The traditional generative analysis holds that main clauses are derived by placing the finite verb in the second position of the clauses, the so-called C-position in (285), followed by topicalization of some constituent into the so-called clause-initial position, the specifier of CP; see Section 11.1 for details.

There seem to be virtually no restrictions on the syntactic function or the categorial status of the topicalized element. The examples in (286) start by showing this for nominal arguments: subjects, direct and indirect objects are all possible in sentence-initial position.

There are, however, two important differences between subject-initial sentences and sentences with an object in first position. First, clause-initial objects can be considered to be semantically marked in that they act as discourse topics or contrastive foci, or have some other special function in the organization of the discourse, while this does not necessarily hold for clause-initial subjects. Second, topicalized objects are often characterized by a special intonation pattern: the objects in (286b&c), but not the clause-initial subjects in (286a), must be accented, as is clear from the fact the latter but not the former can be a reduced pronoun. This suggests that subject-initial sentences may also be syntactically different from constructions with topicalized objects; we will return to this issue in Subsection III.
      Next, the examples in (287) show that it is also possible to topicalize prepositional objects: (287a) illustrates this for a prepositional indirect object and (287b) for the prepositional object of kijken (naar)'to look (at)'.

Complementives can also be topicalized: we illustrate this in (288) by means of three examples with complementives of a different categorial status; they show that noun phrases, APs and PPs can all be topicalized.

Adjuncts can also be topicalized. Example (289a) shows this for supplementives and examples (289b&c) for adverbial phrases. Observe that we did not mark the adverbial phrases for accent; assigning accent is possible but does not seem to be necessary. We will return to this issue in Subsection IV.

The discussion above has shown that topicalization is like wh-question formation in that constituents with various syntactic functions (argument, complementive and adjunct) and of various different forms (noun phrase, AP and PP) can be moved into sentential-initial position. Topicalization differs from wh-movement, however, in that it also allows preposing of clauses; this is illustrated in (290) for a finite clause. We return to topicalization of clauses in Subsection VI. Accent can be assigned at various places within the preposed clause.

The examples in (291) show that it is also possible to topicalize the complement of perfect and passive auxiliaries, a phenomenon known as VP-topicalization. The (a)-examples show that topicalization of the participle is possible both with and without the direct object; the (b)-examples show that subjects are normally not affected. VP-topicalization will also be discussed in Subsection VI. Accent will normally be assigned to the object if it is pied piped by VP-topicalization.

Topicalization involves movement of some constituent into the initial position of the main clause. It resembles the formation of wh-questions in that the movement targets the position immediately preceding the finite verb; this is illustrated again in the (b)-examples in (292). This observation is not trivial; this does not hold for a language like English. We return to this in Subsection VII.

      The (b)-examples in (293) show that topicalization differs from question formation (and relativization) in that it is a root phenomenon. It cannot apply in embedded clauses.

There is no way in which embedded topicalization in examples such as (293b') can be improved. The examples in (294), for instance, show that Dutch does not have the option found in German to have topicalization in embedded clauses with verb-second, as embedded verb-second is categorically prohibited in Dutch. We refer the reader to Haider (1985/2010) and Barbiers (2005: Section 1.3.1.8) for a discussion of embedded verb-second in, respectively, German and a number of non-standard varieties of Dutch; the German example in (294a) is taken from Müller (1998:42) in a slightly adapted from.

The examples in (294) also show that embedded topicalization cannot occur with a phonetically expressed complementizer, unlike what is the case in English examples such as (295a); cf., e.g., Chomsky (1977), Baltin (1982) and Lasnik & Saito (1992). Since there is no a priori reason to think that Dutch topicalization targets a different position than English topicalization, we have added example (295b'), in which the complementizer dat'that' precedes the topicalized phrase.

      Examples (296a&b) show that topicalization is like question formation in that it allows long wh-movement if a bridge verb such as denken'to think' is present. It should be noted, however, that long topicalization is like relativization in that it is possible with a wider range of verbs than question formation; cf. Schippers (2012:105). For instance, the factive verb weten'to know' permits long topicalization (and long relativization), but not long wh-movement. It should further be noted that some speakers prefer the resumptive prolepsis construction in (296c) to the somewhat marked long topicalization construction in (296b).

      That topicalization involves wh-movement is also suggested by the fact that it is island-sensitive, just like question formation and relativization. We illustrate this in (297b) by means of an embedded polar question. For completeness' sake, we have added (297b') to show that the intended meaning can be expressed by means of a resumptive prolepsis construction.

Example (298b) illustrates the island-sensitivity of topicalization by means of an adjunct island. In this case, the resumptive prolepsis construction is not available as an alternative because the verb huilen'to cry' does not license a resumptive van-PP.

This subsection has shown that topicalization exhibits various hallmarks of wh-movement: it targets the clause-initial position, it can be extracted from clauses selected by bridge verbs and it is island-sensitive. What sets it apart from wh-movement and relativization is that it is a root phenomenon; it cannot target the initial position of embedded clauses. We refer to Hoekstra & Zwart (1994), Sturm (1996) and Zwart & Hoekstra (1997) for a discussion of the question as to whether this shows that topicalization targets a different position than wh-movement, as in fact would be claimed in the cartographic approach initiated by Rizzi (1997).