The one thing that all researchers seem to agree on is that the verb and the preposition that heads the PP-complement form a semantic unit, i.e. express a lexically determined meaning; the meaning of the prepositions in isolation is lost. For example, the functional prepositions op and voor in the primeless examples in (315) do not have the locational meaning of the prepositions op and voor in the primed examples. We ignore that (315b') is actually ambiguous between the two readings: this ambiguity disappears when the modifier vlakjust is present.

The meaning of the verbs can also be bleached: while the verb vechtento fight in (315b') implies that the agent is physically involved in the activity of fighting, this is not the most prominent interpretation of the verb vechten in (315b), which merely indicates that Jan is actively involved in some unnamed activity aimed at creating a better world. This ambiguity in verbs such as vechten means that in some cases it is not immediately clear whether we are dealing with a PP-complement or a PP with some other function. In example (316), for instance, the verb vechten can be used with the bleached, “metaphorical” meaning that we also find in (315b), or with the more “literal” reading in (315b'); in the former case the PP functions as a PP-complement, and in the latter as an adverbial purpose phrase of the type that we also find in Ze spaart voor een autoShe is saving money for a car.

Another case, taken from Schermer-Vermeer (2006), is given in (317). Example (317a) involves an adverbial comitative met-PP, as is clear from the fact that this example alternates with (317a'); cf. Section 3.4. In example (317b), on the other hand, the meaning of the verb has bleached and we may therefore be dealing with a PP-complement; a possible argument in favor of this is that this example no longer allows the alternation that we find in the (a)-examples.

The above examples show that the boundary between PP-complements and PPs with some other functions is diffuse. This may be due to the fact that the V + P collocation may be a lexicalized form of an otherwise productive grammatical pattern. Consequently, it may sometimes be difficult to use semantic criteria as evidence for one position or the other; the decision will then have to be made by appealing to a larger number of properties of the construction as a whole; we return to this issue in Section P32.3.3, sub IIB, with reference to a number of concrete examples taken from Schermer & Vandeweghe (2023).

Since the verb and the preposition form a semantic unit, it has been suggested that in order to speak of a PP-complement, the PP must be obligatorily present. This criterion would imply that the PP op zijn vader in (315a) is not a complement of the verb wachten, despite the fact that this example is often given as a prototypical case of a PP-complement. However, it may be possible to claim that PPs that cannot be omitted (without affecting the idiosyncratic meaning of the verb + P collocation) do involve a PP-complement; the fact that the PPs in (318) cannot be omitted in the reading “Jan relies on his luck” may then be considered sufficient to conclude that they are PP-complements; example (318b) without the PP is only acceptable with the meaning “Jan calculates”.

The examples in (315) have shown that clauses with a PP-complement and clauses with an adverbial PP can look very similar on the surface. The two cases can often be distinguished by replacing the PP by adverbial proforms like daarthere and hierhere. If we are dealing with an adverbial PP of place, it is usually possible, but it is not if we are dealing with a PP-complement: daar in (319a) corresponds to the adverbial PPs in the primed examples in (315), but not to the PP-complements in the primeless examples.

The fact that a PP-complement cannot be replaced by an adverb like daar or hier is to be expected, since it would result in the loss of the preposition, which forms a semantic unit with the verb. That it is truly the loss of the preposition that causes the problem with PP-complements can be gathered from the fact that R-pronominalization, which retains the preposition, is possible with PP-complements; the pronominal PPs in (320) are typically interpreted as PP-complements.

However, the possibility of R-pronominalization is not sufficient to conclude that we are dealing with a PP-complement: pronominal PPs like daarvoor/daarop can also be used as adverbial phrases, including locational ones. For instance, the examples in (317) from Subsection I cannot be distinguished by this test.

Since the verb and the preposition form a semantic unit, the preposition of a PP-complement can usually not be replaced by another preposition, unlike in the case of adverbially used PPs of place or time. Some examples are given in (321).

However, this does not provide a foolproof test for determining whether we are dealing with a PP-complement. A first complication is that non-locational and non-temporal adverbial PPs also have a restricted paradigm; for example, the preposition met in comitative PPs such as met zijn buurmanwith his neighbor in (317a) cannot be replaced by any other preposition either (with the possible exception of zonderwithout), which means that this test cannot be used to distinguish between the (a) and (b)-examples in (317).

A second complication is that certain verbs can select different PP-complements. In some cases, like the (a) and (b)-examples in (322), the choice of the preposition hardly affects the meaning of the verbs.

Less problematic are cases in which a different choice of preposition goes hand in hand with a different meaning: for cases like jagen op in (323a), which is interpreted literally as “to hunt”, and jagen naar in (323b), which is interpreted metaphorically with the meaning “to seek”, we can assume that we are dealing with two different lexical entries.

Table 8 presents a small sample of PO-verbs that are compatible with more than one preposition. Note that with these verbs, the paradigms of the prepositions are still quite limited; using any other preposition with these verbs will lead to an unacceptable result or to an adverbial reading of the PP.

Since the verb and the preposition of the PP-complement form a semantic unit, PP-complements differ from adverbial PPs in that they cannot be modified independently. This is illustrated in (324); the fact that the voor-PP in (324b) can be modified by the adverbial modifier vlakjust is sufficient to show that this PP is an adverbial phrase.

Modification is a typical property of locational and temporal PPs, because most other adverbial PPs do not have this property; cf. Section P34.3 for a small number of exceptions. This means that the inability of a PP to be modified is not sufficient to conclude that it is a PP-complement.

So far, we have focused on the fact that the verb and the preposition heading the PP-complement form a semantic unit. It is expedient therefore to emphasize that the verb and the preposition do not form a syntactic unit. In other words, it is not the case that the verb and the preposition display the behavior of a complex (transitive) verb. This is clear from the fact, illustrated in the (a)-examples in (325), that the complement of the preposition op cannot be topicalized and strand the preposition, unlike the object zijn vader of a transitive verb like kussento kiss in (325b).

Note that the string Zijn vader heeft hij opgewacht in (325a) is acceptable if opgewacht is understood as the participle of the particle verb opwachtento wait for/lie in wait for, hence the use of the number sign. A further complication is that some speakers do accept the string in (325a) in its intended sense. Klooster (2001:324) has suggested that these speakers construe the example as in (326), i.e. allow the deletion of the preposed R-word part of pronominal PPs under identity with the left-dislocated phrase zijn vader; cf. Section P36.3 for further discussion.

Although topicalization of the nominal part of a PP-complement is not possible, the formation of pronominal PPs and R-extraction are, as shown in (327), in which the resulting discontinuous pronominal PPs are italicized.

The possibility of R-extraction is a necessary but not a sufficient condition for assuming that a given PP is a complement. That it is not sufficient is clear from the fact that certain adverbial phrases, such as the instrumental met-PP in (328a), and PPs used as complementives, such as op het bed in (328b), also allow R-extraction.

Comitative PPs like met de buurman in (317a) do not easily allow R-extraction given that they involve a [+human] noun phrase; R-pronominalization and R-extraction are normally only accepted in relative clauses such as (329a); cf. Section P36.1. The fact that (329b) can only be construed as involving an inanimate theme can therefore not be used as an argument for claiming that the two PPs in (317a) and (317b) have a different syntactic status.

In the examples in (330), the PP-complement has undergone R-pronominalization. The R-word er in these examples has a function similar to that of the anticipatory pronoun hetit in examples such as dat Jan het al weet dat Peter ziek isthat Jan already knows it that Peter is ill, which is used to introduce the direct object clause dat Peter ziek is; like het, the R-word er functions as an anticipatory pronoun introducing a dependent clause.

Although not all PP-complements can be used as anticipatory phrases, the possibility of such a use is sufficient to claim complement status; cf. Paardekooper (1986). The anticipatory PPs in the examples in (331) can only be interpreted as PP-complements, not as adverbial phrases.

Since anticipatory pronominal PPs are necessarily PP-complements, example (332) shows that the collocation worstelen met must sometimes be analyzed as involving a PP-complement, consistent with our earlier claim that (317b) involves a PO-verb.

A complicating factor that needs to be mentioned is that in many cases the anticipatory pronominal PP can be omitted; a clear example of this phenomenon is given in (333a). In some cases, such as (333b), speakers seem to differ in their judgments about the omittability of the pronominal PP. Example (333c) shows that it is not always possible to omit the pronominal PP.

In fact, examples (334a&b) show that pronominal PPs must be omitted in the nominalized counterparts of the examples in (333a&b). The fact illustrated in (334c) that example (333c) cannot be nominalized is probably due to the fact that the pronominal PP cannot be omitted because nominalization is possible when the noun takes a regular PP: zijn berusting in haar komsthis resignation to her coming.

The fact that the PP can be omitted is surprising, given that the verb and the preposition form a semantic unit. However, there is reason to believe that the PP is still syntactically present even if it is not pronounced. To see this, first consider the examples in (335), which show that the anticipatory pronoun hetit blocks topicalization of clausal objects: (335b) is only acceptable if the pronoun is omitted.

The examples in (336) show that the presence of an anticipatory pronominal PP also blocks topicalization of the clauses in (333a&b). In this case, however, omitting the anticipatory pronominal PP does not improve the result; cf. Vandeweghe & Devos (2003). This can be taken as evidence for the claim that it is still syntactically present.

In (337), we provide a sample of PO-verbs that can be combined with a (finite or infinitival) dependent clause and we also indicate whether or not the anticipatory PP can be omitted in this case. The judgments are our own; it may be the case that other speakers have slightly different judgments. It is not clear to us what factors determine whether the pronominal PP must be overtly realized or can be omitted; cases in which the PP-complement co-occurs with a nominal object are indicate by adding iemandsomeone.

The previous subsections have shown that PP-complements differ from adverbial PPs in several ways: the adpositional head of a PP-complement forms a semantic unit with the verb, is part of a restricted paradigm and cannot be omitted if the PP is pronominalized. The prepositional head of an adverbial PP, on the other hand, has independent meaning, is part of a paradigm, and is suppressed when the PP is replaced by an adverb. This subsection discusses some additional syntactic tests that have been proposed to distinguish PP-complements from adverbial PPs.

We conclude with a brief remark on the distinction between PP-complements and PPs functioning as complementives. Although such predicative PPs can also be said to function as complements of the verb, they are sufficiently different not to be included in this subsection. Here, we will simply assume that PP-complements can be distinguished from PP-complementives by PP-over-V; the examples in (348) show that the former, but not the latter, can be placed after the clause-final verb. The syntactic behavior of PP-complementives will be discussed in more detail in Section P35.2.