If the primed examples in (601) are really ambiguous between a nominal and a verbal reading of the bare infinitive zwemmen, we expect this to come out in the word order of the clause. Since constructions with bare infinitival complement clauses exhibit monoclausal behavior, we expect verb clustering: the verb leren may precede the bare infinitive in embedded clauses and separate it from its dependents (arguments and modifiers). The fact that the verb leren can indeed split the strings goed zwemmen'swim well' and computers repareren'repair computers' in (602) thus shows that we are dealing with clausal infinitival complements in these examples, which is indicated by marking the bare infinitive with the label "V".

The verbal status of the bare infinitives in the examples from the previous subsection also appears from the word order of the clause-final verbal sequence. Because noun phrases cannot follow the verbs in clause-final position, the fact that the bare infinitive follows the clause-final finite verb leren in (602) is already sufficient for concluding that we are not dealing with bare-inf nominalizations but with bare infinitival complement clauses. This word order generalization is especially useful when the verb has no dependent, as in the cases in (603); the bare infinitives following the clause-final finite verb leren must be verbal.

Since bare-inf nominalizations must precede their governing verb in clause-final position, it seems reasonable to assume that the bare infinitives preceding leren are nominal. Nevertheless, we marked them with a question mark because although clause-final verb clusters of the form Vfinite - Vinfinitive normally surface with the finite verb preceding the infinitive, most speakers also allow the inverse order under certain conditions (we will discuss an unambiguous case of this in Section 7.3, sub IC).
      Although in the case of clause-final verb clusters consisting of no more than two verbs precedence of the bare infinitive is only a necessary and not a sufficient condition for concluding that we are dealing with bare-inf nominalization, the word order of clause-final verbal sequences can still be used as a diagnostic with verb clusters consisting of more than two verbs, because in such cases most speakers of Standard Dutch do require verbal bare infinitives to follow their governing verb. The fact that the bare infinitive zwemmen may precede the verb leren in examples such as (604) thus shows that bare infinitivals can indeed be bare-inf nominalizations in the complement of leren, as is indicated by the index N on the preverbal occurrence of zwemmen.

The structural ambiguity of the bare infinitives in the primed examples of (601) is also clear from the contrast with respect to the IPP-effect in the perfect-tense examples in (605): if leren takes a bare infinitival clausal complement, we expect the IPP-effect to arise, but not if it takes a bare-inf nominalization. The primeless examples show that when the bare infinitive zwemmen follows leren, IPP does indeed arise, and we can therefore conclude that the infinitive is verbal in this case. The primed examples, on the other hand, show that when the bare infinitive zwemmen precedes leren, IPP cannot arise, and we therefore conclude that it is nominal in this case.

In the examples in (605) the difference with respect to the IPP-effect was illustrated by means of the intransitive verb zwemmen'to swim'. The same difference occurs, however, with transitive verbs with a bare nominal object like auto rijden'to drive (a car)'. The (a)-example in (606) shows that the infinitive may either precede or follow its governing verb, while the (b)-examples bear out that this affects the occurrence of IPP.

Examples such as (606b') are especially felicitous with bare-inf nominalizations if the object-noun combinations are fixed collocations referring to some conventional activity: aardappels schillen'to peel potatoes', paard rijden'to ride on horseback', piano spelen'to play the piano', etc. Less conventional combinations like computers repareren'to repair computers' in (607) seem acceptable in bare-inf nominalizations, although some speakers may find them somewhat marked.

That infinitives preceding a clause-final verbal sequence of two (or more) verbs are nominal is also clear from the fact that they do not have to be adjacent to the clause-final verbal sequence; the examples in (608) show that like other nominal objects they may scramble to a more leftward position. Observe that examples like these require the infinitive to be assigned contrastive accent, and that even then the (b)-example may be considered somewhat marked by some speakers.

Example (609a) shows the same thing by means of verbs with a bare nominal object like auto rijden'to drive'. The (b)-examples are added to show that the nominal complement of the bare infinitive can be scrambled on its own by focus movement if the infinitive heads a bare infinitival clause, but that this is impossible if it heads a noun phrase. This is consistent with the fact that the nominal complements are never extracted from bare-inf nominalizations.

That less conventional combinations like computers repareren'to repair computers' are acceptable but marked in bare-inf nominalizations is also clear from the fact that focus movement in ?dat Jan computers repareren waarschijnlijk wel nooit zal leren'that Jan will probably never learn to repair computers' may be considered degraded by some speakers.

A final argument for assuming that bare infinitives preceding clause-final verbal sequences of two or more verbs are nominal is that they cannot follow sentential negation expressed by the negative adverb niet'not'; as in the case of other indefinite noun phrases, negation must be expressed by means of the negative article geen'no'. The contrast between the primeless and primed examples in (610) thus confirms that infinitives preceding a clause-final verbal sequence of two (or more) verbs are nominal in nature.

The examples in (611) illustrate the same contrast by means of perfect-tense constructions.

The negation facts are less telling with transitive constructions such as auto rijden'to drive' since the indefinite object auto in the verbal construction cannot follow the negative adverb niet'not' either, and can likewise be preceded by the negative article geen'no'. So, at face value, the two perfect-tense constructions in (612) seem to behave in an identical fashion in this case. We should keep in mind, however, that the article geen is the determiner of the noun phrase geen auto in (612a) but of the bare-inf nominalization geen auto rijden in (612b).

The discussion so far has established six differences between constructions with a bare infinitival clausal complement and a nominal complement in the form of a bare-inf nominalization. These were already announced in Table (599), which is therefore simply repeated here as (613).

The findings in Table (613) are important because they may help us in determining whether a given bare infinitive does or does not belong to the verbal complex. The discussion in this section suggests at least that bare infinitives preceding their governing verb are nominalizations if the bare infinitive is part of a verbal complex of two or more verbs. If correct, this will simplify the description of the word order of the verbal complex considerably; we will return to this in Chapter 7.