Bare al does not occur independently as a subject; the examples in (89) show that it does not occur with either count or non-count nouns.

The (a)-examples with alle + Num/allebei in (90) and (91) seem acceptable, though not particularly good. Example (90b) shows that (90a) improves significantly when a pronoun is added to the left of the quantifier; cf. also Section 21.1.2.2, sub IB. Note that example (91b) is worse than (91a) due to the fact that strong pronouns cannot easily be used to refer to [-human] entities; cf. Section 19.2.1.1, sub V. The (c)-examples, in which a plural pronoun is inserted to the right of the finite auxiliary, are fully acceptable; cf. Subsection III on floating quantifiers for further discussion.

The examples in (92) show that alle(n) and beide(n) with [+human] and [-human] antecedents differ not only in orthography but also in syntactic distribution; while (92a) is formal but acceptable, (92b) is awkward. The contrast between (92b) and (92c) also shows that within the class of non-human referents a distinction should be made between plural count nouns and substance nouns; if alle takes a substance noun such as wijnwine as its referent, the result is completely unacceptable.

The split between allen/beiden and alle/beide in (92) is similar to the one found between the same elements used as quantifiers of pronominal noun phrases; cf. Section 21.1.2.2, sub IIB4. Note that (92b) improves considerably when the plural pronoun zethey is placed to the right of the finite auxiliary; cf. Subsection III on floating quantifiers for discussion.

Bare al does not occur independently as a direct or indirect object, as can be seen in the primeless and primed examples of (93), respectively.

Comparable examples with alle + Num and allebei seem marked but acceptable; moreover, for some speakers there seems to be a slight contrast between the direct and indirect object examples, with the former being slightly worse than the latter; cf. the contrast between the (a) and (b)-examples in (94) and (95). Again, such examples become fully acceptable when a pronoun or noun phrase is added to the left of the quantifier; cf. Subsection III on floating quantifiers for discussion.

The [±human] contrast that we already signaled in connection with the data in (92) reappears in the alle(n) and beide(n) cases in (96) and (97): the [-human] examples in (97) are systematically worse than the [+human] ones in (96). As in the subject examples, the use of substance nouns in (97b&b') is completely impossible. These cannot be saved by adding a pronoun, unlike the count noun examples in the (a)-examples in (97), which become fully acceptable by adding a plural pronoun to the left of the quantifier; cf. Subsection III on floating quantifiers.

It is impossible for bare al to be used in the examples in (98), the prepositional counterparts of the double object constructions in the primed examples in (93), where al is the object of the preposition aan.

Bare al does not normally occur as the object of other prepositions either. An idiomatic example that basically exhausts the possibilities is het ergst van althe worst of all. This example has a more or less productive paradigm in the sense that the superlative adjective can in principle be replaced by any other superlative (cf. English worst/best/most/... of all). We can also mention the fixed expressions al met alall in all, bovenalabove all and vooralparticularly/especially.

For alle + Num and allebei, acceptable variants of the examples in (98) can be constructed, although some speakers consider the examples in (99) less felicitous than their double object counterparts in (94b) and (95b).

The examples in (100a&b) show that alle(n) and beide(n) can occur as the complement of aan when they have a [+human] referent, but that the result is marginal when they have a plural [-human] referent; example (100c) shows that, as before, it is impossible for independent alle to be associated with a substance noun. There is no discernible contrast between the examples in (100) and their double object counterparts in (96b) and the primed examples in (97). The judgments are more or less the same when alle(n) and beide(n) function as the complement of some other preposition.

While the result of independent forms in subject, direct object, and indirect object positions can be systematically improved by inserting a plural pronoun to the left of the quantifiers, a more complex picture emerges when they function as the complement of a PP. The (a)-examples in (101) show that the [+human] examples in (99a) and (100a) become perfectly acceptable with the insertion of the strong pronoun hen, but not with the weak pronoun ze. The (b)-examples show that inserting a plural pronoun does not improve the [-human] cases in (99b) and (100b); this is to be expected if weak pronouns cannot be used when the quantifier is the complement of a preposition, as strong pronouns cannot be used to refer to [-human] entities; cf. Section 19.2.1.1, sub V, as well as Sections 21.1.2.2, sub IB, and 21.1.2.2, sub IIB).

The contrast in acceptability between the strong and weak forms of the pronoun in the (a)-examples suggests that the pronoun and the modifier form a constituent that functions as the complement of a PP; cf. Section 21.1.2.2, sub IIB4. First, scrambling or topicalization of the PP must pied-pipe the quantifier, which shows that the quantifier is PP-internal, as illustrated in (102) for (101a) with allen.

Second, the examples in (103) show that the split pronominal PP waar ... aan is unacceptable when the quantifier is present; this can be derived from the assumption that the pronoun and the modifier form a constituent.

The reason is that R-pronominalization is excluded when the pronoun is part of a larger phrase. This is illustrated for the existentially quantified [-human] pronoun iets by the examples in (104). The (a)-examples show that R-pronominalization is preferred when iets is the complement of a preposition, while the (b)-examples show that R-pronominalization is impossible when the pronoun iets is part of the so-called partitive genitive construction iets hardssomething hard.

The unacceptability of the examples in (103) with alle(n) can therefore also be accounted for by appealing to the constituency of the pronoun and the quantifier alle(n); cf. *... [CP waari ... [PP aan [ti alle(n)]] ...]. However, this analysis leaves unexplained that the pied-piping counterparts of (103) in (105) are not possible either: this fact is surprising in view of the acceptability of the topicalization case in (102b).

At the end of this overview of the use of al, alle + Num and simplex alle as arguments, we would like to draw attention to a form of al which has not figured in the discussion so far, because it does not occur as a modifier of the noun phrase: the quantifier alleseverything, which can only be used independently. As shown in (106), alles, like its English counterpart everything, can occur in all regular argument positions (despite the fact that alles is historically a genitival, neuter form of the quantifier al, whence the -es ending).

The phrase van allesall kinds of things can also be used as a nominal argument, and it seems plausible to assume that this phrase is related to the pseudo-partitive construction van die Nsuch N, which can likewise be used as a nominal argument; cf. Section 20.4, sub I. The two constructions are at least similar in that they are both indefinite, which is clear from the fact, illustrated in (107a&a'), that they can occur as the subject in an expletive er construction. Note that the phrase van alles, like alles in (106a), triggers singular agreement on the verb. Finally, the examples in (107b-d) show that the phrase van alles occurs in all regular argument positions, just like the pseudo-partitive construction.

In present-day Dutch, it is almost impossible to use bare al as a floating quantifier (although it is used that way in at least some Flemish dialects). In idioms one can find relics of this pattern: in (111) al is a floating quantifier associated with the third singular neuter pronoun in the surface subject position. The pattern cannot be productively extended beyond these idiomatic expressions.

Note that in the idiom in (111a), al alternates with the quantifier alles. However, this is the only context in which alles is used as a floating quantifier in contemporary Dutch. As shown in (112), alles can be combined with the singular demonstrative pronouns dit/dat to form the noun phrase dit/dat alles (cf. example (81)), but it cannot be used as a floating quantifier with such pronominal associates.

Alle + Num and allebei are felicitous floating quantifiers with both animate and inanimate noun phrase associates. As shown in (113) and (114), the associate of the floating quantifier can be either a complex noun phrase or a pronoun. However, the primed examples show that the two types of associate differ in that the noun phrase must precede the floating quantifier, while the pronoun can also follow it if the floating quantifier is moved into the clause-initial position. Note in passing that while the predeterminer alle + Num can be written either as a single word or as two separate words (cf. Section 21.1.2.2), there seems to be a tendency to write the two elements as a single word when it is used as a floating quantifier.

The uses of Alle + Num and allebei as predeterminers and floating quantifiers are easy to distinguish. As predeterminers they must be left-adjacent to the determiner, as shown in (115) and (116), whereas as floating quantifiers they usually follow their nominal associate, as in the primeless examples above.

The examples in (115b) and (116b) cannot be interpreted as containing floating quantifiers: the acceptable orders involve modification of the noun phrase, which is also clear from the fact that the quantifier must be left-adjacent to the determiner, i.e. cannot be separated from it by an adverbial phrase. This shows that the floating quantifier cannot be scrambled across certain adverbs, like modal and time adverbs. This does not mean, however, that floating quantifiers can never precede adverbs, since they certainly can (or indeed must) precede °VP adverbials like vriendelijkkindly and zorgvuldigmeticulously in (117).

The primeless examples in (118) show that the floating quantifiers alle + Num and allebei can take as their associates independently quantified noun phrases, both animate and inanimate. Judgments about the primed examples seem to vary from speaker to speaker; the fact that many speakers consider these examples to be marked is probably due to the fact that vele denotes an indeterminate quantity, which conflicts with the precision expressed by the cardinal numeral dertigthirty.

Note that the floating quantifiers differ in this respect from the predeterminers alle + Num and allebei, since the examples in (119) show that it is impossible for the latter to form a constituent with independently quantified noun phrases; cf. Section 21.1.2.2, sub IB.

The contrast between (118) and (119) thus suggests that floating quantifier constructions are not derived from an underlying structure in which the floating quantifier and its associate form as a single constituent. Of course, this argument is not decisive, but in the next subsection, we will see how alle(n) and beide(n) provide additional, more robust evidence against the quantifier-float approach.

When alle(n) and beide(n) are used as floating quantifiers, the orthographic -n forms allen and beiden are found with [+human] noun phrases, as in (120), and the “plain” forms with [-human] noun phrases, as in (121). The use of alle(n) and beide(n) as floating quantifiers is restricted to the more formal registers of the language; the spoken language typically prefers allemaal and allebei. The forms alle(n) and beide(n) behave like the floating quantifiers alle + Num and allebei: as shown in (120) and (121), the associate of the floating quantifier can be either a complex noun phrase or a pronoun, but the two types of associate differ in that noun phrases must precede the floating quantifier, whereas pronouns can follow it when the floating quantifier is in the clause-initial position.

The use of alle(n) and beide(n) as floating quantifiers can be easily distinguished from their use as modifiers within the noun phrase. First, the examples in (122) show that a determiner is incompatible with the modifiers, unlike the floating quantifiers in (120) and (121). Second, in the case of [+human] noun phrases in (122a&a'), the orthographic -n cannot occur on the modifiers, whereas it must be expressed on the floating quantifiers in (120).

However, the fact that alle(n) and beide(n) can be associated as floating quantifiers with noun phrases with which they cannot form a syntactic constituent casts further doubt on the quantifier-float approach. This approach is also difficult to reconcile with the observation that the floating quantifiers alle(n) and beide(n) can be associated with independently quantified noun phrases. Another problem for this approach is that, as in the case of alle + Num and allebei in (118), examples such as (123) are acceptable (although judgments about the primed examples seem to vary from case to case and person to person).

The fact illustrated in (124) that it is impossible for the modifiers alle and beide to form a constituent with a numerically quantified noun phrase thus seems problematic for the quantifier-float approach.