed.’ (11) L’ any que vaig compartir pis amb ell, el Lluís the year that I-shared flat with him the Lluís dinava a les dues. had-lunch-imperf. at the two. ‘The year I shared flat with him, Lluís had lunch at two.’ The modifiers in italics in (10) and (11) enter the derivation in the thematic phase (see Chomsky (1998)) and affect only the predication, and not the temporal reference of the sentence, which is encoded in the functional phase (the phase dominated by the functional category C, according to Chomsky (1998)). It is clear that the interval of truth of (10) does not even overlap the interval designated by the modifier demà (‘tomorrow’), and that the interval of truth of (11) is not confined to the intervals designated by the modifier a les dues (‘at two’), but includes all the period designated by the modifier l’any que vaig compartir pis amb ell (‘the year I shared flat with him’), which in turn includes the intervals designated by the other modifier. The modifier l’any que vaig compartir pis amb ell (‘the year I shared flat with him’) enters the derivation in the functional phase, and modifies the temporal reference of the sentence: it determines its interval of truth. The interval of truth of (10) is not linguistically delimited, because, differently from (11), in (10) there is no modifier associated with a functional category, that is, there is no modifier that determines the temporal reference of the sentence. While in progressive readings there is a presupposed given/familiar time (see section 2.2, and footnote 7), in habituals and prospectives there is no such a presupposition and, consequently, there is no null spatio-temporal argument with an indexical interpretation (recall that this argument corresponds to what sometimes is called Assertion Time12, which is the interval of truth of the sentence). This makes possible that with habituals and prospectives, when there is no modifier associated with the interval of truth, the exact temporal reference of the sentence remains linguistically indetermined. The exact duration of the Event Time of a lexical state combined with imperfective morphology is never linguistically determined (unless we have another event, introduced by another sentence, which interrupts it and determines its end point)13. In these cases, modifiers can only delimit the interval of truth (which is contained in the Event Time). In (11), the modifier l’any que vaig compartir pis Prospective Readings Revisited CatWPL 8, 2000 31 12. The term Assertion Time comes from Klein (1995) and designates the interval of which an assertion is made. 13. Because of their intrinsic nature, the events expressed by lexical states do not have boundaries that can be modified by temporal or aspectual modifiers. It seems that these boundaries can only be introduced by perfective morphology or by another event which is expressed by another sentence. See footnote 10. CatWPL 8 001-137 29/5/2001 17:27 Página 31 It seems to me that the example in (15a) presents a habitual reading constructed from a complex predicate formed with the modal operator of prospective readings. Thus, we have a habitual-prospective reading. This example would be analysed as follows: as I said earlier, after the modal operator binding the event variable of the verb and yielding the complex stative predicate of prospective readings, this complex predicate selects an event argument, and then, in a case like that in (15a), a Q-adverb quantifies over this variable, yielding the habitual reading. The example in (15b) is a case of episodic prospective reading, as those that we have seen before in this paper. As I proposed earlier, in this case the event variable selected by the complex stative predicate would be closed by an existential quantifier. This would yield the episodic reading. Now, having seen that there are cases like (15a) of habitual-prospective readings, in which there is a Q-adverb which quantifies over the event variable, I can go back to the question of the semantic trigger for the operation of existential closure in cases like (15b). As it is possible to have a Q-adverb quantifying over the event variable selected by the complex stative predicate, there need not be and, in fact, there cannot be any semantic trigger that forces the existential closure of this variable. When there is an operation of existential closure, we will obtain the episodic prospective reading, and when there is a Q-adverb quantifying over the event variable, we will obtain the habitual prospective reading. To summarize, according to my analysis, in prospective readings a complex predicate is created by a modal operator that binds the event variable of the verb. This explains the fact that prospective readings only appear with verbs endowed with an event variable. On the other hand, the operator imposes the restriction that the Event Time of the verb is bounded or can be interpreted as bounded, because of the role played by the semantics of this verb in the semantics of the resulting complex predicate. The resulting complex predicate is a stage-level predicate (because of its compositional semantics) and as such, it takes an event variable which can be bound by an existential operator, yielding an episodic prospective reading (see, for example, (15b)), or by a Q-adverb, yielding a habitual prospective reading (see, for example, (15a)). In episodic prospective readings, there is an assertion of a complex event in which the subject is involved. In habitual prospective readings, a complex property is predicated of the subject. 4. The role of temporal/aspectual modifiers In section 2.3 I made a distinction between modifiers which enter the derivation in the thematic phase and modifiers which enter the derivation in the functional phase. I repeat here the examples (10)-(11): (10) El Lluís corria la marató demà, però l’ han the Lluís ran-imperf. the marathon tomorrow but it they-have ajornada. postponed ‘Lluís was to run the marathon tomorrow, but it was postponed.’ 38 CatWPL 8, 2000 Zulema Borràs Barber CatWPL 8 001-137 29/5/2001 17:27 Página 38 (11) L’ any que vaig compartir pis amb ell, el Lluís the year that I-shared flat with him the Lluís dinava a les dues. had-lunch-imperf. at the two ‘The year I shared flat with him, Lluís had lunch at two.’ In section 2.3 I suggested that the preposed modifier in (11) is part of the functional phase, and determines the interval of truth of the sentence. In contrast, the postverbal modifiers in (10) and (11) are part of the thematic phase, and contribute their meaning to the predication, not to the temporal reference of the sentence. Therefore, in (10), the event that takes part in the composition of the complex predicate of the prospective reading is not to run, but to run the marathon tomorrow; and in (11), the event that takes part in the composition of the complex property that is predicated of the external argument, yielding the habitual reading, is not to have lunch but to have lunch at two. These postverbal modifiers do not make any contribution to determine the interval of truth of the sentences in which they appear. De Swart (1998) considers that the basic position of time adverbials and temporal clauses is the sentence final position, because they are PPs which modify the VP, and that when preposed they are often interpreted as topicalized16 (see de Swart (1998:5) for some references). For the cases when they appear in the postverbal position, de Swart (1998) studies what happens when they are focused on. Here I am interested in the preverbal (and dislocated postverbal) and the neutral postverbal positions. Regarding the cases when we have topicalization, it has usually been considered that modifiers in clausal onset position introduce a reference time, a frame time for the main clause (see de Swart (1998) and references therein). See the examples in (16)-(17): (16) Quan jo vaig arribar, el Joan cantava al Liceu. when I arrived the Joan sang-imperf. at-the Liceu ‘When I arrived, Joan was singing at the Liceu’/‘When I arrived, Joan sang at the Liceu (his job was to sing at the Liceu)’/‘When I arrived, Joan was to sing at the Liceu.’ (17) El Joan cantava al Liceu, quan jo vaig arribar. the Joan sang-imperf. at-the Liceu when I arrived ‘Joan was singing at the Liceu, when I arrived’/‘Joan sang at the Liceu (his job was to sing at the Liceu), when I arrived’/‘John was to sing at the Liceu, when I arrived.’ Prospective Readings Revisited CatWPL 8, 2000 39 16. De Swart (1998) does not say anything about topicalized postverbal modifiers. In Catalan and Spanish they are very usual. I believe that everything I will say here about preposed topicalized modifiers holds for postverbal topicalized modifiers too. CatWPL 8 001-137 29/5/2001 17:27 Página 39 In (16) and (17) the modifier quan jo vaig arribar (‘when I arrived’) is dislocated and provides a reference time for the sentence. It happens to be the case that both (16) and (17) can get three readings: a progressive reading, a habitual reading and a prospective reading. Though providing a reference time in the three readings, it seems that the role of the modifier Quan jo vaig arribar (‘when I arrived’) is not exactly the same in the three cases. According to the reinterpretation of Delfitto & Bertinetto’s (1995) arguments that I made in section 3.1, in terms of Delfitto’s (1998) spatio-temporal argument, in progressive readings there is a presupposed given/familiar time which is not present in habitual readings nor, as I suggest, in prospective readings. According to Delfitto (1998), the progressive reading is a case of «observational report on the passing scene» (see Delfitto (1998:46)). This explains the presence of the null spatio-temporal argument which gets an indexical reading (it expresses the now/here of the speaker). In my opinion, this is especially clear when imperfective morphology combines with present tense (see (18)), but the presupposition of a concrete familiar time is also present when it combines with past tense (see (19)). (18) a. Calleu, no feu soroll, que el Joan canta. be quiet not make noise that the Joan sings-imperf. ‘Be quiet, don’t make noise, because Joan is singing.’ b. Calleu, no feu soroll, que ara el Joan canta. be quiet not make noise that now the Joan sings-imperf. ‘Be quiet, don’t make noise, because now Joan is singing.’ (19) Quan vam entrar el Joan cantava. when we-came-in the Joan sang-imperf. ‘When we came in Joan was singing.’ In (19), the modifier quan vam entrar (‘when we came in’) is associated with this temporal argument, and it contributes to determine the given/familiar time that this argument expresses. In (18a), this familiar time coincides with the Utterance Time, and this is why a modifier associated with this temporal argument is unusual, or at least gives a marked reading (see (18b)). This same argument can be made for the progressive reading of the examples in (16) and (17). In contrast, in the habitual reading of (16) and (17), if there was no modifier, the sentences could mean, in the right context, something like, for example ‘When he was alive, Joan used to sing’. Therefore, when the modifier quan jo vaig arribar (‘when I arrived’) appears, it restricts the truth interval of the sentence, but there is no presupposition of a given/familiar time in Delfitto’s (1998) sense. In the prospective reading of (16) and (17)17, if there was no modifier, the sentences could mean something like, for example, ‘Joan was to sing at the Liceu, I 40 CatWPL 8, 2000 Zulema Borràs Barber 17. This prospective reading is an episodic one, because of the modifier. Without this modifier or with other modifiers it could have both an episodic and a habitual prospective reading. CatWPL 8 001-137 29/5/2001 17:27 Página 40 think’, without any presupposition of a given/familiar time. When the modifier quan jo vaig arribar (‘when I arrived’) appears, it simply restricts the interval of truth of the sentence. Summarizing, it seems clear that preposed temporal modifiers are often topicalized, and that they introduce a reference time, a frame time for the main clause. In progressive readings, this reference time corresponds to the givenfamiliar time that Delfitto (1998) assumes with this kind of readings. In habitual and prospective readings, these modifiers just restrict the interval of truth of the sentence. Regarding temporal/aspectual modifiers that appear in an unmarked postverbal position, I want to suggest that they are circumstantial modifiers, in the sense that they contribute a circumstance to the predication, and that, as such, they enter the derivation in the thematic phase (see section 2.3). As part of the predication, these modifiers are associated with the event time of the event expressed by the verb. But they modify the event as an event-type, not as a specific event nor as an existentially quantified event. In habitual and prospective readings this is especially clear, because the event which is asserted and which gets temporal reference according to the temporal information in the functional phase, is not the one expressed by the verb plus its modifiers. The event which is asserted is a complex one created from the material in the VP, by means of an operator (a modal operator in prospective readings and a Q-adverb in habitual readings) binding the event variable of the verb. This dissociation permits to see more clearly the difference between modifiers associated with the event expressed by the verb and those associated with the interval of truth, which is included in the Event Time of the asserted complex event. Consider the data in (20): (20) a. A partir del mes que ve el Joan cantava from of-the month that comes the Joan sang-imperf. tres vegades cada setmana, però al final li ho han three times each week but in-the end to-him it have deixat en dues. left in two ‘From next month on Joan was to sing three times a week, but in the end they left it in two.’ b. ?A partir del mes que ve el Joan actua/actuava from of-the month that comes the Joan acts/acted-imperf. més sovint. more often ‘From next month on Joan is going to /was to act more often.’ (20a) and (20b) contain, in postverbal position, modifiers which usually go with habituals, and despite this, (20a) and (20b) do not have habitual but prospective readings. I believe that this is evidence in favour of the idea that these postverbal modifiers affect only the predication, they modify the event expressed by the verb Prospective Readings Revisited CatWPL 8, 2000 41 CatWPL 8 001-137 29/5/2001 17:27 Página 41 as event-type, but do not directly affect the main event of the sentence, which is the event that can be paraphrased as was to + Infinitive. Some more evidence for the idea that temporal/aspectual modifiers can affect the predication comes from the fact that the combination of predicates with these kind of modifiers is not free, there are some restrictions. I claimed that the postverbal modifiers like the one in (21), which I repeat here, affect the Event Time of the event expressed by the predicate. If this suggestion is on the right track, then we expect to find restrictions between events with a certain kind of Event Time and these modifiers. (21) El Lluís corria la marató demà, però l’ han the Lluís ran-imperf. the marathon tomorrow but it they-have ajornada. postponed ‘Lluís was to run the marathon yesterday, but it was postponed.’ For example, in cases in which imperfective morphology is associated with a lexical state we cannot have a postverbal non-topicalized temporal/aspectual modifier associated with the Event Time. See (22): (22) El Joan estava cansat ahir. the Joan was-imperf. tired yesterday ‘Joan was tired yesterday.’ The only reading that can be obtained for (22) is one in which the modifier ahir (‘yesterday’) is focused on and it is associated with the interval of truth of the sentence, which corresponds to the Assertion Time (see footnote 12). This means that the contrastive focus is on the assertion made by the speaker: this reading could be paraphrased as «It is yesterday that, according to me, Joan was tired, not another day». The fact that we cannot obtain a reading in which the modifier is associated with the Event Time of the event to be tired is easily explained if we assume, as I do, that the Event Time of a lexical state can only be bounded by the Event Time of another event that interrupts it, or by combining the verb that expresses the state with perfective morphology. Therefore, since in (22) the Event Time of estava cansat (‘was tired’) cannot be interpreted as bounded, there cannot be a modifier which expresses a delimited interval and which associates with this Event Time. A related kind of evidence in favour of the hypothesis that temporal/aspectual modifiers can be circumstantials in the sense that they affect the predication, that they enter the derivation in the thematic phase, comes from the restrictions of combination between temporal/aspectual modifiers. Consider the data in (23)-(24): (23) a. *La setmana passada el Joan estava cansat ahir. the week last the Joan was-imperf. tired yesterday ‘Last week Joan was tired yesterday.’ b. Ahir el Joan estava cansat a la tarda. yesterday the Joan was-imperf. tired in the afternoon ‘Yesterday Joan was tired in the afternoon.’ 42 CatWPL 8, 2000 Zulema Borràs Barber CatWPL 8 001-137 29/5/2001 17:27 Página 42 (24) La setmana passada el Joan cantava ahir, però abans the week last the Joan sang-imperf. yesterday but before d’ahir em van dir que s’ havia ajornat. yesterday to-me they-told that it had-been postponed ‘Last week Joan was to sing yesterday but the day before yesterday I was told that it had been postponed.’ As in (22), in (23a) and (23b) the postverbal modifier is associated with the interval of truth of the sentence. The preverbal topicalized modifiers in (23a) and (23b) are also clearly associated with the interval of truth of the sentence. This leads to the hypothesis that the contrast in grammaticality between (23a) and (23b) comes from the restriction that two modifiers associated with the same element (that is, the interval of truth or the Event Time) must point at intervals that have a relation of inclusion. This restriction is fulfilled in (23b) but not in (23a). In contrast, (24) is grammatical despite the fact that the two intervals pointed at by the modifiers have not a relation of inclusion. This can be explained straightforwardly if I am right when claiming that the preverbal modifier in (24) is associated with the interval of truth of the sentence and the postverbal one affects the Event Time as a circumstantial. It could be objected that this just shows that one modifier is associated with the interval of truth of the sentence and the other one with the Event Time of the verb, but not that this Event Time is not referential and that the modifier is a circumstantial. But if the Event Time was referential we would have a progressive reading of the imperfective morphology, and then the interval of truth of the sentence would be included in the Event Time, and it would be a case like the one in (23a). Finally, another kind of evidence that can be adduced in favour of this hypothesis is the cliticisation data. Consider the examples in (25) and (26). (25) A: Tu a les quatre cantaves, oi? you at the four sang-imperf. interjection ‘At four you were singing, weren’t you?’ B: No, jo a les quatre feia la migdiada, així que no I at the four made-imperf. the siesta so that devia ser algú altre, qui has sentit cantar. ought-to be someone else who you-have heard sing-infinitive ‘No, at four I was having a siesta, so it ought to be someone else that you heard singing.’ (26) A: El Pere demà cantava a les cinc,oi? the Pere tomorrow sang-imperf. at the five interjection ‘Tomorrow Pere was to sing at five, wasn’t he?’ B: No ho crec, perquè el Pere no hi canta mai, a not it I-believe because the Pere not temp.-cliticsings never at les cinc, ell sempre canta a la sessió de les set. the five he always sings at the session of the seven ‘I don’t think so, because Pere never sings at that time, at five, he always sings at the seven performance.’ Prospective Readings Revisited CatWPL 8, 2000 43 CatWPL 8 001-137 29/5/2001 17:27 Página 43 In (25A) the imperfective morphology presents a progressive reading, and the modifier a les quatre (‘at four’) is associated with the spatio-temporal argument characteristic of progressives. Both in (25A) and in (25B) this modifier is displaced from its original postverbal position, and there is no reassumptive clitic within the sentence. The imperfective morphology in (26A) presents a prospective reading, and the postverbal modifier a les cinc (‘at five’) is associated with the Event Time of the verb and is a circumstantial, it modifies the predication. The imperfective morphology in (26B) presents a habitual reading. In (26B) the modifier is also associated with the Event Time of the verb, and it is displaced out of its original postverbal position. In contrast to what happens in (25B), there is a reassumptive temporal clitic within the sentence. In works like Matthews (1981), Speas (1990), Rigau (in press) and Pascual (1999) a distinction between two types of modifiers is established: on the one hand, those that contribute to the predication, which establish a narrow relation with the predicate, and on the other those that do not establish any kind of relation with the predicate but with some functional category (sentential tense, aspect, mood, etc.) Pascual (1999), for example, states that modifiers belonging to this last class never agree with the verb, and never cliticise nor are incorporated into the verb, while those belonging to the first class, in some languages can agree with the verb, can cliticise like an argument and can be incorporated into the verb. The only one of these three properties that can be tested for Catalan is the cliticisation property. See the example in (27), (Pascual’s (1999) (15)): (27) Amb aquest ordinador, tots hi hem escrit la tesi. with this-masc-sing computer all pr have-I-pl written the thesis ‘With this computer, all of us have written the thesis.’ This example is parallel to that in (26), in which the temporal adjunt cliticises as the instrumental phrase does in (27). To my way of thinking, the fact that temporal modifiers cliticise like an argument when they seem to be associated with the Event Time of the event expressed by the verb is another argument in favour of my proposal that there are temporal modifiers which behave as circumstantials and contribute to the predication. To summarize, in this section I have suggested that temporal/aspectual modifiers can behave as circumstantials, as modifiers of the predication, and not just as sentence modifiers that contribute to determine the temporal reference of the clause. I have pointed out that this is possible in prospective and habitual readings, because in these readings the event expressed by the verb is not asserted nor existentially quantified, and therefore it can be modified as an event-type. 5. Concluding remarks In the previous sections I pointed out that in prospective readings, like in habitual readings (see Delfitto (1998)), the event asserted is not the event expressed by the verb, but one resulting from a complex event created by an operator binding 44 CatWPL 8, 2000 Zulema Borràs Barber CatWPL 8 001-137 29/5/2001 17:27 Página 44 the event variable of the verb. While in habituals what is asserted is the attribution to the external argument of a complex property created from the material in the VP (which does not include the external argument (see Delfitto (1998))), in prospective readings we have two possibilities. We can have episodic prospective readings and habitual prospective readings. The complex predicate created from the material in the VP (which does not include the external argument) selects, in virtue of its semantics, an event variable. This event variable can be bound by means of an operator of existential closure, yielding an episodic prospective reading, or it can be quantified over by a Q-adverb, yielding a habitual prospective reading. In prospective readings, the operator that binds the event variable of the verb is a modal operator that contributes the meaning paraphrasable as was to + Infinitive. The role that the infinitive plays in this paraphrasis, in prospective readings is played by the event expressed by the verb plus its internal arguments and circumstantial modifiers (the material in the VP). Therefore, the complex predicate resulting expresses a telic event, which is intrinsically bounded at its end by the event expressed by the verb, which is part of the compositional meaning of the complex predicate. Because of this lexical semantics, the complex predicate selects an event variable, which can be existentially closed or quantified over by a Q-adverb. In this paper I have also suggested that in prospective and habitual readings we can find temporal/aspectual modifiers that modify the event expressed by the verb and not the event asserted in the sentence. These modifiers are associated with the Event Time of the event expressed by the verb, but do not contribute to determine the temporal reference of the sentence, since they do not contribute to determine the interval of truth of the sentence, which in prospective and habitual readings, which are stative, is included in the Event Time of the event asserted in the sentence. These temporal/aspectual modifiers behave as circumstantials, they contribute circumstances to the predication, and more precisely, they modify the Event Time of the event expressed by the verb as event-type, not as a token of the event-type. References Borràs, Z. (1999). «Presentava el treball la setmana de l’11, però l’hi han ajornat per motius burocràtics. El valor prospectiu del Present d’Indicatiu i el Pretèrit Imperfet d’Indicatiu». Universitat Autònoma de Barcelona, master thesis. Chomsky, N. (1998). «Minimalist Inquiries: the Framework». MIT, unpublished manuscript. Delfitto, D. (1998). «Aspect, Genericity and Bare Plurals». UiL OTS Working Papers. Delfitto, D. & Bertinetto, P.M. (1995). «A Case Study in the Interaction of Aspect and Actionality: the Imperfect in Italian». In Bertinetto, P.M.; Bianchi, V.; Higginbotham, J.; Squartini, M. (eds.). Temporal Reference, Aspect and Actionality. Vol 1: Semantic and Syntactic Perspectives. Utrecht: Rosenberg & Sellier, Universiteit Utrecht. Gennari, S. (1998). Sequence of tense: where semantics meets pragmatics. Brown University, PH.D. dissertation. Prospective Readings Revisited CatWPL 8, 2000 45 CatWPL 8 001-137 29/5/2001 17:27 Página 45 Klein, W. (1995). «A Time Relational Analysis of Russian Aspect». Language 71.4. Krifka, M. (1992). «Thematic relations as Links between Nominal Reference and Temporal Constitution». In Sag, I. & Szabolcsi, A. (1992). Lexical Matters. Stanford: CSLI. Matthews, P.H. (1981). Syntax. Cambridge: Cambridge University Press. Pascual, M. (1999). The ‘Instrumental Phrase’: is it an adjunct, an argument or a predicate? Universitat Autònoma de Barcelona, master thesis. Ramchand, G. (1996). «Two subject positions in Scottish Gaelic: the syntax-semantics interface». Natural Language Semantics 4: 165-191. Reichenbach, H. (1947). Elements of Symbolic Logic. Reprinted in Free Press [1966]. Rigau, G. (in press). «Els complements adjunts». Ms. Cap. 15. GCC. Speas, M. (1990). Phrase Structure in Natural Language. Dordrecht & Boston & London: Kluwer Academic Publishers. de Swart, H. (1998). «Position and meaning: time adverbials in context». In Bosch, P. & van der Sandt, R.R. (eds.). Focus and Natural Language Processing. Cambridge: Cambridge University Press. Verkuyl, H.J. (1989). «Aspectual Classes and Aspectual Composition». Linguistics and Philosophy 12: 39-94. Vogel, R. (1997). Aspects of Tense. Amsterdam: HIL, DISS, 25. 46 CatWPL 8, 2000 Zulema Borràs Barber CatWPL 8 001-137 29/5/2001 17:27 Página 46