Section 3.1.3, sub IB, has discussed examples such as (100a) and argued that achterop is a compound. The main reason for this is that under R-pronominalization the R-word cannot intervene between achter and op , as shown in (100b), which would be possible if we were dealing with a construction in which achter acts as a premodifier of op , or with a construction in which op de auto is a complement of the preposition achter ; cf. the discussion in Section 1.2.1, sub II.

What we did not discuss there, however, is that there are seemingly similar constructions that behave slightly differently. This will become clear by considering the examples in (101). Although at first sight the clause structures of (100a) and (101a) seem completely parallel, R-pronominalization is excluded in the latter case. The number sign is added to (101b) with er preceding achterop, given that this order is acceptable on the reading “on the back of it”, but on this reading it is clearly not related to (101a).

In order to fully appreciate the difference between (100a) and (101a), one must know that adpositions like achter, voor, boven and beneden can also be used as referring expressions: in a domestic situation, for instance, achter may refer to the rooms in the back of the house, the back yard, etc.; voor may refer to some room in the front of the house; and boven and beneden can be translated as “upstairs” and “downstairs”, respectively. The semantics of the construction in (101) seems to be that achter refers to a certain subpart of the house and that the PP op de plaats further specifies the intended place, that is, that the PP functions as a postmodifier of the referring expression achter.
      The difference between the two constructions can be made clearer by means of the ambiguous example in (102). On the compound reading of bovenop, the sentence expresses that a new shirt is lying on top of the wardrobe. On the postmodification reading, the sentence expresses that the shirt is lying on the wardrobe upstairs. As would be expected on the basis of the data in (100) and (101), R-pronominalization is only compatible with the compound reading: (102b) can only mean that the shirt is lying on one of the top shelves of the wardrobe. Another difference between the two readings is that the phrase boven op de kast can be split on the modification reading, but that this is impossible (for obvious reasons) on the compound reading: (102c) is therefore only compatible with the modification reading. For completeness' sake, we want to note the intriguing fact that, on the modification reading, it seems possible to reverse the order of boven and the PP, as in (102c').

Note that the ambiguity that arises with boven is due to the fact that it can be used both as the first member of the compound bovenop and as a referring expression meaning “upstairs”. A similar ambiguity does not arise with onder, simply because the “downstairs” reading is lacking; the (c)-examples in (103) are uninterpretable as a result.

For the “downstairs” reading, it is the preposition beneden in (104a) that is used. Since a compound reading is not available for *benedenin, it does not come as a surprise that the judgments on the (b)- and (c)-examples in (104) are the mirror image of the corresponding examples in (103).

      We conclude by briefly addressing another potential case of ambiguity, which can be found in (105a). On one reading, this example seems to express that the plane is flying high, with the PP boven de wolken functioning as a postmodifier of the adjective hoog'high', specifying more precisely the position of the plane. On the alternative reading, it is expressed that the plane is flying above the clouds, with the adjective hoog acting as a premodifier specifying the distance between the plane and the clouds. The first reading seems similar to the modification reading that we discussed for boven'upstairs' and beneden'downstairs' above, and the second reading is similar to that of constructions such as 10 km boven de wolken'10 km above the clouds', where a nominal measure phrase is used as a premodifier of the PP. The fact illustrated in (105b&c) that both hoog and boven de wolken can be used as the predicative part of the construction is consistent with this account of the observed ambiguity.

The facts in (106b&c) suggest that (106a) is ambiguous in a way similar to (105a), although it does not seem readily possible in this case to pinpoint a corresponding semantic difference.