PP-complements of verbs are always headed by a functional preposition, i.e. a preposition selected by the verb with relatively little semantic content; cf. Table 29 in Section 32.3.3, sub IIB for a representative set of examples. All PP-complements of verbs allow R-extraction. We illustrate this in (94) for some of the examples in Table 29.

The set of functional prepositions heading PP-complements of adjectives seems to be slightly smaller than the set of functional prepositions heading PP-complements of verbs; cf. Table 29 in Section 32.3.3, sub IIB. All PP-complements of adjectives allow R-extraction. In (95) we illustrate this for some of the examples in Table 29.

The set of functional prepositions that head the PP-complements of nouns is also smaller than the set of functional prepositions that head the PP-complements of verbs (cf. Table 29 in Section 32.3.3, sub IIB). Although the adpositional complements of nouns do allow R-pronominalization, the possibilities for R-extraction (the split pattern) are limited: movement of er preferably targets some position within the noun phrase; movement to a position external to the noun phrase usually leads to a marked result. We illustrate this in (96) for some of the examples in Table 29.

The only cases where R-extraction seems to be acceptable are indefinite and demonstrative noun phrases that take an adpositional complement headed by the prepositions overabout or vanof. Some examples are given in (97).

Furthermore, there are poorly understood restrictions governing the possibility of R-extraction in examples like (97). For example, if the main verbs are replaced by a verb like verbrandento burn, the split pattern gives rise to a degraded result. It is therefore not obvious whether the split pronominal PPs in (97) really function as complements of the nouns, or whether they function as e.g. restrictive adverbial phrases. We will not digress on this issue here; for a detailed discussion of this issue we refer the reader to Section N15.2.2.1 and subsequent discussions.

Section 33.2 has shown that only a few prepositions take adpositional complements. The list is exhausted by vanfrom, totuntil and voorfor in (98). The single-primed examples in (98) show that the adpositional complements of these prepositions can be replaced by pronominal PPs. The doubly-primed examples, however, show that these pronominal PPs cannot be easily split; leftward movement of the R-word seems impossible, regardless of whether the landing site is internal or external to the PP.

Adverbially used locational phrases can usually be replaced by a locational pro-form like daarthere or hierhere in (104b); R-pronominalization, and hence R-extraction, usually yields an unacceptable result.

Note that some speakers report different judgments on relative constructions; they allow the preposition op to be present in examples such as de zolder waar Jan vaak (%op) speelt. However, most speakers appear to prefer the use of the pro-form waar (i.e. the construction without op).

Section 36.2.1, sub II, has already shown that only the temporal prepositions voorbefore and naafter allow the formation of a pronominal PP, as illustrated once more by the contrast between (105b&c) and (105d). However, this does not imply that R-extraction is also allowed with the prepositions voor and na, and the examples in (105b'&c') show that this is in fact not the case; the R-pronoun must be adjacent to (the modifier of) the preposition, and the split pattern leads to unacceptability.

The impossibility of R-extraction is probably due to the fact that temporal PPs usually have an adverbial function, which is included in the general prohibition of extraction from adverbial phrases; cf. the discussion in Section 36.3.1.

Section 36.2.1, sub IIIC, has shown that adverbial PPs headed by non-spatial/temporal prepositions do not allow R-pronominalization. For obvious reasons they therefore do not allow R-extraction either.