A first potential counterexample to the claim that only perfect and passive auxiliaries select a verb in the form of a participle is given in (111a), in which the verb raken'to get' seems to select the participial form of the verb irriteren'to annoy'. There are, however, reasons for assuming that we are dealing with a semi-copular construction of the type in (111b), in which gewond'injured' must be seen as a pseudo-participle as the corresponding verb wonden is obsolete and replaced by the morphologically more complex verb verwonden in present-day Dutch. If this line of reasoning also applies to (111a), the participle geïrriteerd is not verbal but adjectival, and the primed examples in (111) show that this correctly predicts that it behaves just like the pseudo-participle in that it must precede the verb raken in clause-final position. Since the (a)-examples in (111) are more extensively discussed in Section 2.5.1.3, sub IID, we refer the reader to this section for further discussion.

Haeseryn et al. (1997:962) mention cases similar to (111a) with the verb krijgen'to get', but again the position of the participle with respect to the finite verb in clauses-final position suggests that we are dealing with a semi-copular construction; an example such as (112a) can be given a similar analysis as the construction in (112b). We refer the reader to Section A6.2.1, sub I, for a more extensive discussion of this type of semi-copular construction.

In a limited number of cases modal verbs may take a participle as their complement. Since such constructions normally alternate with constructions with an additional perfect or passive auxiliary, Haeseryn et al. (1997:960-2) suggest that they are derived by elision of the auxiliary. Although this seems plausible at first sight, things may not be as simple as that. The following subsections discuss two cases: we start with modal verbs like lijken'to appear', schijnen'to seem' and blijken'to turn out', which may take a te-infinitival clause as their complement, after which we discuss modal verb like moeten'must' and kunnen'can', which may take a bare infinitival clause as their complement.

Haeseryn et al. (1997:963-4) mention a set of collocations consisting of a verb and a participle. Some examples are: (ergens) begraven liggen'to be buried (somewhere)'; (iemand iets) betaald zetten'to get even with someone'; (zich) gewonnen/verloren geven'to admit defeat'; geschreven/vermeld/genoteerd staan'to be recorded', verschoond blijven (van)'to be spared'; opgescheept zitten (met)'to be stuck with'. As Haeseryn et al. notice themselves, there is reason to doubt that the participles are verbal in nature, as they normally precede the finite verb in clause-final position; although acceptability judgment seem to vary from case to case and person to person, placing the participle after the finite verb is always the marked option and in many cases simply excluded. The judgments given here are ours; Haeseryn et al. seem to consider the V-participle order in (123c) fully acceptable.

If the participles in the examples above are indeed adjectival in nature, we immediately account for the fact illustrated in (124) that examples like these do not exhibit the infinitivus-pro-participio (IPP) effect; if the participles marked "A" are indeed adjectival, the participles marked "V" are the most deeply embedded verbs, and we therefore correctly predict that they must surface as a past participle in the perfect tense. This also accounts for the severe ungrammaticality of the primed examples; adjectival participles normally precede the verb cluster.

For completeness' sake, example (125a) shows that adjectival participles may permeate the verb cluster provided they precede the main verb. Example (125b) shows that in this respect they behave just like "true" adjectives. We refer the reader to Section 7.4 for detailed discussion.

Potentially genuine counterexamples to the claim that verbal participles can only be found as complements of perfect and passive auxiliaries are given in (126). These examples suggest that the verb komen'to come' is able to select either an infinitive or a participle. The two constructions are restricted in the sense that the verb selected by komen must be a verb of movement accompanied by a directional phrase like de tuin in'into the garden' or the verbal particle aan, which indicates that the entity referred to by the subject of the clause approaches the speaker. The crucial thing is that the alternative placements of the participles in the primed examples are equally felicitous, which may be taken as evidence for assuming that we are dealing with verbal participles.

Haeseryn et al. (1997: 964-5) claim that the primeless and primed examples in (126) are identical in meaning and simply differ in their geographical distribution: participles are preferred by speakers of the southern varieties, whereas speakers of the northern varieties prefer the infinitive. They further claim that the construction with a participle is more restricted than the one with an infinitive: in the perfect-tense constructions in (127), the verb fietsen'to cycle' must take the infinitival form.

However, the impossibility of the participle gefietst in the primed examples might encourage one to claim that, despite the fact that the participle may follow komen in clause-final position in the primed examples in (126), the participle is adjectival in nature after all. If so, we would predict that the unacceptable examples with the participle gefietst improve when gefietst precedes the verb komen in its participial form (there is of course no reason to expect the IPP-effect if the participle gefietst is adjectival in nature). Judgments on the examples in (128) vary a great deal: some of our informants judge them to be worse than the primed examples in (127), others judge them to be better, whereas some (especially speakers of the southern varieties of Dutch) judge them to be acceptable, provided that the participle gekomen precedes the auxiliary. The marked character of the constructions in (128) makes it impossible to draw any firm conclusion, especially since we have not been able to find examples of this sort on the internet.

Better evidence in favor of assuming that the participle is adjectival in nature is provided by Duinhoven (1997:551-2), who observes the contrast between the examples in (129). The diacritics given here are his and show that although Duinhoven considers the use of the participle marked compared to the use of an infinitive, using the participle is acceptable if it is placed in front of the verb komen. Note that in this case we did find several instances (probably from Belgium) of the order zie ..... aan-V komen on the internet for the verbs wandelen/lopen'to walk', rennen'to run' and vliegen'to fly'. Duinhoven explicitly states that the contrast between the two orders in (129b) shows that the participle is adjectival in nature.

Duinhoven (1997:281ff.) also shows that the construction of komen + participle was very common in medieval Dutch, and actually did not require the addition of a directional phrase or the verbal particle aan. He argues that the participle originally functioned as a manner adverb that modified the verb komen'to come', which is in fact compatible with the fact that the participle is normally optional, also in present-day Dutch.

On this view, the komen'to come' + participle construction is a relic from an older stage of the language, which is under pressure of disappearing, that is, being replaced by the corresponding infinitival construction. For our present discussion it is important that the claim that the participle has or, at least, originally had an adverbial function implies that it is adjectival and not verbal in nature. This means that the komen + participle construction is special and cannot be taken as a straightforward counterexample to our claim that verbal participles are found as complements of perfect and passive auxiliaries only.