Dutch negation is located relatively low in the functional domain of the clause. This is clear from the fact, illustrated in (86), that the negative adverb niet must follow the modal adverbials, which are usually taken to demarcate the boundary between the functional and the lexical domain of the clause.

Sentence negation is nevertheless external to the lexical domain. This intuition is formally expressed in standard predicate calculus by giving negation scope over the proposition come(Jan), which corresponds to the lexical domain. That sentence negation is external to the lexical domain is also clear from the fact that negation behaves like modal adverbials in that it passes the adverbial test for clause adverbials: example (87) shows that example (86) can be paraphrased by placing the negative adverb nietnot in a matrix clause of the form het is adverb zo dat ...

That negation is located low in the functional domain of the clause is further supported by the observation that the negative adverb niet follows all adverbials except the VP adverbials, i.e. adverbials that are part of the lexical domain as modifiers of the predicate expressed by VP in (80); cf. Section 8.2.

Since prepositional objects usually follow VP adverbials, it is not surprising that they also follow the negative adverb niet in examples such as (89a). The same goes for complementives such as the directional PP naar het feest in (89b), which are usually left-adjacent to the clause-final verbs.

Note, however, that nominal arguments cannot follow the negative adverb nietnot when it expresses sentence negation; cf. Kraak (1966). This is surprising, since such arguments can normally follow manner adverbials; cf. the contrast between the two examples in (90).

Apparently there is a surface filter that prohibits a definite nominal argument (DP) in the domain of sentence negation; cf. filter (91a). The rationale for this filter is not immediately obvious, but we believe it to be pragmatic in nature: nominal arguments within the lexical domain of the clause normally express discourse-new information, and it does not seem expedient or informative to negate a proposition with discourse-new information, since this does not result in an update of the background (shared information) of the discourse; cf. Grice’s cooperative principle. This pragmatic account may be supported by the fact that definite noun phrases can follow negative phrases, e.g. Ik heb nooit het boek gelezenI have never read the book; cf. (91b).

A complication for the pragmatic account is that the nominal argument can occur between a modal adverb and the negative adverb niet, as illustrated for the direct object het boek in (92a). At first sight, this seems to contradict the above claim that the direct object in negative clauses with niet must express discourse-given information, but Section 13.3.2 will show that the position between the modal adverb and the negation is a designated position for contrastive foci, which are discourse-given in the sense that their referents are part of a contextually given set. Example (92b) is added for completeness to show that nominal arguments can also undergo nominal argument shift when sentence negation is expressed by the adverb nietnot.

For completeness’ sake, note that the examples in (93) are acceptable, but in these cases we are arguably dealing with constituent negation as the negation takes scope over the noun phrase het boek and the cardinal numeral only; such example are therefore not relevant for our present discussion.

Leaving aside the possible problems surrounding surface filter (91), it seems safe to conclude that sentence negation is external to the vP, but lower than the modal adverbials. Haegeman (1995) argues convincingly for West Flemish that the negative adverb nienot is in the specifier of NegP and not in the head of NegP; the head of NegP is the optional negative preverbal clitic en in the bipartite negation construction in (94a). Example (94b), quoted from Van der Horst (2008:516), shows that a similar clitic was used (almost obligatorily) in Middle Dutch.

Since standard Dutch lost the clitic en around 1600 AD, we will not digress on this issue any further, but simply assume that Dutch nietnot resembles nie(t) in West Flemish and Middle Dutch in that it occupies the specifier position of the functional projection NegP. That Dutch niet is not a head can further be supported by the fact that it can be modified by an amplifying degree modifier: cf. beslist/zeker nietabsolutely/certainly not.

Subsection I has argued that the negative adverb nietnot is located in the specifier position of NegP when it is used to express sentence negation. If this is correct, the overall structure of negative clauses is as given in (95), where we omitted the higher functional projections CP and TP, and where ADVModal stands for clause adverbials such as waarschijnlijkprobably.

This subsection will take the structure in (95) as a starting point and argue that negative noun phrases like nietsnothing, niemandnobody, and geen Nno N also move obligatorily into the specifier of NegP, in order to allow negation to take scope over the proposition expressed by the lexical domain of the clause.

The previous subsections have shown that the specifier of NegP must be filled either by the negative adverb niet or by a negative phrase. We can therefore expect that certain negative clauses can be expressed in two different ways. That this is possible in principle can be seen from the fact that the Dutch example in (104a) can be translated into English by using the negative adverb not followed by the negative polarity item (NPI) anything. However, example (104b) shows that the English strategy is not available in Dutch.

The preference for Neg-movement over the use of niet + NPI is a rather persistent feature of Dutch: the (a) and (b)-examples in (105) illustrate this for a prepositional object and a PP-complement of a complementive adjective, respectively.

The Dutch NPI ook maar iets/iemandanything/anyone can only be used if the clause contains another negative phrase, as in (106a), or if the negation is located in a higher clause, as in (106b). In such examples, however, NPI’s are not involved in expressing sentence negation, but are simply licensed by the negation as emphatic forms of the existential pronouns iets/iemandsomething/someone.

This section has shown that Dutch strongly prefers Neg-movement to the semantically equivalent construction with a negative adverb followed by an NPI. In this respect, Dutch differs markedly from English, where the movement strategy is not found. This is, of course, related to the fact that English has a more rigid word order than Dutch; cf. Broekhuis & Klooster (2010) for an optimality-theoretic account of this difference.

The previous subsections have argued that Dutch has a functional projection NegP external to the lexical domain of the clause, which follows modal adverbials such as waarschijnlijkprobably. If this is correct, the overall structure of negative clauses is as given in (95), with ADVModal indicating the position of the modal adverbials. Since Neg-movement is not restricted to noun phrases, but can also apply to e.g. PPs, it cannot be an instance of A-movement, but should involve A'-movement, which is further supported by the fact that Neg-movement obviously has a semantic motivation: it is needed to assign clausal scope to the negation.

The claim that Neg-movement is A'-movement makes an interesting prediction about the position of nominal arguments in light of Chomsky’s (1986) improper-movement restriction. This restriction prohibits A-movement of a phrase XP after it has been A'-moved: Neg-movement of a nominal argument should therefore block nominal argument shift across modal adverbials such as waarschijnlijkprobably, which Section 13.2 has shown to be an instance of A-movement. The examples in (108), repeated from Section N21.1.4, sub IC, show that this expectation is indeed borne out. The negative pronoun niemand cannot undergo nominal argument shift from the specifier position of NegP, which follows the modal adverb, into the specifier positions of the lexical heads assigning accusative and nominative case, which precede the modal adverb; cf. the representations in example (40) from Section 13.2, sub IA.

The judgments on these examples seem to be confirmed by a Google search (July 22, 2024). We checked our judgments on (108a) using the search string [dat pron niemand waarschijnlijk], where pron stands for the set of subject pronouns; this string does not occur at all. We checked our judgments on (108b) using the search strings [dat waarschijnlijk niemand] and [dat niemand waarschijnlijk]; the first string yielded 255 hits and is thus much more frequent then the second string, which yielded no more than 43 hits, some of which struck us as rather marked. Note that the acceptability contrasts indicated in (108) cannot be explained by claiming that negative noun phrases must occur in the specifier position of NegP, because Subsection IIB has shown that they can easily be moved into the clause-initial position; the relevant examples are repeated in (109).

If the unacceptable orders in (108) are indeed ruled out by the improper movement restriction, the acceptability of the examples in (109) is not surprising because topicalization is clearly a case of A'-movement. The contrast between nominal argument shift and topicalization of negative noun phrases thus supports the claim that Neg-movement is A'-movement (but see Section N21.1.4, sub IC, for an alternative account of the impossibility of nominal argument shift in (108)).

Subsection IV has shown that negative subjects cannot be moved into the regular subject position by nominal argument shift. The fact that negative phrases follow the modal adverbials when they are part of the middle field, while the target positions of nominal argument shift precede the modal adverbials, raises the question whether negative subjects can be crossed by other nominal arguments. The answer is in the negative, as can be seen from the fact, illustrated in (110a), that the direct object het boek cannot shift across the negative subject niemand (under a non-contrastive intonation pattern of the clause), and that the acceptable order is actually ambiguous in that the direct object can be interpreted either as part of the new-information focus or as part of the presupposition of the clause. For completeness’ sake, we have added example (110b) to show that the negative subject cannot easily be pushed up into the canonical subject position by the direct object.

The examples in (111) show that the same pattern occurs for negative subjects and pronominal indirect objects; as indirect objects and subjects are often both [+human], we will use a strong object pronoun to avoid processing problems.

The examples above have shown that negative subjects normally block object shift (although it may be marginally possible for the object to push the negative subject up into the regular subject position in violation of the improper movement restriction). This shows that word-order preservation constraint in (112) also applies in negative clauses.

Recall that the judgments in (110a) and (111a) hold only when the sentences are pronounced with a neutral intonation pattern. The order marked with an asterisk becomes acceptable when the objects are given a contrastive accent: dat het boek waarschijnlijk niemand gekocht heeft and dat hem waarschijnlijk niemand hulp aangeboden heeft are both perfectly acceptable. However, argument inversion in such examples is not the result of A-movement (nominal argument shift) but of A'-movement, i.e. contrastive topic or focus movement discussed in Section 13.3.2. A contrastive accent on niemand in (111b) can also improve the result.