The suggestion that (303b) must be analyzed along the same lines as the easy-to-please construction in (302b) implies that the infinitival clause in (303b) contains a moved empty operator OP which is interpreted as identical to the locational PP in de tuin, as in (304a).

One reason for adopting this analysis is based on the set of examples in (305). Example (305a) shows that there are verbal predicates, like wonento live, which require the presence of a locational phrase (or, alternatively, a qualifying adverb; cf. (322a) in Section 28.5.3). Example (305b) shows that the same restriction applies when the verb wonen functions as the main verb of an infinitival clause. Nevertheless, the locational PP need not (indeed, cannot) be expressed within the infinitival clause if the main clause contains one. By assuming that (305c) contains an empty operator OP, which is coindexed with this locational PP, the apparent violation of the selection restrictions of the verb wonen is solved.

The examples in (306) show that the het and the easy-to-please constructions occur not only in the copular but also in the vinden-construction.

Of course, the analysis suggested in (305c) and (306b) is surprising, since PPs generally do not occur as subjects. An alternative analysis would be to assume that the locational PP in the matrix clause originates within the dependent clause, and that its surface position is the result of movement. In this case, the structure of (305c) is not as indicated there, but as in (307).

However, a first problem with such an analysis is that the movement shown in (307) crosses a clause boundary, which is generally excluded. This is illustrated in (308b), which involves movement of the PP from a complement clause.

Another problem for the alternative analysis in (307) is that the infinitival clause need not be present; a movement analysis along the lines of (307) is not possible for (309a), but its acceptability is fully compatible with the analysis in (305c). The problem that (309a) poses for the alternative analysis would of course be solved if it could be shown that the PP in this example is actually an adverbial phrase. A possible argument for this assumption is that the PP seems to be optional, as shown in (309b).

Note, however, that the pronoun het is referential in (309b), but not in (309a). In (309b), the pronoun refers to something mentioned earlier in the discourse, e.g. watching television. The difference is also clear from the fact that the pronoun het in (309b) can be replaced by the demonstrative ditthis. In contrast, if the PP is present, the pronoun het cannot be construed referentially: example (309a) merely expresses that Amsterdam is a nice place, and het cannot be replaced by a demonstrative. This contrast is illustrated in (310).

The discussion sofar seems to suggest that locational PPs can function as logical subjects of an AP, although it is clear that more research is needed in order to establish this claim more firmly. It is interesting to note, however, that in all the examples so far, the locational PP can be replaced by the locative proforms hierhere and daarthere (cf. (311a)), and that such proforms can indeed occupy the subject position in Swedish; this is illustrated in (311b&c), taken from Holmberg (1986:68)).

The PPs in the examples above all denote a location, but the same construction can be found with PPs denoting a path. If one wanted to discuss two alternative routes from Amsterdam to Groningen, one could use either (312a) or (312b); cf. the examples in (303). The fact that the PP is the logical subject of the adjective is again supported by the fact that the infinitival clause in (312b) can be dropped, as in (312c), and that the pronoun het cannot be replaced by the demonstrative dit; cf. the discussion of the primed examples in (309) and (310).

Finally, note that besides the examples with om + te infinitives discussed above, we also find constructions such as (313) with bare infinitives.

Since the infinitive must precede the finite verb in clause-final position, we are probably dealing with a nominalization functioning as a complementive; if so, the PPs can probably be analyzed as logical subjects of the nominal predicate. To our knowledge, examples such as (313) have not yet been studied.

The examples from Subsection I with a PP subject (if that is the correct analysis) differ in several ways from the cases with a clausal subject. Section 28.5, sub IC, has shown that with a clausal subject in sentence-initial position, the anticipatory pronoun het cannot be realized. In the constructions under discussion, on the other hand, the anticipatory pronoun het must be present if the PP is preposed. This is illustrated in (314).

However, the two examples in (314) exhibit divergent behavior when the infinitival clause is dropped, as in (315): then, examples without het are acceptable when the PP denotes a path, but not when it denotes a location. It is not clear what this tells us, since (315b) without the pronoun het could in principle be understood as a colloquial form of “telegraphic speech”, i.e. as a shorthand for the more elaborate sentence in (315b'). Although Neeleman (1997) argues against this by pointing out that the presumed reduction process should be limited to PP-modifiers denoting a path, let us provisionally assume that the reduction approach is correct, and set aside example (315b) without het as irrelevant.

That the pronoun het must be present with a preposed PP might suggest that the anticipatory pronoun is not associated with the PP at all, but with the infinitival clause. The examples in (316) strongly suggest that this alternative proposal is not tenable. Section 28.5, sub I, has shown that it is possible to prepose the infinitival clause and a predicatively used adjective in isolation when the former functions as a clausal subject of the latter (cf. Subsections C and D), while preposing of the adjective and the infinitival clause as a whole is impossible in this case (cf. Subsection E). The judgments in the examples in (316) show that the facts are exactly the opposite in the construction under discussion; although the (c)-examples are somewhat marked, the contrast with the (a) and (b)-examples is quite sharp.

So, the judgments in (316) strongly suggest that the alternative analysis, according to which the infinitival clause acts as the logical subject of the adjective gezellig/korter, is incorrect.

The examples in (317) show that weather adjectives like warmwarm, heethot, kilchilly, koudcold and natwet typically occur with a PP subject denoting a location, which can also take the form of a proform like hierhere or daarthere. As in (309a), the pronoun het is non-referential in the sense that it does not refer to a discourse entity, and therefore cannot be replaced by a demonstrative pronoun.

This contrasts sharply with examples without a locational phrase: the pronoun het can have either a non-referential or a referential interpretation depending on the context. In the former case, illustrated by the primeless examples in (318), the PP subject has simply been left implicit. In the latter case, illustrated by the primed examples, we are dealing with a referential pronoun acting as the logical subject of the adjective, that can easily be replaced by a demonstrative pronoun.

PP subjects denoting a location occur not only in copular constructions but also in the resultative and vinden-constructions in (319a) and (319b). Example (319c) shows that PP subjects denoting a path also occur in the vinden-construction.

Finally, note that adjectives such as gezellig and the weather adjectives can also occur in constructions of the type in (320). Again, the pronoun het is obligatorily present and cannot be replaced by a demonstrative. Yet the semantic relation seems to be somewhat different than in the earlier examples; instead of referring to a property of the place mentioned in the PP, the adjectives seem to refer to a sensation experienced by the persons denoted by the [+animate] subject of the clause.