This chapter will discuss the semantic and syntactic behavior of the determiners. In the current generative framework, it is generally taken for granted that a determiner defines its own endocentric projection in the structure of the noun phrase; cf. Abney (1987). It is taken to be the head of a so-called determiner phrase (DP), which is located on top of the projection of the head noun, NP. Schematically, example (1a) can be represented in labeled bracketing as in (1b), or as the tree diagram in (1c). Recall that we use the term “noun phrase” in a neutral way, whereas the terms DP and NP are used to refer to the substructures marked as such in (1b&c).

1

a.		de	blauwe	auto
		the	blue	car

b.

[DP [D de] [NP blauwe auto]]

c.

The DP structure of noun phrases formally recognizes the fact that it is the determiner which is the syntactic head, and as such determines the referential/quantificational properties and the syntactic distribution of the noun phrase as a whole (apart, of course, from the semantic selection restrictions imposed by, e.g., the verb on the denotation of the head noun of its complement).
      There are two main types of determiners: articles and pronouns, which will be discussed in 5.1 and 5.2, respectively. Of course, noun phrases can also be introduced by a cardinal numeral or a quantifier like sommige'some'; these will not be discussed in this chapter, but in Chapter 6. Under the generally accepted assumption that a phrase has exactly one head, the claim that demonstrative and possessive pronouns are determiners, and hence occupy the D position of the DP, can be motivated by the fact that they are in complementary distribution with the articles, as well as with each other. It is impossible to simultaneously have, for instance, an article and a demonstrative pronoun in one DP. This is illustrated in (2) for combinations of two types of non-interrogative determiners; obviously, examples containing all three types of determiners are excluded as well.

Note in passing, however, that the claim that articles and pronouns are both determiners is weakened by the fact that this does not seem to be universally valid: some languages, like Italian or Greek, do not exhibit the complementary distribution of the Dutch articles and possessive/demonstrative pronouns; cf., e.g., Alexiadou et al. (2007: 93).
      Personal pronouns are also included in this chapter because there are various reasons to consider them determiners as well. From a semantic point of view they resemble the determiners in having primarily a referring function: their descriptive content is limited and certainly does not exceed that of the possessive pronouns. Furthermore, if it is assumed that personal pronouns are within the NP-domain, it cannot readily be accounted for that they cannot be preceded by an article or a demonstrative/possessive pronoun, whereas this follows immediately if they occupy the D-position; see, e.g., Longobardi (1994) and Alexiadou et al. (2007: 211/9) for more empirical support from Italian and Serbo-Croatian in favor of the claim that personal pronouns are determiners.
      Before we begin discussing the articles in 5.1, we want to make some general comments on the structure of the noun phrase in (1). The NP in this structure can be said to determine the denotation of the noun phrase: it acts like a predicate, and can therefore be represented as a set of entities which have in common that they satisfy the description provided by the NP; the NP blauwe auto'blue car' denotes the set of entities that have the properties of being a car and being blue; cf. Section A1.3. Determiners, on the other hand, are normally used to determine the reference of the noun phrase. A definite determiner like de in de blauwe auto'the blue car', for example, expresses that the denotation set of the NP blauwe auto'blue car' contains exactly one entity and that it is this entity that the speaker refers to. The fact that a definite determiner has this meaning leads us to the relation between language and reality.
      The relation between language and reality has given rise to ardent debates, and we will certainly not try to resolve here all the issues that have been brought up. We want to point out, however, that many of the problems that have been discussed in these debates find their origin in the assumption that language is directly related to reality. Consider example (3). Given the generally accepted idea that a singular noun phrase containing a definite determiner like de refers to a unique entity, this example is problematic because the noun phrase de Nederlandse president'the Dutch president' does not refer to an entity in the real world, which means that at first sight this example cannot be assigned a truth value.

Another problem is that it seems beyond the powers of the language user to determine what reality actually is; if we want to make objective statements about reality, we have to go beyond our personal experience and enter the domain of science. The language user therefore does not refer to reality directly, but to his internalized conception of reality, which is invoked in his speech acts. For example, a sentence like (3) can be seriously uttered by anyone who has the erroneous belief that the Dutch prime minister is the president of the Netherlands, and, consequently, the speaker will also assign a truth value to this sentence. In other words, by assuming that a noun phrase does not refer to entities in the material world but to entities in the speakerʼs internalized conceptualization of the material world, the reference problem in (3) dissolves.
      Next, the question arises of what a language user is actually doing when he or she utters an example such as (3). The definite article de expresses that in the speakerʼs conception of reality there is a unique entity that has the property of being the Dutch president, and it is this entity that the property of being a gifted man is predicated of. Of course, this conception of reality may clash with the conception of reality held by the listener, who is then likely to correct the speaker by saying that the person in question is not the president but the prime minister. What this shows is that language users do not invoke knowledge of reality (which they may be assumed to lack), but knowledge of their internalized conceptualization of reality. Although the conceptualizations of reality may differ among individuals, there is generally sufficient overlap to make communication more or less successful: in fact, one might even argue that the goal of communication is to eliminate discrepancies between the conceptualizations of reality held by the participants in the discourse, by correcting or updating the knowledge encoded by these; cf. Verhagen (2005).
      Often, the participants do not even exploit their full conceptualization of reality in discourse. This can be easily demonstrated by means of the noun phrase de blauwe auto'the blue car'. As we have claimed above, this noun phrase expresses that the set denoted by the NP blauwe auto'blue car' contains exactly one member. Since we can safely assume that every language user is aware of the fact that the set denoted by blauwe auto contains an extremely large number of entities, this knowledge is clearly not relevant. The participants in the discourse rather have a tacit agreement on the question what entities are relevant for the discussion in question; this limited set of entities under discussion is often referred to as the domain of discourse or domain D, which may be assumed to consist of the shared knowledge of the participants on the topic under discussion. And de blauwe auto expresses that, in this limited domain, the set of cars contains just a single member.
      To summarize, we have claimed that there is no direct relation between language and reality. Instead, the two areas are only indirectly related by means of the language userʼs internalized conception of the “real world”, and the assignment of truth-values is only based on the (correct or incorrect) knowledge encoded in this conception. In conversation, the assignment of truth-values is further restricted by domain D, the shared knowledge of the participants on the topic of discussion. This view on the relation between language and reality will be adopted in the discussion in the following sections.