The syntax and semantics of the perfect structure: a comparative approach

Размещено на http://www.allbest.ru/

New Sorbonne University

The syntax and semantics of the perfect structure: a comparative approach

Jacqueline Guйron

Paris, France



Abstract

The article covers some issues that concern the syntax and semantics of the present perfect construction in English and other languages. It states that all Present Perfects may be associated with what is considered the canonical construal in which an assertion is located at the present time but reports the existence of a past situation. However, some Present Perfects may, in addition, have a simple past aorist meaning. The author focuses on pluractional and evidential construals of the Present Perfect in some languages, argue that the Passй Composй construal and the Aorist construal of the perfect construction belong to two different modes of discourse, discours and rйcit, underlines their variability associated with a complex syntactic structure (an auxiliary verb and a verbal participle for Present Perfect, while the Aorist construal is associated with a simple verbal structure). According to the author, the Perfect has both syntactic and analytical ways of realization, thus the analytical syntactic structure of the Perfect is in English, French and German, with both a tensed auxiliary verb and a past partic iple, whereas it is synthetic in Latin, Russian and Arabic as is presented in the past participle alone; in languages with overt aspectual marking, aspect may vary on either the auxiliary, if it exists, or on the participle.

Key words: syntax, Present Perfect, Aorist, semantic meaning, time reference, pluractional construal, evidential construal.



Аннотация

СИНТАКСИС И СЕМАНТИКА СТРУКТУРЫ ПЕРФЕКТА: СРАВНИТЕЛЬНЫЙ АНАЛИЗ

Жаклин Герон

Университет Новая Сорбонна, г. Париж, Франция

Статья посвящена описанию синтаксических и семантических особенностей перфектной модели в разных языках. Показано, что любая форма Present Perfect может быть рассмотрена как утверждение в момент речи о событии, которое имело место в прошлом, однако в некоторых контекстах выявляются случаи передачи аористического значения - законченного действия, совершенного в прошлом. Автор указывает на возможность реализации в некоторых языках с помощью этой формы значений повторяющееся событие и событийность; доказывает, что формы Passй Composй и Aorist функционально относятся к двум разным типам дискурса. В исследовании установлено, что вариативность значений этой формы ассоциируется со сложноорганизованной синтаксической моделью (вспомогательный глагол и причастие основного глагола для Present Perfect, простая глагольная форма для Aorist). Перфект в разных языках может иметь аналитическую реализацию, когда вспомогательный глагол и причастия эксплицируют временную и аспек- туальную отнесенность (английский, французский, немецкий языки), синтетическую реализацию - в синтетической форме глагола (латынь, русский, арабский языки), в языках с открытым аспектуальным выражением - в форме вспомогательного глагола или причастия.

Ключевые слова: синтаксис, настоящий перфект, аорист, семантическое значение, временная отнесенность, повторяющееся событие, результативность.



Introduction

The perfect structure, in particular the present perfect structure, has been the object of much study for English and other languages (cf.: [Kiparsky, 2002; Klein, 1992; McCawley, 1981; McCoard, 1978; Michaelis, 1994; Mittwoch, 1988; Alexiadou, Rathert, Stechow, 2012; Portner, 2003; et al.]). Yet questions remain concerning every aspect of this structure. Some of them are listed below 1.

The acceptability of punctual temporal adverbs. In English punctual temporal adverbs are ungrammatical in the present perfect but acceptable in the past, future, and non-finite perfects.

a. I have seen John recently/*yesterday/*at 3 p.m.

I had seen John yesterday.

I will have seen John today at 3 p.m.

I had hoped to have seen John yesterday.

In French or (South-)German, however, such adverbs are acceptable in the present perfect as well.

a. Jean a parlй а Marie rйcemment/ hier/ а 3 heures.

Der Student hat gearbeitet (gestern).

Syntactic structure.

In English, French, and German, the perfect structure is analytical: it contains a finite auxiliary plus a non-finite past participle. This structure may have either the canonical Past-in-Present interpretation, as in English (1a) (which for simplicity we will call the PC, or Passй Composй, interpretation), or the simple past interpretation as in French (2a) or German (2b) with adverb hier or gestern.

In Latin, Russian, and Arabic, however, the present perfect is synthetic: there is no finite auxiliary, only a participle 2. Yet the structure is still ambiguous, as in French or German, between the PC and the simple past interpretation amavi (I loved/have loved) (Latin) a. Sobaka (uze) vyxodila. (Russian) dog (already) go_out.PAST.IPFV Imperfective Past OR Existential Present Perfect Sobaka vysla.

dog go_out.PAST.PFV

- Aorist Past OR Result Perfect.

jaraa (he ran/has run) (Standard Arabic) 4 run.PFV3.SGM

Aspectual marking. In languages with overt morphological aspect, the aspectual marking of the auxiliary and/or the participle may vary.

In Spanish or French the auxiliary may be IPVF or PFV, but in Portuguese it must be IPFV

The participle may vary for aspect, as in Russian (4a-b) or Bulgarian.

Construal. The most widespread construal of the perfect, which I will consider canonical, is the Existential construal in (6a-b), with its variants, the Resultative construal (6c) or the Hot News construal (6d). Under the Existential construal, as described in Iatridou, Anagnostopoulu, Izvorski [2001], the auxiliary verb defines a Perfect Time Span (PTS), a temporal interval which, starting from the Reference Time - the Utterance Time (UT) in the Present Perfect - stretches backwards to some defined or undefined point of time prior to the Reference or Utterance Time. The situation which the participial Verb Phrase (VP) describes is inserted on one or more points of the Perfect Time Span (PTS) located prior to the Reference or Utterance Time. The point of insertion may be remote from the UT, as in (6b), or so close to the Utterance time that one can speak of a result state pertaining to the UT, as in (6c) and (6d).

a. John has visited London once/ three times this year. (Exist. Perf.)

The earth has been hit by meteors before (and may be again) (Exist. Perf.)

I have broken my glasses. (Result Perf.)

The Orioles have won! (Hot News Perf.)

Iatridou, Anagnostopoulu, Izvorski [2001] propose that the UT sets the right boundary of the PTS in the present perfect while an adverb may set the left boundary. Note, however, that the right boundary is only a pragmatic boundary. The UT is never associated with the actual boundary of an eventuality, event or state. (7a) shows, as discussed in Giorgi, Pianesi [1997], that an event may not be bounded at the UT. Only a state, which lacks boundaries, may be predicated at the UT, as shown in (7b). In (7c), maintenant/ now coincides with the post-state of the event rather than with its end point.

a. *Je trouve la piиce maintenant.

(I find the coin now)

J'aime Jean maintenant.

(I love Jean now)

J'ai maintenant trouvй la piиce/*aimй Jean.

(I have now found the coin/*loved Jean.)

When the participle denotes a state rather than an event, a Universal construal is available which implies that the state is still ongoing at the UT. This construal requires not only a stative participle, but also an adverbial which marks the left boundary of the Perfect Time Span, as in (8b). Otherwise the sentence receives the unmarked Existential construal in which the situation the participle denotes does not continue at the UT, as in (8a).

a. John has been sick this week (but fortunately he is better now).

b. John has been sick since this morning (# but fortunately he is better now).

Pluractionality. The canonical present perfect denotes either a one-time- or an iterated event and accepts cardinal modifiers as in (6a). But in both European and Brazilian Portuguese, the present perfect, unlike the past or future perfect, and unlike the English or French present perfect, can only denote an iterated event (or a durative state). It is incompatible with cardinal modifiers (cf.: [Baptista, Guйron, 2014; Ilari, 2001; Laca, Cabredo-Hofherr, Carvalho, 2010])5. a. Ele tem saido com os amigos (varias vezes *uma vez). \

Lit: he has gone_out with the friends (several times/*one time) b. cf. Eng: He has gone out with friends (three times once).

Evidentiality. The perfect may function as an evidential expression. According to Izvorski [1997], in Bulgarian (and in, e.g., Norwegian), the present perfect structure is ambiguous between the canonical PC construal (10a) and an evidential construal (10b).

Az sam doshal.

be.1. SGPRES come.PART

I have come. (PC = Passй Composй)

I apparently came. (PE = Perfect of Evidentiality)

Under the PC construal (10a), the structure rejects a punctual past adverbial, just like English (1a) (cf. : (11a)). I take this incompatibility to be the defining diagnostic of a canonical PC. In the evidential construal ( 10b), however, a past punctual adverb is fine (cf. : (11b)). Note that the perfect structure is not ambiguous in the third person: the auxiliary is maintained in the PC construal (11a), but absent from the evidential construal (11b).

a. Te sa doshli (?? vchera).

they are come.PART yesterday (They have come (?? yesterday).) b. Te doshli vchera. they come.PE yesterday (They apparently came yesterday.)

Izvorski's [1997] account of this phenomenon is inspired by Iatridou's [2000] analysis of the past tense. Iatridou argues that in English and other languages, past tense can place the event vP denotes either in the past time of the current world, as in (12a), or in a counterfactual world, as in (12b)6.

a. John went to Paris yesterday/*tomorrow.

b. If John went to Paris tomorrow, he could visit Mary.

Iatridou [2000] proposes that past tense defines an EXCLUSION relation between either times or worlds. In (12a), the Topic Time (TT) (or Reference Time) excludes the Utterance Time (UT). In (12b), the Topic world (TW) (or Reference world) excludes the Utterance world (UW). This is schematized in (12'a-b).

(12') a. TT 3 UT (Topic Time excludes Utterance Time) b. TW 5 UW (Topic World excludes Utterance World)

Izvorski [1997] proposes that a temporal category may denote either tense or modality. The present perfect would assert, in its temporal PC construal, that the core eventuality does not hold at the Utterance Time, and in its modal Perfect of Evidentiality (PE) construal, that while the speaker has no direct evidence that the eventuality holds, she does have indirect evidence because traces of the past eventuality are visible at the UT. Izvorksi considers that in the latter use, the present perfect functions as an evidential epistemic modal.

However, a contrast which Izvorski cites, and even her discussion as represented above, shows that evidentiality cannot be classified as a modality, unless the term `modality' is enlarged to include not only situations which can or must be true at future time but those which overt signs suggest are already true at the utterance time. In Bulgarian, in the context (13), the epistemic modal in (13a) is acceptable, but the evidential perfect in (13b) is not.

Knowing how much John likes wine....

a. toj trjabva da e izpil vsichkoto vino vchera. (Modality) he must is drunk all-the wine yesterday

(He must have drunk all the wine yesterday.)

b. # toj izpil vsichkoto vino vchera. (Evidential Perfect)

he drunk.PE all-the wine yesterday (He apparently drank allthe wine yesterday.)

The construal of the Bulgarian perfect as an indirect evidential requires that the context supply some observable result of the core eventuality, which in the context of (13) would be say, empty bottles or someone's account of the event.

The contrast in (13) suggests that the term “evidential modality” is an oxymoron, like “freeze in hell” or “a living death”. In a Kratzerian framework, modal verbs like can or must introduce necessary or possible worlds in which a described situation holds which does not hold in the actual world at the utterance time (cf.: [Kratzer, 2012]). Evidentials, on the contrary, provide direct or indirect evidence that a described situation does hold in the actual world at the utterance time.

It is true that epistemic modality and evidentiality are both speaker-oriented and both suppose the ability to gain knowledge of the world through direct and indirect perception. In fact, humans share this ability with higher animals. However, while evidentiality is backward looking, focusing on what evidence suggests is already present in the discourse world, modality is forward looking: it focuses on situations which, albeit desirable and possible or even necessary, are missing from the discourse world. This particular form of intentionality, which starts from what is known about the discourse world and looks forward toward a possible or necessary addition to that world is exclusive to human beings 7.

Taking into account both Izvorski's [1997] description of the dual role of the perfect in Bulgarian and early work on the link between existential and possessive sentences by Lyons [1968] and others, I claim that the present perfect expresses direct evidentiality and that this construal is based on the existential content of its auxiliary verb.



The syntactic structure of the Perfect

The canonical PC sentence, represented by (1a) in English, contains two verbs each marked with a tense morpheme capable of positioning an event description on a time line [Gueron, 2004]. I assume that a finite tense morpheme (and in some languages a non-finite tense morpheme as well) merges a tense feature with an aspect feature either in the lexicon or in syntax. A grammar may also allow both lexical and syntactic merger of tense and aspect, as with the primary and secondary imperfective in Russian, for example.

In (1a), the present tense feature lexically merged with the imperfective aspect feature of English HAVE derives a Present Time Interval (PTI) which starts before and pragmatically ends at the Utterance Time8. The anterior tense - punctual aspect marker -ED - on the embedded verb places the situation the lexical vP describes at one point or multiple points within that present time interval. Since the PTI starts before the UT and stops at the UT and no part of an event can be placed on the UT itself, this point can only be prior to the UT.

If we assume that a verb has a single temporal-aspectual function, then a structure which involves two times must contain two verbs. It follows that the PC construal is universally associated with a structure containing both a finite auxiliary which denotes a time interval and a participle whose temporal-aspectual affix targets one or more points of that interval.

If the participle denotes an event, the event is over at the Utterance or Reference time even if it is non-telic. If the participle denotes a state, the state is also bounded in time in the unmarked case. But as a state has no internal temporal structure, it may persist throughout the PTI and receive a pragmatic temporal boundary which coincides with that of the UT, deriving the Universal Perfect construal. As we have seen in (7) above, only a state can be predicated of the UT. The boundedness of a state at a time previous to the UT is an implicature of the perfect which an appropriate time adverb can cancel, as in (15d).

a. John has written a letter/three letters. (Here it is/they are.)

John has run this morning. (Now he is resting.)

John has loved Mary (but no longer does).

John has loved Mary since he was a child (*but no longer does).

When, on the contrary, a perfect structure is construed as an aorist tense, then there must be only one verb and one tense-aspect marker in post-syntactic Logical Form (LF) even if the surface syntax is more complex. Under the aorist construal of French (2a) or German (2b), the participle must therefore have raised to the finite Tense node, where merger of the anterior tense feature of the participle with the person and number agreement features of the auxiliary verb allows it to be construed as a simple past tense. The single structure in French (16) is thus associated on the level of LF with two syntactic structures, (17a) and (17b), in English. In French, the participle is generated low in the PC structure and raises to the main tense node under the aorist construal. As English verbs don't raise, the participle is first-merged either in the lower part of the structure where its -ED affix is construed as a punctual aspectual marker or in the higher part, where the same affix is construed as a past tense.

Jean a marchй toute la journйe.

a. John has walked all day. b. John walked all day.

I propose, therefore, that in Russian, Arabic or Latin, the PC construal is based on a syntactic structure containing two verbs, the higher of which is a null copula. Note that these languages have independently been analyzed as containing null copula BE in present tense predications like “Masa krasivaya” for “Masha is beautiful”. Under the PC construal, the null copula defines the present time interval, while the core event is inserted by the aspectual marker of the participle at a temporal point prior to the utterance time. Since there is no overt auxiliary, there is no obstacle to raising the participle to the finite Tense position either in syntax or in LF.

The ambiguity PC/aorist tense is illustrated in (18) in Arabic and Russian. The PC interpretation is read off the full structure with a null auxiliary. The aorist interpretation is read off a structure in which the participle raises from T2 to T1, either in LF as in French or German, in syntax as in Latin, Arabic or Russian, or by first-merge as in English. This syntactic analysis is straightforward given by a grammar with a null copula, verb raising in syntax or LF, and perhaps crucially, some form of feature agreement between the finite verb and the participle 9.

We may ask what allows a morpheme to function as an aspect in T2 and as a Tense in T1. That is like asking what allows a pronoun to be construed as deictic in TP1 and anaphoric in TP2. As discussed in Partee [1973], Tense, like pronouns, includes a [+/- referential] feature. This feature allows a non-referential (non-finite) tense to be calculated with respect to a structurally higher [+/- referential] tense10. Tense also has, or acquires in syntax, an aspectual feature which dictates its punctual or extended temporal dimension. Similarly, a pronoun includes a [+/- referential] feature which allows a non-referential argument to be identified with a structurally higher referential argument. In lieu of aspect, the gender and number features of a pronoun dictate certain classificatory properties.

We consider that Aspect is not a viewpoint on a situation as for Comrie [1976] or Smith [1997]. Nor is it a syntactico-semantic structure, as in many semantic accounts of the perfect. It is a functional morpheme which predicates a verb, and the entire event description headed by the verb, with a point or interval within the reference time line. In Latin, Arabic, Russian or English, the same morpheme has a different function at a different syntactic level:

as a temporal affix in TP1, it defines a temporal point or interval within the deictic discourse time line defined by an auxiliary verb;

as a participial affix in TP2 it predicates the situation that the lexical vP describes of that temporal point or interval.

Again, since the Utterance Time cannot coincide with the temporal boundary of an event, a participle raised from T2 to T1 projects an event on a boundary point within the discourse time line prior to the utterance time, necessarily construed as a point of past time.

A residual problem is how a past participle can raise from T2 to T1, defining an aorist tense, when, as in French or German, T1 already contains an overt auxiliary. I assume that in such cases, either the auxiliary is reduced to its agreement features of person and number, the participle already bearing a tense feature, or that the existential lexical content of the auxiliary has been raised to a higher node with a speaker-oriented function, construed as Evidential, as is plausibly the case in Bulgarian.

I reject all analyses which account for variation among languages in the structure and construal of the perfect by invoking selectional properties of present tense, as in Pancheva & von Stechow's [2004] account of the difference between the German and English perfects, or Schmitt's [2001] analysis of the difference between the Portuguese and English perfects. I assume, with Partee [1973], that Tense is a referential element, like a pronoun. Unlike a predicate, a referential element lacks the propositional content which licenses the selection of complements.

The grammatical function of the canonical present perfect

Why should a grammar which already contains a present tense and a past tense want to combine the two in a single structure? While the past perfect and the future perfect are necessary for temporal interpretation in that they order two events, as illustrated in (19a) and (19b), the present perfect does not order two events. It denotes, rather, the inclusion of a situation occurring at some point in the past within the larger present time interval. So while events occurring in the past and future perfect can be represented on one temporal line as in (19a-b), the present perfect needs a more complex representation as in (19c).

a. John left (E2) after Mary had entered (E1). ST1 ET2 ST

John will have left (E1) by the time you arrive (E2). ST ET1 ET2

John has left. PTI ST

ET

I suggest that the present perfect structure belongs to a mode of discourse distinct from that of the other perfects. Benveniste [1966] described two modes of discourse, Rйcit and Discours. Discours is speaker-centered and contains the deictic elements I, you, now and here. Rйcit is pure narration which does not contain the set of speaker-centered deictic elements. More precisely, even if a Rйcit does contain an I, as in, say, Gide's novels written in the passй simple, the I is a protagonist like any other; it is not associated with a you, a now, and a here.

The same tenses are construed differently in Rйcit and Discours. For example, present tense is unavailable with events in English in Discours in (20a), but it is fine in Rйcit in (20b-c).

a. *John walks into the house now.

John walks into the house. He sees Mary. He sits down on a chair.

Yesterday this guy comes up to me. He asks me where the Town Hall is. I tell him I don't know...

The imperfective tense in French can have a progressive or a habitual meaning in both Discours and Rйcit, but in Rйcit, it can also be construed as an aorist, as in (21b). Here, the tense and aspect features of the verb function independently. The imperfective aspectual feature of the verb projects a past time interval onto the narrative time line, while it's anterior tense feature that places the punctual event description onto one past point of that time line. In other words, the past imperfective functions, with a single verb, in a Rйcit the way the present perfect functions, with two independent verbs, in Discours: an aspectual feature defines a time span while a tense feature places an event description onto a point of time.

a. Jean mourait.

(Jean was dying.) (Progressive)

b. Dix jours aprиs, Jean mourait.

(Ten days later, Jean died.) (Aorist)

I propose that while past and future perfects belong to Rйcit, the present perfect is a deictic speaker-centered tense which belongs to Discours and whose semantic function is evidential. If so, then the Bulgarian present perfect is ambiguous not between a tense and a modal but rather between a direct and an indirect evidential. The present perfect in (10a) is a direct evidential: its existential auxiliary implies direct perceptual evidence of the truth of the underlying proposition p. The Perfect of Evidentiality (PE) in (10b) is an indirect evidential implying indirect evidence of the truth of p perceived in the post-state of a past event.

As for its syntax, which Izvorski [1997] does not discuss, I propose that the Bulgarian PE, which has an aorist reading in (10b), undergoes the same movement in syntax or LF as the aorist perfect in other languages, that is, raising of the verb from T2 to T1 in a complex syntactic structure. However, unlike the French or German structures, in which the copula, reduced to its person and number features, plays no further role in the aorist output, the Bulgarian copula continues to play a semantic role. Presumably it raises to a higher syntactic node in the speaker's CP domain, comparable to the node bearing evidential affixes in other languages, in which its existential content is construed as an evidential operator taking scope over the entire retrospective time span leading from present to past.

Just as evidentials are not limited to languages with evidential affixes, they are not limited to perfect structures. The French sentences of (22), discussed in Kayne [1975], show clear properties of direct evidentials: obligatory imperfective aspect or present tense and the presence of a perception verb or an explicit existential morpheme.

a. J'ai vu Marie qui sortait du cinйma.

(I saw Marie who was coming out of the movie theater.) b. Voilа Marie qui pleure comme une Madeleine.

(Here/there is Mary, crying like a Madeleine.)

The same perception verbs function as indirect evidentials when they take a propositional complement and as direct evidentials with a simple vP or small clause complement.

(direct perception) (indirect evidence) (direct perception) (indirect evidence)

a. I heard Mary sing.

b. I heard that Mary sang.

a. John seems sick.

b. John seems to be sick.

That existential evidentials are not modals in the grammatical system is shown by the fact that existential terms can introduce simple NPs or Small Clauses as well as propositional content, but modals always introduce propositional content.

a. Look! There is a man at the door.

I see John sleeping.

Voilа Marie.

a. *A man must at the door.

*John can sleeping.

*Necessaire Marie.

Evidentiality calls upon the speaker's ability to interpret sense perceptions as evidence of the existence of an entity or an eventuality. This property is shared by human beings and higher animals. Modality presupposes this ability and contributes, in addition, consciousness of the absence of a situation which does not currently exist in the discourse world yet is nevertheless both desirable and possible or even necessary. The consciousness of an absence and the intentionality which it implies to somehow contribute to the realization of the absent situation at a future time are exclusively human properties which (so far) have not been attributed to higher animals. syntactic semantic perfect verb

Existential terms as evidentials. The auxiliaries found over languages in the present perfect, have and be, function in those same languages in existential and possessive structures. Moreover, they maintain in their auxiliary use lexical and aspectual properties of their existential use.

Lyons [1968] argued that possessive structures have the properties of existential structures so that the two are substituable, as in (27).

a. There is a book in John `s hand. b. John has a book.

Benveniste [1966] had already assimilated the present perfect with HAVE in (28a) to the simple possessive structure in (28b) in English and likewise for perfect and possessive structures with BE+Locative morpheme in Armenian (29).

a. John has written a book. (Perfect) (English)

b. John has a book. (Possessive)

a. nora teseal e (Perfect) (Armenian)

(his is seen = he has seen)

b. nora tun e (Possessive)

(his is house = he has a house)

Freeze [1992] claimed that HAVE is derived from BE + a LOC(ative) morpheme in syntax. Kayne [2000] and others developed this idea further11. However, the different aspectual content of BE and HAVE, as shown for English in (30), seems sufficient to account for their construals in various syntactic contexts, without positing a syntactic transformation deriving the verb HAVE from an expression containing the verb BE.

a. John has a child. b. John is a child.

English BE is a punctual auxiliary which links two terms in a relation of partial or full identity. It requires the presence of both a locative morpheme and a locative PP in order to create a Figure-Ground existential construal in English or French. The same construal requires only a contextual PP when, as with Russian est ' or Spanish estar, the finite verb is already lexically endowed with a locative feature.

English HAVE, on the contrary, not only contains a locative feature but is in addition temporally imperfective. In a transitive episodic sentence, its locative content creates an existential Figure-Ground configuration, while its temporal content selects an animate subject whose biography coincides with the present time span, as shown by the contrast in (31).

a. John has a book. (possessive)

b. *The table has a book12. (possessive)

In (32a), existential THERE plus a locative phrase entails the direct perception by the speaker of a Figure-Ground configuration at a specific location at the reference time. In (32b), where the temporal content of HAVE selects an Experiencer subject, the speaker's evidence comes indirectly, from her personal acquaintance with the subject John at the UT. That is why the subject must be alive. A dead man can give no evidence of the truth of p at a point in time included in the PTI.

a. There is a book on the table.

John has read this book.

# Einstein has read this book13.

A transformation merging BE plus LOC to create HAVE cannot account for the temporal content which this verb manifests both in a possessive structure in vP and in a present perfect structure in TP14.

On the impossibility of a universal perfect reading of events

Iatridou, Anagnostopoulu, Izvorski [2001] claim that a PC denoting an event can have a universal reading if the participle has IMPERFECTIVE or NEUTRAL (in the sense of: [Smith, 1997; 2007]) aspect. (33) is a Bulgarian example.

Tja e pilavinoto [ot sutrinta nasam].

she is drunk.NEUT wine-the [from this morning towards now]

(She has been drinking the wine [since this morning].)

Note, however, that Russian (4a) above also has a PC with IPFV /-participle (and null auxiliary) but it derives an Existential reading, not a Universal one, so the claim does not make the correct prediction for Russian.

More importantly, the Bulgarian speaker I consulted15 does not accept (33). The reason is that the PC focuses on the result of a past event (cf.: [Tilkov, Stoyanov, Popov, 1983]), while the adverbial ot sutrinta nasam (that means “from morning towards now”) does not mark the endpoint of the event at the speaker's now. It indicates, rather, an action intended by the subject to last until the speaker's now. There is a confusion between the subject's and the speaker's temporal points of view. That is, the future time at which the subject intended to stop drinking not yet being realized, it cannot be the same moment as the speaker's deictic “now”.

Whenever the right boundary of an event is explicitly provided, as with the adverbial do sega (“until now” in Bulgarian), that boundary, as I have been emphasizing, cannot coincide with the utterance time, but must be placed before the UT time. This is the defining property of the Existential perfect.

To show this, take any universal expression as licensed by an adverbial setting a left boundary (LB) on the present time span, whether it be in the simple progressive present as in French (34a) or the progressive perfect in English (34b). Then, instead of an adverb which specifies solely the Left Boundary of the PTI, select an expression which specifies the coincidence of the Right Branch and the UT, as in (35).

a. Je travaille depuis ce matin (et je continue).

b. I have been working since 8 a.m. (and I am still working).

a. *Je travaille (depuis ce matin jusqu'а maintenant). b. I have been working (from 8 a.m. until now).

The addition of an explicit right boundary makes the French sentence contradictory, since the imperfective present tense describes an event as ongoing, while a temporal boundary places the event in the past. In the English example, the right boundary triggers an Existential rather than a Universal reading. It is this reading that I associate with (33) in Bulgarian. As for now, it means “now at the point of UT” with states, not with events. (36a) refers to the subject's state of loving at the point of UT. (36b) refers to the post-state of an inchoative change of state (like turning on the lights in the room), equivalent to (37b) in English. (37a) is ruled out by the impossibility of predicating an event of the UT 16.

a. Je t'aime maintenant. b. Je te vois maintenant.

a. *I see you now. (Unless performative)

b. I can see you now. I have turned the lights on.

5. Conclusion

propose that the present perfect, like the episodic simple present, is a deictic tense which belongs to Discours, not Rйcit. The deictic effect is triggered by the existential spatial content and imperfective temporal content of HAVE in English or by the temporally imperfective / spatially locative construal of some form of “augmented” BE in other languages. The episodic present tense has an evidential function. It implies that the speaker has direct or indirect evidence for the truth of the existence of p at the present time. Since this implication is licensed by the existential content of the auxiliary verb, it does not exist in a perfect with an aorist construal. Under the aorist construal, I proposed that the auxiliary has been reduced to its agreement features in say, French, or else raised to a speaker-oriented higher syntactic node, as, plausibly, in Bulgarian.

When the sentence lacks a subject, as in “It has been raining”, the discourse context provides direct evidence for the assertion based on the speaker's perception. If the sentence has a subject, the assertion is based on indirect evidence represented by contact between the speaker and the subject at some point during the present time interval (PTI).

The referential subject of a perfect structure has two thematic functions. It is construed both as the Agent of the event described in TP2 and as the Experiencer of the evidential state described in TP1. In order for the same argument to function both as Agent of a past event and as Experiencer of a present information state, the subject must have a biography which spans the entire PTI. The speaker already has a biography, implied by her ability to place events in the past as well as in the present. So the speaker can learn what happened in the past either from evidence visible at present (“It has rained”) or via direct or indirect contact with the subject during the present time interval (“John has won the game”).

NOTES

11 am grateful to Svetlana Vogeleer and Vassil Mostrov for discussion of participles in Russian and Bulgarian, respectively.

The term participle refers here to the Old Russian participle ending in -l, used in Modern Russian both as a finite past and with a perfect meaning, and not to the Modern Russian participle derived from a pe rfective infinitival with a short and a long (adjectival) form.

The abbreviations used in glosses are the following: PFV `perfective', IPFV `imperfective', PST `past tense', PRES `present', PART `participle', M/F `masculine/feminine', 1, 2,... `1st, 2nd,... person', SG `singular', PL `plural'.

Cf.: [Fassi Fehri, 2012].

Baptista & Guйron [2014] attribute this property of the Portuguese perfect to the lexical aspect of the auxiliary verb tener

Generative Grammar distinguishes vP, which contains an overt subject argument, from the syntactically lower VP, which contains only the lexical verb and its internal argument(s). Since an embedded past participle lacks a subject, we analyze it as a TP which directly embeds a VP with no intervening vP. When the participial verb lexically selects a subject as in (1a), “I have seen John”, I propose, following Hoekstra [1984], that the overt subject of HAVE binds/identifies the covert selected subject of see.

The diachronic evolution of the English verb need, Italian bisogna, or French falloir or manquer illustrates the conceptual link between the absence of a desired physical object in the spatio-temporal domain ( il me manque dix euros / I need ten euros), the absence of a desired human being in the psychological domain ( tu me manques / I need you), and the absence of a desired situation in the temporal domain ( il a manquй de tomber / I need to leave now).

The Present Time Interval (PTI) is more general than the Perfect Time Span (PTS) of Iatridou, Anagnostopoulu, Izvorski [2001], which it subsumes. English Present tense HAVE is imperfective in all its uses. Sentence (i) describes a Figure-Ground configuration in which the subject/Ground is distinct from and spatially and temporally more extended than the object / Figure. BE, on the contrary, is punctual (in English). While HAVE includes one spatio-temporal entity in another in vP and one time inside another in TP, BE (unless merged with a locative feature) creates a dimensionless mental representation in which the F igure cannot be dissoc iated from the Ground, as shown in (iii) vs (iv).

John has a book.

John has seen Mary.

John is a doctor.

*John is a book.

See Chomsky [1995] for more details concerning the syntactic structure of the sentence, raising operations and formal feature agreement between constituents.

On the composition of the sentential “Tense-Chain”, see Guйron& Hoekstra [1988].

These authors apply to English and other languages a transformational version of Benveniste's own cross- linguistic analysis: “Or, avoir n'est rien d'autre qu'un кtre а inversй: mihi est pecunia se retourne en habeo pecuniam.” (`Have is nothing more than an inverted be+to: inverted (Latin) to me is money becomes I have money'.) [Benveniste, 1966, p. 197].

Possessive HAVE may not take an inanimate subject, as shown in (ii) vs (i). Note that (iii), with inanimate subject does not qualify as an assertion of episodic possession. Rather, it is construed as an individual-level part- whole predication.

John has a new book. (possessive)

*Philadelphia has a new bench. (*possessive)

This park has many benches. (part-whole predication)

A proposition situated in a mental world has access to metonymy and is not subject to the temporal constraints which make (32c) ungrammatical in the spatio-temporal world, as shown by the contrast between (32c) and (i), or between (ii-a) and (ii-b) below:

(i) Einstein has influenced my physics teacher. (causal reasoning)

e. the works of Einstein have influenced the research/teaching of my physics teacher.

ii) a. *The war gave a book to Mailer. (spatio-temporal event at a point in time) b. The war gave Mailer a book. (causal reasoning over a time span)

See Baptista & Guйron [2014] for more discussion of the differences between the auxiliaries HAVE and BE.

Vassil Mostrov.

Note that my Bulgarian informant, who rejects (33), would accept a resultative perfect like (i):

Tia e pila mnogo ot sutrinta do sega (I sega e piana).

she has drunk a lot since this morning until now (and now she is drunk)

For me, (ii) and even better (iii) are acceptable in English: the insertion of a quantity expression places the whole sentence in the point of view of the speaker, eliminating the contradictory points of view in (33).

She has drunk a lot between this morning and now (and now she is drunk).

She has drunk so much between this morning and now that now she is drunk.



References

1. Baptista M., Guйron J., 2014. Sobre tempo e aspecto em Portuguкs: comparando a PB ao PE. MouraD., Sibaldo M.A., eds. Sintaxe comparativa entre o Portuguкs brasileiro e lmguas crioulas de base lexical portuguesa. Vol. III, Cap. 8. Maceio, Edufal.

2. Benveniste E., 1966. Problиmes de linguistique Gйnйrale. T. 1. Paris, Gallimard. 286 p. Chomsky N., 1995. The Minimalist Program.

3. Cambridge, MA, MIT Press. 420 p.

4. Comrie B., 1976. Tense. Cambridge, Cambridge University Press. 142 p.

5. Fassi Fehri A., 2012. Arabic Perfect and Temporal Adverbs. Alexiadou A., Rathert M., Stechow A. von, eds. Perfect Explorations. Berlin, De Gruyter, pp. 69-99

6. Freeze R., 1992. Existentials and Other Locatives. Language, vol. 68, iss. 3, pp. 553-595.

7. Giorgi A., Pianesi F., 1997. Tense and Aspect. From Semantics to Morphosyntax. New York, Oxford, Oxford University Press. 319 p.

8. Guйron J., 2004. Tense Interpretation and the Argument Structure of Auxiliaries. Guйron J., Lecarme J., eds. The Syntax of Time. Cambridge, MA, MIT Press, pp. 299-328.

9. Guйron J., Hoekstra T., 1988. T-Chains and the Constituent Structure of Auxiliaries. Cardinaletti A., Cinque G, Giusti G., eds. Proceedings of the GLOW Conference in Venice 1987. Dordrecht, Foris, pp. 35-99.

10. Hoekstra T., 1984. Transitivity: Grammatical Relations in Government-Binding Theory. Dordrecht, Foris. 311 p.

11. Iatridou S., 2000. The Grammatical Ingredients of Counterfactuality. Linguistic Inquiry, vol. 31, no. 2, pp. 231-270.

12. Iatridou S., Anagnostopoulu E., Izvorski R., 2001. Some Observations About the Form and Meaning of the Perfect. Kenstowicz M., ed. Ken Hale: A Life in Language. Cambridge, MA, MIT Press, pp. 189-238.

13. Ilari R., 2001. Notas sobre o passado composto em portugues. RevistaLetras, num. 55, pp. 129-152.

14. Izvorski R., 1997. The Present Perfect as an Epistemic Modal. Lawson A., Choe E., eds. SALT 7, pp. 133147. URL: https://pdfs.semanticscholar.org/ d4e5/dac02b2361c186ede4bf89fbd1dca 6019a48.pdf.

15. Kayne R., 1975. French Syntax: The Transformational Cycle. Cambridge, MA, MIT Press. 473 p.

16. Kayne R., 2000. Toward a Modular Theory of Auxiliary Selection. Parameters and Universals. Oxford, Oxford University Press, pp. 107-130

17. Kiparsky P., 2002. The Polysemy of the Perfect. Beaver D., Casillas Martinez L., Clarck B., Kaufmann S., eds. The Construction of Meaning. Stanford, CSLI Publications, pp. 113-135.

18. Klein W., 1992. The Present Perfect Puzzle. Language, vol. 68, pp. 525-552.

19. Kratzer A., 2012. Modality and Conditionals. Oxford, Oxford University Press. 203 p.

20. Laca B., Cabredo-Hofherr P, Carvalho S. de, 2010. When Perfect Means Plural: The Present Perfect in Northeastern Brazilian Portuguese Varieties. Cabredo-Hofherr P., Laca B., eds. Layersof Aspect. Stanford, CSLI Publications, pp. 67-100.

21. Lyons J., 1968. Introduction to Theoretical Linguistics. Cambridge, Cambridge University Press. 519 p.

22. McCawley J., 1981. Notes on the English Perfect. Australian Journal of Linguistics, vol. 1, pp. 81-90.

23. McCoard R.W., 1978. The English Perfect: Tense Choice and Pragmatic Inferences. Amsterdam, North Holland Press. 279 p.

24. Michaelis L., 1994. The Ambiguity of the English Present Perfect. Journal of Linguistics, vol. 30, pp. 111-157.

25. Mittwoch A., 1988. Aspects of English Aspect: On the Interaction of Perfect, Progressive and Durational Phrases. Linguistics and Philosophy, vol. 11, pp. 203-225.

26. Pancheva R., Stechow A. von, 2004. On the Present Perfect Puzzle. Moulton K., Wolf M., eds. NELS34, pp. 469-484.

27. Partee B., 1973. Some Structural Analogies Between Pronouns and Tenses in English. Journal of Philosophy, no. 70, pp. 601-609.

28. Alexiadou A., Rathert M., Stechow A. von, eds., 2012. Perfect Explorations. Berlin, De Gruyter. 442 p.

29. Portner P., 2003. The Temporal Semantics and Modal Pragmatics of the Perfect. Linguistics and Philosophy, no. 26, pp. 459-510.

30. Schmitt C., 2001. Cross-Linguistic Variation and the Present Perfect: The Case of Portuguese. Natural Language and Linguistic Theory, no. 19, pp. 403-453.

31. Smith C., 1997. The Parameter of Aspect. Dordrecht, Kluwer. 469 p.

32. Smith C., 2007. Tense and Temporal Interpretation. Lingua, vol. 117, iss. 2, pp. 419-436.

33. Tilkov D., Stoyanov S., Popov K., 1983. Gramatika na suvremennia bulgarski knizoven ezik. Sofia, BAN. 511 p.

Размещено на Allbest.ru

...