The early versions of generative grammar normally assumed that phrases occupying the postverbal field are base-generated in the middle field of the clause, in line with Koster’s hypothesis that the underlying structure of Dutch is OV in nature, and are subsequently moved into postverbal position by a rule known as extraposition in the case of clauses and PP-over-V in the case of PPs; we simply refer to this rule as extraposition. A problem for this proposal was that it is not in line with Emonds’ (1976) structure preservation principle, which requires movement to target an independently motivated position; cf. Emonds (1976).
      If this was not enough, extraposition also came up against an important empirical problem related to the freezing principle, which prohibits wh-extraction from a moved phrase. At first sight, extraposition of PPs seems to provide strong evidence in favor of a movement analysis, as it only allows wh-extraction if the PP is in preverbal position; if the postverbal position of the PP in (136b) is indeed a derived position, the freezing principle correctly predicts wh-extraction from that position to be impossible.

However, this principle also predicts that wh-extraction from an extraposed clause is impossible, but this is clearly wrong given that it is possible in so-called bridge-verb contexts; cf. De Haan (1979).

The contrast between the extraction possibilities from extraposed PPs and clauses has given rise to the claim that extraposition is not a unitary phenomenon. Barbiers (1995/2000), for example, provides two completely different but compatible analyses for the examples in (136) and (137). That extraposition is not a unitary phenomenon becomes even clearer when we include split extraposition, which has resisted a satisfactory syntactic account for a very long time. Since Kaan (1992), analyses have been developed that give up the idea that split extraposition is derived from a structure in which the split parts form a constituent underlyingly. Koster (2000) and De Vries (1999/2002) have claimed that split extraposition is actually a form of juxtaposition (with or without deletion). For more historical background we refer the reader to Section 9.4, as well as Corver (1991), Kaan (1992), Koster (2000), Baltin (2006), De Vries (2002), and references cited there.