It is often taken for granted that locational and temporal adverbial PPs can be extraposed. That this is justified seems clear from the fact, illustrated in (53), that such PPs normally do not need to be preceded by an intonation break when they occur in postverbal position.

That we are dealing with extraposition in the primed examples in (53) is further supported by the fact that the adverbial PPs can easily be pied-piped under VP-topicalization.

Another argument for assuming extraposition may be that accent can be placed on extraposed adverbial PPs. Note, however, that speakers seem to have different ideas about whether this results in a neutral intonation pattern: while Van den Berg (1978:222) claims that the accent on the PP is the (non-contrastive) sentence accent, as indicated in (55b), Mark de Vries (p.c.) claims that it is contrastive, as indicated in (55b'). We will leave this issue for future research.

Section 12.1, sub IV, has noted that extraposition does not affect the propositional meaning of the construction. Therefore, in order to determine whether we are dealing with extraposition or right dislocation, it may be useful to examine the propositional meaning of the constructions under consideration. Let us first look again at the examples in (53) to show that the structural position of the adverbial phrase in the clause can affect its extraposition possibilities. Under a neutral (non-contrastive) intonation pattern, example (53a) expresses that Jan likes a certain thing, namely eating in the garden: like to do (Jan, eating in the garden). The extraposition example (53a') expresses exactly the same propositional content, and the same is true for the VP-topicalization construction in (54a). However, things are different in the (b)-examples in (53). Example (53b), repeated as (56a), expresses that Jan likes to do a certain thing after dinner, namely to smoke a cigarette: like to do after dinner (Jan, smoke a cigarette). Example (53b') and the VP-topicalization construction in (54b), however, express that Jan likes to do a certain thing, namely to smoke a cigarette after dinner: like to do (Jan, smoke a cigarette after dinner).

This strongly suggests that (53b') cannot be regarded as the extraposition counterpart of (56a) above, but should be regarded as the counterpart of (57a) below, which does express the same meaning as (53b') and (54b).

Assuming that the subject-oriented adverb graaggladly has a fixed position in the structure, this suggests that the structural position of the adverbial PP na het etenafter dinner (i.e. after or before the clause adverbial) determines whether extraposition is possible or not.

This phenomenon seems to be more general: Chapter 8 has argued that clause adverbials like waarschijnlijkprobably or vaakoften are located at the boundary between the lexical and the functional domain of the verbal projection (cf. Section 9.1 for an introduction to these notions): elements appearing to the left of such adverbials are located in the functional domain of the verb, while elements appearing to the right of such adverbials are part of the lexical domain of the clause. Now consider the primeless examples in (58) and their paraphrases in the primed examples.

The extraposition and VP-topicalization constructions in the (a)-examples in (59) correspond in their propositional meaning to example (58a), while the right-dislocation and VP-topicalization constructions in the (b)-examples correspond in their propositional meaning to example (58b). Although the contrast in meaning between the two primeless examples in (59) may not be very sharp, this is certainly the case for the contrast between the two primed examples. If our intuitions are correct, this strongly suggests that adverbial PPs can only be extraposed when they are in a hierarchically lower position than the clause adverbials, i.e. when they are in the lexical domain of the clause.

The hypothesis that adverbial PPs should be in the lexical projection of the main verb in order to be able to occur in extraposed position provides us with two strong predictions: if an adverbial PP can occur to the right of a clause adverbial, it can also occur in extraposed position; if an adverbial PP can only occur to the left of a clause adverbial, it cannot occur in extraposed position. This seems to be more or less the correct generalization. The examples in (60) show that the first prediction is correct for instrumental met-PPs, agentive door-PPs and wegens-PPs expressing a cause/reason; these can follow the clause adverbial vaakoften and extraposition is perfectly acceptable, as is clear from the fact, illustrated in the primed examples, that the postverbal PPs can be pied-piped under VP-topicalization.

That the second prediction is also on the right track is illustrated in (61) by means of an adverbial ondanks-PP expressing concession. Example (61a) first shows that this PP has to precede the clause adverbial vaakoften. We therefore expect extraposition to be impossible, and this is indeed borne out: example (61b) shows that the PP must be preceded by an intonation break when it is in postverbal position (Veld 1993:144). That we are not dealing with extraposition is further supported by the (c)-examples, which show that the PP cannot be pied-piped under VP-topicalization.

Since most adverbial PPs that can occur to the left of clause adverbials can also occur to their right, there are not so many systematic cases that show the same pattern as concessive adverbial PPs. Another less frequent case is the adverbial PP in de regelnormally in (62a). Example (62b) shows that this PP must be preceded by an intonation break if it is in postverbal position, and the (c)-examples show that it cannot be pied-piped under VP-topicalization.

Finally, the examples in (63) show that the clause adverbials themselves cannot be extraposed either if they have the form of a PP; cf. Veld (1993:144). Example (63b) shows that the adverbial PP tot drie maal toeup to three times must be preceded by an intonation break when in postverbal position, and the (c)-examples show that it must be stranded by VP-topicalization.

Note in passing that the intonation break in the word string in (63c) is not obligatory, because this string also allows the frequency adverb to be in the middle field of the clause (in which case the trace of the VP should be placed after the clause adverbial); this is not relevant for the present discussion.

This subsection has hypothesized that adverbial PPs can only be extraposed if they can occur to the right of clause adverbials, i.e. if they are part of the lexical projection of the main verb. Adverbial PPs cannot be extraposed if they can only occur to the left of clause adverbials, i.e. if they are part of the functional domain of the main verb. Sentential adverbial PPs, which seem to be located at the boundary between the two domains, cannot be extraposed either.

Adverbial clauses seem to show more or less the same behavior as their prepositional counterparts, although they might be expected to occur more frequently in extraposed position, as extraposition may be favored in the case of clauses by factors such as those mentioned in Section 12.1, sub V. Their similarity in behavior is illustrated in (64) by a temporal clause with the same function as the temporal adverbial PP na het etenafter dinner in (56): example (64b) shows that the clause does not need to be preceded by an intonation break when it occurs postverbally, and (64c) shows that it can easily be pied-piped under VP-topicalization. The primed examples show that the adverbial clause can also be right-dislocated.

The examples in (65) show that the distribution of adverbial clauses introduced by ondanks is subject to the same restrictions as the adverbial PP ondanks zijn ziektedespite his illness in (61). Example (65a) first shows that the adverbial clause must precede the sentential adverbial vaakoften; placing it in the position indicated by <*> leads to a severely degraded result. Example (65b) then shows that the clause is preferably preceded by an intonation break when it occurs in postverbal position (although this preference seems less strong than in the case of a PP). The (c)-examples show that the clause cannot be pied-piped under VP-topicalization, but must be stranded. All this suggests that the clause cannot be extraposed.

Infinitival temporal adverbial clauses are like their finite counterparts in that they can occur in pre and postverbal position. The postverbal clause can be in extraposed position: it does not need to be preceded by an intonation break, and it can easily be pied-piped under VP-topicalization. The infinitival clause can also be right-dislocated, but we will not illustrate this here.

Some adverbial clauses do not seem to be comfortable in preverbal position. This is the case, for example, with adverbial clauses in conditional and consecutive constructions. The answer to the question whether they are in extraposed or right-dislocated position therefore has to rely entirely on VP-topicalization. We illustrate this in (67) for conditional constructions. Example (67a) shows that the conditional als-clause cannot easily occur in the middle field; it can only occur there as a parenthetical, in which case it should be preceded and followed by an intonation break. The fact, illustrated in (67b), that the postverbal als-clause cannot easily be pied-piped under VP-topicalization suggests that it is right-dislocated.

The examples in (68) illustrate the same for consecutive constructions. Example (68a) shows that the adverbial clause expressing the consequence must be placed in postverbal position: placement of this clause in the middle field positions indicated by <*> is entirely impossible, even as a parenthetical clause. The adverbial clause in (68a) is again preferably preceded by an intonation break, which suggests that it is right-dislocated. This is supported by the fact illustrated in (b)-examples in (68) that the adverbial clause must be stranded under VP-topicalization.

As far as we know, the syntactic behavior of the different semantic types of postverbal adverbial clauses has not been systematically studied. The full story will therefore have to wait for future research; the discussions in Veld (1993: §5.2.8) and De Vries (2002: §7) provide good starting points for a more in-depth investigation.

Adjectival adjuncts are excluded in extraposed position. The following subsections will discuss this for adverbial phrases and complementives.

Adverbially used noun phrases have a temporal meaning. A prototypical example is given in (74a). Example (74) shows that the phrase de hele dagthe whole day can be used in postverbal position, but then preferably preceded by an intonation break; cf. Veld (1993:127). Nevertheless, the fact that omitting the intonation break seems to be marginally possible in slow, careful speech might lead to the idea that extraposition is at least a marginal option. However, the fact, illustrated in (74c'), that pied piping of the postverbal phrase under VP-topicalization leads to a highly marked result (especially without the intonation breaks) suggests that we are dealing with a right-dislocated phrase after all.

The examples in (74) are cases in which the adverbial phrase refers to a time interval that includes speech time. The examples in (75) illustrate the same for a temporal adverbial phrase referring to a point in time following speech time.

Note that lexical items like morgentomorrow and gisterenyesterday, which are listed as adverbs in dictionaries exhibit the same behavior as the nominal phrases in (74) and (75), for which reason we simply treat them as nouns.

Recall from Section 12.1, sub III, that noun phrases can also be used as measure phrases indicating duration when they are selected by verbs such as durento last. Although such measure phrases are often considered adverbial phrases, they differ from the cases discussed in this subsection in that they categorically reject postverbal placement: cf. dat de workshop <een hele dag> zal duren <*een hele dag>that the workshop will last a full day. The pied piping/stranding behavior of these measure phrases cannot be investigated, because constructions of this kind do not easily allow VP-topicalization: *Een hele dag duren doet/zal deze workshop.