A-movement is restricted to nominal categories; cf. Section 13.2, sub III. The fact that focus movement can also affect PPs, as illustrated again in (122b), is therefore sufficient to conclude that we are dealing with A'-movement. Example (122c) further supports this conclusion by providing an example in which an adjectival complementive has undergone focus movement.

The claim that focus movement is A'-movement is consistent with the conclusion that focus movement can target a position to the right of the modal adverbials, because Section 13.2 has shown that nominal argument shift must target a position to the left of the modal adverbials. This contrast between A and A'-movement can be highlighted by the VP-topicalization constructions in (123), which show that stranding of the direct object in a position after the clause adverbials requires it to be contrastively focused.

It can also be illustrated quite nicely by considering the placement of strong (phonetically non-reduced) referential personal pronouns like zijshe and haarher; such pronouns may only occur after modal adverbials if they have an A-accent.

Finally, the fact that nominal argument shift and focus movement target different landing sites is highlighted by the fact that [-human] referential personal pronouns can never occur after the modal adverbials, for the simple reason that they are phonetically obligatorily reduced; to contrastively focus an inanimate entity, the demonstrative deze/diethis/that is used.

Another argument for an A'-movement analysis of focus movement has to do with word order. Section 13.2, sub IC, has shown that nominal argument shift cannot affect the unmarked order of nominal arguments (agent > goal > theme) in Dutch. Neeleman & Van de Koot (2008), Van de Koot (2009), and Neeleman & Vermeulen (2012) claim that focus movement is capable of changing the order of nominal arguments, as illustrated in (126), and that this supports the claim that we are dealing with A'-movement.

The argument is not entirely convincing; the fact that this kind of order preservation does not hold for German nominal argument shift shows that it is not a defining property of nominal argument shift; cf. Section 13.2, sub IC. Furthermore, the judgments made in the studies by Neeleman and his collaborators are controversial, since some speakers of Dutch (including the authors of this work) reject the examples in (126a&b) with the indicated intonation pattern; cf. also Neeleman & Van de Koot (2008:fn.2) and Van de Koot (2009:fn.4). A simpler example, also rejected by some of our informants, is given in (127). In our view, the unclear acceptability status of (126a&b) and (127a) makes it impossible to draw any firm conclusion; it remains to be seen whether such examples should be considered part of the standard variety of Dutch, but we will leave this question to future research.

To avoid confusion, we should note that the examples marked with % are acceptable when the contrastively accented phrases are given a B-accent, in which case we are dealing with a contrastive topic; Subsection II will provide more data showing that topic movement can indeed affect the unmarked order of nominal arguments under certain conditions.

Example (128a) shows that focus movement can change the unmarked order of nominal and prepositional objects: while prepositional indirect objects usually follow direct objects, focus movement of the former can easily cross the latter. Note, however, that this requires the direct object to follow the modal adverb: the examples in (128b&c) show that object shift of het boek has a degrading effect on focus movement regardless of whether the focused phrase precedes or follows the modal adverb; we added the adverb niet to (128c) to make focus movement visible. Note that (128b) becomes perfectly acceptable when the PP is assigned a B-accent, showing that topic movement can cross a shifted object.

This subsection has shown that the claim that focus movement is able to change the unmarked order of nominal arguments in standard Dutch is controversial; whether this property could be used as an argument for the claim that focus movement is A'-movement is also not clear, since order preservation seems to be an accidental property of nominal argument shift in Dutch (not German).

A'-movement such as wh-movement differs from A-movement in that it allows extraction from finite clauses under certain conditions (so-called long movement). Neeleman (1994a/1994b) and Barbiers (1998/1999/2002/2018) have shown that there is also long focus movement: the primed examples in (129) illustrate that foci can target a focus position in the middle field of a matrix clause. The percentage signs are used to indicate that this kind of long focus movement is not normally found in written language, but can be found in colloquial speech; cf. Zwart (1993:200).

That the landing site of the foci in the primed examples is external to the embedded clause is clear from the fact that the foci precede the clause-final main verb of the matrix clause. Section 11.3.3, sub II, has shown that topicalization is impossible in Dutch embedded clauses and this correctly predicts that foci cannot immediately follow clause-final verbs; cf. the structures in example (130).

Although the primed examples in (129) may be objectionable to some speakers, the sharp contrast with the examples in (130) shows that they are at least marginally possible in standard Dutch. This conclusion is also supported by the fact that the primed examples in (129) are clearly much better than the corresponding examples in (131) with the factive verb betreurento regret. This contrast shows that long focus movement is only possible in certain bridge-verb contexts.

There are reasons to assume that long focus movement is like long wh-movement in that it has to pass through the initial position of the embedded clause. A somewhat weak argument for this claim is that the direct object een boeka book in the (b)-examples in (129) can easily cross the subject, as this is a well-established property of A'-movements targeting the clause-initial position. A stronger argument is that long focus movement cannot co-occur with long wh-movement, as shown by the examples in (132): the examples in (132b&c) first show that wh-phrases and foci can be extracted from the embedded clause in (132a), while (132d) shows that they cannot be extracted simultaneously. This would immediately follow if all long movements had to pass through the clause-initial position of the embedded clause: long wh-movement would then block long focus movement (or vice versa), because this position can only be filled by a single (trace of a) constituent (but see Barbiers (2002) for a different account).

The movement analysis of the primed examples in (129) is also used to account for the fact that the primeless and primed examples are propositionally equivalent. It has already been noted in Jespersen (1927:53) that the same seems to hold for negative examples as in (133).

The equivalence of ¬P(x,p) and P(x,¬p), where x stands for the subject of the matrix predicate P and p for the proposition expressed by its complement clause, does not hold for all matrix predicates, but only for predicates that are non-factive. Horn (2001) distinguishes the following subclasses: opinion (to think/expect), perception (to seem/look like); probability (to be likely), intention and volition (to want/choose), and judgment and (weak) obligation (must/to advise). Klooster (2003), in a discussion of Dutch, concludes that the equivalence “obtains in sentences with negated non-factive verbs expressing, together with their negation, a negative attitude towards a hypothetical state's or event's being or becoming fact”. Already early in the history of transformational grammar the two constructions in (133) were related by long movement, now called neg-raising: cf. en.wikipedia.org/wiki/Negative_raising for references and an overview of different syntactic approaches, and Klooster (2003) for an alternative approach.

De Schepper et al. (2014) shows, based on the Corpus of spoken Dutch, that a similar phenomenon exists for pragmatic markers, including the affirmative counterpart wel of the negative adverb niet. Example (134) provides a case with the sentence adverbial eigenlijk; in both cases, the adverb is construed with the embedded proposition. The appendix to the article mentions the following additional cases: even (marker of kindness), gewoonjust, inderdaadindeed, meerrather, ooktoo, sowiesoanyway, tochstill. It is further argued that the matrix predicate must be subjective in the sense that it expresses the speaker’s personal view about the objective content of the embedded clause.

Barbiers (2018) characterizes the set of pragmatic markers as affirmative evaluative (or polarity) adverbs and considers examples such as (134b) as positive counterparts of NEG-raising constructions such as (133b). This can be supported at least for the affirmative marker wel by examples such as (135), where raised wel stands in clear opposition to the non-raised negative adverb niet.

De Schepper et al. (2014) further argues that the matrix predicate must be subjective in the sense that it expresses the speaker’s personal view about the objective content of the embedded clause. This is also the position taken by Barbiers (2018) with respect to speaker-oriented adverbial phrases such as eerlijk gezegdhonestly in (136): a speaker-oriented adverbial such as eerlijk gezegd must have scope over the main clause independently of its position; cf. also De Schepper et al. (2014).

Barbiers further notes a number of systematic properties in pairs such as (136). First, the matrix verb must be non-factive in order to allow a matrix interpretation of an adverbial phrase in the embedded clause: the verb denken ‘to think’ in (136a) cannot be replaced by the factive verb wetento know). Second, an embedded adverbial phrase must have speech act properties in order to receive a matrix interpretation: the adverbial eerlijk gezegd in (136a) cannot be replaced by the speaker-oriented adverb helaas ‘unfortunately’ or by sentence adverbials of any other kind. For example, the frequency adverbial altijd ‘always’ is restricted to its minimal clause.

Third, the reference of the subject of the matrix clause must include the speaker (regardless the position of the adverbial); the subject pronoun ik can be replaced by wijwe’, but not by jij/jullie ‘you’ or hij/zij ‘he/they’ (but see Fortuin 2022:10-11). For further discussion and an attempt at analysis in terms of the Cinque hierarchy of adverbial phrases discussed in Section 8.2, see Barbiers (2018:§2).

The discussion above has shown that long A'-movement of an element into the middle field of the matrix clause is not a phenomenon restricted to focus-movement but is also common in the case of negation and pragmatic markers. However, the three subtypes seem to impose their own restrictions on the matrix predicate, which shows that we are not dealing with a purely syntactic phenomenon, but that semantic and pragmatic factors also play a role; cf. De Schepper et al. (2014), Fortuin (2022) and Barbiers (2022) for further discussion of this issue and the possible consequences for the analysis.

A possible problem for the hypothesis that focus movement is obligatory in standard Dutch is that it is sometimes possible to leave constituents with an A-accent in their original position. This is illustrated by the two examples in (138) and suggests that focus movement is optional. Of course, this conclusion is only valid if the two examples are semantically equivalent, which does not seem to be the case.

Before showing that the two examples in (138) are not fully equivalent, we will first consider example (139a), which clearly has two readings: the contrastive focus can be restricted to the direct object only, in which case the sentence expresses that there are certain other things in the discourse domain that Jan did not read, or it can extend to the verb phrase, in which case the sentence expresses that there were certain things that Jan did not do. The examples in (139b&c) show that the two readings evoke different word orders when the negative adverb niet is present. The clearest case is (139b), where the contrastive focus is restricted to the moved direct object. Example (139c) is somewhat more complicated, as it again allows two readings, one with contrastive focus on the verb phrase and one with contrastive focus on the noun phrase. This can be explained if we assume that in both cases we are dealing with constituent negation: Hij heeft niet de roman gelezen, maar het gras gemaaid he has not read the novel, but mowed the grass versus Hij heeft niet de roman gelezen, maar het gedicht he did not read the novel but the poem.

The crucial point for our present discussion is that (140a) is more suitable for expressing the restrictive focus reading than (140b). The former case evokes alternative propositions expressing that there are persons other than Peter whom Jan is very fond of, whereas (140b) rather expresses that the state of being fond of Peter does not apply to Jan, as is clear from the fact that it can easily be contrasted with an alternative state that does apply (here: to hate Peter), but not with the PP op Marie.

For completeness’ sake, note that the PP in (140a) must precede the negative adverb nietnot: cf. *dat Jan niet op PETER erg dol is. This is expected if it targets the specifier of FocP; cf. the discussion of (120) and (121).

Although constituents carrying an A-accent can apparently remain in situ, the discussion above suggests that this disfavors the restrictive focus interpretation. Of course, more needs to be said about the cases of constituent negation before we can conclude that focus movement is obligatory, but one thing is already clear: because nietnot is not located in the specifier of NegP when it expresses constituent negation, its location does not say much about the position of the contrastively focused phrase that follows it. The next subsection will show that there are reasons to assume that when the negative adverb niet expresses constituent negation, it functions as a focus particle and that the contrastively focused phrase following it normally occupies the specifier of FocP.

Focus adds an additional semantic value (hereafter: focus value) to the regular semantic value (hereafter: ordinary value) of a clause, as indicated again in (141) for the sentence Jan bezoekt MarieJan is visiting Marie.

The function of non-contrastive (new information) focus is that the speaker simply fills an information gap on the part of the addressee by adding a proposition to or selecting a proposition from the focus value of the clause; the speaker does not intend to imply anything for the alternative propositions. Contrastive focus, on the other hand, is counter-presuppositional in the sense that it aims at modifying a subset of propositions that the speaker presupposes to be considered true by the addressee; cf. the discussion of (115) in the introduction to this section. The modification can take different forms; we will slightly adapt Dik’s (1997; §13.4) classification by making the four-way distinction in Table 1, where (PA)S stands for the set of propositions presumably held by the addressee, PS stands for the modified set proposed by the speaker, and expression type refers to the English focus particles prototypically used to express the different subtypes; all subtypes are marked by an A-accent, which is represented by an exclamation mark.

	(PA)S	modified set PS	expression type
correcting	X	Y	not X, but Y!
expanding	X	X and Y	also Y!
restricting	X and Y	X	only X!
selecting	X or Y	X	X!

Correcting focus is the most complex case, because correction involves two simultaneous actions: rejection and replacement. The examples in (142) show that the speaker can perform both actions explicitly, but that he can also leave one of them implicit. The act of rejection is performed by constituent negation, i.e. the focus particle nietnot, in combination with the A-accent, while the A-accent suffices to perform the act of replacement. Note that (142b) is special in that it requires a secondary accent on the negative adverb niet.

Expanding, restricting and selecting focus are illustrated in (143), all of which are marked by the A-accent. Expansion is prototypically expressed by the focus particle ookalso and restriction by alleenonly, while selection is like replacement in that it does not involve the use of a focus particle.

In the primed examples in (143), the focus particles ook and alleen are associated with nominal arguments, but they can also be associated with larger constituents. For instance, in the primed examples in (144), the contrastive foci consist of the verbal projections, and the focus particles are therefore associated with these VPs.

More special cases are the focus particles zelfseven and slechtsmerely, which are not mentioned in Dik (1997), perhaps because they are not necessarily counter-presuppositional. These particles are often similar to the particles ookalso and alleenonly, but in addition they express a subjective evaluation, an extremely high degree, surprise, etc. The response in (145b) expresses that Peter’s presence was not expected, or rather, that his absence was expected.

Much more can be said about the meaning of focus particles, but we will not digress on this here; cf. König (1991), Foolen (1993), Barbiers, (1995a), and Van der Wouden (2002) for more discussion.

For the following discussion, it is crucial to realize that a focus particle and the contrastively focused phrase associated with it can form a constituent (although we will see that they can sometimes also be split). This is clear from the fact that they can occupy the clause-initial position together, as shown in (146) for the relevant examples in (142), (143), and (145). Note that it is not easy to construct similar cases for the examples in (144) with a contrastively focused VP: cf. ??Alleen het boek gekocht heeft hij.

Now that we have established that focus particles can form a constituent with contrastively focused phrases, we can discuss the hypothesis that focus movement is required to assign scope to the contrastively focused phrase. The examples in (147) show that while prepositional objects normally follow sentence negation, they can precede negation when they are contrastively focused. Since we have seen that focus movement targets a position that precedes sentence negation, the fact that the contrastively focused PP in (147b) can follow niet is a possible problem for the hypothesis that focus movement is obligatory.

In (148) we provide similar focus constructions as in (147b), but now with a focus particle present. If such particles can indeed form a constituent with the contrastively focused PP, and if focus movement is obligatory, we correctly predict that the presence of these focus particles requires the prepositional object to be moved across the negation.

The examples in (148) thus support the claim that focus movement is obligatory. Similar examples, in which the contrastively focused PP is embedded in an adjectival complementive, are given in (149). The fact that the PPs must precede the adjective when they are accompanied by a focus particle again shows that focus movement is obligatory; cf. Barbiers (2014).

The examples in (148) and (149) thus strongly suggest that the optionality of the focus movement in examples such as (147b) is only apparent: niet functions not as a sentence negation but as a constituent negation when the contrastively focused phrase follows it, i.e. we are dealing with the counter-presuppositional focus phrase niet op hem, which occupies the specifier of FocP as a whole.

The examples in (150) show that the examples in (148) and (149) alternate with constructions in which the designated focus position is not occupied by the whole contrastively focused phrase, but only by the focus particle; cf. Hoeksema (1989).

Usually both orders are possible, except when the focus particle is associated with a complement clause: as usual, such clauses are placed after the clause-final verbs. We illustrate this in (151a-b) with the focus particles alleen, but similar examples can be constructed for the other focus particles; example (151c) is added to show that the focus particle and the clause do form a constituent.

Barbiers (2010) has proposed that examples such as (150b) are derived by subextraction of the focus particle from the contrastively focused phrase, as in (152a); if this is correct, we can fully maintain the hypothesis that focus movement is obligatory. An alternative hypothesis would be that the focus particle is base-generated in the specifier of FocP as a scope marker (analogous to English constructions such as John hasn’t seen anybody, where the specifier of NegP is filled by the negative adverb not and the object appears in its regular position as a negative polarity item). If this proposal is correct, we need to revise the hypothesis that focus movement is obligatory by stating that the specifier position of FocP must be filled either by a moved focused phrase or by a focus particle with a contrastively focused phrase in its scope.

It is not easy to distinguish between the two hypotheses. Barbiers supports the movement analysis by claiming that the focus particle can be moved further into the clause-initial position; this is demonstrated with ook, but the result seems to deteriorate with the particles alleen and zelfs, as shown in example (153).

It is less clear what the base-generation hypothesis predicts. If the comparison with not in negative clauses such as John has not seen anybody is taken seriously, we would expect the focus particle to remain in its scope position because the adverb/particle not/niet in negative/contrastive focus constructions is not usually found in clause-initial position; this is shown in (154).

The examples in (155a&b) show that we find similar judgments when we move the contrastively focused PP across the (presumed) focus particle; example (155b) is only acceptable when the preposed PP is assigned a B-accent, i.e. when it functions as a contrastive topic, but in this case the adjective would normally be contrastively focused: cf. Op hem is Jan alleen/zelfs boos. The same seems to be true for (155c): cf. Op hem is Jan niet boos (maar woedend) At him Jan is not angry (but furious).

The above examples suggest that ook is not necessarily interpreted as a focus particle; in the (a)-examples in (153) and (155) it has an independent function, i.e. it is not associated with the PP. This is supported by the intransitive examples in (156): ook, zelfs and alleen are clearly focus particles when they immediately precede the contrastively focused phrase, as in the primed examples; an interpretation as a focus particle is much harder to get in the primeless examples, which is especially clear for alleen in (156c), where it is interpreted not as “only” but as “alone”.

The above discussion has shown that the argument in favor of the movement approach in (152a) based on the behavior of ook does not carry over to the other focus particles, while the base-generation approach seems to make a number of correct predictions about the examples in (153) through (156), provided we assume that ook can also be used with a syntactic function other than focus particle. Another argument in favor of the base-generation approach is that focused constituents can be embedded in islands for movement; cf. Hoeksema (1989:117-8). For instance, example (157b) shows that the wh-phrase op wie cannot be extracted from an embedded yes/no question, and (157c) shows that the same is true for the focused phrase alleen op Peter. This makes it unlikely that (157c') is derived by extracting the focus particle alleen from the embedded clause, and seems to refute the movement approach.

Example (157c') is consistent with the base-generation analysis in (152b): the focus particle alleen can be base-generated in the specifier of FocP of the matrix clause, since this results in a configuration in which this particle has the contrastively focused phrase op Peter in its a c-command domain. We tentatively conclude that the base-generation approach is superior to the movement approach.

Scalar focus particles like pasjust, alalready, nogstill, and maaronly must be associated with phrases that denote a linearly ordered scale. The focused phrase is typically a noun phrase containing a numeral or quantifier, as shown in (158). The numeral/quantifier selects a particular value from a contextually defined numerical scale (e.g. from one to twenty), and the particles qualify the part of the scale covered: maaronly indicates that this part is smaller than anticipated, while alalready indicates that this part is larger than anticipated. The fact that the particle and the focused phrase can be placed together in the main-clause initial position shows that they form a constituent.

In example (159a), the particles nogstill and alalready function as temporal adverbial modifiers of the eventuality denoted by Jan werken: the eventuality lasts longer or starts earlier than one might have expected. In example (159b), the particles alalready and pasjust function as adverbial modifiers qualifying the distance between speech time and the start of the eventuality: they characterize the time span as respectively longer and shorter than one might have expected. The adverbial use of the particle maar is restricted to non-stative verbs and expresses durative aspect: Jan praat maarJan talks on and on. Although Barbiers (1995a: §3) has shown that these temporal uses also involve a modification of a linearly ordered scale (the time axis), we will ignore such cases in the following discussion.

The scalar focus particles in (158) modify nominal arguments, but the (a)-examples in (160) show that they can also modify noun phrases embedded in a PP. The (b)-examples further show that it is also possible for such particles to modify the PP as a whole, with apparently the same meaning. The fact that in all these examples the PP must precede the adjective geïnteresseerdinterested, by which it is selected, shows that focus movement is obligatory in these cases.

That focus movement is obligatory is illustrated for direct objects in (161): while the definite noun phrase het boekthe book can easily follow the manner adverb zorgvuldig in (161a), the phrase modified by al must precede it.

Counter-presuppositional and scalar focus particles are similar in that they both trigger focus movement. However, they cannot be considered as belonging to a single category because they behave differently in other respects (although we will see that the judgments on the relevant data are sometimes not very clear).

First, the examples in (162) show that the scalar focus particles differ from the counter-presuppositional ones in that they are preferably adjacent to the focused phrase; cf. the examples in (150). Although examples like the primed ones in (162) are considered perfectly acceptable in Barbiers (2010), we have given them a percentage sign because a Google search (July 25, 2025) on the strings [maar in één ding geïnteresseerd is] and [maar geïnteresseerd in één ding is] revealed that, after omitting linguistic sources, only the former can be found on the internet (29 hits), although there were also 4 cases with the PP is extraposed position ([maar geïnteresseerd is in één ding]). The search results for the strings with the finite verb is preceding maar (i.e. in the verb-second position), as in [is maar in één ding geïnteresseerd] and [is maar geïnteresseerd in één ding], show the same thing: there were 71 hits for the first string and 4 hits for the second string (which can be analyzed as involving string-vacuous extraposition).

Second, the examples in (163) show that the scalar focus particles in (162) can more easily follow their associated focus phrase than the counter-presuppositional particles in (153)/(155). Note, however, that our Google search shows that this word order is not common: the search string [maar in één ding geïnteresseerd] yielded more than two hundred hits (where we stopped counting), while the strings [in één ding maar geïnteresseerd] and [in één ding * maar geïnteresseerd] yielded no relevant results.

Note in passing that the fact that scalar focus particles can either precede or follow their associated focus phrase can lead to ambiguity. Example (164) gives slightly adapted examples from Barbiers (1995a:70); the intended interpretation is indicated by underlining.

Third, although (163b') suggests that scalar focus particles can be stranded in the middle field of the clause, the examples in (165) show that they cannot be topicalized by themselves; the results are clearly more degraded than the comparable examples with counter-presuppositional focus particles in (153).

Barbiers (2010/2014) has shown that scalar and counter-presuppositional focus particles also differ in that the former can be doubled in certain varieties of Dutch, whereas the latter cannot. This contrast is illustrated by the examples in (166): we used the stative verb kennento know to rule out a temporal reading of the second occurrence of maar, which only occurs with dynamic verbs. The percentage sign in the (b)-examples indicates that at least some speakers of the standard variety do not (easily) accept the doubling of scalar focus particles.

Barbiers also observes that counter-presuppositional and scalar focus particles sometimes co-occur (with a slight difference in meaning in the case of ook ... al). The examples in (167) show that in such cases the former precedes the latter. The diacritics in the (b)-examples indicate that some speakers of the standard variety may find these examples somewhat marked.

The examples in (168) show that counter-presuppositional focus particles can also occur in front of the scalar focus particle, with the contrastively focused phrase in its base position; it seems that the counter-presuppositional and scalar focus particles must then be adjacent and be placed in the middle field of the clause.

Barbiers accounts for the data in (167) by assuming that scalar but not counter-presuppositional focus particles can be the head of a functional projection, which we take to be FocP. The primeless examples in (167) can now be derived by moving the contrastively focused phrase into the specifier of FocP, as indicated in (169a), while the primed examples can be derived from this structure by subsequent topicalization of the contrastively focused phrase. The primeless examples in (168) can be derived by placing the counter-presuppositional focus particle into the specifier position of FocP; we follow the earlier conclusion that the focus particle is base-generated as a scope marker in the specifier of FocP; cf. the discussion of (152). The fact that the singly-primed examples in (168) are marked may be due to the fact that these particles are not content-rich enough to undergo topicalization. The fact that the doubly-primed examples are highly degraded follows from the fact that ook al and alleen maar do not form a constituent.

If scalar focus particles do not only occur as the head of FocP, but can also be used to modify a contrastively focused phrase, the doubling of such particles can be derived in a similar way as in (169a); cf. (170a). Since the head of FocP can remain phonetically empty and scalar focus particles are not obligatory, the cases without doubling can be analyzed as in (170b&c); of course, contrastive focus constructions without any focus particle have the structure in (170d).

Recall from the discussion of (162) that there are reasons to assume that scalar focus particles cannot occur in the specifier of FocP, which would be supported by the fact that they cannot occur in structures such as (169b). This restriction would follow immediately if we assume that scalar focus particles are not phrasal in nature, and specifier positions can only be filled by maximal projections.

The claim that scalar focus particles can function as the head of FocP may also explain the contrast between the two primeless examples in (171). Barbiers (1995a:84-5) noted that while the particle maar cannot be interpreted as a modifier of the direct object twee vogels of the embedded clause in (171a), this is possible in (171b), where the direct object is extracted from the embedded clause by topicalization. This follows from the analysis discussed above, according to which the specifier of FocP must be filled in order to assign scope to the contrastively focused phrase; example (171a) is unacceptable in the intended reading because the object has not undergone long focus movement, while (171b) is acceptable in this reading on the assumption that long focus movement precedes topicalization.

The examples discussed in this subsection suggest that scalar and counter-presuppositional focus particles have a different syntactic status: while the latter are arguably heads in all their manifestations, the former show a more projection-like behavior. We will leave this to future research and refer the reader to Barbiers (2014) and Bayer (2019/2023) for more detailed discussion of particle doubling.

The discussion above has shown that the hypothesis that focus movement is obligatory in Dutch can be maintained, provided that we assume that the negative element niet is a focus particle when it expresses constituent negation; this is independently supported by the fact that niet and its associate phrase can be placed together in clause-initial position. Constructions with focus particles separated from their associate focused phrase may be an exception to the general rule, if in such cases focus particles are base-generated in the specifier position of FocP as scope markers (in the same way as niet is base-generated in the specifier of NegP in English negative clauses): on this assumption we have to refine the hypothesis that focus movement is obligatory by granting that the specifier position of FocP must be filled.