The absentive construction has certain semantic similarities to clauses with the aspectual verbs gaanto go and komento come; cf. Section 6.4.1. These verbs can be purely aspectual, but they can also be used in such a way that the lexical meaning of the corresponding main verbs gaan and komen remains active. In this case, aspectual gaan and komen express not only inchoative aspect, but also that the referent of the subject of the clause undergoes some change of location with respect to the deictic center, which is usually contextually determined or taken by default as the “here and now” of the speaker and/or the addressee. On the default interpretation, the examples in (230) express that Jan is going fishing in a place remote from c.q. near to the speaker/addressee to go fishing; the adverbs marked with a dollar sign require the context to provide additional information. See Section 6.4.1, sub I, for a more detailed discussion of the notion of deictic center.

The absentive construction is like the aspectual construction with gaan in that it expresses that the subject of the clause is not present in the implied deictic center; in its default reading, the absentive construction in (231a) expresses that Jan is not in the vicinity of the speaker/addressee, but out of reach. The term “out of reach” is essentially pragmatically determined; it often involves physical distance, but it may also involve other factors. For example, it is not customary to use (231a) when Marie is in a room adjacent to the room where the speaker is, but it is possible to use (231b) when the speaker is in a room adjacent to the bathroom. In the latter case, privacy conventions make Marie sufficiently out of reach of the speaker to justify the use of the absentive, although Haeseryn et al. (1997:1035) claims that even in this case the “physical distance” reading is most prominent.

The absentive construction also expresses that Jan is engaged in the activity denoted by the bare infinitive in a broad sense. The addition in a broad sense is needed to account for the aspectual difference between the absentive construction and the progressive aan het + infinitive construction discussed in Section 1.5.3, sub I.

Example (232a) differs from (232b) in that it does not imply that the eventuality of Jan fishing includes speech time; it covers a wider range of activities, including leaving the deictic center, traveling to the place where the activity denoted by the bare infinitive is taking place, performing the activity itself, and returning to the deictic center; as long as Jan is engaged in any one of these activities, sentence (232a) is considered true. That this is the case is clear from the fact that the speaker can begin to use (232a) at the moment when Jan has left the house (and is thus out of reach of the speaker). This is also shown by the fact that examples such as (233) are very common when the speaker wants to announce that he is leaving to do something (and thus will be out of reach of the addressee).

Note that the use of utterances such as (233) is otherwise very limited in speech for pragmatic reasons; the requirement that the referent of the construction has left the deictic center implies that the speaker and the addressee are not involved in face-to-face interaction. However, such expressions are very common in remote communication (e.g. telephone calls or written messages). In this respect, examples such as (233) differ considerably from perfect-tense constructions such as Ik ben wezen vissenI have been off fishing, which are very common in face-to-face interaction because they do not imply absence at speech time, but at some moment before; cf. Sassen (1977-8) for more discussion.

The discussion above has shown that the absentive construction is typically used when the subject of the clause (i) has left the deictic center, (ii) is out of reach of the speaker/addressee, and (iii) is involved in a broad sense in the activity denoted by the bare infinitive. De Groot further claims that the absentive implies that the subject will return to the deictic center after a certain time, the duration of which is predictable from pragmatic knowledge or previous experience. However, this is disputed in Haslinger (2007: §2), on the basis of the fixed expression in (234a). Given the idiomatic nature of the expression, it is not immediately clear how strong this argument is, so we will leave the issue open. However, Haslinger is right in claiming that the period of time during which the subject will be out of reach need not be predictable, since examples such as (234b) are perfectly natural.

Although De Groot may be overstating the relevance of pragmatic knowledge, he is certainly right to emphasize the relevance of the speaker’s and/or hearer’s world knowledge; the lexical projection of the bare infinitive typically denotes eventualities that have a typical setting or that are typical for the referent of the subject of the clause. If we take the deictic center to include Marie’s house, the speaker can use example (235a) to refer to a series of events beginning with Marie leaving the house, going to the post office, putting the letter in the mailbox, and returning home. And the use of example (235b) seems to imply that Marie is a regular soccer player.

The meaning of the absentive is non-compositional in the sense that it expresses the absence of the subject without there being any overt material that could be held responsible for this aspect of meaning. That this meaning aspect is really present is shown again by the question-answer pairs in (236), which show that the sentence Hij is vissenHe is off fishing can be used to answer not only the question What is Jan doing? in (236a), but also the question Where is Jan? in (236b).

The acceptability of the question-answer pair in (236b) may be due to the fact that the use of the absentive requires extensive knowledge of the subject’s routine, more specifically, that he does not normally perform this activity of fishing in the deictic center. Haslinger (2007: §2), however, claims that it is syntactically encoded, but we will not discuss this proposal here. The reason is that the proposal is crucially based on the implausible assumption that absentive constructions involve control, i.e. have the representation NPi be [PROi .... Vinf]; NPi cannot be an argument of zijn because it is an unaccusative verb and is therefore expected to be in a subject-raising construction: NPi be [ti .... Vinf]. See Subsections III and IV for relevant discussion.

De Groot (2000) and Haslinger (2007) show that there are a number of semantic constraints on the absentive construction. The most important ones are that the subject is agentive and that the eventuality has a certain duration. The bare infinitive is therefore usually an (in)transitive verb denoting an activity or an accomplishment.

Example (238a) shows that unaccusative verbs do not seem to be possible. This also explains De Groot’s observation, illustrated in (238b) (for which he provides a separate semantic account), that examples such as (237a) become infelicitous if we add a directional complementive, because this makes the movement verb unaccusative. Clear evidence for this is provided by perfect auxiliary selection, as in the contrast between Jan heeft/*is gewandeldJan has walked and Jan is/*heeft naar Groningen gewandeldJan has walked to Groningen; cf. Section 1.2.2, sub II.

Another constraint on the use of the absentive is that the eventuality denoted by the bare infinitive is seen as a discrete unit that is carried out in an uninterrupted way. This is clear from the fact that when the infinitive denotes an accomplishment, the referent of the subject is not expected to return to the deictic center until he has completed the eventuality. For example, the two examples in (239) have different implications for the period of time during which Marie will be absent; while in the case of (239a) this period may be short, example (239b) strongly suggests that Marie is taking a sabbatical in a distant place where she will be writing the book.

The nature of the semantic restrictions on the bare infinitives is sometimes unclear, as shown by the fact that De Groot and Haslinger sometimes have different acceptability judgments. We will not go into this question in detail, but leave it to future research, for which the studies of De Groot and Haslinger offer a good basis.

There is no clear evidence that the verb zijnto be in the absentive construction is predicative in nature, i.e. capable of taking arguments. The reason for this is that pronominalization in this case cannot be used as a test to determine whether we are dealing with a main verb. To see this, consider the copular constructions in (240), which show that the problem is that the pronouns het and dat pronominalize not only nominal arguments, but also adjectival and nominal predicates. It follows that we cannot conclude from the acceptability of Marie is dat ook\`1Jan is dat nog niet in the second conjunct of the examples in (240) that the copula zijn is a two-place predicate.

According to the so-called small clause analysis discussed in Section 2.2.4, the complementive and its logical subject are generated as a single phrase, and the surface structure is subsequently derived by raising the noun phrase from its base position into the subject position of the clause. In this analysis, the subject of the clause is an argument of the complementive, and the copular is only needed to express morphologically the tense (present) and agreement features (third-person singular).

Moreover, judgments about pronominalization in absentive constructions tend to vary from speaker to speaker and from construction to construction. Examples such as (242a) are given as grammatical in Haslinger (2007), and some of our informants indeed accept them, although others consider them to be marked. Examples such as (242b) are not discussed in Haslinger’s study, but are again accepted by some of our informants, while considered unacceptable by others.

The variability in judgments makes it difficult to draw clear conclusions about whether pronominalization is possible or not. It may be that speakers who reject pronominalization as marginal or unacceptable simply favor omitting the entire clause, as in the perfectly acceptable and natural examples in (243). Speakers who accept pronominalization in (242), on the other hand, may interpret the second conjunct as a kind of copular construction, with the pronoun functioning as a complementive.

The main finding of this subsection is that the pronominalization test cannot be used in complementive constructions to determine the adicity of the verb, because pronominalization of predicative phrases is also possible. In addition, judgments on the pronominalization of phrases headed by a bare infinitive are somewhat unclear.

The discussion in Subsection III has made it clear that there is no clear evidence for the claim that zijn in the absentive construction is an n-place predicate, i.e. an argument taking verb. This leaves two syntactic analyses: the verb zijn is as a regular copular verb and the bare infinitive is the head of a nominalization that functions as a complementive, as in (244a), or zijn is as a non-main verb combined with the projection of an infinitival main verb, as in (244b).

Semantic considerations seem to favor the analysis in (244b), because bare-inf nominalizations such as een boek kopenbuying a book do not normally denote properties that can be attributed to the referent of the subject of a copular construction (although Haslinger, 2007:41, explicitly claims that phrases such as een boek kopen do denote properties in absentive constructions). Unfortunately, however, the syntactic tests for establishing the categorial status of the infinitive do not yield straightforward results: some support the copular construction analysis in (244a), while others support the non-main verb analysis in (244b).

The two analyses in (244) make different predictions when it comes to word order. If we are dealing with a copular construction, the putative bare-inf nominalization functions as a complementive and is therefore expected to precede the copular in clause-final position; if we are dealing with a construction in which zijn functions as a non-main verb, the bare infinitive must be a main verb and is therefore expected to be able to follow zijn in clause-final position. Haslinger (2007: §2) claims that the infinitive must precede the verb zijn and that the copular construction analysis is therefore the correct one.

It is indeed the case that many of our informants prefer the order vissen is in (245a), but the alternative order is vissen is also accepted by at least some of these informants, which is why we have marked this order with a percentage sign; we also found the order zijn–infinitive on the internet for the intransitive verbs fietsento cycle, logerento stay, sportento do sport, wandelento walk, werkento work, and winkelento shop. Haslinger notes the same in footnote 48 on page 63 but nevertheless claims that the order zijn—infinitive is ungrammatical because speakers who accept that order in examples such as (245a) reject it when the infinitival phrase is more extensive, as illustrated in (245b). However, it seems that the relevant order is much improved when the projection of the infinitive is split, i.e. when the infinitive and its object are separated by the verb zijn, as in (246a). Examples (246b&c) illustrate the same thing for cases found on the internet with the search strings [boodschappen is doen] and [de hond is uitlaten].

It should be noted here that performing a Google search does not reveal much because embedded absentive constructions are rare and difficult to find due to the intervention of other construction types (such as infinitival nominalizations in the sentence-initial position followed by is in second position, e.g. De hond uitlaten is leukWalking the dog is fun). We therefore need a more advanced search in a very large corpus before we can say more about the relative frequency of the object–zijn–infinitive and the object-infinitive-zijn order.

Even if we interpret the fact that many speakers prefer the order infinitive–zijn in examples such as (245) as evidence for the copular construction analysis, we cannot adopt it unreservedly, since it wrongly predicts that the infinitive would also have to precede the clause-final verb clusters in the corresponding perfect-tense constructions, as in (247).

Examples such as (247) are highly dubious, the normal perfect-tense forms are those given in (248); while the string [is wezen vissen] is very common, we have not found any cases on the internet with the string [vissen is geweest] that allow an interpretation as an absentive construction; all other cases involved the progressive forms (e.g. dat Jan aan het vissen is geweest that Jan has been fishing) or forms with the particle uit (e.g. dat Jan uit vissen is geweest), which will be discussed in the next subsection. The fact that the bare infinitives in (248) follow the other verbs in clause-final position militates strongly against analyzing them as nominal complementives. Haeseryn et al. (1997:1033) provides similar judgments and add that examples such as (248) are typically found in the western part of the Netherlands.

The contrast in acceptability between (247) and (248) thus favors the non-main verb analysis of the absentive construction. This analysis is also supported by the fact that these examples exhibit the IPP-effect: wezen has the form of an infinitive which cannot be replaced by the participle form geweestbeen.

A possible problem for the non-main verb analysis is Haslinger’s claim that the sequence wezen + infinitive in (248) can also precede the verb zijn, as shown in (249). This may again open the possibility to analyze the italicized parts as bare-inf nominalizations with the function of complementive.

However, a Google search (November 17, 2023) revealed that the orders in (249) hardly occur. The search for the string is wezen vissen returned about 20 results (not all cases where clear) in embedded context, but only one (Flemish) case of wezen vissen is. We found more or less the same result with the common collocation boodschappen doento shop: the search for the string boodschappen is wezen doen returned 16 results in embedded context, but none for the string boodschappen wezen doen is. This shows that the orders in (249) are not the ones found in standard Dutch (spoken in the Netherlands).

The discussion above strongly suggests that in constructions with more than two verbs the infinitive must be final in the verb cluster, and thus separated from its dependents (if present). This again argues against the copular construction analysis: on this analysis the infinitive heads a bare-inf nominalization, which is impermeable to external elements, such as the verb zijn. Furthermore, example (251b) shows that bare-inf nominalizations are islands for extraction, which would again incorrectly prohibit the clause splitting in (248b).

It is relevant to mention here that Haslinger claims that the object of a transitive verb can be wh-moved, as in (252), because if this is true, it would again show that the copular construction analysis is wrong, and thus support the alternative non-main verb analysis. However, as the percentage sign in (252b) indicates, the judgments are not very clear, since some speakers feel uncomfortable with examples of this kind.

The discussion above has shown that the copular construction analysis is supported by the fact that many speakers prefer the order infinitive–zijn in imperfect examples such as (245). However, the analysis runs into problems in perfect constructions: the copular construction analysis predicts that the bare infinitive should precede zijn, but the reverse order seems to be the one that is actually used. The copular construction analysis also implies that the infinitive functions as the head of a bare-inf nominalization, and thus incorrectly predicts that the infinitive cannot be separated from its dependents. Since the non-main verb analysis does not prohibit the infinitive–zijn order in (245) and is fully consistent with the other facts discussed in this subsection, it should be given more credit than it has received so far. Future research needs to clarify the various uncertainties raised in our discussion.

This subsection concludes by briefly comparing absentive constructions such as Jan is vissenJan is off fishing with constructions such as (253a), in which the absentive meaning is overtly expressed by the particle uitout; cf. Paardekooper (1986:136) and Haslinger (2007: §2.6). We begin by showing that the uit vissen zijn construction is a more special form of the ordinary copular construction in (253b). The discussion will be relatively brief, since we will see that the uit vissen zijn construction is very similar to the uit vissen gaan construction discussed in Section 6.4.1, sub VI.

The claim that the sequence uit vissen in (253a) has the same function as the particle uit in (253b) implies that it behaves like a constituent. That this is indeed the case is clear from the fact, illustrated in (254a), that it can occur in the sentence-initial position. Example (254b) further shows that the sequence cannot be interrupted by the clause-final verb, which in fact also supports our proposal by showing that the infinitive cannot be construed as part of the verbal complex.

Finally, the word order of the phrase uit vissen is fixed: the examples in (255) show that the infinitive vissen cannot precede the particle uit or be separated from it by wh-movement; cf. also Haeseryn et al. (1997:1054).

For these reasons we will analyze the constituent uit + infinitive as suggested in Section 6.4.1, sub VI, according to which the bare infinitive functions as a modifier of the adposition uit. This may also explain why infinitives in the uit vissen zijn construction usually refer to recreational activities. A number of typical examples are given in (256), although there are also cases such as uit werken zijnto be out working and somewhat obsolete expressions such as uit wassen zijnto be out washing, both of which have to do with performing domestic tasks in other people’s homes.

The claim that the constituent uit + infinitive behaves syntactically like a complementive can be supported by the fact, illustrated by the (a)-examples in (257), that it must precede the clause-final verbs and does not trigger the IPP-effect. the absentive constructions in the (b)-examples were discussed earlier and added here for comparison.

We have repeated the absentive construction in the (b)-examples here because the contrasts with the corresponding (a)-examples highlight the fact that they do not exhibit the behavior that we might expect from complementive constructions, and thus strengthen the conclusion from the previous subsection that the absentive construction is not a copular construction.

Finally, for the sake of completeness, note that the bare infinitive accompanying uit does not usually combine with an object or modifier. Haeseryn et al. (1997:1054) gives (258a) as a possible exception, but we have added a number sign to this example because the status of this example is unclear to us. Our informants consider this example to be marked and tend to accept this string only under the irrelevant reading, in which vlinders vangen functions as an afterthought. That (258a) is unacceptable in the intended reading is supported by the fact that its embedded counterpart in (258b) is considerably worse than (258a), which can be attributed to the fact that the placement of vlinders vangen before the clause-final verb precludes the interpretation of this phrase as an afterthought; cf. the perfectly acceptable dat meneer Prikkebeen uit is, vlinders vangenthat Master Prickly Leg is out, catching butterflies.