Finite clauses occur regularly as arguments of verbs: they can be used as subjects, direct objects, or as part of a prepositional object (i.e. a PP-complement of the verb). Indirect objects are usually nominal, which seems to be related to the fact that they typically refer to living entities or institutions, not to propositions. The examples in (8) show that argument clauses are typically placed after the clause-final verbs (if present). The reason for calling the embedded clause in (8c) a prepositional object rather than a direct object is that it cannot be pronominalized by the pronoun hetit, but must be replaced by the pronominal PP eroverabout it. The properties of the three types of argument clauses in (8) will be discussed in more detail in Sections 5.1.2 to 5.1.4.

Finite argument clauses usually take the form of an embedded clause, i.e. a clause with the finite verb in clause-final position, as in the indirect reported speech example in (9a). Possible exceptions to this general rule are found in the direct and semi-direct reported speech examples in (9b-c), where the apparent dependent clause appears in main clause order, i.e. with the finite verb in second position. For this reason, cases of direct and semi-direct reported speech deserve special attention; they will be discussed separately in Section 5.1.2.4.

The examples in (10a&b) show that declarative argument clauses are obligatorily introduced by the complementizer datthat, i.e. unlike English that, Dutch dat cannot be omitted. Example (10c) further shows that Dutch also differs from German in that it does not allow embedded clauses without a complementizer and with verb-second; see Haider (1985) for a discussion of verb-second in embedded clauses in German, and Barbiers et al. (2008: §1.3.1.8) for a number of (especially eastern) Dutch dialects that also have this construction. Note that example (10c) is acceptable as a case of direct reported speech, but this is obviously not the reading intended here.

Interrogative argument clauses are introduced either by the complementizer ofwhether or by a wh-phrase. In speech (but not in writing) it is also common for the wh-phrase to be followed by a complementizer in embedded wh-questions. The complementizer of is used in the northern varieties, while the complementizer dat is more common in the southern varieties, and some (mainly northern) speakers even use the combination of dat; cf. Barbiers et al. (2008: §1.3.1.5) for details on the geographical distribution of these options, and Hoekstra & Zwart (1994), Sturm (1996) and Zwart & Hoekstra (1997) for relevant discussions on the question of whether of dat should be analyzed as a compound or as two separate words.

When two embedded yes/no questions are coordinated by the disjunction ofor, as in (12a), the complementizer of the second clause does not occur as of but as dat, presumably to avoid a sequence of two (homophonous) occurrences of of. That this is a surface phenomenon is clear from the fact, illustrated in (12b), that the second complementizer must be realized as of if we replace the disjunction of by the more formal disjunction dan welor; cf. Haeseryn et al. (1997:547).

There is a small set of cases in which what appears to be an argument clause is introduced by the subordinate conjunction alsif/when; cf. Haeseryn et al. (1997:1136&1153). The primeless examples in (13) show that such als-clauses are especially common in constructions with a subject/object experiencer, although the primed examples show that the experiencer can also remain implicit; note that het is an anticipatory pronoun introducing the als-clause. To our knowledge als-clauses of this type have received little attention in the literature, and it remains to be shown whether they actually function as argument clauses in these cases; for this reason we will examine them in more detail in Section 5.1.2.2, sub IV.

It is important to note that the distinction between declarative and interrogative embedded clauses is formal rather than semantic: the embedded clause in (14a) is called declarative even though we are clearly not dealing with an assertion, and the embedded clauses in (14b&c) are called interrogative even though we are not dealing with true questions. Nevertheless, we will simply adopt the traditional terminology.

The examples in (15) show that finite argument clauses can be introduced by an anticipatory pronominal element (in italics) appearing to the left of the clause-final verbs.

The distribution of anticipatory pronominal elements is complex: Sections 5.1.2 to 5.1.4 will show that in many cases it is optional, but there are also cases in which it must or must not occur. Moreover, the presence or absence of the pronominal element can affect the syntactic behavior of argument clauses: for example, (16b) shows that object clauses allow wh-extraction only if there is no anticipatory pronoun; cf. Bennis (1986: §2).

If the anticipatory pronominal element is optional, its presence may trigger a somewhat different reading: sentence (16a) without the pronoun het presents the proposition expressed by the embedded clause as new information; (16b) with the pronoun, on the other hand, presents the embedded proposition as old information, adding that Jan was the source of the information. In cases such as (17), the presence of the anticipatory pronoun can trigger a factive reading of the object clause: example (17a) simply presents the proposition expressed by the embedded clause as new information, which may or may not be true, whereas (17b) presents this proposition as familiar true information.

A similar contrast can be found in the passive counterparts of the examples in (17) in (18): the impersonal passive with the expletive erthere in (18a) presents the proposition expressed by the embedded clause as new information that may be true or false, while the personal passive with the anticipatory subject pronoun hetit in (18b) presents it as familiar and true; cf. Haeseryn et al. (1997:1138) for similar intuitions. A more detailed description of the distribution of the expletive erthere and the anticipatory subject pronoun hetit will be given in Section 5.1.3, sub III.

Note, however, that the question as to whether a factive reading arises is more complex than the examples in (17) and (18) suggest. Examples (19a&b) show that regardless of the presence or absence of the anticipatory pronoun, the truth of propositions expressed by the clausal objects of typically factive verbs such as betreurento regret is normally presupposed by the speaker, whereas the truth of propositions expressed by the clausal objects of a typically non-factive verb such as bewerento claim is normally left open. It is only with neutral verbs such as vertellento tell, which can be used both as factive and as non-factive verbs, that the presence of the anticipatory pronoun het will normally trigger the factive reading.

Since the semantic effect of the anticipatory pronoun het is sometimes difficult to pinpoint, even with neutral verbs such as vertellen, we will not digress on this issue and leave further investigation to future research. For the sake of completeness, however, it should be noted that the frequency of the anticipatory pronoun het is higher with typically factive verbs such as betreurento regret than with non-factive verbs such as bewerento claim and (especially) neutral verbs such as vertellento tell. This is shown in Table (20) by the results of a Google search (September 4, 2023) on the strings [V-t (het) dat] and [V-de (het) dat].

Haeseryn et al. (1997) claims that argument clauses can also take the form of free relative clauses. This is supported by two facts: (i) that we are dealing with non-main clauses can be seen from the fact that the finite verb appears in clause-final position, and (ii) that we are dealing with arguments can be seen from the fact that free relatives can function as subject, direct object, and part of a prepositional object.

The question, however, is whether free relatives also behave like argument clauses. There are several reasons to answer this question in the negative, and to assume that free relatives are nominal in nature. First, they usually refer to entities, not propositions. This would also explain why free relatives can easily be used as indirect objects, whereas declarative and interrogative argument clauses cannot.

Second, the examples in (23) show that free relatives can occur in the nominal argument positions in the middle field of the clause, which are normally not accessible to declarative and interrogative argument clauses.

Third, the examples in (24) show that the use of the anticipatory elements het and erop is impossible.

Finally, example (25b) shows that extraposition of free relatives leads to an acceptable result when they function as direct objects. Prepositional objects with a free relative clause as complement can also be in extraposed position, whereas this gives rise to a marked result when they have a dat-clause as complement; cf. ??dat Jan wacht op dat Els iets zegtthat Jan is waiting for that Els says something.

The behavior shown in examples (22)-(24) is what we attribute to nominal but not to clausal arguments. What is perhaps not immediately expected is that free relatives acting as direct objects can follow the clause-final verbs, but this would follow if we assume that free relatives exhibit similar extraposition behavior as the regular relative clauses with an overt antecedent (here: iedereen and hetgeen) in (26). However, this proposal leaves unexplained why (25c) is unacceptable with the preposition op stranded in preverbal position, because example (26c) seems to be perfectly acceptable.

Nevertheless, we conclude from the above discussion that free relatives are nominal in nature, and therefore should not be included in our discussion of argument clauses. See Section N16.3.2.2 for a discussion of free relatives.