Os verbos durativos do português

Telmo Móia*

*CLUL | Faculdade De Letras da Universidade de Lisboa, Portugal.

tmoia@letras.ulisboa.pt

ABSTRACT

In this paper, I will discuss the expression of duration and temporal location via predicate-argument combinations in Portuguese. The focus will be on what we can term “durative verbs”– like the counterparts of English last, spend and take (corresponding in Portuguese to at least six extremely common verbs: durar, prolongar-se, arrastar-se, passar, levar and demorar) –, which exhibit a few intriguing grammatical idiosyncrasies. Corpora data from European and Brazilian Portuguese will be used, in an attempt to show that the constructions under analysis reveal a considerable amount of variation and are prone to (linguistic change-evincive) anomaly, having to do mainly with the use of temporal prepositions or preposition-like connectives. Formal analyses – to be made within the Discourse Representation Theory framework – will also be presented. These are meant to underline that the semantic contribution of homonymous elements may be variable across different construction patterns, which undoubtedly correlates with their somewhat unstable grammatical behaviour in contemporary Portuguese.

Keywords: duration, temporal location, verbs, prepositions, Aktionsart, telicity.

RESUMO

Neste trabalho, analisarei a expressão de valores de duração e de localização temporal através de combinações predicado-argumento em português. A ênfase será colocada no que podemos denominar “verbos durativos” – como os homólogos dos predicados ingleses last, spend e take (correspondendo em português a pelo menos seis verbos extremamente comuns: durar, prolongar-se, arrastar-se, passar, levar e demorar) –, os quais revelam interessantes idiossincrasias gramaticais. Serão usados dados de corpora portugueses e brasileiros, para mostrar que as construções em análise estão sujeitas a forte variação e tendem a ocorrer em padrões desviantes, indicativos de mudança linguística, os quais se relacionam principalmente com o uso de preposições temporais ou conectores afins. Serão apresentadas análises formais, no quadro da Teoria da Representação do Discurso, as quais se destinam a evidenciar que o contributo semântico de elementos homónimos pode diferir de construção para construção, um facto que se correlaciona com o comportamento gramatical relativamente instável que estes operadores exibem no português contemporâneo.

Palavras-chave: duração, localização temporal, verbos, preposições, Aktionsart, telicidade.

1. Durative verbs – overall view

The expression of temporal location and duration via predicate-argument sequences, in Portuguese, makes use of different semantic subcategories of predicates. The distinction between durative and non-durative predicates is preeminent (Móia, 2011b: 256). This study focuses on the former – and in particular, on their verbal members -, which are distinguished by their ability to express both strict duration and durative temporal location (idem: 258)^[1].

1.1. Semantic relations expressed by durative verbs

All durative verbs – like Portuguese durar (‘last') – can assert strict duration and durative temporal location. Strict duration applies when they take eventuality-denoting expressions (ev) and predicates of amounts of time (mt), like mais de duas horas (‘more than two hours'), as arguments: e.g. [verb (ev, mt)]. Two cases need be distinguished:

(i) the amount of time is simply associated with the described eventuality, which is assumed to have occurred somewhere in the past, or to occur somewhere in the future, as in (1); in this case, non-anchored duration operates, with [verb (ev, mt)] tantamount to the formal DRS-condition [dur (ev) = mt].

(1) O bombardeamento de Guernica durou mais de duas horas.

the bombing of Guernica lasted more than two hours

‘The bombing of Guernica lasted more than two hours.'

(ii) the amount of time is associated with the part of the described eventuality that goes up to a given anchor point (typically, the temporal perspective point of the sentence, TPpt); in other words, sentences convey the duration that a given (atelic) eventuality has reached at a given point in time (where it still holds)^[2], as in (2); in this case, anchored duration operates, with [verb (ev, mt)] tantamount to the formal DRS-condition [dur (ev´) = mt], where [ev´ ⊇ ev] Λ [beg (ev´) = beg (ev)] Λ [end (ev´) = TPpt], and [ev ◯ TPpt]. Temporal location information is inferred from the association with TPpt, viz. that the eventuality occurs throughout the period of x-time duration abutting TPpt: [t ⊇ ev/ev´], where [end (t) = TPpt] Λ [dur (t) = mt].

(2) O bombardeamento da cidade já dura há mais de duas horas.

the bombing of-the city already lasts there-is more than two hours

‘The bombing of the city has been going on for more than two hours (now).'

On the distinction between anchored and non-anchored duration, and its relevance in Portuguese, see Móia (2006, 2011b).

Durative temporal location applies when verbs take time-denoting expressions (t), like toda a primeira noite da Operação (‘the whole first night of the Operation'), rather than predicates of amounts of time, as internal arguments: e.g. [verb (ev, t)]. They prompt the reading where the eventuality holds throughout the entire interval, i.e. [t ⊇ ev], or – in the more restrictive case of exact location – [t = loc (ev)] (triggering [dur (ev) ≥ dur (t)] and [dur (ev) = dur (t)], respectively, as inferences). This happens in the following example:

(3) O bombardeamento da cidade durou toda a primeira noite da Operação.

the bombing of-the city lasted all the first night of the Operation

‘The bombing of the city lasted the whole first night of the Operation.'

Crucially, as far as location is concerned, durative verbs can only express the durative type. If other types of temporal location – not requiring the location time to be covered in its entirety – are to be expressed, different – non-durative – predicates are used, as ocorrer (‘occur'), in sentence (4), where [verb (ev, t)] is tantamount to the simple overlapping relation [ev ◯ t].

the bombing of-the city occurred during the first night of the Operation

‘The bombing of the city occurred during the first night of the Operation.'

1.2. The set of Portuguese durative verbs

The facts described in the previous section are spelled out in Móia (2006, 2011b). I do not explore in full detail, however, the idiosyncrasies of the various durative verbs of Portuguese, rather focusing (mainly) on durar. I will attempt a relatively more encompassing view here.

Six durative verbs stand out as particularly common in Portuguese: durar, the counterpart of English last or go on for (though sometimes also of take, as noted in Móia 2011b: 259, fn.8, and shown later in example (41)) – cf. (1)-(3); prolongar-se and arrastar-se, (literally, prolong and drag plus an intrinsic unanalysable clitic, se), bothsemantically quite similar to durar, but with some intriguing syntactic-semantic idiosyncrasies (not yet described in the literature, to my knowledge) – cf. (5)-(6); passar, the counterpart of English spend – cf. (7); levar, the counterpart of English take – cf. (8); levar, the counterpart of English spend or go on for, in contexts of anchored duration (analysed in Móia 2011b: 261) – cf. (9) (for the sake of perspicuity, I will sometimes distinguish this form of levar with the subscript “AD”, standing for anchored duration – i.e. levar_AD –, though I do not necessarily claim that this is an altogether different predicate); demorar, semantically quite similar to levar in (8), i.e. a counterpart of English take (with small differences to be described later on) – cf. (10).

(5) O bombardeamento da cidade prolongou-se por mais de duas horas.

the bombing of-the city prolonged SE_{intrinsic clitic} for more than two hours

(6) O julgamento arrastou-se por mais de uma década.

the trial dragged SE_{intrinsic clitic} for more than a decade

‘The trial lasted for more than a decade.'

(7) O presidente passou duas horas a falar sobre direitos humanos.

the president spent two hours at talk_INF about human rights

(8) O funcionário levou dois dias a concluir o relatório.

the officer took two days to complete the report

‘It took the officer two days to complete the report.'

(9) As negociações já levam dois meses.

the negotiations already take two months

(10) O restauro da igreja, muito meticuloso, demorou quase seis anos.

the restoration of-the church very meticulous took almost six years

‘The restoration of the church, which was very meticulous, took almost six years.'

I will focus on just these six verbs henceforth. Other (less frequent) durative predicates, whose specificities will not be studied here are, for instance, perdurar, estender-se or tardar, as in the following examples from the corpus cetempúblico (with glosses provided just for the italicised part, as in any other corpora excerpts hereafter):

(11) «Uma crise cujas implicações económicas (...) perdurarãosem dúvida durante anos.» (ext67040-clt-97a-2); «(...) uma injustiça que perdura há 11 anos!» (ext713809-nd-91b-2); «A confusão perdurou (...).» (ext542218-clt-97a-2)

(12) «(...) os seus militantes estiveram na primeira linha dos confrontos (...) com as forças policiais, que se estenderam por15 horas (...).» (ext255110-pol-93b-1)

[confrontations] SE_{intrinsic clitic} extendend for 15 hours

(13) «A invasão desencadeada por Hanói (...) não tardou duas semanasa chegar a Phnom Penh (...).» (ext949272-pol-95a-1)

[invasion] not took two weeks to arrive to Phnom Penh

(14) Os trabalhos, que terão uma duração de cerca de um ano, vão obrigar ao desvio do trânsito (...). (ext24557-soc-97a-1); «(...) cada utilizador de telefones celulares realiza (...) uma média diária de três chamadas (com uma duração de 3,7 minutos) (...).» (ext68302-eco-98a-3)

[(road) works] will-have a duration of nearly of one year; [(telephone) calls] with a duration of 3,7 minutes

(15) «Depois de negociações prolongadas, o terreno acabou por ser comprado pela Câmara Municipal de Oeiras (...).» (ext475234-eco-93b-1)

[negotiations] prolonged

The verbs put on center in this paper differ in various syntactic and semantic properties. Three are particularly noteworthy: (i) argument structure and subcategorization properties – cf. section 1.3; (ii) sensitivity to telicity restrictions – cf. section 1.5; (iii) compatibility with the semantic relations listed in 1.1, and the form of their temporal complements in each case – cf. sections 2-4.

- durar, prolongar-se, arrastar-se

Verbs durar, prolongar-se and arrastar-se are binary predicates. They relate a non-sentential (typically an NP) external argument, representing an eventuality (ev), with a (temporal) internal argument, representing either an amount of time (mt) or a time interval (t): [verb (ev, mt/t)]. Examples are given in (1)-(3) and (5)-(6) above, the first repeated below, as (16). English last has a similar syntactic behaviour.

(16) O bombardeamento de Guernica durou mais de duas horas.

the bombing of Guernica lasted more than two hours

‘The bombing of Guernica lasted more than two hours.'

Only cases with eventuality-denoting external arguments will be considered here, though external arguments representing common objects or individuals are also possible, with durar (as, for that matter, with last) – e.g. as pilhas duraram mais de dez horas (‘the batteries lasted over ten hours').

Differences concerning the absence/presence of prepositions (or preposition-like connectives) in the internal argument, when it takes the form of a predicate of amounts of time, will be described in sections 2-3. As we will see, these differences separate durar, on one side,from prolongar-se and arrastar-se, on the other.

(17) «Se calhar, se vivêssemos juntos a nossa relação não duraria.» (ext1372256-clt-92b-2); «Se a situação de guerra se prolongarestão preparados para todos os cenários (...).» (ext1143451-nd-98a-1); «O conflito, no entanto, arrastou-se, acabando por ser constituída uma «troika» (...).» (ext289327-pol-93a-1)

[relationship] not would-last; [war situation] SE_{intrinsic clitic} prolonged; [conflict] dragged SE_{intrinsic clitic}

- passar

The verb passar can relate three – or alternatively two – arguments, depending on whether it is considered a control or a raising verb, respectively. See (7), repeated below:

(18) O presidente passou duas horas a falar sobre direitos humanos.

the president spent two hours at talk_INF about human rights

If we take this to be a subject control construction, the verb relates a non-sentential external argument (an NP), representing a common object or individual (x), an internal argument, representing an eventuality (ev), and a second (temporal) internal argument, representing either an amount of time (mt) or a time interval (t). The external argument of passar and the external argument in the embedded eventuality-denoting phrase are co-referential, i.e. passar has the argument structure [verb (X_NPi, ev_{[S NPi ...]}, mt/t)]. The eventuality-denoting internal argument may correspond either to a full (infinitival) sentence, introduced by the preposition a – cf. (18) –, or to a verbless constituent (e.g. an AP or a PP) with propositional content – cf. (19). English spend has a similar syntactic behaviour.

(19) O presidente passou duas semanas {afogado em trabalho / em Roma / com o pai}.

the president spent two weeks drowned in work / in Rome / with the father

‘The president spent two weeks {immersed in work / in Rome / with his father}.'

An alternative analysis would be to consider that passar is a binary predicate – [verb (ev_XP', mt/t)] –, associated, in sentences like (18) or (19), to raising, rather than control. The fact that passar combines with meteorological (zero-argument) verbs – cf. (20) below – seems to go in favour of this analysis. This does not necessarily hold for English spend, which seems incompatible with zero-argument verbs.

today spent two hours at rain_INF

‘Today, it rained for two hours.'

I will not take a stand on the issue of whether passar is a ternary-control or a binary-raising verb in sequences like (18)-(19), since this is immaterial to the goals of this paper.

- levar, demorar

With the durative meaning relevant for this work, verbs levar and demorar can be used in two different ways (a third use, exclusive to levar, will be mentioned afterwards):

(i) as binary predicates, with an argument structure similar to that of durar, prolongar-se and arrastar-se^[3]; this is illustrated for demorar in (10) above, and for levar in

(21) A remoção dos escombros levou quatro horas.

‘The removal of the debris took four hours.'

(ii) as ternary/binary predicates, with an argument structure similar to that of passar in (18); levar and demorar are distinguished from passar, with respect to argument structure, in (at least) three properties – their eventuality-denoting internal argument is always sentential (not a verbless AP or PP, for instance^[4]) and may also be introduced by the preposition para (besides a) – cf. (22)-(23); the main subject of the sentences in which they occur very often correlates with the internal argument (rather the external argument) of the embedded infinitival, which is a clear passive participial construction or a bare infinitival structure with passive meaning/structure^[5] – cf. (24)^[6].

(22) O funcionário levou dois dias {a / para} concluir o relatório.

the officer took two days to / for complete the report

‘It took the officer two days to complete the report.'

(23) Os bombeiros demoraram duas horas {a / para} apagar o incêndio.

‘It took the firefighters two hours to extinguish the fire.'

(24) O incêndio {levou / demorou} duas horas a {ser apagado / apagar}.

the fire took / took two hours to be extinguished / extinguish

‘It took two hours to extinguish the fire.'

The possibility of using levar and demorar in undisputed binary structures with zero-argument verbs in the embedded clause (which favours a raising analysis) seems more restricted, although not impossible, as the following examples show:

today took more time than the usual to stay night closed (= ‘be pitch dark')

(26) Hoje, demorou muito a anoitecer.

today took a-lot to come-the-night (= ‘grow dark')

Structures with expletive subjects, and two internal arguments, like the following, also favour the raising analysis:

(27)«A Expo está praticamente pronta: levou cinco anos a construir (...) 650 mil metros quadrados de superfície urbanizada.» (ext649260-clt-91b-1)

With respect to levar and demorar, it is important to stress that only durative readings (whether involving strict duration or durative temporal location) are relevant in this paper. Very often, especially when the preposition para is used, these two verbs are associated with a different non-durative reading, termed “inceptive”, that I will henceforth ignore. In this reading, the stated amount of time refers to the time span separating a contextually given perspective point from the beginning of the described eventuality, and sentences are equivalent to those containing time adjuncts like passado x-time (‘after x-time'); they express temporal distances (cf. Móia and Alves 2001) rather than duration. Examples:

(28)«(...) o Estrela limitou-se praticamente a esperar pelo adversário no seu meio-campo (...). Levou 38 minutos parafazer o primeiro remate a sério à baliza de Palatsi (...).» (ext230017-des-98b-2)

[it] took 38 minutes to make the first shot in earnest to-the goal of Palatsi

‘Only after 38 minutes did it take the first serious shot on Palatsi's goal.'

(29)«O pessimismo de todos começou a ser combatido bem cedo. Demorou sete minutos parase ver o primeiro remate do Benfica, mas valeu a pena para os espectadores.» (ext317900-des-97b-1)

‘Only after seven minutes was the first shot on goal from Benfica taken.'

Finally, note that the internal argument of demorar – like durar, prolongar-se and arrastar-se (cf. (17)), and unlike levar and passar (or, for that matter, English take,which normally translates demorar) – may be null, with the interpretation “a long time”. In these cases, demorar is better translated (and glossed) as take long:

(30)«Foi então apresentado um pedido de constituição da sociedade (...), cujo deferimento demorou.» (ext13300-eco-92a-1); «O fogo, esse, não demorou a ser extinto, já que às 6h10 os trabalhos estavam concluídos.» (ext75344-soc-98a-1)

[acceptance (of request)] took-long; [fire] not took-long to be extinguished

- levar_AD

The verb levar can also be used as a binary predicate in a construction where it expresses anchored, rather than non-anchored duration, which I separate here for the sake of clarity (given its distinctive properties). In this construction – illustrated in (9), repeated below as (31) –, levar combines with atelic descriptions and corresponds more to English spend or go on for than to English take:

(31) As negociações já levam dois meses.

the negotiations already take two months

‘Negotiations have been going on for two months now.'

Note that, if the verb is changed to the simple past (levaram, instead of já levam), to express non-anchored duration, its meaning changes completely, becoming equivalent to English take.

Levar seems able to convey anchored duration also in constructions with eventuality-denoting verbless constituents with propositional content, where – again – either a control or a raising analysis can be hypothesized:

(32) O jogador já leva dois meses {lesionado / no hospital / sem jogar}.

the player already takes two months injured / in-the hospital / without play

Curiously, however, in these cases, if the verb is changed to the simple past (levou, instead of já leva), to express non-anchored duration, its meaning does not become equivalent to English take,but rather remains equivalent to English spend or go on for. More on this in section 1.5.

1.4.Semantic equivalence between verb-complement combinations and duration adjuncts

It is relevant to underline that the pairing of a durative verb and a temporal complement conveys the same temporal information as can be obtained from the use of a time adjunct applied to an eventuality-denoting sequence (cf. Móia 2006: 43-44; 2011b: passim). The exact correspondences vary with the type of semantic relation expressed. I will just illustrate here the case of non-anchored duration, but, as we will see later, the same applies to all other cases.

Sentences (expressing non-anchored duration) with atelic eventualities – and, typically, durar, prolongar-se, arrastar-se or passar – are equivalent to sentences with time adjuncts headed by the preposition durante₁(‘for'), or, especially in contemporary Brazilian Portuguese, the synonymous preposition por^[7](cf. Móia 2001, 2006).

(33) a. O bombardeamento de Dresden durou três dias. [verb+complement]

the bombing of Dresden lasted three days

‘The bombing of Dresden lasted three days.'

[time adjunct]

Dresden was bombed for / for three days

‘Dresden was bombed for three days.'

Sentences (expressing non-anchored duration) with telic eventualities – and, typically, levar or demorar – are equivalent to sentences with time adjuncts headed by em (‘in'):

(34) a. O funcionário levou dois dias a concluir o relatório. [verb+com- plement]

the officer took two days to complete the report

‘It took the officer two days to complete the report.'

the officer completed the report in two days

‘The officer completed the report in two days.'

1.5. Sensitivity to telicity of Portuguese durative verbs

There seems to be a strong tendency to use different verbs for different telicity values (of the eventuality-describing argument)^[8]. This is the case both in English (cf. e.g. Dowty's 1979 use of the “spend-an-hour/take-an-hour distinction”) and Portuguese (cf. e.g. Móia: 2006, 47-50). In Portuguese, the regular, most frequent, correspondences are given in (35). All sentences presented up to now involve these correspondences.

(35) durar, prolongar-se, arrastar-se, passar, levar_ADÞ descriptions of atelic eventualities;

levar, demorar Þ descriptions of telic eventualities

(36) a. levar Þ descriptions of atelic eventualities
(very common, somewhat informal registers)

b. durar, prolongar-se, arrastar-se Þ descriptions of telic eventual­- ities (relatively infrequent, apparently limited to activity/accom­- plishment ambiguous expressions);

The three other verbs – passar, levar_ADand demorar – are unique in that they seem to combine only with atelic and telic descriptions, in the first two and the third cases, respectively (with hardly any exception).

Let us consider the situations in (36a) and (36b) separately, since they are unalike.

The possibility mentioned in (36a) exists in at least two situations:

(i) when levar combines with non-sentential arguments with propositional – typically stative – content (e.g. an AP or a PP); in this situation, given the stative character of the described eventuality, levar is unequivocally equivalent to passar (or estar)and English spend – cf. (32) above; the structure is considered slightly marginal for some speakers:

the player took/spent two months injured / in-the hospital / without play

‘The player {was injured / was in the hospital / did not play} for two months.'

(ii) when levar combines with clearly activity-describing infinitival arguments, with active structure and headed by a^[10], as in the corpora examples (38), containing the (atelic) verbs or verbal expressions explicar (‘explain'), investigar (‘investigate') and pregar no deserto (‘preach in the desert'); in these contexts (which are relatively common, though slightly unexpected in formal registers, in my opinion), passar can readily occur, with equivalent interpretation:

(38)«(…) [o candidato] levou toda a campanha a explicar que o PRD já não era o mesmo (…).» (ext612007-pol-95a-2); «Calvi levou anos a investigar em Itália (...).» (ext627382-clt-95b-1); «O Tribunal de Contas levou quatro anos a pregar no deserto.» (ext302563-eco-96b-1) [º passou]

took/spent all the campaign at explain_INF that...; took/spent years at investigate_INF in Italy; took/spent four years at preach_INF in the desert

The fact that levar x-time/x-interval a V_INF can combine with activity-descriptions (with the same meaning as passar) leads to inconvenient ambiguities, when descriptions of culminated processes are used: in a non-shifted reading, in which the culmination took place, levar is equivalent to demorar and English take; in a shifted culminated process > process reading (sensu Moens), corresponding to a relatively more informal register, the culmination did not take place, and levar is equivalent to passar and English spend (as in (38)). In other words, the interpretation of levar x-time a V_INF, in these circumstances, is open to doubt (as to whether the culmination occurred or not). Here are two examples, of ambiguous telic/atelic interpretation (dubious as to whether the player recovered or the student finished reading the book, respectively):

(39) O jogador levou {dois meses / o fim-de-semana inteiro} a recuperar.
the player took two months / the weekend entire to recover

the player spent two months / the weekend entire at recover_INF

(40) O estudante levou {dois meses / o fim-de-semana inteiro} a ler Os Maias.
the student took two months / the weekend entire to read Os Maias

the player spent two months / the weekend entire at read_INFOs Maias

The extraordinary possibilities in (36b) are illustrated in the three cetempúblico examples below. They include durar, prolongar-se and arrastar-se combinedwith accomplishment-describing NPs, with nouns leitura (‘reading'), construção (‘construction') and recuperação (‘restoration'), and entailing completion. Obviously, verbs levar and demorar can readily occur here, with the same exact meaning, and are possibly felt as more natural.

the reading of-the court's-verdict lasted/took nearly of one hour

(42) «(...) os arguidos eram profissionais dedicados», frisa o acórdão, cuja leitura se prolongou por cerca de duas horas e meia.» [º levou, demorou]

whose reading SE_{INTRINSIC CLITIC} prolonged for (= ‘took') nearly of two hours and half

(43) «(...) mais de 110 mil contos foi quanto a ARSS gastou na construção desta unidade, verba foi sucessivamente aumentada uma vez que a construção se arrastou por quase quatro anos.» (ext1361688-soc-93a-2); «(...) o Governo decidiu antecipar em dois meses, sem aviso prévio, a inauguração de um edifício cuja recuperação se arrastou durante dez anos.» (ext903341-soc-95b-2) [º levou, demorou]

the construction SE_{INTRINSIC CLITIC} dragged for (= ‘took') almost four years; a building whose restoration SE_{INTRINSIC CLITIC} dragged for (= ‘took') ten years

The technical analysis of these structures is open to debate (as is whether the analysis is the same for all the verbs involved or somehow varies). The eventuality-denoting NPs used in these texts might be taken as descriptions of activities, rather than accomplishments (since they are genuinely ambiguous between the two readings, either referring to a preparatory phase+culmination aggregate, or to [part of] a preparatory phase), with the information that the culmination occurred being contextually/pragmatically elicited^[11]. In this analysis, the use of the durative verbs at stake would be regular, not extraordinary. However, it must be stressed that the culmination did in fact happen and – more vitally – that the amount of time mentioned in the complement is the duration leading up to the culmination (not just of some part of the preparatory phase, as is the regular case in Aktionsart shifts accomplishment > activity; cf. Moens 1987). These sequences represent, therefore, de facto accomplishments and are, to that extent, exceptional and worthy of notice.

In the next three sections, I will make a more detailed grammatical analysis of the selected six durative verbs in the three types of environments illustrated in section 1.1: non-anchored duration, anchored duration, and durative temporal location.

2.Durative verbs expressing non-anchored duration

2.1.Default incompatibility of Portuguese durative verbs with temporal prepositions

One of the most interesting grammatical properties of durative verbs, when expressing non-anchored duration, as noted in Móia (2011b: 259-260) is their incompatibility with the prepositions that head duration adjuncts: durante₁ / por (‘for'), and em (‘in'). In other words, in these contexts (with the remarkable exception of prolongar-se and arrastar-se, that will be discussed in section 2.2), complements of durative verbs have to be bare predicates of amounts of time and cannot be homonyms of duration adjuncts. This is of course predicted, if we assume that durative verbs and durative prepositions have the same semantic role (cf. section 1.4 above) – ascribing the mentioned amount of time to the described eventuality. The presence of both elements is therefore a – normally repudiated – redundancy.

(44) *O bombardeamento da cidade durou durante mais de duas horas.

the bombing of-the city lasted for more than two hours

(45) *O presidente passou durante duas horas a falar sobre direitos humanos.

the president spent for two hours at talk_INF about human rights

(46) *O funcionário {levou / demorou} em dois dias a concluir o relatório.

the officer took / took in two days to complete the report

There is, however, intriguing interlinguistic – and also intralinguistic – variation with respect to (the possibility of) redundant or expletive duration connectives, though only for a very limited set of verbal predicates. For instance, English admits both last x-time and the equivalent last for x-time, as pointed out in Móia (2011b: 260) and illustrated in the examples from the Corpus of Contemporary American English, in (47); French allows durer pendant x-time, besides simple durer x-time, as shown in (48):

(48) «Cette discussion va encore durer pendant quelques mois.»
(europarl.europa.eu, apud www.linguee.com)

It is also curious to note that a few examples of redundant duration connectives with the Portuguese verb durar can be found in corpora (Móia 2011b: 260). These are rare, though, and very marginal – or even totally ungrammatical – to most speakers. In the Portuguese corpus cetempúblico, there are only 3 examples with (adjacent) durar durante x-time, given in (49); in The Brazilian corpus nilc-São Carlos, there are no instances of durar durante x-time and only one example of durar por x-time (or the relevant type)^[12].

(49) «O efeito do LSD (...) costuma durar durantecerca de 10 horas.» (ext1117264-clt-soc-93a-2); «O caos inicial poderá durar durantesemanas ou meses (...).» (ext802044-soc-98b-1); «uma crise para durar durantealguns anos» (ext857789-clt-93a-1)

last for nearly of 10 hours; last for weeks or months; last for some years

(50) «(...) a reação do (...) jogador à divulgação de seu namoro extraconjugal, que durou por um ano e sete meses.» (nilc-São Carlos, Esporte--94a-2)

lasted for one year and seven months

Formally, prepositions like Portuguese durante / por, English for, or French pendant have different semantic roles in time adjuncts and in complements of durative verbs: in time adjuncts, they are essential in triggering the introduction of the DRS-condition [dur (ev) = mt)]; in the complement of durative verbs, they are semantically void (i.e. expletive or redundant), since [verb_DUR (ev, mt)] necessarily entails – by virtue of the lexical meaning of the verb – [dur (ev) = mt)].

2.2. The special cases of prolongar-se and arrastar-se

The verbs prolongar-se and arrastar-se – to my knowledge, not yet scrutinized in the literature – are totally unique among the durative verbs under consideration here, in that their complements are normally headed by prepositions which are identical to those heading duration adjuncts: durante or por.

- prolongar-se

This is a very common temporal predicate. When its internal argument represents an amount of time, it is normally headed by por or, less frequently, durante (cf. footnote 13).

(51) «O livro (...) lançou o nome de Ovídio Martins para o centro de uma polémica que se prolongou por vários anos.» (ext1051689-clt-98a-3)

SE_{INTRINSIC CLITIC} prolonged for several years

(52) «Soeiro estava inconformado com os resultados finais das negociações, que seprolongaram durantesete horas (...).» (ext1064144-soc-93a-1)

SE_{INTRINSIC CLITIC} prolonged for seven hours

The semantic contribution of the prepositions por and durante in these constructions, is – just like in the cases discussed in the previous section – null. Possibly, a grammaticalization process is underway here, changing these prepositions from duration connectives (i.e. triggers of [dur (ev) = mt]) into fixed argumental prepositions: [prolongar-se por/durante (ev, mt)]. The fact that por is the most common preposition used, even in European Portuguese – where it no longer heads common duration adjuncts nowadays, though it used to in the past (cf. footnote 7) –, is a possible indication that some form of linguistic change is operating. Furthermore, it is symptomatic that – just like durar – prolongar-se also takes, albeit infrequently, bare predicate of amounts of time as complements – cf. (53) –, making it clear that the preposition makes no semantic contribution:

SE_{INTRINSIC CLITIC} prolonged years on end

vs.(54) «As enxaquecas podem prolongar-se por horase até dias a fio (...).»
(ext1537993-clt-soc-95b-1)

prolonged SE_{INTRINSIC CLITIC} for hours and even days on end

The verb perdurar is particularly interesting, in this respect: of the 780 sequences containing this verb in cetempúblico, 36 include a por-complement, 21 include a durante-complement and at least 11 a bare predicate of amounts of time as complement.

- arrastar-se

The verb arrastar-se is very similar to prolongar-se in grammatical behaviour though it uses prepositions por and durante in more balanced proportions, and appears to be slightly less frequent^[13].

(55) «O veredicto surgiu no fim de um julgamento que se arrastou por nove anos (...).» (ext88522-pol-95a-2)

[trial] SE_{intrinsic clitic} drag for nine years

(56) «O acordo sobre os dois primeiros diplomas, cujas discussões se arrastaram durante vários anos (...), foi possível com base nos compromissos da presidência (...)» (ext1170162-pol-92b-1)

[discussions] SE_{intrinsic clitic} drag for several years

3. Durative verbs expressing anchored duration

The main facts discussed in this section are detailed in Móia (2011b: 260-262). New information, however, is provided here regarding the use of the analogous verbs prolongar-se and arrastar-se in anchored duration contexts.

The fact that the described eventuality is assumed to be going on at the relevant anchor point – [ev ◯ TPpt] – correlates with two grammatical facts: (i) that only atelic eventualities are allowed; (ii) that the tense form associated with the duration verb has to express overlapping to TPpt; that is, presente, pretérito imperfeito, or futuro imperfeito can be used in this construction (cf. examples below), but not pretérito perfeito simples, for instance.

Constructions with durative verbs expressing anchored duration come in two syntactic patterns, varying with respect to the form of the internal argument:

(i) temporal complement headed by the grammaticalized (preposition-like) connective há (for all types of perspective points, without restrictions) or havia (only for past perspective points, in competition with há, which is much more frequent in contemporary Portuguese) – cf. Móia (2011a) –, i.e. taking the same exact form of a time adjunct – cf. (58); formal issues of compositional analysis, and possible redundancy, along the lines discussed in the previous section, are also raised here

(57) A discussão da proposta {dura / durava} há dois meses. [verb+complement]

the discussion of-the proposal lasts / lasted there-is two months

‘The discussion of the proposal {has been going on for two months now / had been going on for two months then}.'

the proposal is / was to be discussed there-is two months

‘The proposal {has been under discussion for two months now / had been going on for two months then}.'

(ii) temporal complement not headed by há / havia (but conveying similar information)

(59) O trabalho já {leva / levava} dois meses.

the work already takes / took two months

The work {has been going on for two months now / had been going on for two months then}.

Let us consider each of the durative verbs in this subset individually.

- durar

Pattern (i) is used with durar in European Portuguese – cf. (57) –, though not (at least as frequently) in Brazilian Portuguese, which clearly prefers pattern (ii) for durar – cf. examples from NILC-São Carlos in (60). Portuguese cetempúblico includes a small number of texts, non-standard for most speakers, with this pattern (ii) – cf. (61)^[16]. These differences between European and Brazilian Portuguese are described in Móia (in press).

(60) «A guerra na Bósnia já dura21 meses e provocou cerca de 200 mil mortes.» (Mundo--94b-1); «Os mil motoristas e cobradores de ônibus de Foz do Iguaçu suspenderam anteontem a greve da categoria, que já duravauma semana.» (Cotidiano--94a-1)

already lasts 21 months; already lasted one week

(61) «A indefinição jurídica (...) – que já dura seis anos (...) – estará solucionada dentro de poucos meses (...).» (ext1563512-soc-95b-3)

This intralinguistic variation – more precisely, the fact that standard Brazilian Portuguese systematically drops the há connective, and European does the same, though only marginally (in anomalous structures) – seems to be a compelling indication that the connective is regarded as redundant, or expletive, in these contexts. In fact, the presence of há cannot be taken as the trigger for the insertion of the more complex set of DRS-conditions associated with anchored duration here. Rather, it seems to be the overlapping tense form of the verb (e.g. presente or pretérito imperfeito), possibly together with other linguistic cues (like the adverb já, ‘already'), that acts as the relevant trigger, guaranteeing that the intended anchored reading is obtained.

- prolongar-se

Pattern (i) is also used with prolongar-se, in European Portuguese, as illustrated in the following cetempúblico examples:

(62)«A sua eleição seria determinante para o fim (...) da sangrenta guerra civil que se prolonga há ano e meio.» (ext506424-pol-92b-1); «A PSP julga que a actividade deste grupo se prolongava há, pelo menos, dois meses (...).» (ext1087992-soc-95b-1); «Em 1976, numa comarca de província, o juiz ditou a sorte de um casal cujo julgamento se prolongava havia três dias.» (ext1032781-nd-91a-1)

SE_{INTRINSIC CLITIC} prolongs there-is year and half; SE_{INTRINSIC CLITIC} prolonged there-is at-the least two months; SE_{INTRINSIC CLITIC} prolonged there-was three days

Brazilian Portuguese apparently does not use the verb prolongar-se very often to express anchored duration, a use which is, contrariwise, widespread in European Portuguese. Only one text was found in the corpus NILC-São Carlos with this verb, and pattern (i) as well:

(63)«É que entrou na reta final uma negociacão que se prolonga há 14 meses entre dois pesos pesados multinacionais.» (NILC-São Carlos, Dinheiro--94b-1)

SE_{INTRINSIC CLITIC} prolongs there-is 14 months

The following non-standard construction, the only one of its kind found in the corpus cetempúblico, is very interesting, in that it combines prolongar-se with the argumental preposition commonly used in contexts of non-anchored duration (por), but expressing anchored duration (without há):

(64)«Políticos de Brasília identificam nessa troca de embaixadores a origem de uma desavença que já se prolonga por três anos.» (ext978032-pol-96a-1)

already SE_{INTRINSIC CLITIC} prolongs for three years

- arrastar-se

(65) «(...) os homens de Mobutu e os rebeldes deverão conferenciar, (...) a ver se será possível interromper as hostilidades que se arrastam há cinco meses e meio.» (ext15087-pol-97a-1); «O Patronato e o sindicato (...) chegaram (...) a um acordo que põe fim ao conflito que se arrastava hásemanas.» (ext285496-soc-93a-3)

[hostilities] SE_{intrinsic clitic} drag there-is five months and half;

[conflict] SE_{intrinsic clitic} drag there-was weeks

Brazilian Portuguese apparently uses the verb arrastar-se more often than prolongar-se to express anchored duration^[18], and also with pattern (i):

(66) «(...) a oportunidade histórica de finalmente encerrar um conflito sangrento que já se arrasta há tanto tempo.» (NILC-São Carlos, ext285496-soc-93a-3)

[conflict] SE_{intrinsic clitic} drag there-is so-much time

- levar_AD

Pattern (ii) – commonly used with durar in Brazilian Portuguese, as mentioned above – is the standard for durative verb levar, both in European and Brazilian Portuguese, as illustrated in (59), and in the following examples from Portuguese cetempúblico – (67) – and Brazilian NILC-São Carlos – (68):

(67)«O jornal diz que esta «reflexão» já leva quatro anos (...).» (ext303985-soc-96b-1); «A maratona já levava mais de duas horas.» (ext482757-des-95b-1)

[reflexion] already takes four years; [marathon] already took more than two hours

(68) «O processo já levava vários meses.» (Mais--94a-2)

[process] already took several months

These structures with levar are interesting for a number of reasons: first, for evincing the fact that – even in Standard European Portuguese – há is not crucial to triggering the DRS-conditions associated with anchored duration; second, for – somewhat unexpectedly – using the verb levar. Note that, since anchored duration only applies to descriptions of atelic eventualities, it would be predicted not to combine with verbs that are typical markers of telic duration, like levar. This prediction holds for demorar, though not for levar (curiously, a verb that allows for exceptional combination with descriptions of atelic eventualities in non-anchored duration – cf. (37)-(38) above). In fact, levar in (67)-(68), combined with tenses that express overlapping to TPpt (e.g. presente or pretérito imperfeito), seems to be an altogether different verb, closer in meaning to durar and prolongar-se (with the peculiarity that, contrary to those verbs, it never combines with the connective há to express anchored duration). That is why I chose to mark it differently (as levar_AD ) here.

4. Durative verbs expressing temporal location

As said before, durative verbs express location – rather than duration – when their temporal complement refers directly to an interval of the time axis, like the month of March, instead of an amount of time, like two weeks. This means that sentences state “when things happen” rather than merely “for how long they happened” or “in what time they culminated”.

All the durative verbs examined in this paper can express temporal location, but in all cases – even with telic descriptions – the eventuality has to fill the mentioned interval entirely, i.e. [verb (ev, t)] is tantamount to [t ⊇ ev] or [t = loc (ev)], whence duration can systematically be inferred: [dur (ev) ≥ dur (t)] or [dur (ev) = dur (t)], respectively. In other words, durative location lato sensu is always the case^[19]. This restriction correlates with the fact – underlined in (Móia 2011b: 262-264) that these durative verbs only combine with durative complements and with phrases headed by durative connectives.

Durative complements are phrases – prototypically exemplified by universally quantified NPs (e.g. todo o fim-de-semana, ‘all weekend') – that require the whole interval they represent to be involved in the locating relation. A sentence like (69) is thus associated with durative location – [ev ⊇ t] – despite the fact that it includes a non-durative connective (durante₂, ‘during'); if the universal quantifier is suppressed, the sentence conveys a mere overlapping (i.e. non-durative) reading – [ev ◯ t] (a vague reading, compatible with scenarios where it rained all weekend or just in some part of it):

(69) Choveu durante {todo o fim-de-semana / o fim-de-semana inteiro}.

rained during all the weekend / the weekend entire

‘It rained all weekend.'

Durative connectives are those that are systematically associated with durative location – cf. (70). These include Portuguese desde (‘since'), de... a/até (‘from... to/until'), desde... até (‘from...until), and – in most uses – até (‘until') and enquanto (‘while'). Examples of non-durative connectives are Portuguese em (‘in') and durante₂ (‘during'), that readily occur with overlapping – [ev ◯ t] – or inclusive – [ev ⊇ t] – locations.

(70) Choveu das 2 às 5 da tarde.

rained from-the 2 to-the 5 of-the afternoon

‘It rained from 2 pm to 5 pm.'

The combination of durative verbs with durative complements and with phrases headed by durative connectives is illustrated by sequences parallel to those in (69) and (70), but with (homonym) complements, rather than time adjuncts:

(71) A discussão da proposta durou todo o fim-de-semana.

‘The discussion of the proposal lasted the whole weekend.'

(72) A discussão da proposta durou das 2 às 5 da tarde.

the discussion of-the proposal lasted from-the 2 to-the 5 of-the afternoon

‘The discussion of the proposal lasted from 2 pm to 5 pm.'

Let us consider each case separately.

- durative verbs with universally quantified time-denoting NPs as complements

Below are examples (all taken from corpora, except (77)) with this combination. With atelic descriptions, regular, common cases, using verbs durar, prolongar-se, arrastar-se, passar and levar_AD are given successively in (73)-(77). With telic descriptions, less frequent cases, with levar and demorar (entailing that the whole interval, not just part of it, was occupied by the telic eventuality), are given in (78)-(79).

(73) «Os combates duraram todo o dia de quarta-feira e parte do dia de ontem (...).» (ext503346-pol-94b-2)

[fighting] lasted all the day of Wednesday and part of-the day of yesterday

(74) «E prevê que as conversações se prolonguem por todoo mês de Maio.» (ext983283-pol-91a-3)

[talks] SE_{INTRINSIC CLITIC} prolong for all the month of May

(75) «Representantes da seita (...) aceitaram entregar as armas (...), pondo termo a um conflito que se arrastou durante toda a semana (...).» (ext540044-pol-94b-3)

[conflict] SE_{INTRINSIC CLITIC} drag for all the week

(76) «A equipa (...) passou toda a sexta-feira a fazer os preparativos para este encontro.» (par=ext106700-nd-97b-1)

the team spent all the Friday at do_INF the preparations

(77) O Pedro leva quase a semana inteira sem pregar olho.

(78) «Ele está duas horas atrasado para um jantar especial que você levou a tarde inteira a preparar.» (ext1301243-soc-92b-1)

[dinner] took the afternoon entire to prepare

(79) «O furto (...) terá sido planeado minuciosamente e (...) a sua concretização demorou todo o fim-de-semana passado.» (ext1397855-soc-97a-1)

[carrying out of the theft] took all the weekend past

(80) Os combates duraram {??a semana / toda a semana / a semana inteira}.

the combats lasted the week / all the week / the week entire

(81) Os guerrilheiros passaram {a semana / toda a semana / a semana inteira} a combater.

the guerrillas spent the week/ all the week / the week entire at fight_INF

Note, finally, that English verbs last, spend and take can also occur in this type of temporal location environments, as the following examples from the American Corpus of Contemporary American English show: «Take a look at the Southern Plains, very windy, (...) not a drop of precipitation in sight. This is expected to last for the next several weeks.» (CBS_Early, 2011); «I spent the whole month running through the city (...)» (Smithsonian, 2001); «The debriefing of David Sharon began the following morning, and it took the whole day and half the night.» (Frederick Forsyth, The Fist of God).

Below are examples with this combination, for some of the durative verbs under study here. Apparently, it is relatively common for durar (with all connectives), prolongar-se and arrastar-se (except with enquanto), and relatively uncommon, or impossible, for the other verbs^[20]. English verbs last, spend and take also accept this type of temporal complements.

(82) «(...) Portugal ver-se-ia obrigado a alargar a sua soberania (...) através de guerras que duraram até ao primeiro quartel do séc. XX (...).» (ext1211332-nd-95a-2)

[wars] lasted until to-the first quarter of-the century XX

(83) «(...) depois do fim oficial desta guerra civil, que se prolongou de1975 até 1990.» (ext599706-pol-95a-2)

[civil war] SE_{INTRINSIC CLITIC} prolonged from 1975 until 1990

(84) «(...) proclamou unilateralmente a República de Biafra e deu origem a uma enorme guerra civil, que se arrastou até 1970.» (ext1058389-pol-96a-2)

[civil war] SE_{INTRINSIC CLITIC} dragged until 1970

(85) «(...) a construção de um oleoduto (...) vai demorar atéDezembro, segundo disse (...) um responsável do Ministério (...).» (ext1479220-clt-soc-94b-1)

[construction of a pipeline] goes (='will') take until December

These constructions raise some formal (compositional) issues. Connectives like desde or de...a can also head time adjuncts. Thus, the phrases with these connectives are ambivalent between mere time-denoting expressions, when used as arguments, and full adjuncts (also providing the location information), when used adverbially (as in (70) above). Several hypotheses can be explored to account for this fact: (i) that these phrases are genuinely ambivalent, belonging in two different semantic-syntactic categories; (ii) that they are always mere time-denoting expressions, and, when in adjunct position, they co-occur with a null (implicit) locating connective, with the durative value of e.g. throughout (cf. similar proposal, in Móia 2000, of a null locating connective, though non-durative, for e.g. “ambivalent” antes-phrases). I will not attempt to argue for any of these hypotheses here.

(86) *A discussão da proposta durou no mês de Março.

the discussion of-the proposal lasted in-the month of March

(87) *A discussão da proposta durou durante o mês de Março.

the discussion of-the proposal lasted during the month of March

The verbs prolongar-se and arrastar-se are again the exception, as can be seen in (74)-(75) above, since they combine regularly with durante, por, and – somewhat marginally – em, followed by a time-denoting expression^[22]. The prepositions act here also as (grammaticalized) argumental prepositions:

[growth] must SE_{INTRINSIC CLITIC} prolong during the next decade; [festivities] SE_{INTRINSIC CLITIC} prolong during-the month of July; [campaign] SE_{INTRINSIC CLITIC} prolong in-the next three months

(89) «Estas negociações ocorrem alguns meses depois de se terem gorado contactos entre as partes, que se arrastaram durante os anos de 1990 e 1991.» (ext826861-soc-92b-1); «(...) uma série de acções (...), que começaria por uma greve simbólica de aviso no dia 31 de Dezembro (...) e se arrastaria por todo o Verão (...).» (ext654409-soc-93a-2); «Este é um dado inesperado nas negociações que se arrastam nos últimos meses com a Saudi Aramco (...).» (ext53469-eco-97a-1)

[contacts] SE_{INTRINSIC CLITIC} drag during the years 1990 and 1991; [series of actions] SE_{INTRINSIC CLITIC} would-drag during all the Summer; [negotiations] SE_{INTRINSIC CLITIC} drag in-the last months

5. Conclusions

In this paper, the main Portuguese verbs used to express strict duration and durative temporal location – durar, prolongar-se, arrastar-se, passar, levar, demorar – were examined, and a few grammatical idiosyncrasies noted.

As for their argument structure, they all relate (in their relevant senses) an eventuality-denoting expression with a temporal argument (either identifying an amount of time, if duration is involved, or a time interval, if location is involved), with a possible third (object-denoting) element in a control or raising construction (in the cases of passar, levar and demorar). These predicate-argument combinations convey the same information that can be obtained by applying time adjuncts to eventuality-denoting phrases.

As for semantic restrictions, telicity was confirmed to play a crucial role in the selection of the various durative verbs, with durar, prolongar-se, arrastar-se, passar and one form of levar normally combined with atelic descriptions, and demorar and another form of levar normally combined with telic descriptions. A few exceptional cases departing from these correspondences (arguably without any Aktionsart shifts involved) were noted in corpora data, showing that telicity restrictions tend to be more flexible with predicates than with adjuncts.

As for syntactic-semantic restrictions, having to do with compositional analysis and the role of temporal connectives, some intriguing data was collected, illustrating two facts: that redundant temporal connectives (homonyms of those occurring in time adjuncts) may show up in time complements, and that there is a significant amount of variation and anomaly, and some signs of linguist change, affecting this use of connectives.

In non-anchored duration, prolongar-se and arrastar-se stand out from the rest, in that they standardly use prepositions durante₁and por (‘for')in their complements (the latter no longer usually used in European Portuguese to mark common atelic duration in time adjuncts); prolongar-se clearly prefers por, arrastar-se shows no clear preference for either preposition; durar shows signs of linguistic instability, inasmuch as it non-standardly occurs (albeit rarely) with those prepositions (like English last and French durer do with for and pendant, respectively).

In durative temporal location, two contexts prevail: with durative complements, typically NPs with universal quantification; with complements headed by durative connectives like desde, de... a/até, desde... até, até and enquanto. In the first case, prolongar-se and arrastar-se stand out from the rest once more, since they standardly use prepositions durante₂and por (‘during', ‘throughout')in their complements, whereas the other verbs (unless in non-standard rare cases, like durar durante x-interval) do not; the universal quantifier may be only implicit, a case more frequent with some verbs (e.g. passar) than with others (e.g. durar) – cf. passar a semana (inteira) a V_INF vs. durar a semana ^??(inteira). In the second case, verbs differ significantly in frequency (from the extremely common durar desde/até x-interval to the rarer passar desde/até x-interval a V_INF), and also in their compatibility with the different connectives (e.g. enquanto occurs usually with durar, but not with the other verbs).

All in all, the system of durative verbs displays a significant amount of variation and anomaly, indicating that it is an area prone to linguistic change over time.

References

Abusch, D. (1990). Temporal adverbs and the English perfect. In Proceedings of the 20th North East Linguistic Society.         [ Links ]

Brito, A. M. & Oliveira, F. (1997). Nominalization, Aspect and Argument Structure. In G. Matos et al.(Eds.), Interfaces in Linguistic Theory (pp. 57-80). Lisboa: Colibri/APL.         [ Links ]

Dowty, D. (1979). Word meaning and Montague Grammar. Dordrecht: D. Reidel.         [ Links ]

Hitzeman, J. (1993). Temporal adverbials and the syntax-semantics interface. Ph.D. thesis, University of Rochester, Rochester, New York.         [ Links ]

Hitzeman, J. (1997). Semantic partition and the ambiguity of sentences containing temporal adverbials. Natural Language Semantics, 5, 87-100.         [ Links ]

Huddleston, R. & Pullum, G. (2002). The Cambridge grammar of the English language. Cambridge: Cambridge University Press.         [ Links ]

Kamp, H. & Reyle, U. (1993). From discourse to logic. Introduction to modeltheoretic semantics of natural language, formal logic and Discourse Representation Theory. Dordrecht: Kluwer.         [ Links ]

Mittwoch, A. (1988). Aspects of English aspect: on the interaction of perfect, progressive and durational phrases. Linguistics and Philosophy, 11, 203-254.         [ Links ]

Moens, M. (1987). Tense, aspect and temporal reference. Ph.D. thesis, University of Edinburgh.         [ Links ]

Móia, T. (2000). Identifying and computing temporal locating adverbials with a particular focus on Portuguese and English,Ph.D. thesis, Universidade de Lisboa.         [ Links ]

Móia, T. (2001). Sobre a expressão da duração em português europeu e português brasileiro: o uso de sintagmas com a preposição por. In M. E. Soares (Ed.), Boletim da Associação Brasileira de Lingüística, Número Especial, II Congresso Internacional da ABRALIN, Fortaleza, Março de 2001, Anais - Vol. I (pp. 415-419). Fortaleza: ABRALIN.         [ Links ]

Móia, T. (2006). Portuguese expressions of duration and their English counterparts. Journal of Portuguese Linguistics, 5 (1), 37-73.         [ Links ]

Móia, T. (2011a). Expressões temporais com haver: gramaticalização e interpretação semântica. In XXVI Encontro Nacional da Associação Portuguesa de Linguística. Textos Seleccionados 2010. Porto 2010 (pp. 401-419). Lisboa: APL.         [ Links ]

Móia, T. (2011b). Sobre a expressão lexical da duração e da localização temporal em português. In M. Arden, C. Märzhäuser & B. Meisnitzer (Eds.), Linguística do Português. Rumos e Pontes (pp. 251-269). München: Martin Meidenbauer Verlag.         [ Links ]

Móia, T. (in press). Predicados temporais e gramaticalização em português. Presented at 9.º Congresso de Lusitanistas Alemães (9. Deutschen Lusitanistentag), University of Viena, September 2011, to be published in 2015 by Martin Meidenbauer Verlag, München.         [ Links ]

Móia, T. & Alves, A. T. (2001). Sobre a expressão de distâncias temporais no português europeu e no português brasileiro. In Actas do XVI Encontro Nacional da Associação Portuguesa de Linguística (Coimbra, 28-30 de Setembro de 2000) (pp. 699-713). Lisboa: APL.         [ Links ]

Richards, B. (1982). Tense, aspect and time adverbials. Part I. Linguistics and Philosophy, 5, 59-107.         [ Links ]

^[1] For a more detailed discussion of the differences between these two closely related, albeit distinct, notions (as well as the associated distinction between amounts of time and time intervals), see Kamp and Reyle (1993), and Móia (2000: 135ss.).

^[2] On comparable structures in English, namely the possibly TPpt-dependent interpretation of either for- or in-phrases, cf. e.g. Dowty (1979), Richards (1982), Mittwoch (1988), Abusch (1990), Kamp and Reyle (1993), Hitzeman (1993, 1997).

^[3] With levar and demorar the external argumental can also be an infinitival sentence (whereas with durar, prolongar-se and arrastar-se this is normally not the case): «Para Paulo Domingos, votar leva o dia inteiro.» (cetempúblico, ext972797-nd-91b-2); «Funcionamos em equipas de quatro elementos e rodamos por vários palcos. Montar demoracerca de uma hora e meia.» (cetempúblico, ext1381493-soc-98b-1) (glosses: vote_INF takes the day entire; set-up_INF[a stage] takes nearly of one hour and half).

^[4] This is possible, however, for levar in the exceptional, somewhat informal structures, where it is used as a synonym of passar/spend, not of take – cf. section 1.5, example (37).

^[5] Cf. the possibility of adding an agentive phrase headed by por (‘by') to the bare infinitive: «(...) a narrativa é servida por um desenho sumptuoso e barroco que levou dez anos a realizar pelo desenhador Jean-Claude Gal.» (cetempúblico, ext1475848-clt-93a-1) (gloss: took ten years to make by the artist Jean-Claude Gal).

^[6] Apparently, this is not so common in English (at least with a participial passive), but the construction appears in corpora, as the following examples from the Corpus of Contemporary American English illustrate: «Similarly, Christiaan Eijkman and Gerrit Grijn showed in the 1890s that whole-grain rice prevented beriberi (...), but those findings took 10 years to be accepted.» (American Heritage, 2000); «(...) but the necklace is 150 carets of colored diamonds that took two years to assemble.» (ABC_GMA, 1997). Just to have an idea of the prevalence of the passive construction in Portuguese, of the 158 unequivocally relevant results yielded by the search “levou” [] {0,1} “horas|dias|meses|anos”, 107 (68%) have a passive structure (26 participial, 81 bare infinitival) and 51 an active structure.

^[7] Common duration phrases headed by por have been used in European Portuguese since very early times – cf. «E, como ouvesse aly estado por tres anos emteiros, acomteçeo em aquelle lugar meesmo que pasou de aquesta vida outro fraire de gramde perfeiçom (...)» (Crónica da Ordem dos Frades Menores, 13th century, apud Corpus do Português, Davies-Ferreira). In e.g. 19^th century Portuguese literature, they are still common – cf. «Partindo depois para a fronteira (...), guerreou por vinte anos os sarracenos.» (corpus Vercial, id=«História de Portugal II Prosa AH», Alexandre Herculano). However, in contemporary European Portuguese, they are no longer frequent (except in some exceptional cases – cf. Móia 2001: 417). A sequence like foi casado por dez anos,instead of foi casado durante dez anos,is therefore unlikely to occur in Portuguese (as opposed to Brazilian) contemporary texts.

^[8] I will ignore here all the (possibly intricate) issues regarding the determination of the telicity values of eventuality-denoting NPs.

^[9] Móia (2011a) notes the exceptions with durar (p.259, fn.8).

^[11] Cf. also Brito & Oliveira (1997).

^[12] Searches: [lema=”durar”] “durante” and [lema=”durar”] “por”. Both corpora includes several standard productions combining durar por and predicates of amounts of time with the hypernyms tempo and período; in some of these cases, the preposition por can be suppressed (durar por muito tempo, durar por algum tempo; durar por um período de x-time); in other cases, it is mandatory, since it is part of an idiom (durar por tempo indeterminado).

^[13] For instance, the search (in cetempúblico) “se” “arrastou|arrastaram” “por|durante” yields 78 relevants results, 41 (53%) with por and 37 (47%) with durante; a comparable search with prolongar-se yields more than four times the number of results (316), 259 (82%) with por and 57 (18%) with durante.

^[14] Possible 3-argument structures involving levar are those mentioned in (32) above. I will ignore them here for the sake of simplicity.

^[15] In fact, when atelic eventualities expressed via sentences are associated with anchored duration, normally, only constructions with haver-adjuncts, as (58), or similar constructions with fazer (cf. Móia 2011a: 402; no prelo), are used.

^[16] Search [lema=”durar”] “há|havia” yields more than 500 relevant results in the Portuguese corpuscetempúblico, and only 2 in the Brazilian corpus NILC-São Carlos. Contrariwise, search “já” “dura|duram|durava.*” [] {0,1} “minuto.*|hora.*|dia.*|semana.*|mês| meses|ano.*|século.*| milénio.*|tempo” yields only 4 results of durar + bare predicate of amounts of time expressing anchored duration in cetempúblico, and 75 in NILC-São Carlos.

^[17] For instance, whereas the search (in cetempúblico) “se” “arrasta|arrastam|arrastava.*” “há” yields 624 results, the search “se” “prolonga|prolongam|prolongava.*” “há “ yields only 99 (a 6 to 1 proportion). This is stark contrast with the comparable searches involving non-anchored duration (cf. footnote 13), where the proportion is 1 to 4.

^[18] The search “arrast.*” [] {0,2} “há|havia” yields 29 relevant results in the Brazilian corpus NILC-São Carlos (vs. one single result in the comparable search with prolongar-se).

^[19] With telic descriptions, DRS-condition [t = loc (ev)] (which could be termed “exact inclusive location”; cf. Móia 2011b: 256, fn. 4) applies. Differences between durative location (stricto sensu) of atelic eventualities and exact inclusive location of accomplishments are irrelevant, though, for the issues under discussion here, and the term “durative location” (lato sensu) can be used as a cover term for the two situations.

^[21] The non-durative preposition durante₂ (‘during')has a somewhat unstable grammatical behaviour in Portuguese, in this respect, since it occurs redundantly in a few corpora examples after durar – cf. Móia (2011a: 262-263): «(...) a guerra de nervos que durou durante a semana (...).» (ext184147-pol-93a-1) (gloss: the war of nerves that lasted during the week).

^[22] Partial searches (in cetempúblico), namely “se” “prolong.*” seguido de “durante” “o|a|os|as|tod.*|ess.*|est.*|aquel.*”, “em|por” “tod.*”, “por” “ess.*|est.*|aquel.*” and “pelo|pela|pelos|pelas|no|na|nos|nas|ness.*|nest.*|naquel.*” yielded 335 relevant results (60 with durante, 252 with por and 23 with em). Comparable searches with “se” “arrast.*” yielded much smaller numbers: 14 relevant results (5 with durante, 8 with por and 1 with em).