Section 24.3.1, sub III, pointed out that pseudo-participles and a limited number of deverbal adjectives differ from other set-denoting adjectives in that they allow their PP-complement to occur on their left. The same is true for the stranded preposition of a PP-complement; the stranded preposition is sometimes even obligatory on the left. Finally, we have seen that degree modifiers such as ergvery can precede the preadjectival PP-complement and must precede the preadjectival stranded preposition. The relevant examples are repeated in (143).

The examples in (143) show that the modifier erg does not modify the head of the AP: we would expect it to be adjacent to the adjective and thus to follow (the stranded preposition of) the PP-complement. We therefore conclude that it modifies the lexical projection of the adjective as a whole; the structure of the examples in (143) is roughly as given in (144), where PP stands for the base position of the PP-complement of the adjective and P for the position of the stranded preposition.

Now consider the periphrastic majorative and maximative constructions in (145). These examples show that, like the modifier erg, the periphrastic elements meer and het meest can also precede the preadjectival PP-complements (see also Section 26.1.3, sub VC), and must precede the preadjectival stranded prepositions. The same orders arise in minorative and minimative constructions, i.e. by replacing the periphrastic elements meer and meest in (145) by minder and minst, but this will not be illustrated here.

The word orders in (145) show that periphrastic comparative/superlative elements do not modify the head of the AP; these examples can be assigned to the structures in (146), which are similar to those in (144) in all relevant respects. So, the word-order similarity between the examples in (143) and (145) is a first indication that modification and comparison formation can or must be treated on a par.

Obviously, a similar argument cannot be made for morphological comparative or superlative forms, since these are derived with affixes that need to be supported by a stem: this means that the adjectival stem and the comparative/superlative suffix are, by definition, strictly adjacent. Let us now take a closer look at an adjective such as geschikt voor ...suitable for, which allows both the periphrastic and the morphological comparative/superlative form. The examples in (147b&c) show that the stranded preposition can either precede or follow the adjective in the periphrastic construction, just as in (147a), where the adjective is modified by the degree modifier erg.

However, when we use the morphological comparative or superlative form, the placement of the stranded preposition before the adjective leads to unacceptability, as shown in (148).

The striking unacceptability of the order with the stranded preposition preceding the adjective can be accounted for by assuming that the morphological comparative/superlative form occupies the same position as the elements erg, meer, and meest in (144) and (146). Since it is reasonable to assume that the base structures of the examples in (147) and (148) are similar, the surface position of the adjective must be the result of leftward movement. Consequently, the derivations of the examples in (148) start with the base structure in (149a), where the stranded preposition either precedes or follows the adjective. Next, the morphological form of the comparative/superlative is derived by moving the adjective to the position of the affix -er/-st, as in (149b), yielding the morphologically complex forms A-er and A-st. The result of the leftward movement of the adjective is that comparative and superlative forms necessarily precede stranded prepositions.

Leftward movement of the adjective, as in (149b), does not only apply in case of morphological majoratives/superlatives. Consider the examples in (150). In (150a), the stranded preposition of the PP-complement of the pseudo-participle bedachtcautious can either precede or follow the adjective, but when bedacht is modified by the complex modifier zo ... mogelijkas ... as possible, as in (150b), the stranded preposition must follow it; in fact, it must also follow the element mogelijk. The same is true when the adjective is modified by genoegenough, as in (150c).

The pattern in (150) can be explained by assuming that the elements mogelijk and genoeg occupy the same position as the affixes in (149), and that the adjectives must undergo the same leftward movement as in (149b); the base structure of the examples in (150b&c) is as indicated in (151a), and the surface orders are derived by moving the adjective to a position in front of mogelijk/genoeg, as in (151b).

If this proposal is on the right track, it can be seen as a second argument for treating modification and comparative/superlative formation on a par. More evidence for this view will be given in Section 26.3.2.