The term subject can be defined in at least two ways, which has led to a distinction between grammatical and logical subjects. The traditional definition of subject is based on the case of the noun phrase: the grammatical subject is the noun phrase to which nominative case is assigned, like Jan/hijhe in (3a) and ik in (3b). Although the accusative phrase Jan/hemhim in (3b) is traditionally called the object of the verb ziento see, it has an identical thematic relation to the verb lachen as the nominative phrase Jan/hijhe in (3a); if we were to define the notion of subject in terms of this thematic relation, we could say that in both cases the phrases Jan/hij and Jan/hem function as (logical) subjects of the verb lachen. Here we use small capitals for the notion of logical subject and lower case for the notion of grammatical subject. If the two notions of subject refer to the same argument in the clause, as in (3a), we simply use the term subject in lower case.

Like (intransitive) verbs, adjectives denote sets of entities; cf. Section 23.3.2. The members of the denotation set of a verb like lachen can be used as the logical subject of the verb: the two examples in (3) both express that Jan is part of the set denoted by lachen. Similarly, we say that the examples in (4) express that the noun phrase de hondthe dog is part of the denotation set of the adjective dooddead. Therefore, it seems useful to extend the notion of logical subject in such a way that it also covers the relation between the noun phrase de hond and the predicatively used adjective dood.

However, the claim that the noun phrases Jan/hem/hij and de hond are subjects of the verb lachen and the adjective dood, respectively, raises problems related to the intuition that these noun phrases also function as the objects of the finite verbs ziento see and slaanto hit in (3b) and (4b), respectively. First, the noun phrases are assigned accusative case by these verbs, which is especially clear in example (4b): if we passivize the verb slaan, as in (5), the noun phrase de hond appears as the nominative subject of the whole clause. Since we have defined the notion of subject in terms of the thematic relation that the noun phrase has with its predicate, this need not be seen as a serious problem, since case-assignment does not depend on the thematic relations within the sentence; compare the case assignment in the active sentence in (4b) and the passive sentence in (5).

Second, and potentially more serious, the noun phrases in (3b) and (4b) also seem to function as the object of the main verb in terms of their thematic properties: example (3b) implies that we actually see Jan and (4b) implies that the dog is actually hit. However, it has been claimed that the thematic relation between the accusative object and the main verb is secondary to the predicative relation between the accusative object and the adjective; cf. Hoekstra (1984a). An argument for this claim is that comparable examples can be constructed in which the thematic relation between the accusative object and the verb is absent.

This is very clear in the resultative and vinden-constructions in (6): the accusative noun phrases in these examples simply cannot appear when the adjective is absent and so cannot be seen as the thematic object of the verb: it is clearly only an argument of the adjective. The number signs in (6c&d) indicate that the structures without the adjective are possible with the interpretation “Jan finds Marie/the book”, which we can disregard here.

Of course, the examples in (6) are do not show that there is no thematic relation between the noun phrase de hondthe dog in (4b) and the verb slaanto hit, but they do support the claim that adjectives take a logical subject, i.e. that there is a thematic relation between predicatively used adjectives and the arguments in the clause they are predicated of. This claim is also supported by the so-called absolute met-construction in (7): the noun phrase Jan is clearly thematically dependent only on the adjective ziekill.

In the examples in (4) and (6), the adjective is an intrinsic part of the predicate expressed by the VP, which is especially clear in (6), in which the adjective is obligatory. For this reason, we will refer to such cases as complementive adjectives. In other instances, the predication expressed by the adjective is of a secondary nature, i.e. supplementary to the event expressed by the VP. An example is given in (8a): the secondary nature of the predication relation between the adjective kwaadangry and the noun phrase Jan is clear from the fact that the adjective can be dropped without affecting the main proposition expressed by the clause: we only lose the supplementary information that Jan was angry while he performed the action expressed by the VP. We therefore refer to such cases as supplementive adjectives.

Complementive adjectives found in the resultative constructions in (6) are always predicated of the accusative noun phrase, if present. If no such noun phrase is present, the adjective is predicated of the nominative subject of the clause, as shown in (9a&b). These cases involve unaccusative verbs, as is clear here from the use of the auxiliary zijn in the perfect-tense constructions in the primed examples and the acceptability of the attributive use of the past/passive participle in the doubly-primed examples; cf. Chapter V2. Note that the complementive krombent is obligatory in all (b)-examples.

Supplementive adjectives, on the other hand, can be predicated of the subject of the clause when a direct object is present, as is shown in (8a). The noun phrase Jan in (8a) also behaves as a regular subject of the activity verb gooiento throw: the fact that it can appear in a passive door-phrase in (8b) clearly shows that it acts as the agentive argument of this verb.

Complementive adjectives differ from supplementive adjectives in that only the former can license/introduce a noun phrase that is not selected by the verb. Consider the examples in (10). The primeless examples show that weather verbs like regenento rain and vriezento freeze do not select a referential noun phrase like de jongenthe boy as their subject. However, the singly-primed examples show that such a referential noun phrase becomes possible when a complementive (here: resultative) adjective is added, which is compatible with the conclusion drawn on the basis of the examples in (6) that the noun phrase de jongen is licensed as the logical subject of the resultative adjective. The fact that the doubly-primed examples are unacceptable shows that adjectives do not license referential noun phrases when they are used as supplementives; the noun phrase de jongen is not an argument of the adjective kwaad and should therefore be selected by the weather verbs, which the examples in (10a&b) have already shown to be impossible. We will discuss this issue in more detail in Section 28.2.1, sub II.

The examples in (11a&b) show that it is not possible to retain the pronoun hetit in the resultative constructions in (10a'&b'). This supports the widely accepted idea that the pronoun het is not a thematic argument of the weather verb but acts as a placeholder for the empty subject position. The unacceptability of the primed examples, with het interpreted as semantically vacuous weather het is due to the fact that weather het cannot act as the logical subject of an adjective; the number signs indicate that these examples are acceptable if het is interpreted as a deictic pronoun referring to, e.g. het paardthe horse; we can ignore this here.

For completeness, note that in contrast to (12a), example (12b) is fully acceptable under the non-referential interpretation of het. However, it is not clear whether we are dealing with weather het here, since the construction seems to imply a location, which can be made explicit by adding a locational constituent such as buitenoutside; Section 28.5.2, sub III, will argue that het should be seen as an anticipatory pronoun introducing an (implicit) locational subject.

We still need an explanation for the fact that the supplementive in (8a) is predicated of the noun phrase Jan. One possibility is to assume that the supplementive has a phonetically empty subject; this implied subject PRO is interpreted as coreferential with the phonetically realized noun phrase Jan. This would correctly account for the intuition that the supplementive is a kind of reduced clause, i.e. that (8a) can be paraphrased as: Jan gooide het bord tegen de muur, terwijl hij kwaad wasJan threw the plate against the wall, while he was angry. Section 28.3 will discuss this in more detail.