Most of the cardinal numerals corresponding to the multiples of 100 between 1,000 and 10,000 can be realized in more than one way. One way is to use a compound that takes -honderd as its second member: in colloquial speech, this is the only (or at least highly preferred) option for the multiples of 100 between 1100 and 2000.

Alternative forms are regularly used for multiples of 100 between 2,000 and 10,000, as can be seen by the fact that the more common forms in the primeless examples of (20) alternate with the forms in the primed examples. Note that the primed examples express addition even though the additive element en is usually absent.

This alternation is not normally found in colloquial speech for the numbers 2100, 3100, etc. These numbers are usually expressed by the regular compound with -honderd; using the forms in the primed examples of (21) seems possible, provided the cardinal één is included, although it still feels emphatic and artificial.

Compounds with -honderd are never used for multiples of 1000; cf. (22).

As mentioned earlier, there may be reasons to assume that the putative compound forms discussed above are actually phrasal in nature; this is illustrated by the examples in (23). The (a)-examples first show that consecutive numbers in a coordinate structure with of can be used to express “approximation” and that such examples usually have an asyndetic counterpart, as in (23a'). The fact that we find essentially the same in the (b)-examples suggests that the internal structure of the approximate numeral is given as indicated by the brackets. The fact that coordinate structures are not normally used as the first member of a compound thus suggests a phrasal analysis for the numeral; cf. Corver and Tatsumi (to appear:§3) for a slightly different argument based on the same form that also leads to this conclusion.

Occasionally, the first part of a compound-like form is a cardinal greater than 1000, as in the primeless examples of (24). In such cases, speakers tend to use yet another way of expressing the number, as in the primed examples. Both options are limited to fairly round figures; cf. the examples in (28) for other cases.

The use of kommacomma in the primed examples of (24) is related to the fact that in Dutch the comma is used to separate a whole number from a decimal part, instead of the point (as in English). Thus, één komma twee miljoen corresponds to one point two million in English.

The second and third columns of Table (17) have shown that the orthographic conventions are somewhat complex: the compound-like forms derived from honderdhundred and duizendthousand are treated as single words, while the complex forms based on miljoen, miljard and biljoen are written as separate words. This has to do with the fact that the latter forms seem to have noun-like properties that are missing (or at least less prominent) in the case of honderd and duizend. Two of these properties are the ability to be preceded by the indefinite article eena and the ability to take a modifier like halfhalf or kwartquarter; this is shown in the primeless and singly-primed examples in (25). Note, however, that the doubly-primed examples show that all forms allow pluralization, which is also a typical nominal property: the derived forms express a multiplication of the cardinal: n x 100, n x 1000, etc. (cf. Jackendoff 1977:§5.5; Corver & Zwarts 2005). Note in passing that the plural forms have an additional interpretive effect of high-degree quantification, in that Jan heeft honderden boekenJan has hundreds of books implies that honderden boeken is a lot; cf. Postma (to appear).

Note that although we have marked the forms in (25a'&b') with an asterisk, a Google search will reveal their presence on the internet, especially with nouns indicating a measure unit like euro or kilometer. The examples in (25a&b), on the other hand, are unacceptable on the intended reading, but a number sign is used because they are possible with an approximative reading “about a hundred/thousand books”; this reading is not relevant here, since we are dealing here with one of the spurious uses of the indefinite article (cf. Section 19.1.4.4, sub II). Note also that the examples (25b&b') with duizend are in stark contrast to the (b)-examples in (26); this contrast does not appear in the case of honderd, although cases with anderhalf honderd are again not difficult to find on the internet. For completeness, note that (26b') offers yet another possibility in addition to the two (a)-examples in (20).

The acceptability of the doubly-primed examples in (25) might lead to the wrong expectation that juxtaposition with a cardinal greater than one would lead to plural marking on cardinals like miljoen: twee miljoen/*miljoenen boekentwo million books. Booij (2010:199-200) suggests that we are dealing instead with measure nouns of the type meter or kilo; cf. the judgments on example (27b) in its purely quantificational reading.

The above examples have shown that there is clear morphological evidence for the assumption that the cardinals honderd, duizend, miljoen, etc. are nouns: they can all be pluralized. This complements the syntactic evidence in Section 20.1.1.1.

The remaining cardinals are phrase-like in nature. The different subparts of the cardinals are generally optionally linked by the additive conjunction-like element en. Some random examples are given in (28). Note that, unlike honderd and duizend, the element miljoen is obligatorily preceded by the cardinal één.

The examples in (29) show that the additive element en can also be used to link cardinals expressing a round figure with more indefinite numeral-like elements like zoveel (lit.: so many) or nog wat (lit.: yet some); cf. Corver (2012). These formations tend to be used with a proximate reading, which may explain why they can easily co-occur with the spurious indefinite article een; cf. the discussion below (25).

Corver also notes that such indefinites cannot replace any of the constituting parts of the cardinals in the second and third columns of Table (15); cf. the two (b)-examples in (30) for examples illustrating this for nog wat.

This supports Booij’s (2010:204) claim that the compound-like cardinals between 20 and 99 in Table (15) have a different morphosyntactic status than the forms in (16): forms such as negenenzestigsixty-nine are listed as lexical units (i.e. as part of a closed morphological class), while forms such as honderd (en) zesone hundred and six are derived by a productive morphosyntactic process.