Finite direct object clauses differ from nominal direct objects in that they must follow the clause-final verbs in neutral contexts. This is illustrated in (61): while the primeless examples show that nominal direct objects must precede the clause-final main verb, the primed examples show that direct object clauses can follow it.

It seems that object clauses usually follow other constituents of their clause when they occur after the clause-final verbs. This is illustrated in (62) for a prepositional indirect object and a temporal adverbial phrase. Note, however, that the unacceptable orders improve when the object clause is followed by an intonation break, in which case the PP/adverbial phrase would express an afterthought.

There is, however, one systematic exception: direct object clauses are followed by extraposed adverbial clauses. This is illustrated by the primeless examples in (63) for adverbial clauses expressing time and reason, respectively; the number signs in the primed examples indicate that these examples are acceptable only if the adverbial clause is interpreted parenthetically, in which case it must be preceded and followed by an intonation break. Note in passing that example (63a) is actually ambiguous; the adverbial clauses can in principle also be interpreted as part of the object clause, in which case it does not refer to the time at which John told that he was robbed, but to the time at which the robbery took place.

Direct object clauses may also be followed by elements that are not part of the sentence, such as the epithet in (64a) or the afterthought in (64b). Such elements are usually preceded by an intonation break.

Finite object clauses in extraposed position are often introduced by the anticipatory pronoun hetit, which we have indicated here by subscripts; cf. also 5.1.1, sub III.

The examples in (66) show that, in general, direct object clauses do not precede their matrix verb in clause-final position.

There are, however, a number of possible counterexamples to this general rule. First, the examples in (67) show that free relative clauses can generally either precede or follow the clause-final verbs. We have seen in Section 5.1.1, sub IV, that this is one of the reasons to assume that free relatives should be considered not as argument clauses but as noun phrases. So the surprising thing is actually that example (67a) is acceptable, but this can be explained by assuming that free relatives can be in extraposed position just like relative clauses with an overt antecedent: cf. dat Jan de man/Ø prijst [die hij bewondert]that Jan praises the man who he admires.

Second, we find similar order alternations with so-called factive verbs like onthullento reveal and betreurento regret. Although some speakers may judge the primed examples as marked compared to the primeless examples, they seem well-formed and are certainly much better than the primed examples in (66). Barbiers (2000) suggests that the markedness of the primed examples is not related to grammaticality issues, but to the fact that center-embedding (especially of longer constituents) can lead to processing difficulties.

The main difference between the (a)-examples in (66) and (68) concerns the truth of the proposition expressed by the embedded clause; cf. Kiparsky & Kiparsky (1970). Consider the examples in (69). Sentence (69a) shows that the proposition expressed by the complement clause of bewerento claim in (66a) can be denied by the speaker without any problem; the speaker is not committed to the truth of the proposition, but instead assigns responsibility for its truth to Jan. The situation is different with the factive verb onthullento reveal in (69b); by using this verb the speaker presupposes that the proposition “Els is going to emigrate” is true; this can be seen from the fact that the denial in the second conjunct is surprising, to say the least.

The behavior of factive clauses deserves more attention, especially since it has been suggested that they do not function as argument clauses. However, since discussing this would take us too far from the present topic, we will return to it in Section 5.1.2.3.

The examples in (70) show that object clauses can easily be moved into sentence-initial position by topicalization. In keeping with the verb-second requirement in Dutch, the topicalized clause must be immediately followed by the finite verb. Note that topicalization of object clauses is impossible when the anticipatory pronoun hetit is present, as can be seen by comparing the examples in (70) with those in (65).

The unacceptability of the anticipatory pronoun het in (70) can be explained in at least two ways. One possibility is to assume that the examples in (70) are not derived by regular topicalization, but in a way similar to the left-dislocation constructions in (71); cf. Koster (1978b).

If we follow this line of reasoning, the examples in (70) could involve a phonetically empty pronoun pro with the same function as the resumptive demonstrative pronoun datthat in (71). On this analysis, the anticipatory pronoun cannot be realized because it is replaced by the pronoun dat/pro, which is moved into the sentence-initial position; the structures in (72) show that the use of het is blocked because the clause-internal object position is occupied by the trace of the moved pronoun.

The analysis proposed above is challenged in Klein (1979), which points out that the examples in (70) and (71) have different intonation patterns: while the examples in (71) contain an intonation break between the clause and the pronoun dat, indicated by a comma, the clauses in (70) are not likely to be followed by such an intonation break. If one wants to conclude from this that the examples in (70) must be derived by topicalization of the finite clause, one can explain the impossibility of the pronoun het in an alternative way by assuming that the clause must be moved via the regular object position in the middle field of the clause; under this proposal the pronoun het cannot be realized because the regular object position would be filled by a trace of the moved clause. Such an analysis raises the question why finite clauses cannot appear in the regular object position; cf. the discussion in Subsection II. One possibility would be to assume a surface condition that prohibits argument positions from being filled by non-nominal categories; cf. Stowell (1983), Hoekstra (1984a), and Den Dikken and Næss (1993) for proposals to this effect. We will see in Section 5.1.3 that the same question arises with finite subject clauses.

Haeseryn et al. (1997:1155) claims that subject-experiencer verbs like betreurento regret and waarderento appreciate can take an object clause introduced by the conditional complementizer alsif; cf. the primeless examples in (73). Since the claim is simply stated without any motivation, we can only guess why it is proposed; an obvious argument in its favor would be that we can replace the als-clauses with noun phrases that clearly function as direct objects; cf. the primed examples in (73).

However, this subsection will show that there is a strong argument against the hypothesis that we are dealing with object clauses in (73a&b), which concerns the distribution of the anticipatory pronoun hetit and the resumptive pronoun datthat.

The primeless examples in (74) show again that anticipatory pronouns optionally introduce object clauses in extraposed position. The examples in (73a&b) show a different behavior in this respect; the anticipatory object pronoun het is obligatory and not optional when the verbs betreuren and waarderen are followed by an als-clause, as shown in the primed examples in (74).

The examples in (75) show that while the anticipatory pronoun het is replaced by the resumptive pronoun dat when a dat-clause is left-dislocated, the resumptive dat cannot be used with left-dislocated conditional als-clauses.

The primeless examples in (76) further show that resumptive dat is not used when the dat-clause is not followed by an intonation break. The primed examples, on the other hand, show that such constructions without dat are excluded with als-clauses.

Adding an object pronoun like dat or het in the usual object position of the primeless examples in (76) would make these examples ungrammatical, which may be due to the fact that the object position is already occupied by a trace; cf. Subsection III. Adding an object pronoun to the primeless examples in (76), on the other hand, makes these examples perfectly acceptable.

The primed examples in (77) thus suggest that conditional als-clauses and object pronouns have different syntactic functions. This is also supported by the fact that als-clauses in left-dislocation constructions can be associated with the resumptive adverbial element danthen, which also optionally appears in conditional constructions such as Als het regent, (dan) kom ik niet If it rains, (then) I wont come’.

The fact that the object pronoun het/dat cannot be omitted when the resumptive dan is present shows conclusively that object pronouns and conditional als-clauses have different (logical) syntactic functions: object versus an adverbial adjunct with a function similar to that of an antecedent in a conditional-like statement. The implication that dat-clauses and als-clauses have different syntactic functions can be supported by the coordination facts in (79): (79a&b) show that two dat-clauses or two als-clauses can easily be coordinated, while (79c) shows that it is impossible to coordinate a dat-clause and an als-clause. This follows from the independently established restriction on coordination that the conjuncts must have the same syntactic function; cf. Section C38.2, sub IA.

Note in passing that the left-dislocation test can also be applied to other cases in which one might be tempted to analyze a clause, or some other phrase, as a direct object. For example, the phrases introduced by alsof/als in the primeless examples in (80) resemble direct objects in that they cannot be omitted, but the fact that the left-dislocation construction does not allow the resumptive dat, but requires the manner adverb zo immediately shows that we are dealing with adverbial phrases.