The primeless examples in (37) show that all N1s can be preceded by the indefinite determiner eena. This suggests that we are dealing with count nouns, and we therefore expect pluralization to be possible. The primed examples show that this is indeed possible with most N1s, but that the quantifier noun in (37a') resists pluralization. Furthermore, (37b') shows that plural marking is optional for the measure noun liter.

The general pattern in (37) is compatible with the classification in (36): quantifier nouns lack a plural form, while referential nouns allow plural formation. And, as expected, measure nouns show mixed behavior: they may or may not take the plural suffix, depending on whether they have a quantificational or a referential, package-unit reading. However, there are a number of complications. which we will discuss in the following subsections.

The three types of N1s also differ in terms of diminutive formation. The examples in (46c-d) show that referential nouns allow it, while (46a) shows that quantifier nouns do not. As expected, measure nouns again show mixed behavior: diminutive formation is possible when they are interpreted referentially, but not when they are interpreted quantificationally. That the diminutive is derived from the referential and not the quantificational measure noun is clear from the fact, illustrated by (46b'), that they must be pluralized when preceded by a cardinal numeral.

Note that een beetjea bit in een beetje watera bit of water is only an apparent counterexample to the claim that quantificational N1s do not undergo diminutivization: een beetje is a lexicalized formation. This is clear from the fact that it has no counterpart without the diminutive suffix: *een beet water. The plural form preceded by a cardinal numeral or an existential quantifier such as enkele/sommige is also degraded, but the plural form preceded by the quantifier alle occurs frequently on the internet. This pattern is also found when N2 is not present, as in *two/*sommige/alle beetjes helpenall bits help.

The data discussed in Subsections A and B shows that it is necessary to distinguish between purely quantificational and referential N1s. Only the latter allow pluralization and diminutive formation. This distinction seems to be supported by data concerning compounding. The denotation of a nominal compound is mainly determined by its second member, which can be considered as the head of the compound; the first member only has the function of further specifying the denotation of the second; cf. Section 15.4. This is clear from the fact that a tafelaanstekertable lighter is a kind of lighter, not a kind of table. Given this, we predict that only referential nouns can appear as the head/second member of a compound.

The examples in (48) show that our prediction is correct. The first prediction is that the container, part, and collective nouns can occur as the head of a compound, and the acceptability of (48c-e) shows that this is indeed the case, although we should note that perenemmer is a possible, but unattested word. The second prediction is that the quantifier nouns cannot occur as the head of a compound because they have no denotation, and (48a) again shows that this is the case. A problem is that we expect the measure nouns to exhibit mixed behavior, whereas they actually pattern with the quantifier nouns. This suggests that the referential reading of measure nouns is rather marked and only arises under strong pressure from the context.

Note that the quantificational force of container, part, and collective nouns has completely disappeared in the compounds in (48c-e). This also holds for nouns that are normally used as quantifier nouns. For example, in a compound like beestenboelpigsty, the head of the compound is not the quantifier noun boel, but a noun denoting collections of things that do not necessarily belong together. Similarly, the meaning of the second member of compounds like studentenaantalnumber of students is not related to the quantificational interpretation of aantal, but to its referential interpretation; cf. the discussion of example (3) in Section 18.1.1.1, sub I.

The results of Subsections A to C, summarized in Table 4, have shown that a distinction should be made between purely quantificational and more referential N1s. Quantifier nouns belong to the first kind; container, part, and collective nouns all belong to the second kind; and measure nouns are ambiguous between the first and the second kind.

The pattern in Table 4 corresponds nicely with our findings in Table 3: that quantifier nouns are purely quantificational is consistent with the fact that they cannot trigger agreement on the finite verb or a demonstrative; that measure nouns are ambiguous between a purely quantificational and a referential, package-unit reading is consistent with the fact that either they or N2 can trigger agreement; that container, part, and collective nouns are referential is consistent with the fact that they block agreement between N2 and the finite verb or the demonstrative. The fact that all N1s have some quantificational force is consistent with the fact that in all cases N2 can be interpreted as the semantic head of the construction.

Example (49) again illustrates that all N1s can be preceded by the indefinite article een. However, if we are dealing with a quantifier noun, the indefinite article cannot be replaced by a definite article. This is possible with measure nouns, but it results in the loss of the purely quantificational reading: het ons kaas refers to a certain piece or quantity of cheese that can be identified by the addressee. The remaining types of N1s can all be preceded by a definite article. Note that it is N1 that agrees in gender and number with the definite article: the N2s in (49) would all select the article de, not het; cf. the discussion of example (14) in Section 18.1.1.2, sub I.

Note, however, that many noun phrases that normally do not allow a definite article can be preceded by it if they are modified: a proper noun like Amsterdam, for example, normally cannot be preceded by the definite article, but if it is modified by e.g. a relative clause, the definite article is allowed: het Amsterdam *(dat ik zo goed ken)the Amsterdam that I know so well. The examples in (50) show that quantifier nouns can exhibit ambiguous behavior in this respect: some, like boel in (50a), do not allow the definite article in these modification contexts either, while others, like paarcouple of or stootlot of in (50b), are compatible with the article in such contexts.

Note, however, that the article in (50b) is probably not part of the noun phrase headed by N1, but of the noun phrase headed by N2. A reason for this assumption is that the noun paar is neuter (at least in its use as a collective noun), and should therefore select the definite determiner het, not de as in (50b): het/*de paar schoenenthe pair of shoes. This suggests that the construction in (50b) is similar to the quantified constructions in (51), where the article is unquestionably selected by the noun.

The fact that quantifier nouns cannot normally be preceded by a definite article may cast some doubt on the assumption that the element een is a “true” article in constructions with quantifier nouns. The idea that we are dealing with a spurious article should not be dismissed, since there are many contexts in which een clearly does not function as an indefinite article; cf. Section 18.2.1 for another example. For instance, een can also be used in examples such as (52) with a plural noun, where it seems to function as a modifier with an “approximative” meaning. It is tempting to relate this use of een to that in een boel mensen in (49a).

That we are dealing with a spurious article when the noun is purely quantificational can be indirectly supported by the fact, illustrated in (53a), that German ein is not morphologically marked for case when it precedes a quantifier noun, as it would normally be when it is part of a referential noun phrase; cf. (53b), where the noun Paar is referential and the article ein has the dative ending -em.

Another reason to think that the element een in een boel mensen is different from the other occurrences of een in (49) is that it cannot be replaced by its negative counterpart geenno. This is illustrated in (54); note in particular the difference between (54a) and (54e), which form a minimal pair (if we abstract from the agreement on the finite verb).

The data in this subsection suggests that quantifier nouns cannot be preceded by an article. In (49a), the element een is a spurious indefinite article, which may be related to the modifier een in examples such as (52). The other types of N1 occur with both indefinite and definite articles.

Demonstrative pronouns exhibit a pattern similar to that of the definite article. Example (55a) shows that a quantifier noun like boel never occurs with a demonstrative pronoun, while other quantifier nouns, like paar in (55b), are more readily accepted with demonstrative pronouns (especially the distal ones). Note that it is not necessary to modify the QBC in (55b), which may be due to the fact that the demonstratives themselves function as modifiers in the sense that they imply some partitioning of the set denoted by N2; cf. Section 19.2.3.

Recall from Section 18.1.1.2, sub I, that the demonstratives in (55b) do not agree in gender and number with N1, but with N2. This can be easily illustrated by the minimal pair in (56). In (56a) the QBC refers to two shoes that form a pair: the neuter noun paar is therefore referential and the demonstrative agrees with it. In (56b) the QBC refers to a set of two or more shoes: the neuter noun paar is therefore purely quantificational and the demonstrative agrees with N2.

Section 18.1.1.2, sub I, has already shown that QBCs containing a measure noun exhibit the same ambiguity as paar, although some speakers consider the construction in which the demonstrative agrees with N2 to be marked. The relevant examples are repeated in (57a&b).

The examples in (58) show that the remaining types of N1s can occur freely with demonstrative pronouns. The demonstrative pronouns in (58) must agree with N1; replacing them with deze/die leads to unacceptability.

The examples in this subsection have shown that demonstrative pronouns can only occur with a subset of quantifier nouns; if possible, the demonstrative agrees with N2 in gender and number. Container, part, and collective nouns can easily be combined with demonstrative pronouns and agree with them in number and gender. Measure nouns again show a more hybrid behavior.

Example (59a) shows that possessive pronouns always seem to lead to a degraded result with quantifier nouns, regardless of whether a modifier is present or not. The use of a possessor is at least marginally possible with a measure noun like pond in (59b): the measure noun must receive a referential interpretation in this case. Possessive pronouns are easily possible with the referential nouns in (59c-e).

The examples in (60) show that a quantifier noun like boellot of cannot be preceded by a quantifier or a cardinal numeral. The unacceptability of (60a) is not surprising, given that the quantifier noun boel cannot be pluralized (see example (37)), so that it cannot satisfy the requirement that quantifiers like sommigesome and alleall or cardinal numerals such as vierfour take a plural noun. However, appealing to this fact is not sufficient to account for the ban on quantification, because (60b) is unacceptable despite the fact that the distributive quantifier elkeach selects a singular noun.

The unacceptability of the examples in (60) must therefore be related to the quantificational function of the quantifier nouns. This can be explained by appealing to the fact, discussed in detail in Chapter 20, that quantifiers and numerals operate on sets; since quantifier nouns do not denote sets, the quantifier/numeral cannot perform its function. Note that the quantifier/numeral cannot operate on N2 either, since that is precisely the function of the quantifier noun: it is never possible to have two quantifiers or numerals that take scope over the same noun phrase.

The examples in (61) show that container, part, and collective nouns freely co-occur with quantifiers. Note, however, that these nouns have lost their quantificational property, in the sense that in these cases the QBCs refer to concrete cups, pieces, and flocks.

As noted previously, some N1s, such as paarpair, can be used as both a purely quantificational and a referential noun. Given the observations above, we expect the addition of a quantifier to have a disambiguating effect. This is indeed borne out, given that the examples in (62) can only be given a referential interpretation; these QBCs refer to some/all/each of the pairs of shoes in the domain of discourse.

The examples in (38), repeated here as (63), show that the noun paar can also be preceded by a cardinal numeral, in which case the noun can be either singular or plural. In both cases the QBC refers to exactly eight shoes, but the examples differ in the implication that the shoes make up four pairs that belong together: this is implied by (63b) but not by (63a). It is tempting to account for this difference by claiming that the noun paar is purely quantificational in (63a) and referential in (63b). But if this is indeed the case, we must conclude that there is no general prohibition against using a cardinal numeral with purely quantificational nouns.

The examples in (64) show that measure nouns can be preceded by a quantifier. The use of an existential/universal quantifier, which triggers the package-unit reading, can lead to a slightly marked result. The distributive quantifier elk does not trigger this reading and gives rise to a perfectly acceptable result.

Most measure nouns preceded by a numeral can be either singular or plural; cf. Subsection IA, for some exceptions. In the latter case, the quantifier noun is clearly used as a referential noun with a package-unit reading: (65b) refers to four discrete quantities of milk of one liter each; in (65a), on the other hand, it refers to one quantity of milk, further specified as a quantity of four liters. This supports the suggestion above (63) that there is no general prohibition against using cardinal numerals with purely quantificational nouns.

For completeness’ sake, note that some N1s are lexically restricted in the sense that they can only be used if a cardinal numeral is present. An example is given in (66). The N1 man must appear in its singular form.

On the basis of what we have seen so far, we may expect modification of N1 by an attributive modifier to be impossible in the case of purely quantificational nouns; attributive modifiers are used to restrict the set denoted by the modified noun, but purely quantificational nouns do not denote any such set. As shown in (67a), this expectation is indeed borne out. The remaining examples in (67) show that modification of the other N1s is possible.

There are, however, several restrictions on the use of attributive adjectives in constructions of this type. If we are dealing with a measure noun like kilo, the adjective modifying N1 must be of one of the following types: it can be quantificational, as veel in (68a), it can have an adverbial meaning indicating approximation, such as klein in (68b), or it can have a “partitive” meaning, such as half and heel in (68c).

Container, part, and collective nouns exhibit similar restrictions: the examples in (69) to (71) show that quantificational, size and “partitive” adjectives are possible, while adjectives denoting other properties yield marked results.

Recall from Section 18.1.1.2, sub IIB, that an attributive adjective preceding N1 can be used to modify N2. Thus, all kinds of attributive adjectives can precede these N1s, provided that they can be construed with N2: an example such as een smakelijk glas biera tasty glass of beer is acceptable, where the attributive adjective expresses a property of the N2 bierbeer. An interesting case, about which we have little to say, is geef me een nieuw glas biergive me a new glass of beer: in this example, the adjective nieuwnew is construed with N1, but it does not attribute any property to the glass in question; rather, it is interpreted as “another/a fresh glass of beer”.

Table 5 summarizes the results of the previous subsections. It shows that quantifier nouns can maintain far fewer syntagmatic relations than container, part and collective nouns. The latter can be preceded by all kinds of determiners, quantifiers and numerals, and do not exhibit any special restrictions regarding attributive modification. The former, on the other hand, exhibit all kinds of restrictions: the element een preceding quantifier nouns may not be a “true” indefinite article, but something else; definite determiners, quantifiers, numerals, possessive pronouns, and attributive modifiers do not occur at all; demonstrative pronouns may occur with some, but not all, quantifier nouns. Measure nouns again show a more mixed behavior: the notation —/+ indicates that the element in question can be used if the noun has a referential interpretation, but not if it has a quantificational interpretation. The results in Table 5 are consistent with the classification given in (36), which groups the five noun types into the three supercategories in the top row.

A final piece of evidence for the classification in (36) comes from recursive QBCs, i.e. QBCs that embed some other QBC. The examples given so far always contain two nouns, but it is possible to have more complex cases in which a QBC is embedded in a larger QBC, resulting in sequences of three or more nouns. Given that the second part of a QBC must denote a set, it is predicted that the embedded QBC cannot be purely quantificational. The examples in (72) suggest that this expectation is indeed borne out. In these examples, N1 is a quantifier noun and it can be followed by any QBC, provided that the N1 of that QBC is not itself a quantifier noun.

A problem for the claim that quantifier nouns cannot be used as the N1 of an embedded QBC is that the measure noun kilo in (72b) can appear either in its singular or in its plural form; since we have argued above that the measure noun is purely quantificational in the former case, this suggests that QBCs headed by a purely quantificational N1 can be embedded within a larger QBC after all. However, an alternative analysis seems possible. Consider the examples in (73a&b). We have seen that these examples differ in that (73a) simply refers to the weight of four kilograms of cheese without any implication regarding the package units, whereas (73b) implies that we are dealing with four separate package units of one kilo each. This suggests that the structures of the two examples are different, as indicated in the primed examples: in (73a) the numeral vier can be considered part of a complex quantifier vier kilo, whereas in (73b) it modifies the QBC kilo’s kaas.

A similar analysis can be given for the examples in (72b), repeated below as (74): in (74a), the complex quantifier een aantal kilo functions as N1 with kaas functioning as N2; in (74b), on the other hand, een aantal functions as N1 and kilo’s kaas is an embedded QBC.

Independent evidence for the analyses in the primed examples can be found in the examples in (75), which involve quantitative er. The contrast between the examples can be accounted for by the fact that the elided part corresponds to a single constituent in (75b), but not in (75a).

The other examples in (72) are ambiguous in the same way. We will show this for container nouns. Consider the examples in (76). In (76a) the QBC just indicates an amount of sugar, and we are therefore dealing with a complex quantifier vier/een paar zakkenfour/a couple of bags, as indicated in (76a'). In (76b), on the other hand, we are dealing with a number of bags of sugar, and the phrase zakken suiker is therefore a QBC embedded in a larger QBC, as indicated in (76b').

From this we can conclude that (72b) does not provide evidence against the claim that QBCs headed by a purely quantificational N1 cannot be embedded in a larger QBC. The apparent counterexample een aantal kilo suiker can be analyzed as involving a complex quantifier and therefore need not be considered a recursive QBC. Note that the fact that (72a) does not allow an interpretation involving a complex quantifier is consistent with the fact that quantifier nouns cannot be preceded by a numeral either: *vier hoop/hopen mensen*four lots of people.

In (77) we give examples of recursive QBCs, in which N1 is a measure noun. We find the same contrast as in (72): while container, part, and collective nouns can be used as the N1 of an embedded QBC, quantifier nouns cannot. The dollar sign indicates that the examples in (72c&e) are odd due to our knowledge of the world: matchboxes do not normally come in units of a kilo, and it is not common to add up collections of entities until they reach a certain weight. The main difference between the examples in (72) and (77) concerns the measure nouns: a measure noun cannot be followed by another measure noun in the singular. This supports our earlier claim that a QBC headed by a purely quantificational N1 cannot be embedded in a larger QBC: the unacceptability of (77b) is due to the fact that there is no complex quantifier *een kilo ons. Example (77b'), on the other hand, seems to be acceptable despite being marked, because it is difficult to conceptualize and the intended meaning is more easily expressed by the phrase tien onsjes kaasten ounces of cheese.

In (78) to (80) we give similar examples for container, part, and collective nouns. The examples in (78) show that container nouns behave like measure nouns. Example (78e) may again be odd for reasons concerning our knowledge of the world, but otherwise it seems perfectly well-formed.

The part nouns in (79) cannot easily be used as the N1 of a recursive QBC. This is, of course, due to the fact that they can only be followed by a non-count noun, while the referential N1s heading the embedded QBCs are count nouns.

The examples in (80) show that the collective nouns behave like the measure and container nouns.

The previous subsections have repeatedly claimed that all N1s are quantificational in the sense that they indicate a certain amount or quantity of the denotation of N2. In this respect, they behave like cardinal numerals or quantifying adjectives like veelmany/much. As shown in (81), the latter elements can be questioned by the wh-word hoeveelhow many/much. If N1s do indeed have quantifier-like properties comparable to cardinal numerals or quantifying adjectives, we expect that they will also provide felicitous answers to the question in (81a). Example (81b') shows that this is indeed true for quantifier nouns.

In (82) it is shown that the same is true for the measure noun literliter and the container noun glasglass. Note that the N1s can undergo pluralization and diminutivization and can be preceded by a cardinal numeral. This clearly shows that we are dealing with referential nouns.

Similarly, the part and collective nouns in (83) and (84) can be used to answer questions involving hoeveel, although there seems to be an additional restriction: if the part noun plakslice or the collective noun groepgroup is preceded by the indefinite article eena, as in (83b) and (84b), the size of the slice/group must be indicated by diminutivization or addition of an attributive adjective like dikbig or grootbig; this is not necessary if these nouns are preceded by a numeral, as in (83c) and (84c).

QBCs can be either weak or strong noun phrases. On the weak reading, exemplified by the primeless examples in (85), these noun phrases get a non-specific indefinite interpretation, i.e. they simply refer to a set of new discourse entities. In the strong reading, exemplified by the primed examples, these noun phrases get a partitive interpretation, i.e. they refer to a subset of a larger set of entities already given in the domain of discourse. The primed and primeless examples in (85a&d) show that while the indefinite article is always possible in the weak reading of QBCs, it sometimes leads to a degraded result in the strong reading.

All N1s indicate a certain amount or quantity. The difference between quantifier nouns and the other types of N1s is that quantifier nouns indicate an indefinite amount or quantity, whereas the other types indicate an often conventionally or contextually determined definite amount or quantity. The difference is brought out clearly in constructions with the preposition perper. This preposition can be followed by a cardinal numeral like vierfour but not by a quantifier like veelmany/much, which indicates an indefinite amount or quantity.

The examples in (87) show that the same difference can be found between quantifier nouns like boel and hoop, which indicate an indefinite amount of quantities, and the other N1s, which indicate a (conventionally or contextually determined) amount or quantity.

The result is often marked when per is followed by a plural noun, although acceptability may vary depending on the context and ease of conceptualization; an example such as (88c) gives rise to a perfectly acceptable result in the following example found on the internet: Deze speculaas weegt ca. 125 gram per plak en wordt per twee plakken verkochtThis spiced biscuit weighs about 125 grams and is sold in sets of two pieces. The fact that (88a) is fully acceptable with the singular form of kilo shows that a phrase like twee kilo does not function as a plural noun phrase; it simply refers to a definite quantity. The marked status of the plural form kilo’s shows that the noun phrase twee kilo’stwo kilos is plural: it refers to two discrete entities of one kilo each.

If N1s are quantificational, one would expect them to co-occur with quantitative er. However, the primeless examples in (89) show that only the quantifier and the measure nouns bear this out. Note that the measure noun in (89b) must be followed by the sequence of + numeral, which is probably due to the fact that this makes the quantifier less definite. Given the requirement that the phonetically empty noun is [+count], it is not surprising that measure nouns like liter give rise to a degraded result, since they usually combine with non-count N2s. Given that part nouns also combine with non-count N2s, we could in principle give a similar account of the unacceptability of (89d), but the unacceptability of (89c&e) shows that there is more involved than just a count/non-count distinction: the unacceptability of (89c-e) is clearly related to the referential status of the N1s.

Note that the intended assertions of the unacceptable examples can be expressed by the examples in (90), where N2 is simply left implicit. This is impossible with quantifier nouns like paar, which is typically preceded by the unstressed indefinite article eena; the noun paar in (89a) can only be interpreted as a collective noun, which is typically preceded by a numeral in this context.

Example (91a) shows that QBCs headed by the quantifier noun aantal can trigger either singular or plural agreement on the finite verb (cf. Section 18.1.1.2, sub I), and the same is shown for the measure noun kilo in (91b). The primed counterparts with quantitative er, on the other hand, are compatible with plural agreement only, showing that in these constructions the verb agrees with the phonetically empty N2 that we assumed to be present in these constructions. We have added example (91c) to show that in constructions without quantitative er, agreement is always triggered by N1.

The primed examples in (92) show that N1 differs from N2 in that it cannot be replaced by a nominal gap licensed by quantitative er: this holds both for the quantifier noun aantal in (92a'), as well as for referential nouns like the part noun stukpiece in (92b'). The two types of N1 exhibit divergent behavior when it comes to replacing the phrase N1 + N2: the doubly-primed examples show that this is easily possible when N1 is a referential noun, but not when it is a quantifier noun. This difference in behavior has nothing to do with the semantic distinction between the two classes of N1, but is simply be due to the fact that the quantifier noun aantal cannot be preceded by a numeral/weak quantifier. That the indefinite article eena cannot license quantitative er is also illustrated by the fact that (92a'') cannot be used with the interpretation I heb nog een stuk chocolaI still have a piece of chocolate either.

Cardinal numerals can be modified by variety of modifiers; cf. Section 20.1.1.5. In this subsection we will deal with the modifiers in (93): the modifier minstensat least in (93a) indicates that the cardinal numeral provides a lower bound, while hoogstensat most in (93b) indicates that it provides an upper bound. The modifiers in (93c) have an approximative meaning. The primed examples show that these modifiers cannot be used with quantifiers like veelmany or weinigfew.

The examples in (94) show that most N1s can be preceded by the modifiers in (93). The only exceptions are quantifier nouns, which is not surprising since they indicate an indefinite amount/quantity, just like the quantifier veel in the primed examples in (93); cf. Subsection IIIC. The examples in (94e&e') show that modification of collective nouns is possible only if the collection consists of a default number of entities; if this is not the case (as with a flock or a group), the resulting construction is unacceptable. Note that there are also modifiers that have a more restricted use: for example, ruimmore than seems to be more or less restricted to measure and container nouns; cf. the contrast between ruim een kilo/zak krentenover a kilo/bag of raisins and ??ruim een reep chocola/team voetballers.

The fact that the modifiers in (94) cannot immediately precede N1, but must precede een, suggests that it is the whole phrase een N1 that acts as a quantifier. This will be clear from the fact that in examples such as (95a) the modifier must be adjacent to the modified numeral. Finally, in cases where N1 is preceded by a numeral, it is the numeral that is modified, not N1; this explains why, unlike (94e'), (95b) is acceptable.

The examples in (96) show that cardinal numerals and quantifiers can take scope over nominal phrases of different sizes: in the primeless examples their scope is limited to one conjunct, whereas in the primed examples they can have both conjuncts in their scope.

These examples differ in the scope of the attributive modifier/numeral: in the primeless examples the numeral/quantifier has scope only over the noun immediately following it, whereas in the primed examples it has scope over both nouns. This difference is clearest with the numeral vierfour in the (a)-examples: (96a) refers to a set of people with cardinality 8, whereas the phrase in (96b') refers to a set of people with cardinality 4. The difference is less clear with the quantifier veelmany in the (b)-examples, because (96b) entails (96b'). However, the same does not seem to hold in the other direction: in a situation with 90 men and 4 women, (96b') might be appropriate, whereas (96b) is not.

The quantifier noun hoop has the same property as the quantifier veel: while (97a) entails (97b), the entailment does not hold vice versa. This shows that in (97b) the quantifier noun must also be assumed to take scope over the conjoined phrase dieven en inbrekers.

Measure nouns, on the other hand, act like cardinal numerals: the full noun phrase in (98a) refers to a total quantity of two kilos of potatoes and vegetables, while in (98a') it refers to a total weight of one kilo. More or less the same applies to the container noun glasglass in the (b)-examples: (98b) refers to two glasses, one filled with gin and one with tonic, while (98b') refers to a single glass filled with a mixture of gin and tonic. Collective nouns also behave in this way, but this is not illustrated here. Finally, the (c)-examples show that part nouns like stukpiece cannot take scope over both conjuncts: this is because QBCs with these nouns must refer to “homogeneous” entities.

This subsection has compared the behavior of the three types of N1s with the behavior of numerals and quantifiers. Quantifier nouns were shown to pattern with quantifiers. Container, part, and collective nouns, on the other hand, tend to pattern with cardinal numerals, despite the fact that the latter, but not the former, license quantitative er. Measure nouns again exhibit mixed behavior.