The most striking difference between supplementives and complementives is that the former can be freely added to almost any clause with an activity verb (cf. Subsection IV), while the latter occur only with a restricted set of verbs (cf. Section 28.2). Furthermore, the optionality of the supplementive adjectives in (146) suggests that they should be characterized as adjuncts.

The complementive adjectives in (147a-c), on the other hand, must be present, which suggests that they act as complements to the verb. The obligatory presence of the complementives is also due to the fact that they are needed to license the noun phrases de hond/het gras. The fact that the complementive is optional in (147d) is of course related to the fact that the noun phrase de hond can be licensed either as the logical subject of the complementive or as the patient argument of the transitive verb slaanto hit. See Section 28.2.1, sub II, for further discussion.

The bond between the complementive adjective and the verb is thus much stronger than between the supplementive adjective and the verb, which is also reflected in the fact that the meaning of the complementive is an essential part of the core proposition expressed by the clause. The semantic contribution of the supplementive, on the other hand, is rather “supplementary” to the core proposition: the semantic relation between the supplementive and the remainder of the clause is often characterized by the notion of “simultaneity”. In (146a), for instance, it is expressed that the event of leaving and the state of being angry apply simultaneously to the referent of the noun phrase Jan. Crucially, neither of the two predicates is really dependent on the other: the supplementive merely provides additional information about the subject or the direct object.

Complementive and supplementive adjectives are predicated of either the subject or the direct object of their clause. With complementives, the predication relation is always determined unambiguously. If there is no accusative object, as in the copular construction in (147a) above, the complementive is predicated of the subject of the clause. But if there is an accusative object, the complementive must be predicated of that object: example (147b) cannot be interpreted as meaning that it is Marie who is considered kind.

The examples in (148), on the other hand, show that ambiguity can arise in the case of supplementive adjectives. Example (148a), for instance, can express either that Marie is drunk or that the guests are drunk (but not both). Although some speakers may prefer one of the two readings, we can easily demonstrate that we are dealing with genuine ambiguity by replacing one of the animate arguments by an inanimate one: the supplementive is predicated of the subject in (148b) and of the object in (148c), and in both cases the result is perfectly acceptable.

The examples in (149) provide two more cases, in which the selection restrictions of the adjective (i.e. the selection of a [±animate] logical subject) determine whether the supplementive is predicated of the subject or the object.

The examples in (150) show that the supplementive adjective must follow the noun phrase it is predicated of: although (150a) is fully acceptable, it differs from (148a) in that dronken cannot be predicated of the direct object de gasten; it is Marie who is drunk. The acceptability contrast between the two examples in (150b&c) also shows that supplementive adjectives can only precede the direct object in (149) if it is predicated of the subject.

Another difference between complementives and supplementives is that only in the latter case can the noun phrase the adjective is predicated of be left unexpressed. For example, if we passivize example (149a), as in (151a), the supplementive adjective dronken can be felicitously used regardless of whether the passive door-phrase is present or not. Note, however, that this only holds if the supplementive cannot be predicated of the subject of the passive construction for semantic reasons; example (151b) cannot be interpreted to mean that the implied agent was drunk.

In the northern varieties of Dutch, complementives are immediately left-adjacent to the verbs in clause-final position in the unmarked case; cf. the discussion of example (61) in Section 28.2.2. The examples in (148) have already shown that this does not hold for supplementives; their position can be farther to the left. Furthermore, the examples in (152) show that in a clause containing a supplementive as well as a complementive adjective, these are always strictly ordered: the former must precede the latter.

In some cases, as in (152a&b), the supplementive adjective must precede not only the complementive adjective, but also adverbial phrases like al gauwalready quickly and al snelalready quickly. In other cases, however, the supplementive can follow the adverbial phrase; in (152c), for instance, the supplementive natwet follows the frequency adverb altijdalways.

In other cases, the supplementive is not even able to precede the adverbial phrase, as in (154). The factors that determine the relative position of these adverbial phrases and supplementives will be discussed in Subsection III.

The relationship between the supplementive adjective and the clause is usually described as one of “simultaneity”; cf. Haeseryn et al. (1997:1184). This correctly describes the interpretations assigned to the examples in (156): (156a) is interpreted to mean that the event of going home and the state of being content apply simultaneously to the referent of the noun phrase Jan; (156b) is interpreted to mean that the event of being ironed and the state of being wet apply simultaneously to the shirts.

However, the examples in (157) show that the notion of simultaneity is sometimes not precise enough; cf. Broekhuis (1995). The interpretation of (157a), for example, does not seem to be that the activity of speaking incomprehensible nonsense and the state of being drunk are simultaneously applicable to Jan at a given moment. Rather, we are dealing with a conditional relation, in which the supplementive (and the noun phrase it is predicated of) acts as the antecedent (= the when-part) and the remainder of the clause acts as the consequent (= the then-part), i.e. “when Jan is drunk, then he talks rubbish”. The same thing holds for (157b), as indicated by the English paraphrase.

Of course, the notion of simultaneity is also applies to examples such as (157), since the English rendering includes the temporal connectives when ..., then ...; the purely conditional paraphrase if ..., then ... is not applicable. This distinction between the two types of conditional clauses is not so easy to make in Dutch, because they both involve the frame als ..., dan ....

The semantic difference between the examples in (156) and (157) is related to several other differences. The first has to do with the intonation patterns of the examples: the supplementive adjectives in (157) are followed by a rise in the intonation contour, whereas this is impossible in (156a). Second, when a clause adverbial like waarschijnlijkprobably or altijdalways (cf. Section 30.2.2) is present, it is easy to show that the difference in interpretation between (156) and (157) correlates with a difference in word order. The examples in (158) show that the supplementive adjectives tevreden and nat from (156) must follow the clause adverbial (although we will see shortly that the use of the frequency adverb altijd results in the loss of the pure simultaneity reading found in (156)).

The examples in (159), on the other hand, show that the supplementives dronken and ziek from (157) usually precede the clause adverbials (the unacceptable word orders may improve somewhat if the supplementive is explicitly presented as belonging to the new information of the clause, e.g. as an answer to the question Wanneer kraamt Jan altijd onbegrijpelijke onzin uit?When does Jan always talk incomprehensible nonsense?).

For ease of reference, we will refer to the supplementive adjectives in (159), which precede the clause adverbials, as supplementive-I, and to the supplementive adjectives in (158), which follow them, as supplementive-II.

We have already mentioned in passing that the frequency adverb altijd in (158) precludes the simultaneity reading of the examples in (156). Instead, the examples are also interpreted as conditionals, albeit that the logical implications of (158) are of a different kind than those of (159): while in (159) the clause acts as a consequent, it acts as an antecedent in (158). The respective implications are represented in (160) and (161); the arrows ⇒ and ⇏ are interpreted as “always implies” and “does not imply”, respectively.

A minimal pair is given in (162): in (162a), the adjective precedes the clause adverbial altijdalways and we are therefore dealing with a supplementive-I, which is also clear from the validity of the logical implication in (162a''); in (162b) the adjective natwet follows the clause adverbial altijd and we are therefore dealing with a supplementive-II, which is also clear from the validity of the logical implication in (162b'). Note that in these examples the adjective gladsmooth is not a supplementive but a resultative adjective.

Replacing the universally quantified adverb altijdalways in (158) and (159) by an adverb like meestalgenerally would have the semantic effect that the implications in (160) and (161) do not always hold, but only generally. In other words, the arrow ⇒ would not be interpreted as “always implies”, but as “usually implies”. In our representations, the semantic contribution of the clause adverbial is thus captured in the interpretation of ⇒; a formal semantic representation would involve an operator quantifying over time. In the following examples, we will mainly use the adverb altijd, because this allows us to keep the semantic representations of these examples as simple as possible; they are simple material implications, without modal operators that evoke possible worlds in the sense of logic.

The examples in (162) also show that supplementive-I and supplementive-II both precede the complementive, in this case the adjective gladsmooth. The examples in (163) show that they also precede PP-complements of the verb (unless the latter are assigned focus accent and moved leftward).

The discussion above has shown that the use of supplementive-II (i.e. placement of the adjective after the quantified adverb altijd) implies that if the proposition expressed by the verbal part of the clause is true, the predicate expressed by the adjective should also applies (but not vice versa), while supplementive-I implies that the reversed situation holds.

Supplementive-I and supplementive-II can co-occur in a single clause. As the examples in the previous subsection already suggest, the former must precede the latter in neutral cases; cf. the examples in (164). The fact that supplementive-I and supplementive-II can co-occur also makes it clear that they do not have the same grammatical function, but are instantiations of two different grammatical functions. For completeness’ sake, the primed examples in (164) provide the valid implications of the examples in the primeless examples; the reverse implications do not hold.

In (164), supplementive-I and supplementive-II are both predicated of the nominative subject Jan. The examples in (165) show that it is also possible for the two supplementives to be predicated of different arguments in the clause: in (165a) supplementive-I natwet is predicated of the direct object de overhemdenthe shirts, while supplementive-II opgewektcheerful is predicated of the subject Jan; in (165b) it is supplementive-I that is predicated of the subject, while supplementive-II is predicated of the direct object.

While supplementive-I and supplementive-II may co-occur, it is not possible to have two supplementives-I referring to two different arguments in the clause. The (a)-examples in (166) are uninterpretable. It is less clear whether the same thing holds for supplementive-II: although the (b)-examples are marked, they are clearly better than the (a)-examples.

Note that (166a) improves when the supplementive opgewekt is followed by an intonation break, which is indicated by a dash in (167a), but nat then seems to be interpreted as a supplementive-II. This may have to do with the fact that frequency adverbs are not always interpreted as clause adverbials, but can occasionally also be interpreted as VP adverbials; cf. Section 30.2.3, sub III. Example (166a') also seems to improve somewhat when nat is followed by an intonation break.

The examples in (166) and (167) show that the two kinds of supplementive distinguished here can easily co-occur but that it is less clear whether or not multiple occurrences of the same kind of supplementive are allowed. Determining the latter requires further investigation of various factors like the kind of supplementive, the type of clause adverbial, and intonation; we leave this issue to future research.

The distribution of supplementive-I and supplementive-II may depend on certain properties of the clause they are part of. These are discussed in the following two subsections.

We conclude this subsection on the distribution of supplementive-I and supplementive-II by considering their behavior in topicalization and wh-constructions.

Adjectives referring to an “inherent” or “permanent” property of the modified noun phrase do not produce a felicitous result when used as a supplementive. This is illustrated in (184) by the contrast between the stage-level adjective vermoeidtired and the individual-level adjective intelligent, which denote a temporary and a more permanent property, respectively.

The contrasts in (184) are probably related to the fact that similar contrasts can be observed in their paraphrases in (185). The conditional paraphrases in (185b-c) associated with (184b-c) are equally odd: apparently, both the when and the then-clause of a conditional when(ever) ... then-sentence must denote temporary situations in everyday use. Something similar holds for the simultaneity reading of (184a), which can be paraphrased as in (185a).

Note that example (186) is not a counterexample to our claim, as the adjective intelligent is not predicated of the noun phrase Jan in this case. Instead, the adjective modifies the VP: we are dealing with a manner adverb; cf. Section 30.2.3.

Note that we are not claiming that individual-level adjectives can never be used in conditionals; the examples in (187) show that they can. The difference between (185b-c) and (187a) is that the latter does not involve quantification over time, since the frequency adverb nooitnever is replaced by the negative marker nietnot. Note also that the appropriate translations of the examples in (187) are material implications and involve the logical connective if ..., then ..., not the temporal connective when ..., then ...; cf. the discussion in Subsection IIIA, below (157).

The verb must denote an action: if a supplementive is added to a clause with a stative main verb, such as kennento know, the use of a supplementive adjective yields a severely degraded result.

Furthermore, the use of a supplementive often leads to an unacceptable result when the verb is intransitive, as in (189a). In contrast, unaccusative verbs yield fully acceptable results, as shown in (189b).

Since the addition of a predicative locational PP can turn intransitive motion verbs such as lopen into unaccusative verbs, we expect the addition of such a PP to (189a) to license the presence of a supplementive. That this expectation is borne out is shown in (190a). Example (190b) shows that, for some as yet unclear reason, the predicative PP naar de directeur cannot be topicalized when a supplementive is present (which is unproblematic when the supplementive is absent).

The addition of a locational adverbial phrase like over straat also seems to improve example (189a), but it may well be that in this case the marginal status of (191a) is the result of interpreting the adjectives razend/vrolijk as manner adverbs. Example (191b) shows that this is especially possible with the adjective vrolijkmerry; this sentence does not express that the subject of the clause is merry, but that the laughing/chattering makes a merry impression. See Section 30.2.3, sub I, for further discussion.

Finally, the primeless examples in (192) show that supplementives can easily be used with transitive verbs and verbs that take a PP-complement. However, if the direct or the prepositional object is omitted the result deteriorates, unless the sentence contains a progressive auxiliary like zittento sit.