The examples in (36) show that the cardinal numerals, with the exception of éénone, are usually followed by a plural count noun.

The generalization that cardinals are usually followed by a plural noun also applies to the cardinal nulzero; cf. Haeseryn (1997:432). The examples in (37) show that it differs in this respect from the negative article/quantifier geenno, which can be followed by either a plural or a singular noun.

Section 19.1.5.1, sub IA, argues that the fact that geenno can be followed by a plural or a singular noun is related to the fact that the negation expressed by geen takes sentential scope: the primed examples in (38) can be seen as the denial of the propositions expressed by the primeless examples. Note that the plural is used when there is no presupposition about the actual cardinality of the set of children; cf. Section 19.1.1.1 for examples.

The fact that the cardinal nulzero cannot normally be followed by a singular count noun can perhaps be accounted for if we assume that it differs from geenno in that it expresses constituent (and not sentential) negation. If so, we can assign nul kinderen the meaning not (one or more children) and thus expect it to impose the same selection restriction as the complex cardinal één of meer, which cannot be followed by a singular noun either; cf. (39). We leave it to future research to investigate whether this line of reasoning is feasible.

Although the generalization that cardinals (except één) are followed by a plural noun is descriptively correct in general, we will discuss a number of exceptions in the remainder of this subsection.

There do not seem to be many special restrictions on the determiner preceding a cardinal numeral: (56a-c) show that definite articles, demonstratives, and possessive pronouns all yield an equally acceptable result. When the noun phrase is indefinite, as in (56d), we can assume that the empty indefinite article ∅ is present, as in the case of other plural indefinite noun phrases.

The indefinite article eena cannot be combined with a cardinal, which is of course related to the fact that noun phrases containing this article are inherently singular: adding the cardinal éénone would therefore be redundant and adding any other cardinal would lead to a contradiction. Example (62) in Subsection III will show that the complex demonstrative zo’nsuch a, which is the result of the contraction of zo and een, cannot be followed by a cardinal either.

Example (58a) shows that the cardinal éénone is special in that it can never be preceded by the definite article de/het or a demonstrative or possessive pronoun (unless it is used as a noun, as in de één is gelukkig\`1 de ander nietsome people are happy, some are not). However, it is generally accepted that the cardinal één can follow a demonstrative pronoun when it is inflected; example (58b), taken from Haeseryn et al. (1997), clearly demonstrates this. Note, however, that inflected ene also occurs in noun phrases with a more deictic meaning, in which case it is in opposition to the adjective andereother, as in (58c).

It seems premature to decide on the basis of meaning alone that ene is a cardinal numeral in (58b); we should first make a comparison of the two uses of ene in (58b) and (58c). We will leave this to future research.

The introduction to this chapter argued that cardinal numerals (and quantifiers) can be assumed to be generated in the specifier position of Num, indicated by an em dash in structure (59a). This predicts that cardinals generally follow the determiner and precede those elements that are part of the lexical NP-domain, i.e. the head noun itself and its attributive modifiers, as in (59b). This seems to be confirmed; example (59c) shows that the cardinal cannot precede the definite article and example (59d) that it cannot follow the attributive modifier(s) of the head noun.

However, more needs to be said about the order in (59d), since this order is acceptable when the attributive adjective is assigned contrastive accent, as in (60a); cf. Kobele and Zimmermann (2012:239) for a similar observation in German. This is an option when the active discourse domain (domain D) contains various bouquets, each consisting of eight roses: (60a) can then be used to distinguish the yellow bouquet from the bouquets containing roses of a different color. This order of the cardinal and the attributive modifier requires that the roses be part of domain D, which correctly predicts that this order is not found in indefinite noun phrases such as (60a'). In (60b&b') we give some comparable examples from the linguistic literature: again, the marked order requires that the referents of the noun phrase be part of domain D, so that the indefinite noun phrase in (60b') yields an infelicitous result.

The unacceptable example in (59c) cannot be saved by assigning it a special intonation pattern: cardinals never precede the definite article. The same applies to the D-linked demonstrative and possessive pronouns in (61). Note that the primed examples become acceptable when the cardinal is preceded by alle as in alle vier zijn dochtersall four of his daughters; cf. Section 21.1.2.2.

Haeseryn et al. (1997) claims that the non-D-linked demonstrative pronouns zo’n and zulke behave differently. The primeless examples in (62) first show that these demonstratives cannot be followed by a cardinal: for (62a) this is of course not surprising, since we have seen in (57) above that the article eena, which here acts as part of the complex determiner zo’nsuch a, cannot be combined with a cardinal either; for (62b), on the other hand, no such explanation seems to be available. The data becomes even more puzzling when we consider the primed examples, which are certainly better than the primeless ones; similar cases can be easily found on the internet.

Although some Dutch speakers reject examples such as (62b'), the fact that the primed examples are considered perfectly acceptable by some speakers poses a problem for the assumption that the non-D-linked demonstratives are located in the D-position of (59a), and suggests that they are actually lower in the nominal projection than the cardinal numerals.