The perfect tense situations presented in Figure 25 usually arise when the speaker is giving a second-hand report. If Els promised the speaker yesterday that she would read the paper under discussion today, the speaker can utter example (397) at noon to report this promise. In this case it is irrelevant whether Els has actually read the paper, is still reading it, or will start reading it later in the day.

That the present perfect can also refer to an eventuality overlapping or following n is a direct consequence of our claim that Dutch does not express the binary feature [±posterior] within its verbal system. This finding opts for the binary tense theory over Reichenbachian approaches to the verbal tense system, since the latter have no means of expressing it and must therefore treat such cases as special or unexpected uses of the present perfect.

The choice between past and present perfect is often related to the temporal location of some other event. Consider the examples in (398): the present tense in example (398a) requires the examination to be part of the present-tense interval (and indeed strongly suggests that it will take place in the non-actualized part of that interval), whereas (398b) strongly suggests that the examination is part of the past-tense interval preceding speech time n.

Similarly, an example such as (399a) can be used to inform the addressee that the window in question is still open at speech time n, whereas (399b) does not have this implication, but can be used e.g. in a story about a burglary that happened in some past-tense interval.

The interpretation of example (397) can be limited by pragmatic considerations. In the context given above, the split-off point of the possible worlds precedes present-tense interval i and thus also speech time n. However, if the speaker is able to monitor Els’ actions during the actualized part of the present-tense interval ia, then the split-off point of the possible worlds coincides with n, and in this case example (397) would normally be used to refer to the situation shown in Figure 28, in which eventuality k precedes n; cf. Verkuyl (2008).

That k usually precedes n in the situation sketched above is illustrated in (400a). Recall that 1.5.4.1, sub II, referred to this preferred reading of (400a) to explain why the present in (400b) cannot normally be used to refer to an event preceding n.

Examples such as (401a), in which the completion of eventuality k can be located in the non-actualized part i◊ of the present, and indeed must be located there if the speaker knows that Els has not yet finished the paper at speech time n, may help us better understand how the more restricted interpretation in Figure 28 arises. As will be discussed in more detail in Subsection III, temporal adverbial phrases can modify the precise location of eventuality k within interval j; the temporal adverbial phrase om drie uur indicates that the completion of the eventuality of Els reading the speaker’s paper will take place before 3 p.m.; cf. also Janssen (1989). The reason why example (401b) does not normally refer to eventualities following n in the situation sketched in Figure 28 is that the relevant point in time at which eventuality k must be completed is taken by default to be speech time n; that this is indeed the default reading can be seen from the fact that making this point in time explicit, e.g. by adding the adverb nunow, can only be done if the speaker intends to emphasize that the relevant evaluation time is indeed the speech time.

Although an account along these lines seems plausible, the examples in (402) show that it cannot be the whole story. In these examples, the adverb vandaagtoday again modifies j and the adverbial phrase tot drie uuruntil 3 p.m. restricts the location of eventuality k to some subinterval of j preceding 3 p.m. However, in situations in which the speaker is able to monitor eventuality k, present-perfect examples such as (402a) are normally used when k is completed before speech time n (i.e. it would be odd to use this example at 1 p.m.), whereas simple present examples such as (402b) are normally used when k is completed after n.

The fact that (402a) cannot have a future interpretation suggests that something is missing. We return to this in Subsection IIIB, which aims to fill this gap by showing that Aktionsart also restricts the temporal interpretation of the perfect tenses.

As in the case of the simple tenses, the temporal interpretation of the perfect tenses can be restricted by adverbial modification. However, it seems that the situation is somewhat more complex, since also Aktionsart can restrict the interpretation of the perfect tenses: more specifically, atelic predicates differ from telic ones in that they allow a future interpretation of the perfect only under very strict conditions.

So far, we have tacitly assumed that the eventuality denoted by the lexical projection of the main verb occurs only once. Although this may be the default interpretation, the examples in (416) show that this is not necessary: example (416a) expresses that in the actualized part of the present-tense interval i, denoted by vandaagtoday, the speaker has eaten three times before speech time n. Similarly, in (416b) the frequency adverb vaakoften expresses that within the actualized part of the tense interval i, denoted by the adverbial phrase dit jaarthis year, there have been many occurrences of the eventuality denoted by the phrase naar de bioscoop gaango to the movies.

As expected, the default interpretation of examples such as (416) is that the eventualities precede speech time n. However, this default reading can easily be overridden. An example such as Als ik vanavond naar bed ga, heb ik drie maaltijden gegeten: ontbijt, lunch en avondetenWhen I go to bed tonight, I will have eaten three meals: breakfast, lunch and supper can easily be uttered at dawn or at noon by someone with an eating disorder who wants to express his good intentions.

The fact that the present/past-tense interval can contain multiple occurrences of the eventuality denoted by the lexical projection of the main verb is fully exploited in habitual constructions such as (417). These examples differ from the simple present examples in (385) in that they tend to locate the habit in the actualized part of the present-tense interval ia; for example, there is a strong tendency to interpret example (417b) to mean that Jan has quit smoking. However, it is certainly not necessary to interpret perfect habituals in this way, as can be seen from the fact that example (417a) can easily be followed by ... en hij zal dat wel blijven doen... and he will continue to do so.

Unlike their present-tense counterparts in (395), it does not seem possible to interpret the perfect-tense examples in (418) generically: the examples in (418a&b) are acceptable only if the subject refers to a (set of) unidentified true gent(s); example (418c) can at best lead to the interpretation that a certain whale has become a fish (which is an impossibility in our world).

Present-perfect clauses introduced by alswhen seem to allow both conditional and hypothetical readings, just like the simple-present examples in (387) from Section 1.5.4.1. The conditional reading illustrated in (419) is again the default. This example is ambiguous in that a teacher could say this sentence either to his students in general, to indicate that those who have fulfilled the condition expressed by the antecedent of the sentence may go, or to a particular student when he does not know whether that student has fulfilled the condition.

The hypothetical reading of this sentence arises when the discourse participants know that the antecedent is not fulfilled in the actualized part of the present-tense interval, e.g. when the teacher addresses a particular student of whom he knows that he has not yet put his things away; cf. the gloss and rendering of (420).

The fact that contextual information is needed to distinguish between the two readings of the antecedent clause Als je je spullen opgeruimd hebt, mag je weg clearly shows that pragmatics is involved. However, it is possible to favor one reading by using an adverbial phrase. As in the corresponding present-tense examples, the conditional reading in (419) is favored by adding an adverb such as altijdalways to the consequence: Als je je spullen opgeruimd hebt, mag je altijd wegIf you have put away your things, you can always go. The same applies to the addition of alalready to the antecedent, since this locates the eventuality denoted by the lexical projection of the main verb of the antecedent clause in the actualized part of the present-tense interval and thus blocks the hypothetical reading: Als je je spullen al opgeruimd hebt, mag je wegIf you have put your things away already, you can go. The addition of strakslater to the antecedent, on the other hand, favors the hypothetical reading, as it suggests that the speaker knows that the condition is not yet fulfilled at speech time: Als je straks je spullen opgeruimd hebt, mag je wegIf you have put your things away later, then you can go.

Past-perfect utterances allow both conditional and counterfactual readings, like the simple-past examples in (390) from Section 1.5.4.1. The default conditional reading can be found in (421a), which refers to some general rule that was valid in the relevant past-tense interval. The conditional reading is not easy to get when the pronoun je is interpreted referentially, as in (421b), which instead seems to prefer a counterfactual interpretation. This preference is again be pragmatic in nature; since the eventuality is located in the past-tense interval, the speaker and the addressee can be expected to know whether the addressee has fulfilled the condition mentioned in the antecedent.

It is important to note that the use of the simple past of the verb mogento be allowed in the consequence does not necessarily imply that the leaving event denoted by the lexical projection of the main verb in the consequence is located before speech time n. In fact, the preferred interpretation of counterfactuals of the form in (421b) is that in possible worlds in which the condition mentioned in the antecedent is fulfilled, the leaving event would coincide with or follow speech time n. This will be clear from the fact that the use of the adverb gisterenyesterday is not possible in (422a). This again shows that the past-tense interval can include speech time n and thus overlap with the present-tense interval; cf. the discussion in Section 1.5.1, sub IC. For completeness’ sake, note that this restriction on adverbial modification is lifted when the consequence is in the perfect tense, as in (422b).

An interesting fact about conditionals and hypotheticals is that the als-phrase alternates with constructions without als, in which the finite verb is in the first position of the clause: the antecedent in (422) can also take the form Had je je spullen opgeruimd, dan ... In the case of antecedent clauses of this form, counterfactuals are often used to express regret (“I wish I had put away...”) or a wish (“I wish I/he had put away ...”); for obvious reasons, the regret reading is more likely to occur when the counterfactual involves the speaker himself. For completeness, note that the consequences (here: dan had ik/hij weg gemogen) can easily be left implicit: cf. Had ik mijn spullen maar opgeruimd ...!

If the counterfactual involves the addressee, as in (424), the resulting structure is easily construed as a reproach. However, the construction is peculiar in that it is not possible to express the subject of the antecedent overtly, which strongly suggests that we are formally dealing with an imperative; cf. also the discussion of examples (177) and (178) in Section 1.4.2, sub I.

The counterfactual examples in this subsection all have in common that the speaker/hearer can be assumed to know whether the condition given in the antecedent is fulfilled (or not), which makes the conditional reading of these examples uninformative: the speaker could simply have given (or denied) the addressee permission to leave. The counterfactual reading, on the other hand, is informative because the speaker is informing the addressee about the situation that would have occurred if he had fulfilled the condition expressed by the antecedent; Grice’s maxim of quantity therefore favors this interpretation. This shows that this maxim is involved in triggering different kinds of irrealis meanings of past-perfect constructions.

Like the simple past in example (395), the past perfect in (425) can be used to express that the nominal description een echte heera true gent is not applicable to a specific person. Again, imagine the situation of a pregnant woman getting on a bus. All the seats are taken, and no one seems to be willing to oblige her by offering her a seat. An elderly lady gets angry and utters (425) to the boy next to her. Knowing that he had no intention of giving up his seat, she implies that the description een echte heer does not apply to him.

This section has shown that, as in the case of the simple tenses, the default reading of the perfect tenses is that the time interval j in which the eventuality denoted by the lexical projection of the main verb must take place is identical to the complete present/past-tense interval i: the completion of the eventuality can take place before, during, or after speech time n/n'. In many cases, however, the interpretation is more restricted and can sometimes have non-temporal implications. This section has shown that this can easily be derived from the interaction between temporal information (tense and adverbial modification), modal information encoded in the sentence (the theory of possible worlds) and pragmatic information (Grice’s maxim of quantity).