Indefinite noun phrases are not identifiable to the speaker and/or the addressee. This also holds for modified noun phrases. The effect of adding a PP-modifier is simply that the relevant set is smaller in size: in example (44c), for instance, the speaker expresses that the entities he has in mind are part of the subset of tulips that have the property of being from Amsterdam.

The noun phrases in (44) can be pronounced with neutral intonation or with an accent on the PP. In the former case, the speaker is simply referring to a (possibly empty) set of entities with the desired properties. In the latter case, as in (45), two sets are (implicitly or explicitly) contrasted: the first set is characterized by having the property denoted by the nominal head but not the property denoted by the PP, whereas the second set has both properties. Example (45b), for example, contrasts the set of girls with red hair with the set of girls with hair of a different color.

Indefinite constructions of this type can also be used generically with both singular and plural noun phrases. In example (46a), for instance, the predication is said to hold for all watches from Switzerland, while in (46b) it is claimed that all tomatoes from tropical countries are delicious. Again, it is not asserted that the predications hold for the larger sets of watches and tomatoes.

In quantified noun phrases, restrictive PP-modifiers again perform the function of restricting the denotation of the NP. As the PPs are part of the NP-domains and thus fall within the scope of the quantifier, they serve to restrict the domain of the quantifier. In other words, in (47a) the PP voor mijn moederfor my mother first restricts the set of possible presents, and then the noun phrase as a whole is quantified by enkelesome. As a result of these scope relations, the sentence in (47a) says nothing about the total number of presents I am looking for (which may be many), but only about the number of presents for my mother. Similarly, in (47b) the predicate tasty is not assigned to all tomatoes, but only to those from Italy, while in (47c) not all books are reduced in price, but only those with a red dot.

In definite noun phrases, the function of the PP is to enable the hearer to uniquely identify the intended referent set in the given discourse situation. Example (48a), for example, can be used in a context where one might be expected to bring presents for one’s mother. Note that, as in the case of indefinite noun phrases in (45), stressing the PP implies the existence of yet another set of presents: in (48b), where the PP is contrastively stressed, the hearer is assumed to know that the speaker has bought presents for a number of people.

When the noun phrase is singular, the referent set is said to contain exactly one referent; thus, the most likely reading of (49a) is one in which the neighbor has only one car. This effect seems to be lost, or at least less prominent, in contrastive contexts: (49b) simply expresses that a car was confiscated, and that it happened to be my neighbor’s car (not someone else’s).

The examples in (50) show that, in the case of locational prepositions, the PP-modifier can be contrasted in several ways. First, consider the contrast between the PPs in the (a) and the (b)-examples: the PP in (50a) is not contrastive and only serves to distinguish the book on the table from any other book, whereas the PPs in the (b)-examples are contrastive and thus imply a second set of books that are not on the table. In (50b), the contrastive accent is placed on the complement of the preposition, de tafelthe table, and the alternative set of books is consequently identified by its relation to an object other than the table. In (50b'), on the other hand, the contrastive accent falls on the preposition, and the alternative set of books is consequently identified by having a different orientation with respect to the table.

Generally speaking, when a singular definite noun phrase is modified by a PP, the nominal complement of the preposition is also definite, as shown in (51a&b). Given that definiteness typically indicates identifiability, this is not surprising; if the entity itself is identifiable, then the properties referred to by the PP are also identifiable. Alternatively, one could argue that since the function of the modifier is to enable the addressee to identify the referent, we again expect the presence of a definite noun phrase in the PP.

As shown in (52), plural definite noun phrases can be modified by a PP when the noun phrase complement of the preposition is indefinite. In this case, the DP refers to a contextually determined set of entities, and the function of the PP-modifier is to restrict this set to those entities that have the property expressed by the PP.

Note that the referent set of (52b) varies with the choice of determiner: if we are dealing with the definite determiner de, the set consists of houses on a contextually determined street corner; if we are dealing with the indefinite determiner een, the set consists of the subset of houses on any street corner. For a similar reason, example (52a) with a definite noun phrase complement is unacceptable because it forces a reading in which there is a particular (identifiable) hole in each of the buckets.

The (a)-examples in (53) show that singular indefinite noun phrases have the inverse property of not allowing PP-modifiers if the noun phrase complement of the preposition is definite: once a particular property is identifiable or known, so must be the entity referred to by the noun phrase as a whole. The (b)-examples are acceptable with a definite PP-modifier only if it is established knowledge that there is more than one house on the street corner in question.

If this line of reasoning holds, we should conclude that in inalienable possession constructions, the addition of an indefinite van-PP to a definite noun phrase can make the definite noun phrase sufficiently “indefinite” to act as the modifier of an indefinite noun phrase. This is illustrated by the examples in (54), which are acceptable in any context.

Combinations of a demonstrative determiner and a restrictive PP-modifier are not very common. This is not surprising, since in most cases the use of a demonstrative, whether deictic or anaphoric, suggests the identifiability of the referent, thus rendering the use of a restrictive PP superfluous. This explains why the sentences in (55) are marked with neutral intonation in neutral contexts.

However, we will show below that there are a number of special cases where using a PP-modifier will lead to an acceptable result.

Restrictive PP-modifiers cannot easily be used to modify constructions with possessive pronouns, personal pronouns or proper nouns. In all cases, the referent of the noun phrase is assumed to be identifiable independently of the information provided by the PP-modifier, which is therefore superfluous. However, there are certain cases where it is possible to add a restrictive PP-modifier. These will be discussed in the following subsections.

When the noun phrase is indefinite, the implication is that the additional information provided by the PP applies to all members of the referent set of the noun phrase, regardless of the specificity of the noun phrase. In (70a), for example, we have a non-specific indefinite noun phrase, and it is claimed that the speaker is looking for presents; as additional information, it is added that they are intended for the speaker’s mother. In (70b) we have a specific indefinite noun phrase, and the PP-modifier adds as additional information that the property of being impressed by the view applies to the person the speaker has in mind.

Non-restrictive PP-modifiers can also be used if the indefinite noun phrase is interpreted generically. Perhaps the result in (71a) is somewhat marked, but as far as this sentence is acceptable, it expresses that all cuckoo clocks come from Switzerland. Sentence (71b) seems perfectly acceptable and expresses that all Lapps live in the north of Sweden.

Non-restrictive PP-modifiers fall outside the scope of the quantifier in quantified DPs; consequently, they do not serve to restrict the domain of the quantifier, but provide additional information about an already quantified set. Moreover, the truth conditions of the predication are not affected by the presence or absence of the modifier. In (72a) the speaker is looking for some presents; the PP merely adds the information that they are all intended for the speaker’s mother. In example (72b), the speaker states that there are many tourists in the hotel; the additional information provided by the PP is that they are all from Germany. Similarly, in the generic sentence in (72c), the additional information asserted is that all the tomatoes under discussion are from Italy. Finally, in (72d), each book is said to be reduced in price; from a semantic point of view, the information provided by the non-restrictive PP-modifier is superfluous and is added merely for the pragmatic reason of removing potential doubt on the part of the hearer.

If the DP is definite, the function of the PP is again to provide additional information about the (possibly singleton) referent set of the noun phrase. This means that this referent set is assumed to be uniquely identifiable to the speaker without the information in the non-restrictive modifier. Thus, the most likely reading of (73a) is one in which the car in question has already been introduced into the discourse and is therefore identifiable by the addressee. The salient new information is that it has been confiscated; additional and typically new information about ownership is provided by the non-restrictive PP-modifier van mijn buurmanof my neighbor. Since the identifiability of the intended referent is not determined by the PP-modifier, the neighbor may actually own more than one car. Similarly, (73b) is felicitous in a context where the speaker has already mentioned the present without mentioning the benefactive. The latter information is now added as additional information which does not (cannot even) serve to identify the intended referent. The cat mentioned in (73c) is identifiable for the speaker, and two things are said about it at the same time, namely that it was purring and that it was lying between soft cushions (which might well be the cause of the purring event).

As can be seen from example (74), non-restrictive PP-modifiers can also be used in generic contexts. Again, the implication is that the additional information provided by the PP-modifier is new to the hearer and applies to all members of the referent set of the noun phrase.

It is quite common for DPs with a demonstrative determiner to be followed by a non-restrictive PP-modifier; the use of the demonstrative, which suggests identifiability on the basis of textual or contextual information, does not conflict with the function of non-restrictive modifiers. The sentences in (75) are therefore all perfectly acceptable, both with and without the PP-modifiers.

Similarly, in contrastive contexts the PP-modifier does not serve to distinguish between two intended referents: the identity of the intended referents is assumed to be known from the context; the information given in the PP-modifiers is additional information about these referents. That this additional information is nevertheless relevant can be seen in example (76b), where the two non-restrictive PPs provide the reason for the claim made in the main clause (see also (75c)).

Non-restrictive PPs can easily be used to modify constructions with possessive pronouns, personal pronouns, and proper nouns. In all cases, the referent of the noun phrase is assumed to be identifiable independently of the additional information provided by the PP-modifier. The PP-modifier can omitted without affecting the grammaticality, felicity or truth conditions of the proposition.