Clausal arguments usually follow other postverbal arguments. This is illustrated in (124) for a direct object clause: it follows the prepositional indirect object aan Peter and the PP-complement tegen Peter. The relative orders of the arguments in these examples clearly illustrate the mirror effect, as they are the inverse of the order in the middle field of the clause, where the direct object usually precedes the prepositional object; cf. Jan heeft dat verhaal aan Peter verteldJan has told that story to Peter and Jan heeft die dingen tegen Peter gezegdJan has said those things to Peter.

That argument clauses follow prepositional objects is not only true for object clauses but also for subject clauses, which resemble object clauses in that they originate as internal arguments; cf. Section 5.1.3, where it is shown that subject clauses do not normally occur in intransitive constructions. We illustrate this with the passive counterparts of the (a)-examples in (124); cf. Dit verhaal werd aan Peter verteldthis story was told to Peter and Die dingen werden tegen Peter gezegdthese things were said to Peter.

Because subject and object clauses usually originate as internal arguments, there are few cases in which they co-occur. In the rare cases where this happens, it is not easy to determine the relative order of the subject and the object, because De Haan (1974) has shown that the subject clause then tends to be placed in clause-initial position, as in (126a). He even claims that the subject clause cannot occur in postverbal position, which may be an overstatement, since the sharp contrast between the two (b)-examples suggests that the primeless example is relatively well-formed.

The contrast between the two (b)-examples could be seen as another instantiation of the mirror effect, but it should be noted that the postverbal placement of the subject clause requires that the subject position be filled by the anticipatory pronoun hetit. This suggests that the subject clause is not really an argument of the verb, but an apposition in right-dislocated position; cf. Section 12.1, sub IB. This suggests that we should leave cases such as (126b) aside from our present discussion. Another reason for doing so is that the subject may actually be nominal in nature, since the clause can easily be replaced by the noun phrase het feit dat hij niet klaagtthe fact that he does not complain.

Argument clauses usually precede adverbial clauses. This is illustrated in (127) for adverbial clauses expressing time and reason, respectively; the primed examples are marked with a number sign because they are possible if the adverbial clause is interpreted parenthetically, in which case it must be preceded and followed by an intonation break. Note that the strings in the primeless examples are ambiguous in speech, as the adverbial clauses need not be interpreted as part of the main clause, but can also be interpreted as part of the object clause, in which case they refer to the time/reason of the resignation.

The order restriction illustrated in (127) is again an instantiation of the mirror effect. This becomes clear as soon as one realizes that object clauses are normally part of the focus (new information) of the clause, and that adverbial clauses tend to precede the focus of the clause when they are in the middle field of the clause; we illustrate this in (128) with the non-specific indefinite nominal object iets belangrijkssomething important. The mirror effect thus correctly predicts the primed examples in (127) to be excluded.

Subject clauses behave in a similar way to object clauses, and for the same reasons as before. The examples in (129) illustrate this with the passive counterparts of the (a)-examples in (127) and (128). Note that the strings in the primeless examples are again ambiguous in speech, as the adverbial clauses can also be interpreted as part of the object clause.

The examples in (130) provide similar cases to those in (127), but now with a temporal and a locational PP instead of a clause. However, the judgments here are less clear: the primed examples in (130) all seem acceptable, with a preference for the order in the doubly primed examples, where the adverbial PPs precede the object clauses. This may be related to the fact that in speech the PP in the singly-primed examples tends to be interpreted as part of the object clause. The best order seems to be those in the primeless examples, where the PPs are in preverbal position.

We see that the examples in (130) do not exhibit the mirror effect found in the earlier examples. This is perhaps not surprising given that direct object clauses also tend to follow adverbial phrases of other categories. For instance, example (131b) is only acceptable with an afterthought intonation contour; the modal adverb waarschijnlijk must be preceded by a distinct intonation break and bear accent.

The same may be true for example (132b), although the more prominent interpretation is that the adverb morgentomorrow is interpreted with the object clause as a backgrounded right-dislocated phrase: it is not the telling but the resignation that will take place tomorrow.

Note, however, that Section 12.3 has shown that postverbal adverbs like waarschijnlijk and morgen are also right-dislocated. If this is correct, we have to conclude that the object clauses in the primed (b)-examples in (131) and (132) are not extraposed, but also right-dislocated. This seems to be confirmed by the fact that VP-topicalization cannot pied-pipe the object clause when the adverb is present.

This, in turn, makes it plausible that the doubly-primed examples in (130) also involve right dislocation and should therefore be ignored for our present purposes. The discussion above has shown that the determination of the relative order of extraposed argument clauses is not a trivial matter and should be given more attention than it has received so far.

The plot graph in Figure 1, taken from Hoeksema (2014), shows that PP-complements usually occur left or right adjacent to the verb/particle in clause-final position; PP-complements in scrambled and topicalized positions are less common; they make up only about 20% of all cases.

The choice between examples with preverbal and extraposed PP-complements, as in (134a&b), is influenced by two factors in particular. The first factor is the weight of the PP: shorter PPs in terms of the number of syllables are less frequent in extraposed position than longer PPs. This is also visualized in Figure 1, which shows the correlation between the weight of PP-complements and their distribution in topicalized, A'-scrambled, regular and extraposed position; here our concern is with the last two cases. The second factor is related to the status of the nominal complement of the PPs. One of the striking findings in Figure 2 is that PPs with a pronominal complement hardly ever extrapose, while this seems to be easily possible in all other (i.e. non-pronominal) cases; note that for various reasons about 50 cases were removed from the earlier set when compiling this plot graph.

One of the reasons why extraposition of complement-PP with a pronominal complement is rare might be that pronouns are simply less weighty in the sense discussed earlier, since with the exception of the plural object pronoun jullieyou they are all monosyllabic. They are also lighter in the sense that they are not usually accented; to our ear, extraposition is much better when the pronoun is accented, as shown in (135), but the effect of this can of course only be studied in a spoken corpus.

The corpus does not include cases with pronominalized PP-complements such as erop in (136a), which are clearly unacceptable in extraposed position. This is probably not due to their weight but rather to the fact, illustrated in (136b), that such PP-complements are usually split and that the resulting stranded prepositions must occur left-adjacent to the clause-final verbs; cf. Section P36.3.4.

Figure 2 also shows that the definiteness of the nominal complement of the PP does not really affect the choice between preverbal and postverbal position. Proper/bare nouns, on the other hand, seem to occur more often in preverbal position, but it does not seem too far-fetched to say that this is related to the fact that they are often lighter; cf. Hoeksema (2014:234-5). If so, perhaps the two factors can be reduced to one, the weight of the PP (an option not considered in Hoeksema. Hoeksema also notes that an effect of the lexical combinations themselves (i.e. V + PP) may manifest itself: certain combinations favor the PP in preverbal position while others favor it in postverbal position. He leaves this as a suggestion for future research. For completeness’ sake, note that Hoeksema’s data also shows that extraposition of PP-complements has become more frequent since about 1800, which may be partly explained by the fact that the average length of PP-complements has also increased since then. To conclude, we can say that extraposition is quite a common phenomenon for PP-complements, provided they are not too light.

Note that extraposed PP-complements are usually adjacent to the clause-final verb position. This leads to the so-called mirror effect (Koster 1974); elements that in the unmarked case are left-adjacent to the clause-final verbs in the middle field of the clause occur right-adjacent to these verbs when extraposed. Subsection IA has already illustrated this for the order of postverbal prepositional objects and argument clauses. The examples in (137) show that the mirror effect also occurs in the case of adverbial clauses/PPs.

However, prepositional object clauses introduced by the anticipatory PP erop differ from postverbal PP-complements in that they tend to follow postverbal adverbial phrases; an example such as (138b') is acceptable only with an intonation contour associated with afterthoughts, i.e. with a distinct intonation break before the adverbial phrase, which receives a contrastive accent. This finding is not surprising in view of our conclusion in Section 12.1, sub IB, that clauses introduced by an anticipatory pronominal element are not extraposed but right-dislocated.

That prepositional object clauses introduced by an anticipatory PP are not extraposed but right-dislocated is also clear from the fact, illustrated in (139), that they obligatorily follow postverbal adjectival and nominal adverbial phrases like waarschijnlijkprobably and morgentomorrow, which were shown to be right-dislocated in Section 12.3.

A'-scrambling and topicalization of PP-complements lead to a marked information structure of a clause: Sections 11.3.3 and 13.3.2 show that topicalization and A'-scrambling generally have an effect on the information structure of a clause: they involve leftward movement of e.g. foci and topics. This implies that they can only apply to phrases that have these discourse functions. Figure 2 shows that leftward movement of PP-complements is possible, regardless of the nature of the nominal complement. However, there seems to be a contrast in the case of topicalization and scrambling: while pronouns, definite noun phrases and proper nouns allow both topicalization and A'-scrambling, the latter is rare in the case of indefinite noun phrases and bare noun phrases (i.e. without a determiner).

Hoeksema (2014) found that complement-PPs with indefinite noun phrases do not scramble at all, and claims that this is to be expected in view of the fact that indefinite objects also resist scrambling. This sort of proposal overlooks the fact that object scrambling and PP-complement scrambling are distinct: the first is A-scrambling while the second is A'-movement; cf. Section 9.5, sub IIIA, for more detailed discussion. The reason for this restriction in scrambling of complement-PPs must therefore lie in the fact that indefinite objects are typically used to introduce a new entity into the domain of discourse and thus cannot be used as contrastive foci or topics: these are typically found in the middle field of the clause. Something similar may hold for bare nouns, given that they are often non-referential; cf. Section N18.1.2.3. We will not digress further and leave this issue to future research.