Some, but not all, speech-act nouns can also be followed by an infinitival complement clause, which is sometimes optionally introduced by the complementizer om. Table 18 gives an indication of the possibilities.

nominal head	complement
declaratives: mededeling ‘announcement’	—
yes-no questions: vraag ‘question’	—
wh-questions: vraag ‘question’	de vraag hoe het probleem op te lossen ‘the question how to solve the problem’
requests/suggestions/orders: verzoek ‘request’	het verzoek (om) te (mogen) vertrekken ‘the request to be (allowed to) leave’
wishes, ideas, suggestions: advies ‘advice’	het advies (om) te blijven ‘the advice to stay’

The primed examples in (643) show that declarative speech-act nouns do not easily accept infinitival complement clauses. Note that PRO stands for the implied subject of the infinitival clause.

As for questions, only wh-questions can be realized as infinitival complement clauses. This is simply a reflex of the fact that speech act verbs do not take infinitival yes/no questions either, which may be due to the fact that there is no linker that can mark infinitival clauses as questions: while the complementizer ofwhether can formally mark a finite clause as an embedded yes/no-question, such a specialized marker is missing for infinitival clauses.

Speech-act verbs like vragento ask do take infinitival wh-questions: Jan vroeg [wat te doen]Jan asked what to do. This is probably related to the fact that the complement is now explicitly marked as a question by the wh-phrase in the clause-initial position. Example (645) shows that the speech-act noun vraag can also take infinitival wh-questions as its complement. Note that the implied subject PRO can either receive an arbitrary interpretation or be construed as coreferential with (a referent set containing) the agent of the speech-act noun.

Dependent wh-clauses seem to be restricted in that wh-phrases with an adverbial function usually give rise to a marked result. A notable exception is hoehow, which occurs frequently in this construction; our Google search (May 13, 2021) for the string [de vraag hoe te] yielded over 200 different instantiations. For comparison, we want to mention that we found about 100 relevant instantiations of the string [de vraag wanneer te], and at most two potentially relevant instantiations out of 97 unique hits for the string [de vraag waarom te]. The examples in (646a&b) are taken from the internet; (646c) is a constructed example.

The same is true for interrogative pronominal PPs of the form waar + P. Our Google search for the string [de vraag waar+P te] for the prepositions op, in and over yielded a total of about 30 hits, most of which involved the same kind of verbs. The relevant examples are given in (647): in addition to investerento invest, we found beleggento invest, and in addition to schrijvento write, we found other communication verbs such as twistento dispute.

The facts in (645) to (647) more or less mirror what we find in the verbal domain, so we can conclude that we are dealing here with inherited arguments. It is not surprising that speech-act nouns derived from verbs denoting the act of asking permission or giving directions easily accept infinitival complements: these verbs typically combine with infinitival clauses anyway. In (649), the implicit subject PRO is interpreted as coreferential with, respectively, the agent and the goal of the speech-act noun (both of which may be left implicit), just as it would be construed as coreferential with, respectively, the agent and the goal of the corresponding verb.

The verb vragento ask can be used with the same function as verzoekento request and bevelento order and so the derived noun vraagquestion can be used with the same function as verzoek and bevel. However, it differs from these nouns in that it requires explicit mention of the antecedent of the PRO subject.

Nouns derived from speech-act verbs like adviserento advise or verzekerento assure also give rise to a perfectly acceptable result. In (650a) the implicit PRO subject is coreferential with the implied agent of the noun phrase, which in turn is coreferential with the subject of the clause; the complementizer om must be omitted, just as in the corresponding verbal construction. In (650) the implicit PRO subject is coreferential with the goal of the speech-act noun.

As declarative speech-act nouns, proposition nouns denoting statements or facts seem to give rise to a marked result when combined with an infinitival complement.

There is a very small set of non-derived nouns that can take an infinitival complement. A typical example is the noun ideeidea in (652a): the implicit PRO argument can be interpreted arbitrarily or refer to the agent/possessor of the idea. Another possible case is wenswish in (652b), provided one accepts that this noun is non-derived and can be used as input for deriving the verb wensento wish instead of the other way around.

The fact that the noun feitfact in (653) cannot be combined with an infinitival clause shows that not all non-derived nouns that take a finite clause can take an infinitival clause. The difference between (652a&b) and (653b) may be related to factivity: infinitival clauses cannot be factive, which would fit nicely with the earlier observation that declarative speech-act and proposition nouns cannot take infinitival complements either.

The examples in (654) show that although some adjectives allow infinitival complements, these complements are not necessarily inherited by the deadjectival noun. As in the case of the finite complement clauses in (640) and (642), the unacceptability of complement clauses with deadjectival nouns can be accounted for by the fact that these nouns do not denote abstract content.

Consider in this respect also the sentence pairs in (655). As shown in (655a), the adjective bangafraid can take an infinitival complement, preferably introduced by om. However, the deadjectival noun bangheidfear in (655a'), though acceptable without complement, cannot inherit the clausal complement. In the case of the near-synonyms angstigafraid and angstfear, on the other hand, the reverse is true: the denominal adjective angstigafraid in (655b) cannot take a clausal complement, while the non-derived noun angstfear in (655b') is a (relational) proposition noun that accepts an infinitival complement clause. The (c)-examples in (655) show, however, that there are also cases in which both the adjective and the derived noun take an infinitival complement; if it were true that deadjectival nouns do not inherit the argument structure of their base, this would have to be accidental.