Consider the locational phrases in (86). That the modifiers of the PPs, dichtclose and hooghigh, are adjectives is clear from the fact that they can be modified by the degree modifiers zeervery and vrijrather, as shown in the primeless examples. The primed examples also support this claim by showing that these modifiers allow comparative and superlative formation.

The modifying adjectives and PPs in (86) form constituents, which is clear from the fact that they can be placed together in sentence-initial position; cf. the constituency test. Some examples corresponding to the primeless examples in (86) are given in (87); note that we have replaced the proper noun Peter in (86a) by an indefinite noun phrase in order to facilitate the topicalization of the modified PP.

The examples in (88) make clear that it is the adjective that modifies the PPs in the examples above, and not vice versa, by showing that omitting the adjective yields an acceptable sentence, while omitting the PP is impossible.

Example (89a) provides another argument for the claim that it is the adjective that modifies the adpositional phrase, and not vice versa: postpositional phrases such as het bos ininto the forest are normally used only as predicative complements (cf. Section P35.2.1), so there can be no doubt that in these cases the adjective functions as a modifier of the PP. The fact that (89b) has the same idiomatic reading as Jan zat in de putJan was down-hearted can be taken as a final argument for this claim.

Sometimes it is not so easy to determine the direction of modification. In (90a), for example, neither the adjective ver nor the PP van de bewoonde wereld can be omitted, so we cannot determine in this way what functions as the modifier in the complex phrase ver van de bewoonde wereld; we can only conclude that it is the adjective ver that modifies the PP by appealing to analogous examples such as (90b).

In (91a) it is even more difficult to determine the direction of modification, since the primed examples show that both the adjective and the PP can be omitted. An argument for the claim that it is the adjective hoog that acts as a modifier of the PP boven de stoel is that (91b) shows that it can be replaced by a nominal measure phrase such as twee metertwo meters. The unacceptability of (91b') clearly shows that the nominal measure phrase must be analyzed as a modifier and thus shows that an analysis of (91a) according to which the adjective acts as a modifier of the PP is at least possible. Whether the alternative analysis is also possible must be left to future research.

Adjectives do not only modify complete adpositional phrases, but can sometimes also modify particles, as in dichtbijclose, veraffar away and ver wegfar away. The fact that these cases involve adjectival modification is obscured by the fact that at least the first two combinations are usually orthographically represented as a single word, which may be related to the fact that the adjective is usually obligatory: omitting it in the examples in (92) results in unacceptability.

Since we have seen in (90a) that the same can hold for complete adpositional phrases, the examples in (92) do not provide conclusive evidence for compounding. That we are not dealing with compounding is clear from the examples in (93). Given the so-called right-hand head rule, according to which the rightmost member determines the properties of the compound, the compound analysis would wrongly predict that the modification possibilities of the complex forms in (93) reflect those of the particles; however, the primed examples show that modification by heel/zeer is a property of the adjectival part, not of the particle. The structure of the primeless examples should therefore be [[Mod A] prt], not [Mod [A-prt]].

Further evidence against the compounding analyses is provided by the examples in (94): the primed examples show that the adjectives dicht and ver can undergo comparative and superlative formation, which would of course be impossible if the adjectives were part of compounds.

The examples in (95) provide a number of modifiers for which adjectival status is not easily established, since they categorically resist modification by a degree modifier and comparative/superlative formation. The adverbs in (95b-d) can also occur in attributive or predicative position with related but not identical meanings.

Consider the temporal phrases in (96). The adjectives kortshort and langlong and the PPs following them form a constituent. This is clear from the fact, illustrated in the primed examples, that they can be placed together in sentence-initial position; cf. the constituency test.

That the modifier is adjectival in nature is clear from the fact that it can be modified by a degree modifier. Comparative and superlative formation, on the other hand, seem to give rise to marked results.

The complementizers voordatbefore and nadatafter are compounds consisting of a temporal preposition and the complementizer datthat; cf. Section P33.4.1, sub II. The examples in (98) show that dependent clauses introduced by these complementizers can be modified in the same way as the PPs in (96). The primed examples show that the adjective and the dependent clause form a constituent.

The examples in (99) again provide a number of modifiers for which adjectival status is not easily established, since they categorically resist modification by a degree modifier and comparative/superlative formation; cf. example (95).

Dutch has a large number of idiomatic prepositional expressions that can be used in complementive position; some examples are given in (100). We are dealing with idioms here, as can be seen from the fact that attributive modification of the nominal complement of the PP is usually excluded; cf. Section P32.3.3, sub I, for further discussion.

The idiomatic PPs in (100) share several similarities with scalar adjectives. First, they denote some (mental) property of the subject of the clause, just as adjectives like gelukkighappy or tevredensatisfied do. Second, the primeless examples in (101) show that these PPs can be modified by a degree modifier, which can be zeer, i.e. the modifier generally assumed to occur only with adjectives. The primed examples are added to show that the adjectives and the PPs in the primeless examples form a constituent; cf. the constituency test.

Finally, the PPs in (100) seem to be eligible for comparative/superlative formation just like the scalar adjectives. This is shown in (102).