En esta aportación presento una propuesta metodológica que consiste básicamente en la utilización de un repertorio de películas clásicas y contemporáneas como recurso di-
1. Consideraciones previas: un espíritu libre no debe aprender como esclavo
It has been argued on a number of occasions that there is a relation between the presence of the infinitival marker te in infinitives, and the presence of a TP (tense) projection (see e.g. Bennis & Hoekstra 1989b). If the infinitive lacks te, it lacks a TP, and hence an independent time reference. In such cases, the time reference of the infinitive depends on the tense in the matrix clause. This can be illustrated with the help of a perception verb such as zien (‘see’), as in (27a-b):
(27) a. Ron zag [Hermelien dansen].
Ron saw Hermione dance-INF
‘Ron saw Hermione dance.’
b. Ron ziet [Hermelien dansen].
Ron sees Hermione dance-INF
‘Ron sees Hermione dance.’
In both (27a) and (27b), Hermione’s dancing is necessarily simultaneous with Ron seeing Hermione dance.
In a te-infinitive, on the other hand, the temporal reference of the infinitive is not necessarily identical to that of the matrix clause:
(28) Harry zegt [Hermelien te bedanken].
Harry says Hermione to thank-INF
‘Harry says that he will thank Hermione.’
In (28), the event of Harry saying something does not necessarily coincide with him thanking Hermione.
This observation begs the question whether all te-infinitive are tensed, and, if not, how tensed te-infinitives can be distinguished from their tenseless counterparts. In Cremers (1983), it is argued that some te-infinitives are timeless to the extent that they lack independent time reference. More specifically, Cremers argues that there are two types of te-infinitives, which differ from each other in terms of the type of complements that they select:
(29) a. verbs taking a VP-complement
(e.g. proberen ‘try’, durven ‘dare’, dwingen ‘force’, weigeren ‘refuse’)
b. verbs taking a CP (i.e. sentential) complement
(e.g. zeggen ‘say’, denken ‘think’, beweren ‘claim’, beseffen ‘realise’) The VP-complements in (29a) lack an independent time reference, so that a verb like
proberen will form a temporal unit with its infinitival complement. According to
Cremers, the two types of te-infinitives behave differently with respect to the future auxiliary zullen (‘will’). The examples in (30a,b), taken from Cremers (1983:181), indicate that VP-complements cannot contain this auxiliary, while CP-complements can:
(30) a. * Jacoba probeert jou te zullen bezoeken. (VP-complement)
Jacoba tries you to will visit
b. Jacoba zegt jou te zullen bezoeken. (CP-complement)
Jacoba says you to will visit
‘Jacoba says that she will visit you.’
However, the problem with this test is that the presence of zullen does not correlate with the presence of an independent tense domain, nor does its absence correlate
with the absence of such a domain.15 While (30a) shows that VP-complements do
not allow zullen, they can, as IJbema (2002:103) notes, be independently modified by a temporal adverb. Consider (31a,b), where the VP-complement in (31b) has a time reference that is different from that of the matrix clause:
(31) a. Jan weigert een boek te (*zullen) kopen.
John refuses a book to will buy
‘John refuses to buy a book.’
b. Gisteren weigerde Jan nog volgend jaar op vakantie te gaan.
yesterday refused John still next year on holiday to go ‘Yesterday John refused to go on a holiday next year.’
However, IJbema (2002:106) notes that this temporal adverbial in a VP-complement can refer only to the future, and not to the past:
15As Barbiers (1995) and others have argued, zullen is most appropriately viewed as a modal rather than
as a temporal auxiliary. For instance, constructions with zullen allow both a deontic and an epistemic modal reading:
(i) Deontic: (ii) Epistemic:
Je zult je kamer opruimen. Jan zal nu wel in Amerika zijn. you will your room clean John will now surely in America be ‘You must clean your room.’ ‘John must be in America by now.’
(32) * Vandaag weigert Jan het boek gisteren te hebben gekocht.
today refuses John the book yesterday to have bought
‘Today John refuses to have bought the book yesterday.’
VP-complements differ from CP-complements in this respect. As (33a,b) show, the latter do allow time adverbials that refer to both future and past time:
(33) a. Gisteren beweerde Jan nog volgend jaar op vakantie te gaan
yesterday claimed John still next year on holiday to go ‘Yesterday John claimed to go on holiday next year.’
b. Vandaag beweert Jan haar gisteren te hebben ontmoet.
today claims John her yesterday to have met
‘Today John claims to have met her yesterday.’
The future time reference found in VP-complements is not a tense specification to the extent that it relates the time of an event to the time of an utterance. Rather, it signals that the event is not realised. This observation leads IJbema (2002:108) to replace Cremers’s classification by a classification in terms of “irrealis” vs. “realis” complements. IJbema’s classification is given in (34):
(34) a. verbs selecting an irrealis complement
(e.g. proberen ‘try’, beloven ‘promise’, besluiten ‘decide’, denken ‘think’)
b. verbs selecting a realis complement
(e.g. beseffen ‘realise’, beweren ‘claim’, meedelen ‘announce’, zeggen ‘say’)
The verbs in (34a) select an irrealis complement, which can refer to future time only. The verbs in (34b) select a realis complement, which can have an independent time reference. Thus, a realis complement can refer to an event that temporally precedes or follows the event described in the matrix clause.
As regards the theoretical interpretation of these observations, IJbema (2002:109) gives the classification in (34) the following syntactic implementation:
(35) a. irrealis infinitival complements [VP besluiten [Mood irrealis te [… [VP V ]]]]
b. realis infinitival complements
[VP zeggen [T(past) te [Mood irrealis [… [VP V ]]]]]
In the structure in (35a) the infinitival marker te is projected in the Moodirrealis node,
given that irrealis complements lack a T(past) projection. Since, as IJbema assumes, future reference is not an instantiation of tense, this accounts for the observation that irrealis complements allow future reference only. The structure in (35b) shows that
realis complements do project a T(past) node, since these can have past reference; te is projected in the head of this functional projection.
IJbema’s classification and syntactic implementation are not unproblematic. First of all, it is doubtful whether the occurrence of a temporal adverb such as gisteren is a suitable diagnostic for determining the presence of a T(past) projection. As IJbema (2002:102) acknowledges, gisteren can also appear as a modifier in the nominal domain, as is illustrated in (36), in which gisteren is part of an attributive AP, which is again part of the DP:
(36) Het gisteren nog zwijgzame elftal staat vandaag de pers te woord.
the yesterday still silent squad stands today the press to word
‘The squad, which were silent yesterday, are speaking with the press today.’ If gisteren is used as a diagnostic for the presence of TP, then we would be forced to
extend the presence of a TP projections to include nominal and adjectival domains.16
The classification in (34) is also problematic on empirical grounds. IJbema makes a distinction between irrealis and realis complements on the basis of the possibility of independent past-time reference, using the gisteren diagnostic. This predicts that verbs such as besluiten (‘decide’) and denken (‘think’), which IJbema classifies as irrealis complement taking verbs, cannot take a te-infinitive that contains gisteren. This prediction is not borne out, as the following examples show:
(37) a. Na lang nadenken besloot Jan zich gisteren te hebben vergist.
after long think decided John himself yesterday to have erred
After long consideration, John decided he had been mistaken yesterday.’
b. Na uren studeren dacht Jan zich gisteren te hebben vergist.
after hours studying thought John himself yesterday to have erred
‘After hours of studying, John thought he had been mistaken yesterday.’ This suggests that the possibility of gisteren with a te-infinitive is not indicative of realis status — or, in terms of IJbema’s analysis, of the presence of a TP projection.