Like adpositions, als and dan can take a noun phrase as their complement. They differ from adpositions in that they usually do not assign case; in a sense they are “case-transparent”, which we will illustrate here by their occurrence in comparison constructions. The noun phrase complement of als/dan in (417) has the same case as the noun phrase in the main clause with which it is compared: when it is compared with the subject of the clause, as in (417a), it is assigned nominative case; when it is compared with the object of the clause, as in (417b), it is assigned accusative case. The percentage sign in the (a)-examples is used to indicate that non-nominative pronouns are fairly common in speech, but are generally considered substandard. Nevertheless, their occurrence shows that for many Dutch speakers als and dan can also function as a regular preposition.

Constructions of the type in (417) will be discussed in more detail in Section , where we will argue that als and dan do not take a simple noun phrase as their complement but a reduced finite clause; examples such as Ik zie haar vaker dan hij/hem will be analyzed as roughly indicated in (418); note that the non-reduced counterparts of these examples are also acceptable (albeit somewhat awkward).

Als-phrases can also be used in other constructions, as shown in (419a) and (419b), where they have the function of a complementive and a supplementive, respectively. In such cases, we cannot check whether als assigns case, since the bare nominal complements cannot be replaced by a pronoun. However, that we are not dealing with an ordinary adpositional construction is shown by the fact that sometimes als takes not only nominal but also adjectival complements, which is certainly not usual for adpositional phrases.

Like als and dan, behalve is “case-transparent”: if the phrase headed by behalve excludes entities from the set denoted by the subject of the clause, as in (420a), its complement is assigned nominative case; if it excludes entities from the set denoted by the object, as in (420b), its complement is assigned accusative case.

The same thing holds for the form uitgezonderd; as in (420a), the pronoun in (421a) is assigned nominative case. Note that the more colloquial phrasal adposition met uitzondering van in the primed examples assigns objective case.

We refer the reader to Section for a more detailed discussion of such cases; there it will be argued that behalve and uitgezonderd are like als/dan in comparative constructions in that they do not take a simple noun phrase as their complement but a reduced clause.

The phrase van die + NP can be used in regular nominal positions; it is used as a direct object in (422a) and as a subject in (422b). In the latter case, the verb agrees with the noun in number, which clearly shows that we are dealing with a spurious PP in these examples. Note that the expletive er is used in (422b), which shows that the phrase van die + NP functions as an indefinite noun phrase.

In general, the noun phrase in this construction is plural. The only exception is the case in which the noun phrase is headed by a substance noun. The cases in (423) show that the determiner die/datthat then agrees in gender with the mass noun, just as in the case of regular demonstratives.

The sequence van die + NP is often preceded by a numeral or a quantifier: cf. vier/sommige van die lekkere broodjesfour/some of those nice sandwiches. Such cases are discussed in detail in Section , where the claim that the sequence van die + NP is actually a noun phrase is also discussed more extensively.

As in the case of the van die + NP construction, number agreement on the verb is triggered by the nominal complement of wat voor. Again, this suggests that voor does not function as a regular adposition; cf. Section for a more detailed discussion of this construction.

The examples in (425) show that manner adverbials in Dutch can have the form of an AP such as luidloudly or a PP such as op luide toonloudly/in a loud tone. The two cases are nearly synonymous and can be questioned with the interrogative manner adverb hoehow: Hoe sprak Jan?How did Jan speak?.

Corver (2022b: §4) discusses the use of complex A+P formations in the same function, where the prepositional part can be at least opon and uitout. The examples in (426) give an example for each case.

Formations of this kind have not been discussed extensively in the theoretical literature to date, and in fact it is far from clear what the analysis of such forms would look like. Given that the complex forms consist of two easily identifiable parts, viz. the adjectives hard/lang and the adposition op/uit, two possible analyses spring to mind: we are dealing with an adjectival structure or an adpositional structure. Assuming that we are dealing with a morphologically complex word, the first analysis should be rejected because it would violate the independently motivated right-hand head rule, according to which the rightmost member (here the adposition) in a morphologically complex word determines the category of the full form. The syntactic behavior of complex forms also suggests that they are not adjectives. First, the use of these forms in predicative position leads to a degraded result; cf. Het/Dat is hard/??hardop It/that is loud. Second, the impossibility of attributive inflection in the det-inf nominalizations in the primed examples in (427) shows that they are not attributive adjectives but adverbial phrases; cf. Section N14.3.1.

This leaves us with the second possibility; we are dealing with an adpositional structure. However, there are also strong reasons to think that we are not dealing with a complex adposition. First, because luidop and languit do not take a nominal complement, we should conclude that the complex form is an intransitive adposition, but these are typically used as verbal particles, not as adverbial phrases. An even larger problem may be that the adjectival part of the combination can be combined with a degree adverbial or even comparative morphology, as in the following examples with rechtop (zitten)to sit upright and vrijuit (praten)to talk freely; note that we switch to different forms because the results vary from case to case.

Since the first member of a compound cannot be modified in this way, this would show that we are not dealing with a complex word, but with a phrase. Unfortunately, it is not easy to show beyond doubt that the degree modifier modifies the adjectival part (i.e. [[modifier A] P]) and not the complex as a whole (i.e. [modifier [A P]]); cf. Schermer (2024). Crucial support for this analysis is provided by the fact that for many speakers the adjectival part can also undergo comparative (and superlative) formation. As this is usually excluded in morphologically complex forms, this provides quite strong evidence for a phrasal analysis; we are dealing with a PP headed by op. The structure in (429) is still not very satisfactory, since it leaves open the relation between the adpositional head and the adjectival phrase.

Corver (2022b) argues that the most likely analysis is that we are dealing with a postpositional phrase, i.e. with the AP as the complement of the adposition. However, since adpositions usually do not take an adjectival complement, he proposes that the AP is part of a larger nominal complement with a silent (phonetically empty) nominal head expressing a manner component, represented as manner in (430). The traces in these representations indicate that the NP in the postpositional phase originates in the complement position of the preposition, as discussed in Section 32.2.6; the reader can ignore this here.

Corver (2022b) presents this proposal as a preliminary analysis that raises many new questions. Postpositional phrases, for instance, are generally transparent to movement; the examples in (431) with the directional postpositional PP het bos in show that the nominal complement of the postposition can be scrambled or topicalized to the left.

This leads to the expectation that the noun phrase in the PP-structures in (430) can also be extracted from the PP by leftward movement, but the interrogative examples in (432) show that this is not borne out, which of course requires an explanation.

This unexpected behavior can be explained by the syntactic function of the PP. The PP het bos in in (431) functions a complementive, which generally allow extraction; this is illustrated in the (a)-examples in (433) for R-extraction from a complementive op-PP in a relative clause. Manner adverbials on the other hand do not easily allow extraction, as is illustrated in the (b)-examples for R-extraction from a prepositional manner adverbial in a relative clause; relativization favors pied piping of the preposition (although stranding is at least marginally accepted by some speakers).

Another potential problem is that there are A+P forms that cannot be used as manner adverbials and therefore cannot be assigned a structure similar to those in (430); cf. Schermer (2024: §4) for examples. This is illustrated by the examples in (434). Corver analyzes kortafcurtly as a manner adverbial in the same way as the forms hardoploud and languitsprawled in (426). However, the examples in (434b&c) show that kortaf differs from these cases in that it can be used in attributive and predicative position; cf. also Corver (2022b:fn.25), where it is mentioned that kortaf also allows the comparative form kortaffer (which is more frequent on the internet than korteraf). These examples show that kortaf can in fact be a set-denoting adjective, and consequently can be analyzed as a supplementive PP in (434a).

The review in Schermer also notes that the status of the empty noun manner is unclear. Corver (2022b:63) claims that the forms such as hardoploud and languitsprawled can be paraphrased by P A wijze. The examples in (435) show that this may be true for hardop, although the result is somewhat clumsy, but it is clearly false for e.g. languit. It is thus clear that it cannot be it is simply elided (as is also clear from other cases, in which an appeal to silent nouns has been made; cf. Section NQ for some other cases.

The examples in (434b&c) thus show that there are apparent A+P forms that should be analyzed as lexical items, i.e. adjectives, and that we have to decide from case to case whether or not we are dealing with complex forms and, if so, what their internal structure is. We leave this to future research, and refer the reader to Schermer (2024: §4) for a more detailed discussion of a larger set of potential A+P forms.