A notable exception to the ban on permeation of verb clusters are bare objects in N + V collocations like paardrijden'to ride a horse' and pianospelen'to play the piano' in (173).

Examples of this type need not be a problem for the claim that verb clusters are impermeable provided that we assume that collocations like paardrijdenor pianospelenare compounds if the bare noun permeates a cluster. There are, however, various reasons not to follow this suggestion. First, bare nouns permeating larger verb clusters need not be adjacent to their associate main verb, as shown in (174). The acceptability of the linear order V3–Noun–V2–Main1 shows that assuming a compound analysis is not sufficient to explain why bare nouns may permeate verb clusters.

Second, the examples in (175) show that the bare noun cannot be pied-piped when the verb undergoes verb-second. Examples such as (175) contrast sharply with examples such as Peter stofzuigt graag'Peter likes to hoover' where stofzuigen'to hoover' is a compound. The compound analysis of paardrijden and pianospelen calls for a separate explanation for the impossibility of pied piping.

Third, participle formation cannot be based on the presumed compounds paardrijden and pianospelen, as is clear from the fact that the prefix cannot precede the bare noun in (176). Examples such as (176) contrast sharply with examples such as Peter heeft gestofzuigd, in which stofzuigen'to hoover' is a compound.

The examples in (174) to (176) show that the compound analysis of paardrijden and pianospelen does not fully solve the problem, and actually creates a number of new problems. The alternative analysis is that there is in fact no general ban on permeation of verb clusters by nominal arguments of the main verb. The alternative finds support in the fact that certain varieties of Standard Dutch spoken in Flanders do also allow permeation of the verb cluster by bare (singular or plural) objects that do not form a collocation with the verb. In West-Flanders permeation is even possible by indefinite and definite objects, but it is not clear to us whether this can be considered part of the regional variety of Standard Dutch or whether it should be considered a dialectal property. The examples in (177) are taken in a slightly adapted form from Barbiers et al. (2008: Section 2.3.1), to which we refer the reader for further discussion of the regional spread of these forms of permeability of verb clusters.

Although adjectival complementives normally precede the verb cluster as a whole, many (but not all) speakers accept permeation of the cluster if the adjective is monosyllabic. In other words, there is a sharp contrast between example (178a) and (178b). Example (178c) further shows that in order to be able to permeate the verb cluster the adjectival phrase must be simple, in the sense that it cannot be modified by, e.g., a degree adverb or be otherwise complex.

It has been suggested that the acceptability of permeation of the verb cluster in examples such as (178a) is due to complex predicate formation, that is, incorporation of the adjectival complement into the verb, as a result of which a compound-like element is created; cf. Neeleman (1994b). There are various reasons not to follow this suggestion. The most important one is that bare adjectives that permeate larger verb clusters need not be adjacent to the verb which they are assumed to form a complex predicate with. The acceptability of the order V3–Adjective–V2–Main1 in (179a) shows that assuming an incorporation analysis is not sufficient to explain why bare adjectives may permeate verb clusters. On the basis of the incorporation analysis we would furthermore expect that the adjective could be pied-piped under verb-second; the fact illustrated in (179b) that this expectation is not borne out thus forces us to assume additional stipulations in order to account for this.

      Verbal particles, which are also analyzed as complementives in Section 2.2, are even better suited to illustrate that there is no absolute ban on permeation of verb clusters. All speakers of Dutch accept examples of the type in (180).

Again, it is often suggested that the permeation of the verb clusters in examples such as (180) is due to the fact that we are dealing with compound-like verbs. That this is not evident is clear from the fact that particles that permeate verb clusters do not need to be adjacent to their associate verbs (Bennis 1992), and from the fact that they must be stranded when the verb undergoes verb-second.

The examples in (182a) further show that many speakers also allow postpositions to permeate verb clusters, and (182b) shows the same holds for the second part of circumpositions like over ... heen'over'; see van Riemsdijk (1978) and Section P5.2.2 for more discussion. This is, however, not generally accepted for stranded prepositions like op in (182a), although southern speakers are more permissive in this respect.

Barbiers et al. (2008: Section 2.3.1) further show that especially West-Flemish speakers allow complex PP-complements to permeate verb clusters.

Adverbs are normally not allowed to permeate verb clusters. Given that manner adverbs must be directly construed with the main verb, they are best suited to illustrate this fact. An example of an adverb modifying a verb phrase is given in (184b).

The percentage signs again indicate that permeation is not rejected by all speakers; it is acceptable for many speakers from West-Flanders; see Barbiers et al. (2008: Section 2.3.1).

The previous subsections have shown that there is no general ban on permeation of verb clusters: there is a clear tendency to avoid it, but there are many exceptions and there is a considerable regional variation; more detailed information on regional variation can be found in Sections 5.2.3 and 6.2, as well as Barbiers (2008:ch.2). There have been attempts to account for some of the cases by assuming that they involve compound verbs or (syntactically created) complex predicates, but we have seen that this still does not fully account for all the facts and sometimes even creates new problems. Furthermore, it is not easy to extend such accounts in order to account for permeation of verb clusters in some of the more permissive varieties of Dutch like West-Flemish, which also allows definite objects and adverbs to permeate verb clusters. Regardless of whether these varieties should be considered as dialects or as instantiations of a regional variety of Standard Dutch, this is quite telling since we have reasons for assuming that the situation in West-Flemish corresponds to the older stages of current Standard Dutch. The limited amount of permeation we found in the northern variety of Standard Dutch has arisen by a gradual reduction of the set of elements that could permeate the verb cluster; we refer the reader to Hoeksema (1994) and Van der Horst (2008) for a more detailed discussion of this diachronic development.