Consider the following situation. John goes for a walk and meets someone he has never seen before. Given that domain D is largely determined by agreement between the participants in the discourse, one would assume that the conversation between John and the other person begins with a tabula rasa. However, the fact that John could not utter example (23a) without sounding silly shows that certain entities cannot be introduced into the discourse by means of an indefinite noun phrase: (23a) suggests that there is more than one sun that could be relevant in this context, and this conflicts with the knowledge that we normally ascribe to people. Therefore, the use of a definite noun phrase, as in (23b), is preferred.

This shows that the use of definite noun phrases does not depend entirely on domain D, but may also reflect the speaker's intuitions about the extra-linguistic knowledge that can normally be ascribed to individuals in his society. Or, to put it another way, some entities like the sun, the moon, etc. can be evoked in any conversation without being explicitly part of domain D; simply mentioning the sun is sufficient for any speaker to identify the entity to which the noun phrase refers.

Definite noun phrases can also be used as the possessum in inalienable possession constructions of the type in (24). They are special in that their referent need not be part of domain D, i.e. they can be used without any prior mention of Marie’s arms. There is some individual and regional variation in whether a definite article or a possessive pronoun is preferred, but both options seem to be available in standard Dutch; cf. Scholten (2018) for a detailed discussion.

The acceptability of the sentences in (24) with the possessive pronoun haar is not surprising, because the hearer can simply assume that the pronoun haar is coreferential with the inalienable possessor Marie (as indicated by the indices) and thus will be able to identify the intended reference of the noun phrase haar hand/armen as Marie’s hand/arms. It has been claimed that the inalienable possessor also enters into a syntactically encoded coreference relation with the definite noun phrase de hand/armen, which also enables the hearer to identify the intended reference of this noun phrase as Marie’s hand/arms. If this is correct, it suggests that the requirement that the referent of the definite noun phrase must be part of domain D can be waived by syntactic means; cf. Section 19.1.4.2 for further discussion.

Anyone listening to sentence (25) will conclude that the noun phrase de kleertjesthe clothes refers to the baby’s clothes. This is because the verb aankledento dress can be paraphrased as “to put clothes on someone”. The fact that the baby’s clothes are semantically implied by the description of the event in the first conjunct obviously makes it unnecessary to introduce the baby’s clothes with an indefinite noun phrase.

The appeal to the meaning of the verb aankleden does not account for the fact that the noun phrase de kleertjes in (25) can be replaced by the noun phrase de luiersthe diapers, as in (26). After all, the verb aankledento dress cannot be paraphrased as “to put a romper on someone”.

The fact that the definite article is acceptable in the second conjunct shows that language users have richly structured schemata for certain events. Speakers of Dutch know that babies usually wear diapers, so the event of dressing a baby typically evokes the idea of diapers, which therefore need not be introduced by an indefinite noun phrase. These structured schemata are available not only for events, but also for entities. Speakers of Dutch know that a wedding involves a bride and groom, a best man, a civil servant, etc. Therefore, these entities do not need to be introduced by an indefinite noun phrase, but can be referred to directly by a definite noun phrase, as in (27a). Similarly, for many people the mere mention of a house is enough to evoke an image of a building with a garden, a front door, a chimney, etc., and as shown in (27b) these entities can be referred to directly by a definite noun phrase.

The acceptability of examples like (26) and (27) is, of course, due to the fact that parts of the speaker’s and listener’s conceptions of reality are culturally determined, and therefore have sufficient overlap to invoke the desired inferences in these examples.

Occasionally, however, inferences are not socially determined. The use of a definite noun phrase must then be seen as an invitation to the listener to establish a relation between the referent of the definite noun phrase and a known entity in domain D. Consider an example such as (28a). Although it is not typically assumed that houses have dogs, the listener is invited to connect the referent of the noun phrase de hond with the previously mentioned house (or, alternatively, with Jan). The most likely interpretation is that the dog lives in the house (or that Jan has a dog with him). Replacing the definite noun phrase de hond with an indefinite one, as in (28b), would not force the listener to adopt such an interpretation; in this case, the referent of een honda dog may just as well be completely unrelated to the referents in domain D.

This brief discussion of the use of definite and indefinite noun phrases shows that a simple description in syntactic and/or semantic terms is not possible. It is not the case that entities are always introduced into domain D by the use of indefinite noun phrases. They can also be evoked by the lexical meaning of words or be made available by common knowledge, including generally available structured schemata of events and entities. The most we can say is that the use of a definite noun phrase indicates that the speaker assumes that the listener is able to assign the intended referent a proper place in domain D by associating it with some referent that is part of domain D. A full description of the distribution of indefinite and definite noun phrases must therefore appeal to notions from linguistics, semantics, pragmatics, and cognition. Since this would clearly take us too far afield here, we refer the reader to Keizer (1992b:§5) and Alexiadou et al. (2007: part II) for good overviews of the contributions these fields have made.