Section 15.1.2 has shown that the noun phrase can be divided into two subdomains, the NP-domain and the DP-domain: the former is headed by the noun and determines the denotation of the noun phrase, while the latter is headed by a determiner or a quantifier/numeral and determines the referential and/or quantificational properties of the noun phrase. Thus, the internal structure of the noun phrase as a whole can be represented as in (1), where the determiner (D) and the noun (N) are the heads of the projections DP and NP, respectively, and the dots indicate the possible positions of other elements. In this chapter, as well as in Chapter 17, we will concentrate on the projection of the noun, i.e. the NP-domain.

Each NP contains an obligatory head N and, optionally, one or more other elements which can be further categorized according to their function, i.e. according to whether they function as complements or as restrictive modifiers. Complements are elements whose presence is required by the semantics of the head noun; the idea is that these complements are obligatory arguments of the nominal head, comparable to the complements of verbs. Restrictive modifiers, on the other hand, are not required by the semantics of the head noun.

Complements are generally closer to the nominal head than the restrictive (as well as the non-restrictive) modifiers. Example (2a) illustrates this for the nominal complement tomatentomatoes and the adjectival modifier gebruikelijkusual in prenominal position, example (2b) for the PP-complement van Janof Jan and the PP-modifier in het ziekenhuisin hospital in postnominal position, and example (2c) for the clausal complement dat Jan ziek geworden was and the restrictive relative clause dat net binnenkwam in postnominal position.

These word-order facts lead to the generalization in (3), according to which complements and modifiers are inserted at different levels within the NP: complements are immediate sisters of the head noun, whereas modifiers are adjuncts inserted at some higher level in the NP. According to this assumption, the structures of the noun phrases in (2) are as shown in (4).

The difference between complements and restrictive modifiers is often difficult to determine, as the two may be categorially identical. This is not so much the case in prenominal position, where the modifier typically takes the form of an AP and the complement takes the form of a noun phrase or a PP, as in (5a), but this problem does arise in postnominal position, where complements and modifiers can both take the form of a PP or a clause, as in (5b&c).

Although the present chapter is mainly concerned with complements, it is necessary to first look more closely at the difference between complements and modifiers: Section 16.2.1 focuses on the difference between prepositional complements and modifiers in (5b), and Section 16.3.3 focuses on the difference between clausal complements and modifiers in (5c). Modification within the NP is the main topic of Chapter 17.

The notion of complementation is usually associated with the verbal domain. Verbs have argument structures that specify the number and thematic roles of their arguments. The arguments of verbs can be divided into (i) internal arguments or complements and (ii) external arguments. The former in a sense complete the predicate, so that it can be predicated of the latter. In the lexical frames in (6), the external argument is underlined to distinguish it from the complements. The semantic arguments of the verb are usually realized as syntactic arguments: internal arguments typically appear as objects of the verb, while the external argument corresponds to the subject.

Nouns can also function as predicates, and can therefore take arguments. This is shown in (7), where the nominal noun phrase geniegenius is predicated of the noun phrase Jan, which therefore functions as the external argument. Since the usual labels for semantic roles refer especially to the roles of arguments in the event structure denoted by verbal predications, we will simply call the external argument of nouns the referent (Ref), i.e. the entity to which the property denoted by the nominal (or adjectival) predicate applies.

The syntactic mapping of the external arguments of nouns is more complicated than that of verbs. When the noun is used as the head of a nominal predicate in a copular or vinden-construction, the mapping is rather straightforward: in the two sentences in (7) the external argument corresponds to the noun phrase Jan, which functions as the (nominative) subject and (accusative) object of the clause, respectively. However, when the noun is used as the head of a noun phrase in argument position, it typically tends to be syntactically avalent: rather than behaving like a predicate with one or more arguments, the noun phrase of which it is a part acts as an argument of some other predicate. Correspondingly, such noun phrases do not denote a property, but typically have a referential function: they identify the entity or set of entities of which something is predicated. For instance, in (8), the noun phrase de man has the same function as Jan in (7), i.e. it acts as the external argument of the nominal predicate.

This does not mean that nouns heading a noun phrase in argument position do not have a predicative function: such nouns can be said to predicate something of their referential argument, i.e. of the entity or set of entities referred to by the noun phrase; cf. Williams (1981). This is clear from the fact that the predicate calculus represents the noun phrase een man in (9a) as ∃x man (x): there is an entity x such that the predicate man applies to x. Correspondingly, (9a) is usually assigned the semantic interpretation in (9b), which involves the conjunction of two predicates: there is an entity x such that both the predicates man and walking in the street apply to x.

The discussion above has shown that nouns always have an external argument, but that this argument is not syntactically realized when the noun is the head of a noun phrase that functions as the syntactic argument of some other predicate, as in (9). The external argument of the noun can (and must) be syntactically realized only when the noun is the head of a noun phrase that functions syntactically as a predicate, as in (7). This is expressed by generalization II in (10).

Complementation is not a typical property of non-derived nouns, i.e. nouns with a monomorphemic stem. However, there are at least two classes of non-derived nouns that usually require the presence of an argument. The first is the class of relational nouns, which includes kinship nouns like vaderfather, broerbrother, nichtniece, and nouns denoting physical properties of objects like vormshape, gewichtweight or kleurcolor; cf. Section 15.2.3 for more examples. These relational nouns can only successfully fulfill their referential function if they are related to some other entity. This is illustrated in (11a) for the noun vader in unexpected contexts: this example is only fully acceptable if there is a noun phrase carrying the “child” role assigned by the noun vader. In (11b) the noun vormshape cannot refer independently but requires the syntactic realization of the noun phrase referring to a physical object that has a shape. Note that, according to generalization II in (10), the “Ref” role only needs to be syntactically realized if the noun phrase headed by vader functions syntactically as a predicate. The complementation of relational nouns will be discussed in more detail in Section 16.2.2.

Other non-derived nouns that can take complements are the so-called picture and story nouns. In (12a), the noun schetssketch assigns a theme role to de Amstel (it is the object depicted) and an agent role to Rembrandt (he is the maker of the painting); in (12b) something similar holds for the story noun gedicht. The complementation of picture/story nouns will be discussed in more detail in Section 16.2.5.

Other cases which may involve some kind of complementation are the cases in (13), all of which take the PP headed by aan, although scala also occurs (less frequently) with van; cf. https://taaladvies.net/scala-van-of-aan-artikelen. That we are dealing with complementation is supported by the fact that the PP can only be omitted if its content is recoverable from the context. Hoeksema (2013) notes that scala sometimes also occurs with bare plurals (cf. een scala maatregelen/kleuren), but this seems to be formal and less common; such cases may be instantiations of the (innovative) binominal construction of the kind discussed in Section 4.1.2.

The examples in (14) are more incidental cases of non-derived nouns that seem to take a PP-complement: an argument for assuming that the noun is non-derived is that there is (synchronically) no corresponding verb; arguments for assuming argument status of the PPs are that the prepositions seem semantically vacuous and that they are selected by the noun; cf. Hoeksema (2013).

While non-derived nouns typically do not take complements, derived nouns do. The arguments of these derived nouns are usually “inherited” from the input stem. Consider, for example, the transitive verb behandelento treat and the noun behandelingtreatment, which is derived from the former by adding the suffix -ing. As shown in (15), the verb and the noun can take the same arguments: an agent and a theme. The main difference between the verbal and the nominal predicate is that the agent is the external argument of the verb, while it is an internal argument of the noun: the external argument of the noun is assigned the “Ref” role. In accordance with generalization II in (10), the argument bearing the agent role can be expressed within the noun phrase (while the “Ref” role need not be syntactically expressed).

The same applies to nouns derived from intransitive verbs, as shown in example (16). Here, too, the agent argument of the verb is inherited by the noun as an internal argument, with the addition of a new external argument assigned the “Ref” role. Again, the argument with the agent role can be expressed within the noun phrase.

In the nominalizations in (15) and (16), the external argument of the verb has become an internal argument of the derived noun. However, in the case of agentive er-nouns, i.e. deverbal person nouns derived by the suffix -er, the external argument of the verb is similar to the external argument of the noun: the argument assigned the “Ref” role of the derived noun schrijverwriter in (17b) corresponds to the argument assigned the agent role by the predicate een boek schrijven in (17a). Accordingly, the argument corresponding to the argument assigned the agent role by the verb cannot be expressed within the noun phrase; it can only be realized as the subject or object of a clause in which the noun phrase is used as a syntactic predicate; cf. JanRef is de schrijver van het boek Jan is the writer of the book.

Inheritance of argument structures can also be found with deadjectival nouns. For instance, in example (18) the external argument of hoog, which we call “RefA” to distinguish it from the “RefN” role of the noun, functions as an internal argument of the noun hoogteheight.

Since adpositions cannot easily be used as the input for nominalization, we will not discuss them here; we refer the reader to Section 15.3.4 for examples.

This subsection has shown that nominalization of verbs and adjectives generally involves the internalization of the external argument of the input form; only with agentive er-nouns does the external argument of the verb seem to correspond to the external argument of the derived noun. This is expressed as generalization III in (19).

Finally, note that it is not always immediately obvious whether we are dealing with a non-derived noun with an argument structure of its own or a derived noun that has inherited the argument structure of its input. At first glance, the verb aanvalattack in (20a) seems to be a good example of a deverbal noun, but on closer inspection it seems more plausible that it is a non-derived noun with an argument structure of it own, because the presumed input verb aanvallento attack takes a nominal and not a PP-complement.

The internal arguments of a noun can be realized in several ways. Within the verbal domain, agentive arguments appear as nominative noun phrases in active clauses, and optionally as agentive door-PPs in passive clauses. Arguments with the semantic function of theme typically appear as accusative noun phrases in active transitive clauses, and as nominative noun phrases in passive and unaccusative clauses. Arguments with the semantic function of goal/benefactive can be realized as a dative noun phrase or as an aan/voor-PP. The remaining arguments are realized as PPs. However, since a noun does not assign case, we expect that its internal arguments will typically be realized as PPs. However, this does not exhaust the possibilities.

This subsection deals with word order restrictions on the internal arguments of nouns. We will start with the relative order of the head noun and its arguments. This is followed by a brief review of the relative order of the internal arguments themselves.

The seven generalizations formulated in the previous subsections are repeated in (55).