The fact that verbs take arguments is closely related to the fact that they function semantically as n-place predicates. An intransitive verb like lachen'to laugh' in (1a), for example, functions as a one-place predicate, which can be represented in predicate logic as in (1a'). A transitive verb like lezen'to read' in (1b), on the other hand, takes two arguments and thus functions as a two-place predicate, which can be semantically represented as in (1b'). See Section 1.1, sub I, for more discussion.

The semantic representation in (1b') suggests that the two arguments of the transitive verb lachen have more or less the same status; the subject noun phrase Jan and the direct object noun phrase het boek'the book' are both needed to saturate the predicate lezen and thus to complete the predication. In another respect, however, their relation to the verb is asymmetrical; the direct object is needed to create a complex predicate het boek lezen 'to read the book' that can be predicated of the subject Jan. In other words, the verb phrase leest het boek in (1b) has the same semantic status as the intransitive verb lachen in (1a), and objects can thus be said to be internal to the one-place predicate that is predicated of the subject of the clause. For this reason objects will be called internal arguments or complements of the verb, whereas the subject is normally an external argument; see Section 1.2.2, sub I, for more discussion and Williams (1980/1981) for the original definitions of these notions.

The previous subsection claimed that subjects are normally external arguments. The addition of normally is needed because in present-day linguistics the notions of internal and external argument are used not only to refer to the function of arguments in the saturation of the predicate denoted by the verb, but also (and perhaps even primarily) to the thematic roles that these arguments may have; in the prototypical case an external argument refers to the agent or the cause of the event, whereas an internal argument instead refers to a theme, a goal/source, an experiencer, etc; see also Section 1.2. Since there are cases in which the subject of the clause does not refer to the agent/cause, but rather to one of the thematic roles that are typically assigned to internal arguments, this means that the notion of subject cannot be equated with that of external argument. For example, the subject of the passive clause in (2b) is not an external but an internal argument of the verb lezen'to read', just like the direct object of the active clause in (2a).

Section 2.1 will show that there is a group of so-called unaccusative verbs that have the defining property that their subject is not an external agentive argument, but an internal theme argument. That something like this may well be the case can be readily illustrated by means of the examples in (3), given that the thematic role of the subject of the one-place predicate breken in (3b) seems identical to that of the object of the transitive verb breken in (3a).

Therefore, the notions of subject and object will from now on be strictly reserved for, respectively, the nominative and non-nominative arguments in the clause, whereas the notions of internal and external argument will be used for arguments of the verbs carrying certain thematic roles.

External arguments are typically nominal in nature, but this does not necessarily hold for internal arguments (complements) of the verb. The examples in (4) show that complements may also be prepositional or clausal in nature; for each example, we give the complement of the verb in italics and the phrase that is predicated of the subject of the clause in square brackets.

The strings consisting of the verb and its complement are constituents. This can be made clear by means of the complex verb constructions in (5): the primed examples show that the phrases within brackets can be placed in clause-initial position by means of topicalization, which is sufficient for assuming that they are constituents (cf. the constituency test). Since these constituents are headed by a verb, they will be referred to as a verb phrase or verbal projection (VP).

The examples in (6) are somewhat more complex than run-of-the mill transitive clauses like Jan sloeg de hond'Jan hit the dog' in that they contain not only a verbal predicate but also an additional predicate in the form of an adjectival, a prepositional or a nominal phrase. Such examples are therefore said to involve secondary predication: the secondary predicates are italicized and the secondary predications are given within curly brackets. The fact that the secondary predicates are predicated of the direct objects of these examples suggests that the latter do not function as internal arguments of the verbs. The complements of the verbs are instead the secondary predications; these are therefore part of the predicates that are predicated of the subjects of the clauses, which is indicated again by means of square brackets. We will refer to the secondary predicates in (6) as predicative complements or complementives.

This chapter is organized as follows. Section 2.1 starts by discussing more extensively the classification of verbs proposed in Section 1.2.2, sub II, which is based on the number and types of nominal arguments that the verb takes; this section is therefore mainly concerned with arguments that surface as subjects or nominal object(s) of the clause. This is immediately followed by a discussion of secondary predicates in Section 2.2; the reason is that such predicates take a nominal external argument that likewise surfaces as the object or the subject of the clause.
      The traditional definition of (in)transitivity in terms of the number of nominal arguments implies that the term intransitive verb can also be used for verbs like wachten op 'to wait for' that take a prepositional instead of a nominal complement. However, such verbs differ from the core cases of intransitive verbs at least as much as transitive verbs in that they also take an internal argument, which happens to be syntactically realized, not as a noun phrase, but as a PP. We will discuss such prepositional object verbs separately in Section 2.3. Section 2.4 continues by raising the question as to whether there are also verbs taking a (non-predicative) AP-complement. Since clausal complements raise a large number of additional questions they will not be discussed in this chapter: Chapter 5 will be entirely devoted to this topic.
      Section 2.5 concludes the current chapter on argument structure with a discussion of so-called causative psych-verbs like ergeren 'to annoy' and inherently reflexive verbs like zich vergissen 'to be mistaken' ; it will show that these verbs exhibit special behavior in various respects.