The examples throughout this study amply illustrate that the core case of complementation of an adpositional phrase involves a noun phrase. The only restriction on nominal complements seems to be of a semantic nature. For example, if we use the spatial preposition binnen'inside' or buiten'outside', the nominal complement must have dimensional properties that are compatible with these prepositions. Example (4a) is acceptable because a city can be conceived of as an entity with an interior and an exterior. Example (4b) is semantically anomalous because een dak'a roof' is normally not used to divide space in this way; we really have to force an exceptional interpretation on this example in order to make it acceptable. An example such as (4c), finally, is completely unacceptable because space is not involved at all. It seems that, as long as the semantic restrictions imposed by the adpositions are met, any noun phrase can be used as a complement.

The set of prepositions that may take a bare singular noun phrase as their complement is rather limited. Some examples are given in (5): the (a)-examples involve the locational prepositions in'in' and op'at', and the (b)-examples the directional prepositions naar'to' and van'from'; examples (5c&d) show that the reduced phrasal directional prepositions richting'in the direction of' and the temporal prepositions voor'before' and na'after' can also take a bare noun phrase.

The interpretation of such PPs is often rather special. The PPs op school and op kantoor in (6), for example, are not spatial in nature, but express that Jan has a certain occupation; note that the location in these examples is expressed by the deictic locational pro-form hier'here'. These readings are virtually lost if a determiner is used; the noun phrases then preferably refer to the actual objects, and the PP is interpreted spatially.

Not only is the number of prepositions that can enter this construction limited, there are also a number of badly understood restrictions on the nominal complement. For example, given the acceptability of the examples in (6), one might expect that (7a) would also be acceptable with the meaning “Jan is enrolled as a student here (= at this university)”. This example is, however, severely degraded; the only way to express the intended meaning is by using the definite determiner de preceding the noun, as in (7b).

      In the examples above, the bare singular nominal complements give rise to interpretations involving occupations, but they may also give rise to event interpretations, which may or may not be idiomatic in nature. Apart from the fact that the use of a definite determiner is excluded, the meaning of (8a) seems fully compositional. Example (8b), on the other hand, has the more specialized interpretation that Jan has gone to bed in order to get some sleep (which will be lost if the article het'the' is added). Note that in cases such as (8c), the presence or absence of the definite determiner does not seem to make much difference for the interpretation.

      The contrasts between the Dutch and English examples in (9) suggest that bare singular nominal complements are not as common in Dutch as in English; see for more English data Quirk et al. (1985: 277-9), from which we have also taken the primed examples. These contrasts strongly suggest that the collocations consisting of a preposition and a bare noun phrase are listed in the mental lexicon.

      Finally, note that the Latinate preposition per in (10) is characterized by the fact that it can only take bare noun phrases.

The preposition in in (11) often occurs with the verbs zijn'to be' or brengen'to bring' in more or less fixed collocations, many of which denote mental states.