When negation is present in a predicative construction, it is often not clear a priori whether it modifies the complementive or the clause. Consider the near-synonymous sentences in the primeless and primed examples of (287).

Since the copular verb zijnto be does not express a meaning that can be negated (its presence is mainly motivated by the need to express the tense and agreement features of the clause), semantic considerations do not seem at first glance to help determine whether niet modifies the whole clause or just the AP. This subsection will show that there are reasons to assume that niet modifies the whole clause.

A syntactic reason is that the constituency test shows that the negative adverb and the clause are separate constituents: the (a) and (b)-examples of (288) show that while topicalization of the adjective alone is fully acceptable, pied piping of the negative adverb leads to unacceptability. The (c)-examples show that the negation behaves in this respect like the clause adverbial zekercertainly in examples like Dit boek is zeker leukThis book is certainly amusing. We conclude that the negative adverb expresses sentence negation.

A semantic reason for accepting the conclusion that the negative adverb niet acts as a sentence negation is that this would also account for the fact that the two (a)-examples in (287) are not equivalent: example (289a) is not a contradiction. The felicity of this example is due to the fact that Jan is niet aardig applies to a larger range of the implied scale of kindness than Jan is onaardig: it also includes the neutral zone. The result is that example (289a) locates Jan’s degree of kindness in the neutral zone, i.e. in the overlapping part of niet aardig and niet onaardig. This is illustrated in the graphical semantic representation in (289a).

The semantic difference between the two (a)-examples in (287) can also be expressed in the logical formulas in (290): in the former, the negation expressed by niet has sentential scope, while the scope of the negation expressed by the prefix -on is restricted to the adjective.

Note, however, that the inclusion of the neutral zone is lost if the negative element niet is modified by an absolute modifier like absoluutabsolutely or helemaaltotally. Example (291a) expresses that Jan is quite unfriendly, and example (291b) that Jan is quite friendly.

Negation can also be used when an amplifier such as ergvery is present, as in (292a). In (292b) we show the range of the scale implied by niet erg aardig in a graphical representation. Example (292c) provides the semantic representation of niet erg aardig. If the amplifier expresses an extremely high degree, such as afgrijselijkterribly, the result is less felicitous: the amplifiers given in (21) above then yield marked results.

Despite the fact that Jan is niet erg aardig has the meaning in (292c), the intended range of the scale can be further restricted by using accent. If the amplifier is accented, as in (293a), the most salient interpretation is that Jan is kind, but only to a lesser extent; in other words, the degree to which Jan is kind is located somewhere between the neutral zone and the point where the range denoted by erg aardig begins. If the adjective has an accent, as in (293b), the most salient interpretation is that Jan is unkind, i.e. we are dealing with a form of litotes; cf. Subsection IIB.

This difference in interpretation between (293a) and (293b) is partly semantic in nature, and can be technically explained by assuming that they differ in the scope of negation. In (293a) we are dealing with constituent negation; the scope of negation is limited to the degree modifier ergvery. This means that only the clause d > dn is negated, so that the sentence is assigned the semantic representation ∃d [aardig (Jan, d) & ¬(d > dn)], which is equivalent to ∃d [aardig (Jan, d) & (d ≤ dn)]; this correctly picks out the range between the neutral zone and the range denoted by erg aardig. When we are dealing with sentence negation in (293b), the sentence is assigned the interpretation in (292c). That the most salient interpretation of (293b) is that Jan is unfriendly does not follow from the scope assignment to the negation, but could be explained by appealing to Grice’s (1975) maxim of manner: if the speaker wants to express that Jan is friendly, but not very friendly, he can do so more precisely by using (293a), so that (293b) can be seen as a pragmatically infelicitous means of referring to this range of the scale.

The presence of a downtoner in the scope of sentence negation usually leads to an unacceptable result. One possible explanation for this is that the intended range on the implied scale can be indicated more economically by niet aardignot kind; cf. (294b). However, it seems unlikely that (294b) is the correct graphical representation of the meaning of (294a); the meaning we would expect is given in (294d), which corresponds to the graphical representation in (294c). If (294c) is indeed the correct representation, then sentence (294a) can be excluded by appealing to Grice’s (1975) maxim of quantity, because it produces an uninformative message in the sense that niet vrij aardig refers to two opposite sides of the scale.

This account of the infelicity of (294a) is consistent with the fact that (294a) becomes more or less acceptable when it is used to deny a presupposition or statement made earlier in the discourse, as in (295), in which case the speaker can propose an alternative on either the left or the right side of the scale.

Example (295b) sounds a bit marked, but becomes fully acceptable when an accent is assigned to the degree modifier vrij. In this case, niet expresses constituent negation; the first conjunct of example (296a) is then assigned the semantic representation in (296b), which is semantically equivalent to the representation in (296b'); this option may be preferred to (295b) for pragmatic reasons because it unambiguously places Jan’s kindness at the right end of the scale, as shown in the graphical representation in (295c).

The use of constituent negation in (296) is similar to the use of the (stressed) marker wel, which can thus be seen as the positive counterpart of niet. As shown in (297b), the presence of wel does not affect the part of the scale to which the adjectives onaardigunkind and aardigkind refer. Its main function is to contradict a presupposition or statement made earlier in the discourse; for example, (297a) is only acceptable if the presupposition is that Jan is not kind.

Finally, let us look briefly at the absolute adjectives. In these cases, the negative adverb niet simply indicates that the property denoted by the adjective does not apply. Like approximative and absolute modifiers, the negative adverb itself can be modified. However, while the examples in (279) and (284) have shown that the first two can be modified by both alalready and nogstill, the examples in (298) show that the negation can only be modified by nog. Example (298a) can only be used felicitously if we are emptying bottles, and (298b) if we are filling them. Note in passing that al can be used here as a clause adverbial, meaning “it is already the case that the bottle is not empty”; this is irrelevant in the present context.

If the adjective is modified by an absolute modifier such as helemaalcompletely, the combination of negation and modifier is more or less equivalent to an approximative: example (299a) is more or less synonymous with De tafel is vrijwel rondThe table is almost round. Approximative modifiers lead to a strange result in the presence of negation, as shown in (299b).

Example (299b) is marked with a percentage sign because it is acceptable when used to deny a presupposition or statement made earlier in the discourse; cf. (300). This use of negation is similar to its use in the examples in (295) and (296) with a scalar adjective modified by a downtoner.

Subsection I has shown that the scope of the negative adverb niet and the affirmative marker wel can be limited to the degree modifier of an adjective, in which case they are used to deny a presupposition or statement made earlier in the discourse. This subsection discusses other uses of niet and wel with restricted scope.

Negation can also be expressed by quasi-negative phrases, in which the negation is in some sense hidden in the meaning of the phrase: weiniglittle in (313a), for example, can be paraphrased with an overt negative as niet veelnot much. Other quasi-negative modifiers are given in (313b&c); in all these cases, the use of the modifier suggests that the property denoted by the adjective does not apply. The dollar sign assigned to (313c), in which niets alternates with the more colloquial form niks, indicates that such examples are old-fashioned and not accepted by all speakers; some other examples are Zij/Dat is niets aardigShe is not at all kind /That is not at all nice, Zij is niets lui; ze werkt hardShe is not lazy, but works hard and Zij is niets bang; ze durft allesShe is not afraid, but dares everything.

The modifier weiniglittle is also compatible with a downtoning interpretation. Example (314a) shows that in this interpretation weinig can also be negated, and that the resulting meaning is more or less equivalent to that of the amplifier zeervery. The modifiers in (313b&c) do not allow a downtoning interpretation, and the examples in (314b&c) show that negation of these modifiers is excluded.

The modifier weinig can only be used with scalar adjectives that have an absolute antonym; example (315a) is unacceptable, since the antonym of aardig is also gradable; cf. erg onaardigvery unkind. The modifiers in (313b&c) can be used in these contexts.

The examples in (316) show that the modifier weinig also differs from the modifiers in (313b) in that it cannot be combined with an absolute adjective; the modifier niets cannot be used in this context either.

We have little to say about the distribution pattern described above because, to our knowledge, it has not been considered in the literature.

A special case are the negative polarity elements al te and bijster. In colloquial speech, these elements usually occur in the scope of negation; cf. Klein (1997). The construction as a whole is downtoning in nature; some examples are given in (317).

The combination al te also occurs without negation, often in more or less fixed expressions like Dit gaat me al te ver\`1Dit is me al te veelThats too much for me’, Hij maakt het al te gortigHe goes too far, and the proverb Al te goed is buurmans gek (lit.: all too good is neighbor’s fool, which may be close to the American saying If you make yourself an ass, don't complain if people ride you). These cases can be considered relics of the older use of al te as a regular amplifier; in more formal registers, al te can still be used without negation. In older stages of Dutch, bijster could also be used as an amplifier in positive contexts, but this use seems to have died out in colloquial speech.

The negative adverb niet and the adjective do not form a constituent, which is clear from the fact that topicalization of the AP strongly favors stranding of the negative adverb niet; cf. (318).

That niet and the adjective do not form a constituent can also be seen from the fact that the negation can be external to the clause containing bijster/al te, as in the (a)-examples of (319), or expressed on another constituent in the clause, such as nooitnever in the (b)-examples.

Furthermore, when the AP is used in attributive position, the negation can be located external to the noun phrase, as in the (a)-examples in (320), expressed by the article, as in the (b)-examples, or placed within the noun phrase, as in the (c)-examples. The noun phrase containing the modifier is enclosed in square brackets.

Subsection I has shown that there are also modifiers that cannot occur in the scope of negation. This is especially true for downtoners, as shown in the (a)-examples of (321). However, Subsection I has also shown that the use of negation becomes fully acceptable when the downtoner is used contrastively, as in the (b)-examples, in which case we are dealing with constituent negation. The (c)-examples show that downtoners can also occur in the scope of negation in yes/no questions.

The acceptability of the (b) and (c)-examples suggests that the impossibility of a downtoner in the scope of negation in the declarative (a)-examples is not due to some inherent semantic property of downtoners, but has a pragmatic reason; this was already suggested in our discussion of (294) and (295) in Subsection I.