It has been claimed that PO-verbs select at most one PP-complement; cf. e.g. Haeseryn et al. (1997:1179) and Neeleman & Weerman (1999: §5.5). However, there are many cases that could plausibly be analyzed as PO-verbs with two or more PP-complements. One possible case has already been discussed, namely the verbs of exchange in (310), repeated here as (405), which also constitute a possible counterexample to the hypothesis put forward in the introduction to this section on PP-complements, according to which a verb can take at most two complements.

However, we have seen that there are reasons to assume that the voor-PPs in these examples are adverbial phrases and not complements of the verbs. First, as was also noted in the introduction to Section 2.3, it is possible to paraphrase the examples in (405) by an ... en pronoun doet dat XP clause, in which XP is generally assumed to be an adjunct (the adverbial test): if this test is indeed conclusive, the examples in (311), repeated here as (406), show that the voor-PPs are adverbial phrases.

Second, we have seen in Section 2.3.2, sub IE, that PP-complements in the middle field of the clause cannot normally precede the direct object, as shown again in (407): the judgments presuppose a neutral (i.e. non-contrastive) intonation pattern.

The fact, illustrated in (408), that the voor-PPs in (405) can precede the direct objects provides further support for the claim that they are not PP-complements but adverbial phrases. We conclude that the examples in (405) are just apparent counter examples for the claim that PO-verbs can take at most one PP-complement (as well as for the claim that a verb can take more than two complements).

Many examples with potentially two PP-complements contain a so-called comitative met-PP, which typically refers to a co-agent of the activity denoted by the verb. At first glance, the examples in (409a&b) can be plausibly analyzed as cases involving the PO-verbs praten overto talk about and praten metto talk with. If these analyses are correct, one might conclude that example (409c) is a case in which pratento talk takes two PP-complements.

However, the ... en pronoun doet dat XP test suggests that, unlike the over-PP, the comitative met-PP is not entitled to PP-complement status, as shown by the acceptability contrast between the two examples in (410).

Since the examples in (409) do not contain a direct object, nothing can be inferred from the word order of these examples. In order to apply the word-order test, we need to construct examples with both a direct object and a comitative met-PP, and see if the PP can precede the direct object. Consider the examples in (411).

Since the met-PPs in these examples can precede the direct objects, it seems plausible to conclude that comitative met-PPs should never be considered complements, and thus that examples like (409c) and (411b) are not counterexamples to the claim that PO-verbs take at most one PP-complement. Moreover, the hypothesis that a verb can take at most two complements would also dictate that the comitative met-PP is an adjunct; if the met-PP were a complement of the verb, an example such as (411b) would contain three complements.

The claim that a PO-verb can take at most one PP-complement is not generally accepted; cf. De Schutter 1974: 227-8. It has recently been challenged in a series of papers by Vandeweghe & Devos (2003), Vandeweghe (2005/2011/2020), Colleman & Delorge (2010), and Vandeweghe & Colleman (2011). These papers argue that examples such as (409c) contain two PP-complements. This double PP-complement construction is claimed to occur typically with verbs of human interaction (communication, negotiation, etc.); the comitative met-PP in (409c) is assumed to be selected by the verb besprekento discuss, since it denotes an activity that requires at least two [+human] participants (in the prototypical case). The more recent papers further argue that the word-order test used in (411) only reflects a tendency and is not generally valid. Vandeweghe & Colleman (2011) investigated the behavior of the 11 PO-verbs with a direct object in the Twents Nieuws Corpus, and found that 70 (11.9%) of the 585 attestations appeared in the PP-NP order in the middle field of the clause. We have omitted two of their cases from our Table 11: bespreken metto discuss with, because it is for this case that we want to determine whether it takes a PP-complement or not (the results for this verb will be given later), and veranderen (in), because we would analyze this as a verb with a complementive; cf. note 12 in Vandeweghe & Colleman (2011). This resulted in a lower rate of PP-NP orders (7%), which is entirely due to the omission of bespreken met, since no PP-NP orders were found for veranderen (in).

example	translation	NP-PP-order	PP-NP order
baseren op	to base on	54	13
beschermen tegen	to protect against	92	0
beschuldigen van	to accuse of	64	0
danken aan	to owe to	47	3
herinneren aan	to remind of	50	0
herkennen aan	to recognize by	37	15
verdenken van	to suspect of	50	0
vergelijken met	to compare with	73	0
vervaardigen uit	to create out of	5	5
Total: 508		472	36

Admittedly, the results suggest that the test is not absolute and that specific factors can affect the order of the nominal and the prepositional object. This is not surprising, as De Schutter (1976) and Broekhuis (2004) have already noted that the preferred NP-PP order can be overridden by information-structural considerations. For instance, example (412a) shows that PP-complements can precede a direct object when the latter introduces new information into the discourse domain, in which case it is typically realized as an indefinite noun phrase or as a noun phrase preceded by a demonstrative pronoun. Example (412b) shows that the same is true for cases in which the direct object is a negative phrase; moreover, the use of zulke in this example seems to favor a contrastive focus interpretation of the PP, especially when it is stressed. Crucially, however, it is not possible to place the PP-complement before a direct object with a definite article, as shown in (412c); such examples are at best marginally possible and definitely require that the PP-complement is given a contrastive accent: ?dat Peter op deze feiten de nieuwe theorie baseerde.

In order to conclude that the word-order test is invalid, it is necessary to show that the verbs that allow the inverted order also allow this order if the direct object is definite and the PP-complement is not given special emphasis. Vandeweghe & Colleman do not indicate whether such examples have been found; all the examples in the article are of type (412a&b), and the same goes for the constructed examples in Colleman & Delorge (2010), which are all of type (412a). We conclude that the PP-NP orders reported in Table 11 are not underlying orders, but result from leftward focus movement of the kind discussed in Section 13.3.2.

It should also be noted that it is not always clear what the results indicate: it may be that the PPs we find with herkennento recognize and vervaardigento create are simply misanalyzed as PP-complements. This is difficult to prove, since no examples of these cases are given in Vandeweghe & Colleman, but that this may well be so is suggested by the fact that the aan-PP with herkennen in Jan herkende Marie aan haar rode jasjeJan recognized Marie by her red jacket is neither obligatory nor semantically implied by the verb. For instance, an example such as (413a) does not necessarily imply that Marie and the problem have specific characteristics by which Jan could or should recognize them. This contrasts sharply with example (413b), which does imply something that Jan could or should have waited for.

Similarly, an example such as (414a) does not imply that Peter has turned something into piano sonatas; it is even the case that the uit-PP can only be used in very special circumstances, e.g. if the sonatas contain recycled musical material. This again contrasts sharply with an example such as (414b), which implies that there are possible threats against which the population must be protected. We conclude that the uit-PP in (414a) is not a PP-complement; it is probably an adverbial or complementive phrase.

We conclude from the discussion so far that the raw numbers in Table 11 do not tell us much without a thorough investigation of the individual cases with the PP-NP order; the examples in (412) and (413) clearly suggest that they cannot be used as evidence for the claim that verbs can take more than one PP-complement.

Recall that Table 11 does not include bespreken metto discuss with, which appears in the PP-NP order in 35% of the attestations found by Vandeweghe & Colleman (34 out of 96). We tend to interpret this as evidence for adjunct status of the met-PP, and indeed there is independent evidence for this conclusion involving modification. Vandeweghe (2011/2014) notes that met-PPs can sometimes be modified by the element samentogether, and claims that this element can only be added if the met-PP functions as an adverbial phrase; the met-PP in (415a) is an adverbial phrase, whereas the met-PP in (415b) is a PP-complement. We added the primed (a)-example to show that the phrase samen met Marie can be placed in the sentence-initial position and therefore should indeed be considered as a single constituent; cf. the constituency test. Since Section 2.3.1, sub IV, has shown that modification is excluded in the case of PP-complements, we may accept the conclusion that we are dealing with an adverbial PP in (415a), although it remains to be shown that the met-PP in (415b) should be analyzed as a PP-complement.

The above means that we now have a test which can help us determine the syntactic status of the met-PPs in (409) and (411): if the met-PPs can be modified by samen, we can be sure that we are dealing with adverbial phrases; if this is impossible, we may or may not be dealing with PP-complements. Our judgments on the examples in (416) therefore clearly point in the direction of adjunct status for the met-PPs. This means that we can safely conclude that the hypothesis that PO-verbs take at most one PP-complement still stands.

However, this subsection has also shown that there are still many cases in which it is not immediately obvious whether we are dealing with a PP-complement or not; example (415b) may or may not be a PP-complement, or may be even be ambiguous. In order to be able to say more about such cases, we need to refine the tools currently at our disposal.

The examples in (417) show that, unlike in English, modal verbs can be used as main verbs in Dutch; cf. Section 5.2.3.2. They also show that modal verbs can sometimes take a PP-complement. The modal kunnencan can even take prepositional phrases headed by different prepositions; the PP-complement in (417a) is headed by buitenwithout while the one in (417b) is headed by tegenagainst; the difference in meaning suggests that these Vmodal + PP collocations are listed in the lexicon.

There are also cases in which the modal verb takes a particle and a PP-complement. An example is opkunnen tegento be up to in (418); it is not clear whether in such cases we are dealing with true modal verbs. To our knowledge, examples like these have hardly been studied, and at the moment we have little to say about them.

There is a large set of fixed and idiomatic verbal expressions that include PP-complements. Some examples are given in (419). A larger sample of these expressions is given in Table 12.

preposition	verbal expression	translation
aan	aandacht besteden aan deelnemen aan gebrek hebben aan grenzen stellen aan	to pay attention to to participate in to lack to put limits on
achter	haast/spoed/vaart zetten achter	to speed up
bij	baat hebben/vinden bij belang hebben bij	to benefit from to have an interest in
in	belang stellen in trek hebben in troost zoeken in	to take an interest in to feel an appetite for to find solace in
met	akkoord gaan met de draak steken met contact opnemen met	to agree with to make fun of to contact
naar	oren hebben naar navraag doen naar	to rather like to inquire about
op	acht geven/slaan op invloed uitoefenen op vat krijgen op	to take heed of to influence to get a grip on
over	de baas spelen over uitsluitsel geven over een vonnis vellen over	to boss about to give a decisive answer about to pass judgment on
tegen	wrok koesteren tegen een aanklacht indienen tegen represailles nemen tegen rancune hebben tegen van leer trekken tegen	to hold a grudge against to lodge a complaint against to take reprisals against to bear a grudge against to pitch into/to lash out
tot	aanleiding geven tot toenadering zoeken tot zijn toevlucht nemen tot	to give cause to to try to approach to resort to
tussen	het midden houden tussen een wig drijven tussen een onderscheid maken tussen	to stand between to drive a wedge between to distinguish between
uit	troost putten uit conclusies trekken uit	to find solace in to draw conclusions from
van	een afkeer hebben van afstand doen van last hebben van werk maken van	to have an aversion to to renounce to suffer from to take pains over
voor	partij trekken voor de tijd nemen voor het veld ruimen voor	to take sides with to take one’s time about to give way to
zonder	het stellen zonder (buiten)	to do without

In many cases the PPs feel like modifiers of the nominal part of the expression, which suggests that we are not dealing with PP-complements of the verb. The fact that the PPs can often easily be placed before the nominal part of the verbal expression strengthens this position, since we have seen that true PP-complements in neuter (non-contrastive) contexts usually cannot.

Still, assuming that the PPs are selected by the verbal expressions as a whole seems plausible. This is especially clear when the verbal expression can be replaced by a simple verb, as in the cases in (421).

There may also be cases in which the nominal part of the PP-complement is part of an idiomatic expression. Two possible examples are given in (422), although it should be noted that we could also be dealing with complementive constructions. The complementive analysis is particularly plausible for the PP in (422b) because it can also be used in a copular construction; cf. the contrast between *Jan is over zijn nek and Die winkel is over de kopThat shop is broke.

Also noteworthy are the verbal expressions in (423), in which the PP seems at first glance to be selected by a te-infinitive, as in te kampen hebben metto have to contend with, te lijden hebben vanto suffer from and te maken hebben metto have to deal with. Note, however, that te-infinitives do not normally precede the verbs in clause-final position. The fact that the te-phrases in (423) have to precede the verb hebben in clause-final position therefore suggests that we are actually dealing with PPs headed by te.

Finally, there is a set of more or less fixed expressions that contain non-referential het as subject or direct object. That het is non-referential in the examples in (424) is clear from the fact that it cannot be replaced by other (pro)nominal phrases.

We will not take as stand on the syntactic analysis of the above constructions, especially because they are probably lexically listed and are therefore not part of the core grammar, i.e. syntactically unproductive.