It is generally claimed that the movement of interrogative noun phrases is obligatory; if the movement does not apply, the interrogative meaning is usually lost. The examples in (8), for example, are not true questions, but rather receive an echo interpretation: sentences like these, in which the question word is heavily emphasized, are used when the speaker has not heard properly what the addressee has just said, to express the speaker’s astonishment at what he has just heard, or in teacher-student interaction as test questions.

Nevertheless, strings like those in (8) are also occasionally used as true wh-questions when given a more interrogative intonation pattern (with a drop in pitch after the question word); the reader is referred to V13.3.1.1, sub I, for further discussion.

A clear and systematic exception to the general rule that wh-phrases must be moved into the clause-initial position can be found in so-called multiple questions, which contain more than one wh-phrase. In this case, the requirement that a wh-phrase be moved is overridden by the fact that only a single constituent can be placed in the clause-initial position. Normally, the wh-phrase that is superior (≈ closest to the target position) is wh-moved. The effects of this so-called superiority condition can be seen most clearly in embedded clauses like those in (9). Example (9a) shows that when both the subject and the direct object are wh-phrases, it is the subject that occupies the clause-initial position; moving the object instead, as in (9a'), leads to a severely degraded result. The b-examples show that the same contrast holds to a slightly lesser degree for examples in which both the direct and the (bare) indirect object are questioned; it is clearly preferred that the indirect object undergoes wh-movement, not the direct object. Finally, the (c)-examples show that when the indirect object is periphrastic, the direct object is preferred displaced.

The gradual increase in acceptability of the primed examples in (9) is probably related to the fact that the order of subject and direct object in the Dutch middle field is solidly fixed, whereas it is not completely impossible to have an accusative DP preceding a dative one, and it is certainly not uncommon to have a periphrastic indirect object preceding the direct object.

The examples in (10) show that judgments are less clear in the case of main clauses. All speakers agree that the primeless examples are preferable to the primed ones, but many speakers find the latter significantly better than the primed examples in (9). It is not clear what causes this acceptability contrast.

It seems that the wh-phrases in the primed examples in (10) need to be of the same kind in order to be able to violate the superiority condition; for example, if one of the two DPs headed by a demonstrative pronoun is replaced by an interrogative personal pronoun, as in (11), the results seem to get worse. Again, it is not clear what causes this effect.

Wh-movement does not have to target the initial position of the minimal clause containing the moved argument, but can also target the initial position of some higher clause. For this to be possible, the clause containing the wh-phrase must be the complement of a limited set of so-called bridge verbs, generally a verb taking a propositional complement like the verbs of saying or thinking; cf. Section V11.3.1.2, sub IV for further discussion.

It has been argued that in many languages there is an asymmetry between subjects and objects (as well as other non-subjects) with respect to this kind of so-called long wh-movement. While objects can undergo long wh-movement, subjects cannot, unless the language has some special feature that makes this movement possible: Whoi do you think (*that) ti came, for example, shows that dropping the complementizer that makes extraction of the subject possible in English. In early generative grammar, this led to the empirical generalization that a complementizer cannot be followed by a subject trace, which was formulated as the complementizer-trace filter in (13), where C and ti stand for the complementizer and the subject trace, respectively; cf. Chomsky & Lasnik (1977).

In later stages of the theory, the filter in (13) evolved into a more general, cross-linguistic principle that excludes the sequence of an overt complementizer and a subject trace (unless the language has a special mechanism for licensing the subject trace, such as the French que/qui rule). At first glance, Dutch seems well-behaved with respect to this principle: while the examples in (12) are perfectly acceptable, example (14a) is marked (though not as bad as its English translation with the overt complementizer that). On closer inspection, however, it turns out that the degree of acceptability of such examples is influenced by the type of noun phrase: D-linked noun phrases such as welke jongen in (14a) do not allow this movement easily whereas non-D-linked noun phrases such as wie in (14b) do.

A possible reason for the difference in acceptability of these two examples may be that, despite appearances, the traces of the two wh-phrases do not occupy the same position in the clause. This can be made clearer by considering embedded clauses that do not contain a definite object, such as those in (15).

As can be seen in (15b), the example with wie requires the embedded clause to contain the expletive er. Since the expletive normally precedes the indefinite subject (cf. Gisteren heeft er iemand gelogenYesterday someone lied) and can therefore be assumed to occupy the regular subject position, we can conclude that the subject trace does not occupy the regular subject position of the clause in (15b); cf. Section 22.1.4 for further discussion. If this is a general property of non-D-linked interrogative personal pronouns, the same must hold for (14b). If we now reformulate the complementizer-trace filter in (13) such that it expresses that a complementizer cannot be followed by a trace in the regular subject position, Dutch would behave in accordance with this generalization. The main point of this subsection is that the ability of the embedded subject to undergo long wh-movement depends on D-linking. We end our discussion here and refer the reader to Section V11.3 for a more detailed discussion of (long) wh-movement.

In the unmarked case, the initial constituent of a main clause is the subject. As we have already seen in the discussion of example (1) in Section 22.1.1, almost all noun phrase types can function as clause-initial subjects, the only exception being weak noun phrases; the latter usually occur in the expletive construction, in which case it is not the subject itself but the expletive erthere that fills the clause-initial position. A noteworthy property of clause-initial subjects is that they can appear as weak (i.e. phonetically reduced) pronouns; the masculine third-person singular form –iehe is an exception in that it must be right-adjacent to the finite verb in second position. The second-person plural pronoun is special in that it has no weak subject form in most varieties of Dutch.

In this respect, clause-initial subjects differ markedly from topicalized object pronouns, which must always be realized in their strong form, as shown in (17). This also explains why the neuter object pronoun het cannot be topicalized; since het is always pronounced in its weak form (cf. Section 19.2.1.1, sub V), it is usually the neuter demonstrative ditthis or datthat that is topicalized instead.

Thus, subject and object pronouns differ in that the latter must be stressed in the clause-initial position, whereas the former need not. This difference between subject and object pronouns has been used to argue that, despite appearances, clause-initial subjects are not topicalized, but rather occupy the regular subject position. This may also explain why the expletive erthere, which is generally assumed to occupy the subject position, can also be used clause-initially; cf. Section 22.1.4 for examples. The conclusion that clause-initial subjects are not topicalized, if correct, has far-reaching theoretical implications in the sense that it breaks with at least two traditional assumptions about Dutch syntax. First, it presupposes that the finite verb in subject-initial main clauses does not occupy the C(omplementizer)-position, as this would result in a verb-first order: [CP ... C + Vfin [IP subject ... [VP ... tVfin (V) ]]]. Instead, it should be placed in the lower I(nflection)-position, which in turn implies that the I-position is to the immediate right of the regular subject position: [IP subject I + Vfin [VP ... tVfin (V)]]. This implies a second radical break with the more traditional view of Dutch syntax, according to which the I-position is located in the right periphery of the clause, following the base positions of the verb(s); cf. Section V9 and especially Zwart (1997/2011: Part III) for a discussion of some of the ramifications of this change of view.

The term topicalization suggests that this wh-movement process plays a role in determining the information structure of the clause by moving the discourse topic into the first position of the clause. One potential problem with such a claim is that clause-initial subjects need not be topics. However, it seems that this problem can be set aside, since we have seen in Subsection A that there are reasons to assume that such subjects are not actually topicalized, but occupy the regular subject position of the clause. Therefore, it seems possible to maintain that topicalization applies for information-structural reasons. However, we will see in the following subsections that the preposed phrase need not be a discourse topic, but can also be presented as a contrastive or emphatic focus; cf. Neeleman & Van de Koot (2008).

The (b)-examples in (6) above have shown that topicalization cannot target the initial position of an embedded clause. However, this does not mean that it is impossible to topicalize a constituent that is part of an embedded clause; topicalization can also target the initial position of a higher main clause. In (22) we give an example of such long topicalization of a direct object, which is perfectly acceptable when the object is assigned contrastive accent.

The examples in (23) involve long topicalization of a subject. In these cases, there is a clear contrast between definite and specific indefinite noun phrases, on the one hand, and non-specific indefinite noun phrases, on the other. Only the latter are fully acceptable, provided that the topicalized phrase is emphatically stressed.

Again, this can be explained by the generalization in (13) that a complementizer cannot be followed by a trace in the subject position. When a non-specific indefinite noun phrase is topicalized, it is not moved from the regular subject position, which is occupied by the expletive, but from some position following it. If we are dealing with a definite or specific indefinite noun phrase, the expletive is not present and the movement proceeds from the regular subject position, resulting in unacceptability.