<?xml version="1.0" standalone="yes"?>
<Paper uid="W01-1306">
  <Title>Telling apart temporal locating adverbials and time-denoting expressions</Title>
  <Section position="4" start_page="1" end_page="1" type="metho">
    <SectionTitle>
2 Basic semantic and syntactic
</SectionTitle>
    <Paragraph position="0"> characterisation The semantic characterisation of TDEs and TLAs, already briefly sketched above, is next formulated.</Paragraph>
    <Paragraph position="1"> Time-denoting expressions represent intervals or sets of intervals. They do not, just by themselves, locate the eventualities (or other entities) described in the structures to which they apply. Formally, within a DRT- framework, a given time-denoting expression TDE simply introduces a time discourse referent t, together with a condition of the form TDE (t). TDEs may (but need not) occur as part of a TLA [?] cf. Paulo died in 1980 vs. 1980 was a splendid year.</Paragraph>
    <Paragraph position="2"> Temporal locating adverbials contribute a more complex meaning. More specifically, I assume that they play two concurrent roles (cf. Moia 2000a, for details): (i) they define the location time of the sentence [?] t [?] out of the interval represented by the complement TDE of the temporal preposition [?] t c [?], i.e. that state that some relation R holds between the two intervals: R (t, t</Paragraph>
    <Paragraph position="4"> may coincide with t (with e.g. in- or during-phrases), or set one of its boundaries (with e.g. since- or untilphrases); (ii) they contribute to defining a relation R' (e.g. overlap or inclusion) between the location time t and the located eventuality ev (which is described in the basic sentence) [?] R' (ev, t).</Paragraph>
    <Paragraph position="5"> These differences in denotation go, quite naturally, on a par with differences in distribution. An obvious one involves the (im)possibility of direct combination with eventuality descriptions. Being mere designators of intervals, or sets of intervals, TDEs [?] as such [?] are expected not to combine directly with descriptions of eventualities, this combination requiring the presence of a locating particle, such as during or in. This clearly tells 1910 apart from in 1910, for instance: (7) Portugal became a Republic *(in) 1910.</Paragraph>
    <Paragraph position="6"> Conversely, other contexts are apparently exclusive to TDEs. In these contexts, temporal expressions headed by e.g. in, during, since or until cannot (in principle) occur. Four of such contexts are described as follows: (i) arguments of temporal nominal predicates like period, year, or month - cf. (8a) - or of non-temporal predicates (like terrible or disaster) representing properties that can be predicated of time stretches - cf. (8b); (ii) arguments of equative be, in constructions where the other member of the equation is an (undisputed) TDE - cf. (8c); (iii) (normally prepositioned) complements of temporal predicates, like schedule (for), date (from) or date back (to) - cf. (8d-e); (iv) complements of temporal prepositions like since  or until:- cf. (8f): (8) a. (*In) 1910 was a very tense year in Portugal.</Paragraph>
    <Paragraph position="7"> b. (*In) the year 1980 was a disaster to me.</Paragraph>
    <Paragraph position="8"> c. (*In) 1910 was the year in which Portugal became a Republic.</Paragraph>
    <Paragraph position="9"> d. The elections are scheduled for (*in) June.</Paragraph>
    <Paragraph position="10"> e. This bridge dates from (*in) the 12th century.</Paragraph>
    <Paragraph position="11"> f. Portugal has been a Republic since (*in) 1910.</Paragraph>
    <Paragraph position="12">  Many temporal expressions only occur in the typical contexts of TDEs, and can therefore, rather uncontroversially, be classified as such. Some examples in Portuguese and English are: o seculo XX / the 20th century, os anos 70 / the seventies, o ano de 1910 / the year 1910, Janeiro / January, 11 de Outubro / October 11, as tres horas / three o' clock, or o meio-dia / noon. However, as was said above, many other expressions may occur, with the same surface form, in the typical contexts of both TDEs and TLAs, thus raising the already mentioned categorisation issue. Let us now consider these apparently ambivalent expressions in some detail.</Paragraph>
  </Section>
  <Section position="5" start_page="1" end_page="8" type="metho">
    <SectionTitle>
3 Apparently ambivalent time-denoting
</SectionTitle>
    <Paragraph position="0"> / temporal locating expressions A first group of apparently ambivalent phrases is composed of sequences that can freely occur with or without a temporal locating preposition, without semantic variation, when combined with eventuality descriptions. Some examples, in Portuguese, are the names of days of the week e.g. sabado [Saturday] - and expressions with the deictical adjective passado [last], like a semana passada [last week]. See the following Portuguese sentences (whose parenthesised elements may or may not be expressed): (9) O museu foi reaberto (em) a semana passada.</Paragraph>
    <Paragraph position="1"> &amp;quot;the museum was reopened (in) the week past&amp;quot; [The museum was reopened last week.] (10) O museu estara aberto (no) sabado.</Paragraph>
    <Paragraph position="2"> &amp;quot;the museum will-be open (on-the) Saturday&amp;quot; [The museum will be open on Saturday.] In these examples, if the parenthesised elements are omitted, the expressions in italic are used as complete TLAs; otherwise (i.e. if the preposition em is explicit), they are used as mere TDEs (as is, for that matter, the case if they are preceded by an explicit preposition like desde [since]). In general, the (non-)mandatory character of the preposition in this type of constructions seems to be syntactically or lexically determined, varying from language to language for parallel expressions. Note, for instance, that the grammatical English equivalent of (9) is obligatorily non-prepositioned, while the counterpart of (10) is prepositioned.</Paragraph>
    <Paragraph position="3"> A second (larger) group of apparently ambivalent phrases includes expressions that systematically occur without any (explicit) temporal locating particle in adverbial position. English expressions like last week, or referentially dependent English and Portuguese adverbs like today / hoje, yesterday / ontem, tomorrow / amanha, now / agora, and then / entao belong in this group. Accordingly, the main difference with respect to the first one (mentioned above) is the ungrammaticality of the use of a temporal preposition like English in or Portuguese em in sentences where these expressions occur adverbially:  (11) a. Ana got a new job (*in) yesterday.</Paragraph>
    <Paragraph position="4"> b. A Ana arranjou um novo emprego (*em) ontem.</Paragraph>
    <Paragraph position="5"> (12) a. Ana is in his office (*in/at) now.</Paragraph>
    <Paragraph position="6"> b. A Ana esta no escritorio (*em) agora.</Paragraph>
    <Paragraph position="7"> This contrasts with the possibility of using the same expressions as complements of other temporal prepositions, like English since or until, and their Portuguese counterparts desde and ate: (13) a. Ana has been working in this company since yesterday.</Paragraph>
    <Paragraph position="8"> b. A Ana trabalha nesta empresa desde ontem.</Paragraph>
    <Paragraph position="9"> (14) a. Ana was in her office until now.</Paragraph>
    <Paragraph position="10"> b. A Ana esteve no escritorio ate agora.</Paragraph>
    <Paragraph position="11">  The non-prepositioned adverbial cases illustrated in sentences (9)-(12) above are treated by many authors [?] regarding English and other languages [?] as involving a null preposition with a value close to that of in (or on or at, depending on the complement). Examples are Asher et al. (1995: 109) for the French adverbial hier [yesterday], and Kamp and Reyle (1993), for the English adverbials with last: &amp;quot;from a semantic viewpoint last Sunday functions rather like a prepositional phrase, with an empty preposition whose semantic contribution is the same as that of on in on Sunday. This is what we assume the syntax of the adverb last Sunday to be&amp;quot; (p. 623).</Paragraph>
    <Paragraph position="12"> At this point, it must be stressed that, under this &amp;quot;null preposition analysis&amp;quot;, the expressions at stake are not ambivalent: they are mere time-denoting phrases in all the contexts in which they occur. The property that distinguishes them from other &amp;quot;canonical&amp;quot; TDEs (like 1910) is the fact that they can [?] or must [?] occur with a null locating preposition, when they are used adverbially.</Paragraph>
    <Paragraph position="13"> I will embrace this analysis and will, furthermore, advocate - departing in this case from the analyses currently adopted in the semantic literature - that several other expressions that are normally treated as &amp;quot;canonical&amp;quot; TLAs should be put on a par with those just mentioned, i.e. they should also be treated as mere time-denoting expressions that in adverbial contexts occur with a null locating preposition. These are namely the structurally complex expressions headed by the English particles before, after, between, when, ago, or from (and possibly others, similar to them), and their counterparts in Portuguese, and possibly other languages (cf. Table 1 below). Notice that these complex expressions already contain an explicit temporal particle (prepositional or similar). Thus, when occurring in adverbial position with no other explicit preposition, I will postulate the existence of a  second invisible temporal particle: (15) a. John graduated [?] in before 1980.</Paragraph>
    <Paragraph position="14"> b. John graduated [?]  in/at three years ago.</Paragraph>
    <Paragraph position="15"> I think that at least four groups of these structurally complex expressions [?] listed below [?] can be distinguished (attending to the process by which they define time intervals). The question of how, precisely, intervals are defined via these complex expressions, i.e. of what specific properties these intervals have, will be addressed only superficially, since this question is far too complex to be tackled within the limits of this paper. In the sets of sentences given below, examples a illustrate the occurrence of the italicised expressions in a characteristic context of a TLA, and examples b their occurrence in a characteristic context of a TDE. I. Phrases expressing anteriority or posteriority, in particular those headed by English before and after, and Portuguese antes and depois:  (16) a. Ana got married after 1980.</Paragraph>
    <Paragraph position="16"> a'. A Ana casou depois de 1980.</Paragraph>
    <Paragraph position="17"> b. Ana worked in this company until after 1980.</Paragraph>
    <Paragraph position="18"> b'. A Ana trabalhou nesta empresa ate  depois de 1980.</Paragraph>
    <Paragraph position="19"> These expressions define intervals (t) by explicitly defining one of their boundaries (t  (18) a. Europe was at war between 1914 and 1918.</Paragraph>
    <Paragraph position="20"> a'. A Europa esteve em guerra entre 1914 e 1918.</Paragraph>
    <Paragraph position="21"> b. This painting dates from between 1914 and 1918.</Paragraph>
    <Paragraph position="22">  b'. Este quadro data de entre 1914 e 1918. These expressions define intervals (t) by explicitly defining both their boundaries (t  (20) a. The president hired a new secretary when he returned from Brazil.</Paragraph>
    <Paragraph position="23"> a'. O presidente contratou uma secretaria nova quando regressou do Brasil.</Paragraph>
    <Paragraph position="24"> b. The president scheduled a meeting for when he returns from Brazil.</Paragraph>
    <Paragraph position="25">  b'. O presidente marcou uma reuniao para quando regressar do Brasil.</Paragraph>
    <Paragraph position="26"> These expressions define intervals (t) via one of a series of possible relations R with the eventuality expressed in the sentence they  b. The law will be in force until two years from now.</Paragraph>
    <Paragraph position="27"> b'. A lei estara em vigor ate daqui a dois anos.</Paragraph>
    <Paragraph position="28"> These expressions define intervals (t) by measuring time, or counting temporally ordered entities, from anchor points (Apt)  . Two examples of the relevant logic conditions (expressed in the language of DRT), for cases involving time measurement, are as follows:  (19b) applies to structures with plural complements, like between the two world wars.</Paragraph>
    <Paragraph position="29">  t may either be the &amp;quot;loc&amp;quot; (in the sense of Kamp and Reyle, 1993) of ev, its resultant state, or its preparatory phase, for instance; in more complex cases, an eventuality derived from ev by Aktionsart shift may be involved (cf. e.g. Moens, 1987 and Carecho, 1996). These complex issues are obviously beyond the scope of this paper.</Paragraph>
    <Paragraph position="30">  Cf. Moia (2000a [chapter 7] and 2000b) for a more thorough analysis of these expressions.</Paragraph>
    <Paragraph position="31">  enquanto while, as long as ao longo de throughout desde, a partir de since, from ate until, by de...ate / a from...until / to II. heads of (structurally complex)  traditionally classified as temporal locating It must be stressed that by analysing all these four groups of expressions as mere TDEs, the set  The condition [beg (t  ) [?] t] must be read as: &amp;quot;the interval t (which is the interval denoted by the expression as a whole) is an interval surrounding (up to a contextually dependent limit) the beginning of t  &amp;quot;. This definition is motivated by the fact that these phrases do not behave as punctual, but rather as extended ones (cf. Moia, 2000a).</Paragraph>
    <Paragraph position="32">  I take for to be a temporal locating preposition in structures like for the last two months or for those two months (cf. Moia, 2000a [chapter 5]).  The discontinuous particle de...a can form a TLA [?] e.g. de 1980 a 1985 (from 1980 to 1985) [?] or a TDE [?] e.g. daqui a cinco dias (five days from now). of temporal particles that are traditionally classified as temporal locating is partitioned into two main groups, illustrated in Table 1. So far, the analysis under discussion was essentially motivated by the multiple categorial behaviour of the expressions at stake, namely their ability to appear in the contexts of both TLAs and TDEs. The point to stress is that this analysis provides a uniform treatment of such phrases, avoiding a double categorisation, and, as a consequence, (arguably) conferring a stronger generalisation power to the system. In the next section, I will briefly show that independent motivation exists for the two components of the advocated analysis: (i) the assignment of the categorial-denotational status of TDEs to the relevant expressions; (ii) the postulation of a higher invisible temporal locating preposition.</Paragraph>
    <Paragraph position="33"> 4 Further motivation for the analysis As for the assignment of the categorial-denotational status of TDEs to the relevant expressions, several arguments are supplied in Moia (1998) to support a time-denoting analysis of before- and after-phrases which can be adapted, in a broader perspective, to the group of expressions under analysis. Among these arguments, I underline the possibility of anaphoric reference to the intervals denoted (in adverbial or non-adverbial contexts)  : (24) a. Every student who graduated [before the stock market crashed] t is presently employed. It was not so difficult to get a job [then] t b. This painting does not date from [before 1300] t . There were no paintings like this</Paragraph>
    <Paragraph position="35"> As for the null locating preposition, the following facts can be taken as favourable arguments: i. The systematic optionality of the postulated locating preposition with a subgroup of  Other advantages of the proposed analysis seem more specific to the before- and after- type of phrases (cf. Moia 1998, where I argue for an analysis in which these phrases, when occurring adverbially, do not directly assert anteriority or posteriority, but rather inclusion - marked by null in - in an interval defined along the lines of (17)).</Paragraph>
    <Paragraph position="36"> ambivalent expressions - e.g. Portuguese sabado or a semana passada (cf. (9)-(10), repeated below): (25) a. O museu esteve encerrado (em) a semana passada.</Paragraph>
    <Paragraph position="37"> b. O museu estara aberto (no) sabado.</Paragraph>
    <Paragraph position="38"> ii. The (explicit) presence of the postulated preposition in comparable forms of other language variants, or of other languages - cf. e.g. em antes (literally, in before), in certain Northern Portuguese dialects, or in between, in English: (26) a. &amp;quot;...que as espigas, em antes de ir para o canastro, tinham de se por a secar&amp;quot; (corpus of oral discourse from the North of Portugal) b. Berlin was well known for its good living in between the two world wars.</Paragraph>
    <Section position="1" start_page="8" end_page="8" type="sub_section">
      <SectionTitle>
(Collins Cobuild English Dictionary)
</SectionTitle>
      <Paragraph position="0"> iii. The (explicit) presence of the postulated preposition before semantically equivalent structures with the predicate period: (27) a. That happened [?] in [before Christmas].</Paragraph>
      <Paragraph position="1"> b. That happened in [the period before Christmas].</Paragraph>
      <Paragraph position="2"> (28) a. Ana was travelling [?] in [between June and October].</Paragraph>
      <Paragraph position="3"> b. Ana was travelling in [the period between June and October] Marginally, but quite symptomatically, it must be emphasised that the presence of null prepositions seems to be motivated also within other temporal subsystems, like duration and temporal quantification: (29) a. Ana only slept [?] for about half an hour.</Paragraph>
      <Paragraph position="4"> b. Ana went to Paris [?] in every week.</Paragraph>
      <Paragraph position="5"> (cf. e.g. Rothstein, 1995) In addition to the arguments already presented, the following (crucial) one can be adduced: the advocated analysis correctly predicts the possibility of recursion of the particles treated as heads of complex TDEs and the impossibility of recursion of the true temporal locating particles (given their role in defining the - unique - location time of the sentence). Compare the grammaticality of expressions like those in (30) with the ungrammaticality of e.g. those in (31): (30) a. before two months ago a'. antes de ha dois meses b. (in the period) between two years ago and six months ago b'. (no periodo) entre ha dois anos e ha seis meses c. between two months before the start of the war and the second week of the fighting c'. entre dois meses antes do comeco da guerra e a segunda semana de combates d. More than three months elapsed between when the bridge was built and when the first vehicles crossed it.</Paragraph>
      <Paragraph position="6"> d'. Passaram-se mais de tres meses entre quando a ponte foi construida e quando os primeiros automoveis a cruzaram. (31) *since in, *in since, *until throughout Obviously, the occurrence of more than one of these heads in a sequence is limited by e.g. (i) specific properties of the connectives, which prevent, for example, in normal circumstances, the sequences before after, or in X-TIME ago, and (ii) discourse simplicity, which imposes the avoidance of some theoretically possible forms if simpler equivalent ones are available. Note still that, given the appropriate conditions, these rather complex expressions may occur as the complement of an explicit temporal locating preposition, thus forming a particularly complex TLA: (32) A: I will give up smoking two weeks from now.</Paragraph>
      <Paragraph position="7"> B: I don't believe you. You will go on smoking until long after {that / ? two weeks from now}.</Paragraph>
    </Section>
  </Section>
class="xml-element"></Paper>