The verb laten is polysemous: it can be causative “to make” or permissive “to let”. When the subject of laten refers to a person, as in (729), we are usually dealing with a causer, i.e. an agent capable of performing an unspecified action with a specific effect. In the causative reading of the examples in (729), the action performed by the causer causes the eventuality referred to by the infinitival clause to occur. In the permissive reading, the causer refrains from performing an action that could have prevented that eventuality from occurring. Cases such as (729) are sometimes called indirect causation; cf. Haeseryn et al. (1997:1015ff).

When the subject of laten is inanimate, as in (730), we are usually dealing with a cause: such subjects do not perform some unspecified action, but have as their immediate effect that the eventuality referred to by the bare infinitival clause occurs; we are dealing with direct causation.

Causer and cause subjects differ in that the permissive reading is generally only possible with the former. The contrast can be illustrated by the examples in (731); whereas (731a) normally has a permissive reading, example (731b) can only be interpreted as causative.

Having a causer subject usually implies that the subject is capable of consciously affecting the eventuality expressed by the bare infinitival clause. This may account for the contrast between the examples in (732). Under normal circumstances, the psychological state of longing for a vacation cannot easily be deliberately induced in a person. However, it is quite normal for such a state to be triggered by something. Note that it is easy to remove the markedness of (732a) by adding an adverbial cause-PP: cf. Jan laat Marie met zijn verhalen naar vakantie verlangenJan makes Marie long for a holiday with his stories, in which it is claimed that Jan’s stories are the trigger for the emergence of the psychological state of longing for a holiday in Marie.

The examples in (733) show that AcI-constructions with doen are mostly used to express direct causation, although Haeseryn et al. (1997) notes that speakers from Belgium are often more permissive here than speakers from the Netherlands.

Doen as a direct causation verb usually has a cause and not a causer subject. This is illustrated by the following pair from Haeseryn et al. (1997:1017); example (734a) expresses that the subject of the sentence triggers certain memories of the speaker’s brother, while (734b) expresses that the psychiatrist consciously tries to make the speaker think of his brother (e.g. as part of a therapy).

As a result of this semantic difference between AcI-constructions with laten and doen, we need not be surprised that the frequency of causative doen is much lower than that of causative laten. However, it may also be due to the fact that doen is mainly found in more or less fixed expressions and in the more formal register; the idiomatic examples in (735) are selected from the list given in the digital Van Dale dictionary Dutch-English; for cases from the formal register, we refer the reader to the discussion in Haeseryn et al. (1997:1015ff).

It is not clear a priori whether the polysemy of laten, i.e. the ambiguity between the causative and the permissive reading, indicated in the glosses by to make and to let, justifies the postulation of two different verbs laten, or whether we are simply dealing with a single verb with different context-dependent readings. The first option seems difficult to substantiate, since the behavior of laten does not seem to be affected by the specific reading associated with it. A possible difference is related to the fact, illustrated in (736), that the subject of transitive bare infinitivals can normally be left implicit. At first glance, this seems to strongly favor the causative reading.

However, it is not clear what this proves. For one thing, it could simply be the case that this preference for the causative reading in (736b) is a byproduct of the fact that the causative reading focuses more on the actualization of the eventuality denoted by the bare infinitival verb (here: the walls being painted) than on the question of who performs that eventuality. The permissive reading, on the other hand, by its very nature (giving permission/not hampering someone) focuses on the agent(s) involved in this eventuality. Moreover, since the speaker will not normally make someone steal his favorite book, examples such as (737b) show that subjects of bare infinitivals can sometimes also be left implicit in permissive constructions, again with the effect that the focus of the construction is on the actualization of a specific state of affairs (here: the speaker’s favorite book being stolen).

The causative/permissive verbs latento make/let and doento make behave like perception verbs in that they can occur in AcI-constructions. As the examples in (738) show, this means that laten and doen are argument taking verbs; they can add a causer/cause argument to those selected by the embedded main verb, viz. the subject of the main clause (here Marie and de zonthe sun). This shows that we are dealing with a main verb according to our definition.

However, the examples in (739) show that laten and doen differ from the perception verbs in that their bare infinitival complement cannot be pronominalized. In this respect, laten and doen behave more like typical non-main verbs like the aspectual verb gaan: cf. Jan gaat wandelen versus *Jan gaat dat. The number sign “#” in (739b) indicates that examples of this kind are perfectly acceptable in contexts where the verb doen can be rendered by English to do, as in Jan doet datJan does that.

Note in passing that Dutch has the imperative form Laat dat!Stop that/Do not do that!. The verb laten in this idiomatic form is not causative/permissive but rather obstructive, does not syntactically select an obstructor (the speaker is contextually defined as such), and does not allow a bare infinitival complement; cf. *Laat dat liedje zingen! (intended meaning: “Do not sing that song!”).

The examples in (740) show that laten and doen-constructions exhibit monoclausal behavior: the primeless examples show that the bare infinitives are part of the verbal complex and can be separated from their arguments, and the primed examples show that these constructions exhibit the IPP-effect in the perfect tense.

The question remains whether causative/permissive latento make/let and causative doento make can take a bare-inf nominalization as their complement. Section 5.2.3.3, sub III, has shown that a phrase headed by a bare infinitive with an overt subject cannot be analyzed as a bare-inf nominalization, for the simple reason that the subject of the input verbs of such nominalizations is either left implicit or expressed by a van or a door-PP. This leaves us with those constructions in which the subject is implied. Analyzing such constructions as involving bare-inf nominalizations seems a priori implausible, because laten and doen normally do not allow nominal complements at all, which was already shown in Subsection II by the unacceptability of pronominalization in the examples in (739). That we are not dealing with bare-inf nominalizations in such cases is also clear from the fact that the bare infinitives cannot precede the clause-final verb sequences in examples such as (741), regardless of whether the subject is overtly realized or left implicit.

Since the verbs laten and doen are not able to take a nominal complement, it seems that we can a priori exclude the possibility that the direct object Jan/hem in (742a) is an internal argument of laten; we can therefore safely conclude that it functions as the subject of the bare infinitival clause. The subject of the bare infinitival clause is generally taken to be assigned accusative case by the causative verb. That the case in question is accusative is difficult to establish on the basis of the Dutch example in (742a), but it may be supported by the fact that this case is used in its German translation in (742b), taken from Drosdowski (1995:739).

Since there is no case assigner in the embedded infinitival clause, it seems inevitable to attribute case assignment to the verb laten, but again there is little independent evidence for this. One way to establish this would be through passivization: the fact that the accusative subject of the infinitival clause in the English example in (743a) becomes the subject of the matrix clause in the corresponding passive construction in (743b) can be taken as evidence for “exceptional case marking” of the subject of the infinitival clause by the matrix verb to expect.

However, this kind of evidence is not available for Dutch AcI-constructions, because passivization of such examples is always impossible. The (a)-examples in (744) show this for a construction in which the bare infinitive is monadic (i.e. intransitive or unaccusative), and the (b)-examples for a construction in which it is transitive; cf. Section 5.2.3.3, sub IVB, for a more detailed discussion of the impossibility of passivization in AcI-constructions.

Since the examples in (745) show that the verb laten can be passivized when it takes a complementive, the unacceptability of the primed examples in (744) remains puzzling; cf. Bennis & Hoekstra (1989b) and Petter (1998: §4) for two different proposals to account for the unacceptability of the primed examples in (744).

Note in passing that Coopmans (1985) mentions that some (dialect?) speakers do allow constructions of the type Het hek is laten schilderen; such examples are not relevant in the present context, because it is not the presumed subject of the infinitival clause (which is assumed to be assigned accusative case by the verb laten) that is promoted to the subject of the matrix clause, but the object (which, under standard assumptions, should receive case from the bare infinitive). This construction is not widespread, as shown by a Google search (November 1, 2023) on the string [is laten V] for the transitive verbs wassento wash, strijkento iron and vervento paint yielded no relevant result.

The discussion above has shown that there is no clear evidence that the subject of the bare infinitival clause is assigned case by the verb laten; the main reason for assuming this is that it solves the problem that subjects of infinitival clauses cannot normally be assigned case by any element internal to infinitival clauses.

The verb laten is like the perception verb horento hear in that it allows the subject of the bare infinitival to remain implicit. The examples in (746) show that, for this to be possible, the bare infinitival clause must be sufficiently “heavy” in the sense that the bare infinitival must have at least one argument that is overtly expressed; this means that while monadic (intransitive and unaccusative) verbs do not easily allow non-realization of their subjects, transitive and PO-verbs do. Non-realization of the subject of the infinitival clause is often easier with causative than with permissive laten, for the reasons given in Subsection I.

As in the case of horento hear, it is sometimes possible to realize the subject of the bare infinitival as an agentive door-PP. The examples in (747) show that this possibility is limited to constructions that allow non-realization of the subject.

The examples in (748) show that, contrary to what we see in AcI-constructions with ziento see, passivization of the infinitival clause is never possible. We do not shown this for vertrekkento leave, because unaccusative verbs cannot be passivized.

As in the case of the perception verb horento hear, the possibility of expressing the agent by a door-PP may give rise to the idea that the non-realization of the subject is the result of a passive-like process; cf. Section 5.2.3.3, sub IVC. Petter (1994/1998: §4) objects to such an analysis on the ground that examples such as (749a) allow non-realization of the noun phrase despite the fact that the verb weten usually resists passivization: cf. Marie weet het antwoordMarie knows the answer versus *Het antwoord wordt geweten. She further observes that the omitted noun phrase cannot be replaced by an agentive door-PP, but can be replaced by an aan-PP; this is shown by (749b). See Petter (1994/1998:141-2) for additional (cross-linguistic) evidence against the idea that the omission of the subject results from a passive-like process.

The choice between the door-PP and the aan-PP seems to be determined by the embedded infinitive: verbs such as zingento sing in (750a) are only compatible with door-PPs, verbs such as ziento see in (750b) are only compatible with aan-PPs, and verbs such as lezento read in (750c) have both options. To our knowledge, the properties that determine which verbs go with which type of PP have not yet been investigated, so we must leave this topic for future research. Example (750c'') shows that the two types of PPs are mutually exclusive, even for verbs that allow both types; changing the word order does not improve the result.

The data above suggests that there are at least two types of causative/permissive constructions. The first is an AcI-construction: it takes a bare infinitival complement with an overt subject that can be replaced by a door-PP. The nature of the second type is less clear, but probably involves a dative noun phrase that can be replaced by an aan-PP. Petter suggests that the dative phrase does not originate as the subject of the bare infinitival complement (which should therefore be analyzed with a PRO-subject), but as an internal (goal) argument of laten. We leave further exploration of this topic to future research, while noting that Dutch laten would not be the only case with this option; French faire, for example, is also compatible with both a par-PP and an à-PP (although it does not allow an accusative noun phrase); cf. Broekhuis & Gronemeyer (1997) for data and references.

It is also possible to find constructions with doento make, in which the subject is left implicit. However, given the idiomatic nature of many causative doen-constructions, it does not seem to useful to discuss whether this is a productive process. That example (751) is idiomatic is clear from the fact that the subject of the infinitival clause must be left implicit.

The previous subsections discussed AcI-constructions with causative/permissive laten. The discussion suggests that laten behaves in most respects like the perception verbs in AcI-constructions. This subsection discusses a number of facts about the behavior of laten in two seemingly related constructions, and examines the extent to which we find similar facts with the perception verbs.