In copular and vinden-constructions, a nominal predicate denotes a non-singleton set, and it is claimed that the logical subject of the predicate is part of that set. In the copular constructions in the (a)-examples in (92) the logical subject of the predicate is the subject of the clause, and in the vinden-constructions in the (b)-examples it is the accusative object of the clause. These examples also show that the nominal predicate and its logical subject agree in number. Example (92c) is added to show that, for some unknown reason, resultative constructions with a “truly” nominal predicate are not common; resultative verbs generally take a nominal predicate introduced by als or tot, which will be discussed in Subsection III.

The (a) and (b)-examples in (92) have illustrated the use of indefinite noun phrases as predicates. Definite noun phrases can be used in the same way. The main difference is that while the use of an indefinite noun phrase expresses that its logical subject is part of the denotation of the NP, the use of a definite noun phrase implies that the logical subject exhausts it. For example, (93a) expresses that Peter is only one member out of the set of students supervised by Marie, while (93b) implies that Peter is the only student supervised by Marie (within the given domain of discourse). The (b)-examples in (93) thus differ from the (a)-examples in that they do not express a set/subset relation, but an identification relation.

The fact that we are dealing with an identification relation does not mean that both DPs are referring expressions. That this is not the case will become clear when we consider the referential behavior of the coordinated DPs in (94). The plural marking on the finite verb in (94a) shows that the coordinated subject de dokter en de burgemeesterthe physician and the mayor is also necessarily plural. This shows that there is a one-to-one relation between the number of articles and the number of referents: each conjunct refers to a separate person. However, this is not the case in (94b), in which the coordinated DP functions as a predicate: there is only one person who can be identified as both the physician and the mayor of the village. Example (94b') shows that predicative definite noun phrases can also be coordinated by conjunctions other than enand. Note that the adverbs tevensalso and ookalso in (94b&b') emphasize the fact that the persons denoted by the coordinated predicative noun phrases are actually the same; cf. also the discussion of the examples in (103) below.

Examples like (94b&b') clearly show that definite nominal predicates do not refer to any entity in the domain of discourse; the function of the definite articles is to express that the sets denoted by the nominal predicates dokter van dit dorp and burgemeester van dit dorp have only one member. The construction as a whole expresses that the members of these two singleton sets are identical and can be identified with the referent of the subject of the clause, i.e. Jan. See Alexiadou et al. (2007: prt II, §2.3) for further discussion.

The examples in (95) show that in constructions expressing an identification relation it is often possible to swap the positions of the two noun phrases. At first sight it is not clear what this tells us. It may be the case that the two word orders differ in underlying syntactic structure, i.e. differ in terms of which noun phrase performs which syntactic function (subject or complementive). However, it may also be the case that the two orders have the same underlying syntactic structure and that one of the two orders is derived by topicalization, an option that can be argued independently on the basis of copular constructions such as Aardig is hij nietNice, he is not.

The possibility of having two different underlying structures seems to be refuted by the examples in (96) with embedded clauses: if the proper noun(s) need not act as subject but can also act as predicate, the primed examples should be perfectly acceptable, which they are not. However, they are not as bad as one might expect: especially when the proper nouns are given emphatic accent, the result is reasonably acceptable, and it at least yields quite a contrast with examples such as *dat aardig Jan niet is, in which aardig is clearly a predicate.

Another prediction would be that in the vinden-constructions the two noun phrases are also interchangeable. However, as can be seen in (97), there is a strong preference for the proper noun(s) to precede the definite noun phrase: the primed examples in (97) are unacceptable with neutral intonation, and at best marginally possible with strong emphasis on the proper nouns. Therefore, it seems safe to conclude that in these examples the proper noun acts as the logical subject and the definite noun phrase acts as the predicate.

For completeness’ sake, note that the test in (97) is only available if the nominal predicate is evaluative; if it is non-evaluative, the vinden-construction always yields a degraded result; cf. Els is de (beste) voorzitterEls is the (best) chairperson versus Ik vind Els de *(beste) voorzitterI consider Els to be the best chairperson.

That the definite noun phrase functions as the complementive can also be supported by subject-verb agreement: if the clause contains a first-person or second-person personal pronoun, the verb must agree with the pronoun and not with the nominal predicate. This is illustrated by the examples in (98), in which the copular verb appears as the second-person form ben(t)are and not as the third-person form isis. Similar facts concerning number agreement can be found in (99).

The (b)-examples in (98) and (99) show that the agreement facts in Dutch are different from those in their English renderings, where the inversion of subject and predicate leads to verb agreement with the preposed predicate. This is illustrated again in example (100), which also shows that the pronoun appears in its subject form in Dutch, but in its object form in English. Den Dikken (2006:§4, fn.43) argues that these differences show that predicate inversion results from topicalization in Dutch, as we have assumed throughout, but not in English.

The discussion above has shown that there is a fixed division of labor between the two noun phrases in predicative constructions expressing identity. This leads to the question of what determines whether a given noun phrase acts as a logical subject or as a predicate. As in the case of regular predicative constructions, this seems to be related to inclusion relations. While proper nouns usually refer to a specific entity in the domain of discourse, the referents of definite noun phrases are primarily presented as members of a larger set denoted by the NP. For example, the definite noun phrase de beste leerling van de klasthe best pupil(s) of the group does not simply refer to a specific individual, but to an individual who is characterized as a member of a larger subset denoted by the NP leerling van deze klaspupil of this group. The facts reviewed above suggest that it is always the noun phrase presented as part of a larger superset that is taken as the predicative part of the construction.

It seems that this can be made even more precise. The examples in (101), which involve two definite noun phrases, suggest that it is the noun phrase presented as part of the largest superset that is taken as the complementive. In the most plausible extra-linguistic context to use the primeless examples in (101), the set of tulips will be considerably smaller than the total set of flowers exhibited at the exhibition, and the primed examples unambiguously show that as a result it is the noun phrase de mooiste bloem van deze tentoonstelling that functions as the complementive, since example (101b') is at best marginally acceptable with a strong emphatic accent on the noun phrase deze blauwe tulp.

Of course, there are still many cases in which it is not so obvious which of the two noun phrases should be considered the subject or the predicate of the construction; the two primeless examples in (102) seem to be at least semantically equivalent, which is supported by the fact that the word orders in the embedded clauses in the primed examples suggests that the two definite noun phrases can both be considered as either the subject or the predicate of the copular construction. However, the difference between (102a') and (102b') is that in the former the noun phrase de voorzitter van deze vergadering is taken to be a referential expression, the referent of which is known to both the speaker and the addressee, whereas in the latter it is taken to be a property assigned to the referent of the referential noun phrase de decaan van de universiteit. In short, the assignment of referential or predicative status depends on the common ground of the discourse; the speaker is taken to know the referent of the noun phrase de voorzitter van deze vergadering in (102a') but not in (102b'), whereas the inverse holds for de decaan van de universiteit is not: the sentences are intended to assert that the two noun phrases have the same referent.

The primeless examples in (102) seem semantically ambiguous, but the two interpretations are generally associated with two different intonation patterns. If the sentences are interpreted with the first noun phrase as a referential expression, it can be pronounced with a neutral, continuous intonation contour. The alternative, inverse, interpretation requires a special phonetic cue: the second noun phrase can be preceded by a brief intonation stop and/or assigned emphatic accent. The inverse interpretation can also be blocked by adverbs like ookalso or tevensalso: the second noun phrase in the examples in (103) can only be interpreted as the predicate.

In some cases, the nominal predicate must be introduced by als, tot, or (less frequently) voor. This seems to be at least partly an idiosyncratic property of the selecting verb. Verbs like beschouwento consider and benoemento appoint, for example, obligatorily take such a predicate: in (104a) the als-phrase cannot be omitted without shifting the meaning of the verb to something like “to observe”; in (104b) the tot/als-phrase can be omitted, but is then semantically implied.

The examples in (105) are some more or less fixed collocations with the verb houden and the preposition voor; in the (a)-example voor is followed by a noun phrase and in the (b)-example by an adjective.

The use of a nominal predicate introduced by tot is the normal way of expressing a result. While bare adjectival predicates like boos or dood can be used in all kinds of resultative constructions, the possibility of using a bare nominal predicate seems to be restricted to the verb maken.

We should add that, in line with what we have already observed with respect to the examples in (104), many cases of the kind in (106b') seem to be lexically restricted. An example such as (107a) is unacceptable, even though it seems to make perfect sense semantically; cf. the acceptability of (107b) with the adjectival complementive fit.

All of the examples discussed above involve a set/subset relation. However, the predicative als-phrase can also be used to express an identity relation, in which case the noun phrase is typically definite.

Finally, example (109) shows that noun phrases introduced by als can also be used as supplementives. Note that the noun phrase in this example does not contain an article, just like the noun phrase voorzitterchairman in (104b). This is a typical property of predicatively used nouns denoting a profession or social function; we will discuss this in detail in Section 22.2.2, where we will also discuss examples such as (109) in more detail.