The part of the clause preceding the finite verb in second position should be characterized as a position rather than a field, since it can be occupied by only a single constituent; cf. Section 11.3 for a detailed discussion. This is clearly not the case for the postverbal field, which can be occupied by an (in principle) unlimited number of constituents; the primeless examples in (7) provide cases where the number of postverbal constituents ranges from 0 to 2, and it is certainly not difficult to construct or find examples with more than two postverbal constituents; the primed examples are added to show that there can be only one constituent in the clause-initial position.

In the early stages of generative grammar, the examples in (7b-d) were derived from (7a) by a movement rule known as extraposition, which moves the PP from a preverbal position to the postverbal field. A problem with this proposal, already noted in Koster (1973), is that it is not consistent with Emonds’ (1976) structure preservation principle, which requires movement to target an independently motivated position; this principle is satisfied by wh-movement, since there is clearly an identifiable clause-initial position, but this is not obviously the case for extraposition because we are dealing with a set of positions; if the postverbal position of the phrases in (7) is indeed derived by movement, we may thus be dealing with a set of rules, each of which may have its own properties. Nevertheless, we will stick to the notion of extraposition as a descriptive term to refer to constructions with clause-internal postverbal constituents.

The primeless examples in (7) show that adverbial and argument PPs can occupy different positions in the clause: clause-initial, preverbal and postverbal. The examples in (8) further show that extraposition affects the linear order of these PPs: the (a)-examples first show that in the middle field of the clause adverbial PPs usually precede argument PPs (when the clause is pronounced with a neutral intonation pattern), while the (b)-examples show that in postverbal position the order is usually inverted; Koster (1974) called this phenomenon the mirror effect, because the clause-final verb position in a sense mirrors the word order.

At first glance, it may seem difficult to determine whether extraposition can also be applied when the main verb is in verb-second position and consequently there are no verbs in clause-final position. However, there are several ways to determine this indirectly. First, we can appeal to the mirror effect discussed in the previous subsection: since the examples in (8) have shown that adverbial phrases precede PP-complements in the middle field of the clause, but follow them in extraposed position, the acceptability of the word order in (9b) shows that at least the adverbial PP can be extraposed, and thus most likely both.

Second, we can take advantage of the fact that certain elements, like complementives and verbal particles, are usually left-adjacent to the clause-final verbs. The primeless examples in (10) illustrate this by showing that, although the PPs in the primeless examples can be placed either in pre or in postverbal position, they crucially cannot be located in the position indicated by the asterisk, i.e. in between the complementive/particle and the clause-final main verb. From the fact that these PPs can follow the complementive/particle in the primed examples, we can again deduce that extraposition does not depend on the presence of a clause-final verb, but applies across the board; cf. Koster (1974).

At first glance, it seems relatively easy to determine whether a given element is extraposed by considering its position with respect to clause-final verbs, complementives, or particles. However, this is only seemingly so, as it is necessary to distinguish between different types of postverbal constituents, which can be easily illustrated by the placement of noun phrases. Example (11a) first shows that nominal arguments cannot be extraposed: placing the nominal object de directeur to the right of the clause-final participle gesprokenspoken is excluded. However, the (b)-examples in (11) show that it is possible to place this noun phrase to the right of the participle if the regular object position is filled by another noun phrase; the comma indicates that the postverbal noun phrase is generally preceded by an intonation break.

The postverbal noun phrases in the (b)-examples are different from other extraposed phrases. For example, the fact that the regular object position is filled by the noun phrase dhr. Jansen shows that the postverbal noun phrase is not selected by the verb, but that we are dealing with a parenthetical constituent that is not an integral part of the clause; cf. Klein (1977) and De Vries (2009). This conclusion is supported by the fact that the postverbal noun phrase is preferably separated from the preceding clause by an intonation break; we are dealing with an apposition, i.e. an addition intended to clarify some potential obscurity in the preceding clause. Note that the postverbal noun phrase can be used to provide either discourse-new or discourse-old information. We will follow De Vries and refer to the former as afterthought right-dislocation and the latter as backgrounding right-dislocation; the two cases differ prosodically in that an additional accent is assigned to the former but not to the latter, as indicated by the small caps in (11b).

Afterthoughts and backgrounded phrases are easy to recognize when they are associated with arguments, as the latter are normally obligatory. However, it is more difficult when they are associated with optional constituents, such as the adverbial comitative met-PP in dat Jan graag (met Peter) schaaktthat Jan likes to play chess (with Peter). Examples such as (12), in which the adverbial met-PP is realized in the middle field of the clause, are of course straightforward: the postverbal met-PP can only be right-dislocated, as is also clear from the fact that it must be preceded by an intonation break.

When the preverbal adverbial PP is not present in the middle field, as in the examples in (13), we have to rely entirely on intonation. Recognizing an afterthought still seems relatively easy because it is signaled by an additional contrastive accent; moreover, freestanding afterthoughts can often be preceded by appositional markers such as en wel. However, it can be quite difficult to distinguish an extraposed PP from a backgrounded PP, as this hinges crucially on the intonation break, which can be quite difficult to detect in casual (fast) speech.

This makes distinguishing between extraposition and backgrounding in constructions such as (12) a rather delicate matter; our judgments on the examples given here and later in this chapter are based on our own intuitions about whether an intonation break is necessary, possible or obligatory in slow, careful speech. One fact that may help to distinguish extraposed from backgrounded phrases is that backgrounding right-dislocation does not affect the intonation contour of the clause. If the postverbal phrase is assigned a (non-contrastive) sentence accent, as in (14a), we can safely conclude that we are dealing with extraposition (sentence accent is indicated by italics). However, if the sentence accent is assigned to (a constituent preceding) the clause-final verb, as in (14b), it is again not obvious whether we are dealing with extraposition or backgrounding. Afterthought right-dislocation in (14c) is again relatively easy to recognize: it does not affect the placement of the sentence accent, and the afterthought itself is assigned an additional accent.

A syntactic test that can be helpful in distinguishing the different types of postverbal phrases is VP-topicalization. The examples in (15a&b) show that ordinary extraposed constituents like clausal and prepositional direct objects are usually pied-piped under VP-topicalization.

°Stranding of clausal and prepositional direct objects is only possible if they are right-dislocated, i.e. preceded by an intonation break. In the case of the clausal object, this is only perfectly acceptable if the anticipatory pronoun het is present (due to the fact that the verb vertellento tell needs a direct object), while in the case of the PP the anticipatory pronominal PP er ... opfor it may be absent (because wachtento wait can also be used without a PP-complement).

The examples in (17) show that right-dislocated phrases have a tendency to strand; the (b)-examples show that pied piping of afterthoughts requires us to use quite distinct, long intonation breaks (indicated by em-dashes), and even then some speakers tend to reject it; the (c)-examples show that pied piping of backgrounded phrases leads to a downright bad result. In both cases, stranding is easily possible.

It is not a priori clear that the markedness of pied piping in (17c') is of a syntactic nature, as De Vries (2002:292) suggests that pied piping of backgrounded phrases may be incompatible with the focus/topic interpretation assigned to topicalized phrases. What is important for us at this stage, however, is that extraposed phrases seem to be preferably pied-piped under VP-topicalization, while backgrounded right-dislocated phrases tend to be stranded, and that some speakers allow both options in the case of afterthoughts (given the right intonation contour).

This subsection has shown that it is often not possible to conclude on the basis of the postverbal placement of a constituent alone that we are dealing with extraposition; we may also be dealing with, e.g., an afterthought or a backgrounded phrase. Moreover, distinguishing between extraposed and backgrounded phrases may be hazardous, as the intonation break that characterizes the latter may be quite difficult to detect in casual (fast) speech; we therefore had to appeal to our own intuition about the use of intonation breaks in slow, careful speech. Finally, we proposed VP-topicalization as a means of distinguishing extraposition from right dislocation: extraposed phrases tend to be pied-piped, while backgrounded phrases tend to be stranded under VP-topicalization. For a more detailed discussion of right dislocation, we refer the reader to Section C37.3.

Leaving aside the cases in which extraposition is impossible or obligatory, the question arises as to what determines whether extraposition takes place or not. To our knowledge, this question has not received much attention in the literature. One factor that may play a role is the information structure of the clause. Consider the examples in (18). Although it is not easy to detect a clear difference in meaning between the two orders in the active clause in (18a), the impersonal passive constructions in the (b)-examples show that the absence of the expletive er has a degrading effect on extraposition when the middle field of the clause is empty. Since Bennis (1986) has shown that the expletive signals the absence of presuppositional material, the PP-complement in (18b) should be part of the new-information focus, while the PP-complement in (18b') should be part of the presupposition of the clause. From this we can conclude that presuppositional material must precede the clause-final verb, while non-presuppositional material can also follow it; cf. also Haeseryn et al. (1997:1366). That extraposed phrases are part of the new-information focus of the clause is supported by the fact that under a neutral (non-contrastive) intonation pattern they tend to receive sentence accent (indicated by italics); cf. Zwart (2011:63-4).

Although example (18b) shows that PPs presenting discourse-new material can occur preverbally, there are cases in which discourse-new material must be extraposed. The examples in (19) illustrate this for an adverbial clause of reason. Although we have seen in examples (7) to (10) that adverbial clauses can occur in preverbal position, the clause in (19a&b) is preferably placed in clause-final position. This preference for extraposition may be due to prosodic reasons, as clauses and other long phrases produce an awkward intonation contour when they precede the clause-final verbs; cf. Truckenbrodt (1995) and De Vries (2002:260). This is especially true when the adverbial clause immediately precedes a verb with sentence accent, as evidenced by the fact that the result is much better in (19b'), in which the adverbial clause is followed by other material. This seems to be a more general phenomenon; see Haeseryn et al. (1997:1366) for similar cases where a clause immediately precedes a negative adverb niet with sentence accent.

Another factor that may affect the placement of constituents that optionally undergo extraposition is related to processing: there is a tendency to minimize the distance between the finite verb in clause-initial position and the non-finite verbs in clause-final position, and to reduce the complexity of the middle field. Extraposed material is therefore more likely to be found in long and complex sentences; cf. e.g. Haeseryn et al. (1997) and Coussé (2008: §2.5).