Sequences of the form A + van + noun are mostly more or less fixed collocations. The noun in the van-PP is never preceded by a determiner and cannot be modified by an adjective. If the adjective is gradable, it can be modified by a degree modifier, but not by an approximative/absolute adverb like vrijwelalmost or helemaalcompletely. From this we can conclude that the sequences in (335) are scalar.

The examples in (336) show that the sequence A + van + N can be used in copular verb and vinden-constructions. However, the sequence cannot easily be used in resultative constructions such as (336c), which suggests that the complex APs are individual-level predicates; cf. Section 28.2.1, sub II, ex. (27). The fact, illustrated in (336d), that the sequence cannot be used as a supplementive either, is consistent with this conclusion; cf. Section 28.3, sub IV. In this respect, the examples in (335) differ from sequences such as rood van opwindingred with excitement (to be discussed in Subsection B below), which typically function as stage-level predicates.

An exception to the two generalizations above is the sequence gelijk van lengte in (337): it can be modified by vrijwelalmost or helemaalcompletely, but not by the degree modifier zeer. It can function as a stage-level predicate as is clear from the fact that it can be used in resultative constructions. The exceptional behavior of the sequence gelijk van lengte may be related to the fact that it is the only case that alternates with the sequence with the preposition in: gelijk in lengteequal in length.

The sequences in (335) form a constituent, which is clear from the fact that the whole sequence can be placed in clause-initial position; cf. the constituency test. This is shown in (338).

The sequence is not easy to split. The examples in (339) show that wh-movement or topicalization of the adjective leads to a marked result when the van-PP is stranded; these examples improve somewhat when the adjective is heavily accented.

Movement of the van-PP cannot strand the adjective either, as shown in (340); the (a)-examples involve PP-over-V, and the (b) and (c)-examples are derived by leftward movement of the PP.

Attributive use of the sequence A + van + N is impossible. Since we have just established that the van-PP must be right-adjacent to the adjective, this is correctly excluded by the head-final filter on attributive adjectives, which requires the adjective to be immediately adjacent to the modified noun; cf. Section 27.3, sub IB.

In general, the sequence A + van + N denotes a property of human (and other animate) beings. Consequently, it is expected not to occur in the partitive genitive construction, since this construction usually denotes [-human] entities; see the contrast between iets leukssomething nice and *iemand leuks (lit.: someone nice), discussed in Section 29.2.3. Whether this fully explains the impossibility of the sequence in the partitive genitive construction is not clear, however, since the AP knap van opzet in (342b), which is exceptionally predicated of a [-human] noun phrase, cannot enter the partitive genitive construction either.

In constructions such as (343), the van-PP does not express a restriction on the adjective, but instead indicates the cause of the property denoted by the adjective; this cause is often a mental state of the argument of which the adjective is predicated, or something external that can affect the physical state of the argument of which the adjective is predicated. The examples in (343) are similar to those discussed in Subsection A: they denote properties of human beings, and often have an idiomatic flavor. They differ, however, in that the noun must be preceded by a definite determiner if it denotes an external cause, as in (343c); if the noun denotes a mental state, the determiner is usually absent in this type as well (the percentage sign indicates that isolated cases with the article present can be found on the internet).

Modification of the adjective by a degree modifier generally leads to a degraded result, although modification by the absolute modifier helemaalcompletely is easily possible. This would suggest that the construction in question is not gradable and may even be absolute, which may be related to the fact that the construction is often metaphorical in nature; someone who is gek van angstcrazy with fear is not crazy, but very much afraid. Perhaps this can be supported by the fact that the use of a degree modifier in (344a) produces a much better result than in the other cases; one can literally turn red from excitement, while one does not literally turn green from envy.

Additional support for the proposed explanation comes from the examples in (345), which show that comparative formation leads to similar judgments as modification by a degree modifier such as ergvery, shown above.

The sequence A + van + N(P) can be used in copular constructions, but not, for reasons that remain unclear, in the vinden-construction. The fact that the sequence can also be used in resultative and supplementive constructions is consistent with the fact that the adjectives involved are °stage-level predicates: in contrast to the adjectives discussed in Subsection A, they denote transitory properties.

The sequences in (343) form a constituent; this is illustrated by the examples in (347), which show that the entire sequence can be placed in clause-initial position; the constituency test.

As in the constructions discussed in Subsection I, wh-movement and topicalization of the adjective yield a degraded result with a stranded van-PP. Note that the high degree of unacceptability of (348b) may be due to the fact that the AP gek van angst is not gradable.

However, the (a)-examples in (349) show that PP-over-V leads to an acceptable result, although Dutch speakers have different preferences regarding the placement of the PP; some prefer preverbal placement of the van-PP, while others strongly prefer postverbal placement (even to the point of claiming that preverbal placement is unacceptable). The (b) and (c)-examples show that leftward movement of the PP leads to unacceptable results.

Since PP-over-V is possible, we might expect that it would be possible for the sequence A + van + N to be used attributively with the van-PP in postnominal position, but (350) shows that this expectation is not borne out.

Note that the constructions in (351a&b) are acceptable. The unacceptability of (351c) suggests that such examples must be interpreted literally.

Possibly the examples in (351a&b) involve a third type of construction, since the AP is predicated not of a person but of a body part. This suggestion is supported by at least two facts: first, example (351a) shows that the noun opwinding can at least optionally be preceded by a definite article; second, the examples in (352) show that the van-PP can be easily moved leftward into some clause-internal or clause-initial position; cf. the (a) and (b)-examples in (349).