The simple present expresses that eventuality k takes place during present-tense interval i. This can be represented as in Figure 7, where index j is taken by default to be synchronous to the present i of the speaker/hearer (j = i). The solid part of the line below k indicates that the preferred reading of an example such as Ik wandelI am walking is that eventuality k overlaps with speech time n.

In many languages, including Dutch, the implication that k is true at speech time n can easily be canceled by e.g. adverbial modification: the simple present example (268) with the adverbial phrase morgentomorrow can be used to refer to a future eventuality k.

Of course, this is to be expected from the definition of the present in (265a), which states that the present-tense interval i must include speech time n, but does not impose any restrictions on j or k. Although we will return to this issue briefly in Subsection 5, we will defer a more thorough discussion of it to Section 1.5.4, where we will show that this use of the simple present is a characteristic property of languages that do not express the future within the verbal tense system, but by other means, such as adverbials.

The default reading of the present perfect is that eventuality k takes place before speech time n, i.e. eventuality k is located in the actualized part of the present-tense interval ia (j = ia). In addition, the present perfect presents eventuality k as a discrete, bounded unit that is completed within time interval j and therefore cannot be continued after n; this is represented in Figure 8 by the short vertical line after the solid line below k.

The acceptability of a present-perfect construction such as Ik heb gisteren gewandeldI walked yesterday can now be fully understood: (i) the present-tense form hebhave of the auxiliary indicates that we are dealing with a present-tense interval i, i.e. a time interval including speech time n; (ii) the fact that the definition of the present in (265a) and the definition of the perfect in (267b) do not impose any restriction on the location of j with respect to n means that j can precede, follow or overlap with n; (iii) since j can precede n, the adverb gisterenyesterday is eligible as a modifier of j; (iv) the use of gisteren thus implies that eventually k is located in ia, the actualized part of present-tense interval i.

In many languages, including Dutch, the default interpretation that k takes place before speech time n can easily be canceled by e.g. adverbial modification: the present perfect example (269) with the adverb morgentomorrow can be used to refer to some future eventuality k. Again, this is to be expected, since neither the definition of present in (265a) nor the definition of perfect in (267b) imposes any restriction on the location of the present j of eventuality k with respect to n; Section 1.5.4 will return to this issue.

There is considerable debate in the literature as to whether perfect-tense constructions should be considered temporal or aspectual in nature. The position taken by individual linguists often depends on the specific tense and aspect theories they endorse. Since the characterization of the perfect tense in binary (and Reichenbachian) tense theory does not appeal to the internal temporal structure of the event, this allows us to adopt a non-aspectual view of the perfect tense. The non-aspectual view may also be supported by the fact that the use of the perfect tense does not affect the way in which the internal structuring of eventuality k is presented; rather, it is the interaction of the perfect tense and the Aktionsart (inner aspect) that should be held responsible; cf. Section 1.2.3 for a discussion of the different kinds of Aktionsart. This will become clear when we consider the contrast between the atelic (states and activities) and telic (accomplishments and achievements) eventualities in (270).

All examples in (270) present the eventualities as autonomous units which (in the default reading) are completed at or before speech time n. This does not imply, however, that the eventualities cannot be continued or resumed after n. In fact, this is quite natural in the case of atelic verbs: an example such as (270a) does not imply that Jan will not love Marie after speech time n, as is clear from the fact that it can easily be followed by ... en hij zal dat wel altijd blijven doenand he will probably continue to do so forever. Similarly, example (270b) does not imply that Jan will not continue to work on his thesis after speech time n, as is clear from the fact that (270b) can easily be followed by ... en hij zal daar vanmiddag mee doorgaan... and he will continue to do so in the afternoon. The telic events in (270c&d), on the other hand, do imply that the events have reached their implied endpoint and therefore cannot be continued after speech time n. The examples in (270) thus show that the internal temporal structure of the eventualities is not affected by the perfect tense, but is determined by the Aktionsart of the verbs/verbal projections in question. From this we conclude that the perfect is not aspectual in nature, but part of the tense system; cf. Verkuyl (2008:20-27) for a more detailed discussion.

The future expresses that eventuality k takes place after speech time n, i.e. eventuality k is located in the non-actualized part of the present-tense interval (j = i◊).

The implication of Figure 9 is that eventuality k cannot occur during ia, but example (271) shows that this implication can be easily canceled in languages like Dutch: the second sentence explicitly leaves open the possibility that Jan has already read the paper at speech time n. This will be one of the main reasons for denying that zullenwill functions as a future auxiliary in Dutch. We will return to this in Sections 1.5.2 and 1.5.4.

The interpretation of the future perfect is similar to that of the future, but differs in two ways: (i) it is not necessary that the eventuality k has started after n, and (ii) it is implied that the state of affairs is completed before the time span i◊ has come to an end.

The implication of Figure 10 is again that eventuality k cannot take place/be completed during ia, but example (272) shows that this implication can be easily canceled in languages like Dutch by e.g. adverbial modification. We will leave this non-future reading aside for the moment, but return to it in Sections 1.5.2 and 1.5.4.

The main difference between the future and the future perfect is that the former focuses on the progress of the eventuality (without considering its completion), whereas the latter focuses on the completion of the eventuality k in j. This difference is often subtle in the case of states and activities but transparent in the case of telic events. Whereas the future tense in example (273a) expresses that the melting process will start/take place after speech time n, the future perfect in (273b) simply expresses that the completion of the melting process will take place in some j located in i◊; the future perfect leaves entirely open whether the melting process started before, after, or at n.

In (274), similar examples are given with the accomplishment die brief schrijven: (274a) places the entire eventuality after the time n, while (274b) makes no claim about the starting point of the eventuality, but simply expresses that the eventuality will be completed after n (but within i◊).

For the moment, we will ignore the difference between future and future perfect with respect to the starting point of the state of affairs, but we will return to it in Section 1.5.2, where we will question the claim that zullen is a future auxiliary.

In the tense representations given in the previous subsections, we have distinguished between (i) the present i of the speaker/hearer and (ii) the present j of eventuality k. Although the latter is always included in the former, it is easy to show that the distinction must be made. This may not be so clear in examples such as (275a), where j seems to synchronize with the entire present-tense interval i of the speaker/hearer. However, adverbial phrases of time can cause j to synchronize with a subpart of i: the adverb vandaagtoday in (275b) refers to a subpart of i including n, while morgentomorrow in (275c) refers to a subpart of i located in i◊.

That it is the present j of k and not the present-tense interval i that is affected by adverbial modification is also clear from the fact that it is possible to have conjoined present-tense examples such as (276) in which the two adverbial phrases refer to distinct subdomains within i.

Furthermore, entailments are computed on the basis of j and not on the basis of the present-tense interval i. Example (275b), in which j synchronizes with a subpart of i that includes n, says nothing about the speaker’s whereabouts yesterday or tomorrow, even if these time intervals are construed as part of present-tense interval i. The fact that entailments are computed on the basis of j and not i is even clearer in example (275c), in which j synchronizes with (a subpart of) i◊; this example says nothing about the speaker’s whereabouts at speech time n, clearly showing that the state of affairs need not hold for the entire present-tense interval i. It is only in cases such as (275a), without a temporal modifier, that we (by default) conclude that the state of affairs holds for the entire present-tense interval i.

The simple past expresses that eventuality k occurs during past-tense interval i. This can be expressed as in Figure 12, in which the solid line below k refers to the time interval during which the eventuality occurs. The solid line below k indicates that the default reading of an example such as Ik wandeldeI was walking is that eventuality k takes place at n'.

By stating that j = i, Figure 12 also expresses that the simple past has no implications for the time before or after the relevant past-tense interval i: the eventuality k may or may not hold before/after i. Consequently, an example such as (281) says nothing about the speaker’s feelings before yesterday, or today and later. This implies that the simple past cannot shed any light on the question as to whether speech time n can be included in the past-tense interval i.

The default reading of the past perfect is that eventuality k takes place before n', i.e. k is located in the actualized past-tense interval ia (j = ia). In addition, the past perfect presents the eventuality as a discrete, bounded unit that is completed within the present j of k, i.e. that cannot be continued after n'; this is again represented by the short vertical line after the solid line below k. Since k precedes n' and n' precedes n, k also precedes n, which implies that examples of this type cannot shed any light on whether speech time n can be included in the past-tense interval i.

The future in the past expresses that the eventuality k takes place after n', i.e. k is located in the non-actualized part of the past-tense interval (j = i◊).

The future in the past examples in (282b&c) show that speech time n can be included in the past-tense interval. We have already seen above that this refutes the definition of past in (278b) and supports the revised definition in (280b).

The interpretation of the future perfect in the past is similar to that of the future in the past, but requires the state of affairs to be completed within time span i◊.

The difference between the future in the past and the future perfect in the past is parallel to the difference between the future and the future perfect discussed in Subsection A: in future in the past examples such as (283a) the focus is on the progress of the eventuality, which in its entirety is placed after n', whereas in future perfect in the past examples such as (283b) the focus is on the completion of the eventuality and no particular claim is made about the starting point of the event.

Similar examples with the achievement die brief schrijven are given in (284): the future in the past in (284a) locates the entire eventuality after n', whereas the future perfect in the past in (284b) does not seem to make any claim about the starting point of the eventuality, but simply expresses that the eventuality will be completed after n' (but within i◊).

The examples in (285) with the adverbial phrase morgentomorrow show that the future perfect in the past provides evidence for the claim that speech time n can be included in the past-tense interval. We have already seen that this refutes the definition of the past tense in (278b) and supports the revised definition in (280b).

The use of adverbial phrases of time in sentences with a past tense can introduce a so-called supratemporal ambiguity; cf. Verkuyl (2008:118-123). This ambiguity is especially visible when the adverbial phrase occupies the sentence-initial position, as in (288).

The two sentences in (288) have a regular “real event” interpretation in the sense that the sentence is about Marie’s departure or about Marie having black hair at the time indicated by the adverbial phrase; in such cases the adverbial phrase functions as a regular temporal modifier of the time interval j that includes eventuality k. However, there is also a supratemporal interpretation in which the eventuality itself plays no particular role apart from being the topic of discussion. In this interpretation, the speaker of (288a) expresses that his latest information about Marie’s departure goes back to five o’clock: “according to my information at five o’clock, the situation was such that Marie would be leaving”. Similarly, (288b) can be interpreted as a correction of an error signaled by the speaker, e.g. in a manuscript; “An hour ago I read that Marie has black hair (but now it is mentioned that Jan likes her auburn hair)”.

Past-tense clauses are compatible with future eventualities on a supratemporal reading. Consider a situation in which the speaker is discussing Els’ plans for some time interval after speech time n. He can then compare the information available at two different moments: sentence (289a), for example, compares the information the speaker had yesterday with the information he has just received. The first conjunct of (289a) also illustrates that past-tense clauses with a supratemporal reading are compatible with locating the eventuality k after speech time, as the speaker’s talk is located in the speaker’s future. Example (289b) shows that it is even possible to make the future location of k explicit by using a second adverbial phrase such as morgentomorrow, especially if the particle nog is added right after gisterenyesterday; cf. Boogaart & Janssen (2007) for similar cases.

The fact that the past tense in the first conjunct of the examples in (289) is compatible with locating the eventuality after speech time n, and that the adverbs gisterenyesterday and morgentomorrow can be used in a single clause is exceptional. That this is possible, however, is not surprising when we realize that speech time n can be included in the past-tense interval. As shown in Figure 16 for example (289b), the first conjuncts in sentences such as (289) give rise to completely coherent interpretations. The notation used aims at expressing that the adverbial phrase gisteren is a supratemporal modifier denoting a time interval including the virtual speech-time-in-the past n', whereas the adverbial phrase morgentomorrow functions as a regular temporal modifier of j.

The claim that speech time n can be included in the past-tense interval also has important consequences for the description of the so-called sequence of tense phenomenon, i.e. the fact that the tense of a dependent clause can be adapted to coincide with the past tense of the matrix clause. Sequence of tense is illustrated in (290): example (290a) is unacceptable if we interpret the adverb morgentomorrow as a temporal modifier of the eventuality k, while example (290b) is perfectly acceptable in that case.

The unacceptability of (290a) is usually taken to represent the normal case: past tense is incompatible with adverbial phrases such as morgen that locate the eventuality after speech time n, and therefore (290a) cannot be interpreted as a modifier of the then-present j of the eventuality k; on this view, the sequence-of-tense example in (290b) is unexpected and must therefore be regarded as a special case. If, on the other hand, we assume that speech time n can be included in the past-tense interval, the acceptability of (290b) is expected; the eventualities in the main clause and the embedded clause are both seen as belonging to past-tense interval i, which happens to include speech time n. The real problem on this view is the unacceptability of example (290a), since the system predicts that this example is also possible in the intended reading.

The claim that the past-tense interval can include speech time n can also explain the contrast between the two examples in (291). In (291a) the eventualities expressed by the embedded and the main clause are both considered to be part of the past-tense interval, and as a result we cannot determine from this example whether the speaker believes that Els is (still) pregnant at speech time n; this may or may not be the case. In (291b), on the other hand, the eventuality of Els being pregnant is presented as part of the present-tense domain, and the speaker therefore implies that Els is still expecting at speech time n; cf. Hornstein (1990: §4.1) for similar intuitions.

This contrast in interpretation can be seen in the examples in (292). Since sequence-of-tense constructions do not imply that the eventuality expressed by the embedded clause still exists at speech time n, the continuation in (292a) is perfectly natural; it is suggested that Els has given birth and hence is now a mother. In (292b), on the other hand, the continuation leads to a semantic anomaly because the use of the present tense in the embedded clause indicates that the speaker believes that Els is still pregnant.

A similar account can be given for the observation in Kiparsky & Kiparsky (1970:162-3), illustrated in (293), that for some speakers factive and non-factive constructions differ in that the former usually display optional sequence of tense, whereas the latter tend to have obligatory sequence of tense. Again, the reason for this is that the use of present tense suggests that the speaker believes that the eventuality expressed by the embedded clause holds at speech time n. We used a percentage sign in (293b) to indicate that some speakers seem to accept the use of present tense in non-factive constructions like this.

The discussion in the previous subsection has shown that the claim that the past-tense interval can include speech time n correctly predicts that sequence of tense is not required, and may even be impossible, if the right conditions are met. As we noted earlier in our discussion of (290), this in a sense reverses the traditional problem; it is not the sequence-of-tense example in (290b) that poses a problem, but the fact that the past tense in simple clauses such as (290a) blocks the use of adverbial phrases such as morgentomorrow that locate the eventuality after speech time n.

Note, however, that under certain conditions the past tense can actually be combined with adverbs such as morgen. This is the case, for instance, in the question in (294b) provided by Angeliek van Hout (p.c.). The two examples in (294) differ in their point of perspective: (294a) expresses that the speaker assumes, on the basis of his knowledge at speech time n, that the addressee will come tomorrow, whereas (294b) expresses that the speaker assumes this on the basis of his knowledge at virtual speech-time-in-the-past n'. Some speakers report that (294b) feels a bit more polite than (294a), which may be related to this difference in perspective; by using (294b) the speaker explicitly leaves open the possibility that his information is outdated and consequently that the conclusion he draws from it may be wrong, i.e. the question can be seen as a request for reaffirmation; cf. also Haans & De Hoop (2023), which comes to a similar conclusion (formulated in Reichenbachian terms).

Past tenses can also be combined with the adverb morgen in questions such as (295b). Again, the difference is one of perspective. In question (295a) the speaker is simply asking for some information available at n; he expects that there will be a visitor tomorrow and he wants to know who this visitor is. Example (295b) is used when the speaker is aware of the fact that he had information about the identity of the visitor at some virtual speech-time-in-the-past n', but is not entirely sure of the accuracy of his recollection (which is typically signaled by the string ook al weer).

Another example, taken from Boogaart & Janssen (2007:809), is given in (296). Example (296a) simply states the speaker’s intention to leave tomorrow, whereas example (296b) leaves open the possibility that there are reasons, not known at some virtual speech-time-in-the-past n', that may prevent the speaker’s intention to leave from being realized.

A final example that seems closely related to the one in (296b), also taken in slightly adapted form from Boogaart & Janssen, is given in (297b). Examples like this can be used as objections to a command or request by showing that it is inconsistent with some previous commitment or plan.

The examples in (294) to (297) show that there is no inherent prohibition on combining past tenses with adverbs such as morgentomorrow; they show that there is no need to build such a prohibition into tense theory. Of course, this still leaves the unacceptability of simple declarative clauses such as Jan kwam morgenJan came tomorrow, but Section 1.5.4 will solve this problem by arguing that this example is excluded not because it is semantically incoherent, but for pragmatic reasons: Grice’s maxim of quantity prefers the use of the simple present/future in cases like this.

The previous subsection has shown that it is possible to combine past tenses with adverbs referring to time intervals following speech time n. Similarly, it seems possible to combine present tenses with adverbs such as gisterenyesterday referring to time intervals preceding speech time n. Subsection A 2 has already discussed this for present-perfect constructions such as (298), and has shown that it is fully allowed by our definitions in (287); since neither the definition of present in (287a) nor the definition of perfect in (287c) imposes any restriction on the location of j (or k) with respect to n, the adverbial gisterenyesterday can be analyzed as an identifier of j on the assumption that the time interval referred to by gisteren is part of a larger present-tense interval i that includes speech time n.

For the same reason, we would expect it to be possible to combine adverbs such as gisteren with the simple present: the definition of present in (287a) imposes no restriction on the location of j (or k) with respect to n. This means that we expect examples such as (299c) to be possible alongside (299a&b). Although the examples in (299a&b) are certainly more common, examples such as (299c) occur regularly in speech and can be easily found on the internet.

The acceptability of examples such as (299c) is not surprising, and so needs no special explanation. The only thing we need to consider is why the frequency of such examples is relatively low: one reason that might come to mind is simply that examples of this kind are blocked by the perfect-tense example in (299b), because the latter is more precise in that it presents the eventuality as completed.

Present-tense examples such as (299c) are particularly common in narrative contexts as an alternative to the simple past, which is why this use of the simple present is often referred to as the historical present. The historical present is often said to result in a more vivid narrative style (Haeseryn et al. 1997:120), which is easily understood from the perspective of binary tense theory. First, it should be noted that the simple past is usually preferred to the present perfect in narrative contexts because it presents the story not as a series of completed eventualities but as a series of ongoing events. However, because the simple past presents the story from the perspective of a virtual speech-time-in-the-past, it maintains a certain distance between the events being discussed and the listener/reader. The vividness of the historical present results from the fact that the simple present removes this distance by presenting the story as part of the actual present-tense interval of the listener/reader.

The historical present has become an often-used convention in the narration of historical events, even if the events are more likely to be construed as part of a past-tense interval; cf. (300). Again, the purpose of using the historical present is to bridge the gap between the reader and the narrated events by presenting them as part of the reader’s present-tense interval.

This use of the historical present is therefore not very special from a grammatical point of view, since it only involves the pretence that n' = n, and we will therefore not elaborate on this use. The conclusion that we can draw from the above discussion is that the stylistic effect of the so-called historical present confirms our main claim that the choice between past and present tense is a matter of perspective.

This subsection concludes our discussion of the choice between present and past by showing that tense not only determines the perspective on the eventuality expressed by the lexical projection of the main verb, but also affects the interpretation of so-called non-rigid designators such as de minister-presidentthe Prime Minister; cf. Cremers (1980) and Janssen (1983). Non-rigid designators are noun phrases that do not have a fixed referent, but have referents that change over time: whereas the noun phrase de minister-president refers to Jan Peter Balkenende in the period July 2002–February 2010, it refers to Mark Rutte in the period October 2010–July 2024.

That the choice of tense can affect the interpretation of the noun phrase can be illustrated by the examples in (301). The interpretation of the present-tense example in (301a) depends on the actual speech time n; if uttered in 2008, it is an assertion about the then Prime Minister Balkenende, but if uttered in 2022, it is a declaration about Prime Minister Rutte. Similarly, the interpretation of the past-tense example in (301b) depends on the location of the virtual speech-time-in-the past n': in a discussion about the period 2002 to 2010 it will be interpreted as a declaration about Balkenende, but in a discussion about the period 2010 to 2024 it will be interpreted as a declaration about Rutte. Crucially, example (301b) need not be interpreted as an assertion about the person performing the function of Prime Minister at speech time n.

The examples in (301) show that present/past tense fixes the reference of non-rigid designators; we choose their reference at n/n'. Now consider the examples in (302), in which the index now on the noun phrase is used to indicate that the intended referent is the one who performs the function of Prime Minister at speech time n. The number sign indicates that example (302a) is not very felicitous if one wants to express that the current Prime Minister attended high school when he was young. This follows directly from the claim that the reference of non-rigid designators is determined by tense; the past tense indicates that the description de minister-president can only refer to the person who performs the function of Prime Minister at virtual speech-time-in-the-past n'. Example (302b), on the other hand, can felicitously express the intended meaning, since it simply presents the Prime Minister’s school days as part of the present-tense interval: the person referred to by the description de minister-president at speech time n is said to have attended high school during the actualized part of the present-tense interval.

Note that we are not claiming that it is impossible to interpret a non-rigid designator from the perspective of speech time n in past tense sentences, but this is only possible if the description happens to refer to the same person at n and n'. This is illustrated by the fact that the two examples in (303) are both perfectly acceptable.

The discussion above has shown that present/past tense not only determines the perspective from which the eventuality as a whole is viewed, but also affects the interpretation of noun phrases that function as non-rigid designators. Before concluding this subsection, we should mention that Cremers (1980:44) has claimed that the judgments on the above examples hold only when a non-rigid designator is used descriptively, because in certain contexts such noun phrases can also be used as proper names. For instance, an example such as (304b) can easily be used in a historical narrative to refer to Queen Beatrix, even if the story is told or written during the reign of King Willem-Alexander.

Since the previous subsection already mentioned that historical narratives often use the historical present, an alternative account for the interpretation in (304b) might be that this use of the present affects the interpretation of non-rigid designators; pretending that n' = n simply does not block the possibility of interpreting the non-rigid designator with respect to n'. We leave this issue to future research.

The previous subsections have shown that in Te Winkel/Verkuyl’s binary tense system the present and past tenses are structured in a completely parallel way. The present subtenses are located in a present-tense interval including speech time n, whereas the past subtenses are located in a past-tense interval including a contextually determined virtual speech-time-in-the-past n'. The subtenses locate the eventuality with respect to n/n'.