The examples in (22) show that topicalization of a predicative adpositional phrase referring to a (change of) location is easily possible.

Scrambling of locational complementives generally leads to degraded results. As shown in (23), the locational PP usually immediately precedes, i.e. is left-adjacent to, the verb(s) in clause-final position.

However, there are at least two exceptions to this general rule. First, A'-scrambling is possible if the locational complementive is assigned emphatic focus (indicated by small caps) or is preceded by a focus particle such as zelfseven. Thus, while leftward movement is excluded in the neutrally pronounced example (24a), it is possible in the examples in (24b&c).

Second, example (25b) shows that scrambling is also licensed if the located and the reference object are quantified. At first sight, the two variants in (25a&b) seem to have the same meaning in (25c): for every vase (in the domain of discourse), there is a rose such that Jan puts that rose in it. The main difference is that the order in (23a), in which the noun phrase een roos precedes the PP seems to prefer emphatic focus on the nominal object.

However, the claim that the two orders in (25a&b) express the same meaning appears to be false: the two orders in (26a&b) clearly express different meanings. The most prominent reading of (26a) is that one of Janʼs fingers is put in all bowls (e.g. Jan is tasting the content of each bowl by using the index finger of his right hand), while (26b) is most easily interpreted as involving more fingers, with each finger put in another bowl. If these intuitions are correct, we can conclude that the two orders differ in the relative scope of the two quantified nouns, as is formally expressed in (26a'&b').

The conclusion that the order of the noun phrase and the PP (26a&b) reflects the relative scope of the two phrases also explains the fact that placing the PP left-adjacent to the verb leads to a somewhat marked result in (25a) if the sentence is pronounced without emphatic focus on the noun phrase: the resulting interpretation that there is only one rose that is put in each vase does not refer to a plausible situation.

In conclusion, the discussion in this subsection strongly suggests that scrambling of locational PPs is excluded unless (i) they are assigned emphatic focus or (ii) the scope of the located and the reference objects is reversed.

Although the precise judgments on the acceptability of the primed examples in (27) seem to vary somewhat among speakers, the general consensus seems to be that PP-over-V of complementive adpositional phrases denoting a (change of) location yields a marked result.

It should be noted, however, that the ban on PP-over-V is lifted when we are dealing with particle verbs or certain prefixed verbs. Example (28) illustrates this for the particle verb neerleggento put down; we will return to this fact in Section 35.2.1.2, sub II and III.

R-extraction from predicative adpositional phrases denoting a (change of) location is easily possible. The examples in (29) illustrate this by means of a relative clause, in which the relative R-pronoun waar corresponds to the complement of the locational adposition.

The previous subsections have shown that while locational complementives do allow topicalization, they cannot be scrambled or extraposed. Furthermore, we have seen that these complementives allow R-extraction. Subsection III will show that directional complementives exhibit more or less the same behavior.

Topicalization of directional complementives requires that contrastive accent be assigned to a specific part of the adpositional phrase; the part that must be accented depends on whether we are dealing with a prepositional, postpositional, or circumpositional phrase. Example (31a) shows that topicalization of a prepositional phrase requires that the nominal complement of the PP be contrastively accented (indicated by small caps). Topicalization of a postpositional phrase, as in (31b), requires contrastive accent on the postposition; if the accent is on the nominal complement of the postposition, the complement can be topicalized, while the postposition must be stranded in its original position. Topicalization of a circumpositional phrase requires contrastive accent on the second part of the adposition, as in (31c).

The examples in (32) show that, as in the case of the predicatively used locational PPs, scrambling is generally impossible with directional PPs. Note that while (32a') can be saved by giving the naar-PP a contrastive accent, this does not improve the use of the postpositional examples in (32b'&c').

For completeness’ sake, note again that the nominal complement of the postposition in the (b)-example leads to an acceptable result: cf. dat Jan de heuvel welgemoed op reedthat Jan drove up the hill in good spirits.

The examples in (33) show that, as in the case of the predicatively used locational PPs, PP-over-V of directional adpositional phrases leads to a degraded result.

R-extraction from directional prepositional phrases gives rise to a degraded result. This is illustrated in (34a) by means of a relative clause, in which the relative pronoun corresponds to the complement of the directional preposition naar; cf. Section 36.2.1, sub IB, for a more detailed discussion. R-extraction from postpositional phrases is also excluded; example (34b) is allowed, but strongly favors a locational reading of the adposition op, which shows that we are dealing with the preposition op, not the postposition. Example (34c) shows that R-extraction from circumpositional phrases, on the other hand, is easily possible.

Example (35b) shows that (34a) becomes grammatical when the element toe is added; this is not surprising in view of the acceptability of (34c), since we are then probably dealing with the circumposition naar ... toe in (35a). It is not clear whether the unacceptability of (34a) is due to some (unknown) syntactic constraint, or whether (34a) is blocked by (35b), because the latter is preferred for some reason.

In order to obtain a directional reading in (34b), waar must be replaced by the relative pronoun diethat as in (36b). This is remarkable, given that the use of the relative pronoun die in prepositional and circumpositional constructions like (36a&c) produces ungrammatical results.

Note that the judgments on the (b)-examples in (34) and (36) are somewhat idealized and actually vary somewhat among speakers. Furthermore, there are some postpositional constructions in which the R-pronoun can also be used. Since a detailed discussion of these would lead to a lengthy digression, we postpone this issue to Section 36.2.2.

The acceptability of (36b) is most likely related to the fact that the postposition op and its nominal complement need not be adjacent: the nominal complement can be topicalized, as in (31b'), repeated here as (37a), and it can be scrambled, as in (37b). Moreover, in the southern varieties of Dutch, it can be separated from the postposition by incorporating the latter into the verb cluster, as in (37c).

Although the nominal complement of the postposition seems to behave as an independent constituent in this respect, it is important to note that it is assigned case not by the verb but by the postposition. Note that examples such as (38a) cannot be passivized; indeed, the fact that verbs of traversing are unaccusative a priori militates against the assumption that the noun phrase de heuvel is assigned accusative case by the verb rijdento drive in (38a).

That the postposition op assigns case to the noun phrase de heuvel in (38a) is also supported by the fact that it is the located object de autoʼsthe cars that is assigned accusative case by the verb in its causative counterpart in (39b). As expected, passivization results in the promotion of the located object to subject, not of the reference object de heuvelthe hill, as can be easily verified by (the lack of) number agreement between the noun phrases and the finite verb in passive construction (39b).