A common exception to the general rule is when the antecedent of die is mentioned in the immediately preceding discourse, as in example (514a), repeated here as (526a). We have added indices to this example to indicate the intended interpretation of the distal demonstrative pronouns. Note that these are common demonstratives in that their form and number properties are determined by their antecedent. However, they are not easily replaced in informal speech by their proximate counterparts deze and dit; cf. onzetaal.nl/taaladvies/deze-die-dit-dat/.

Example (527) shows that, unlike referential personal pronouns, the distal demonstrative die cannot be bound; it is disjoint in reference to any c-commanding antecedent in the same sentence, and thus behaves in this respect like the referential expression het meisjethe girl. See Chapter 23 for further discussion of the binding properties of referential expressions, and Section 19.2.2.1, sub V, for similar but somewhat less strict restrictions on the interpretation of the possessive pronoun diens.

The distal demonstrative die differs from the referential personal pronouns in contrastivity. If the question in (528a) is answered by (528b), it strongly suggests that the person answering the question has seen another person who might be relevant in the given context; a natural continuation would be a report of what that other person said. The answer in (528b'), on the other hand, is neutral in this regard.

Recall that weak object pronouns cannot occur in the clause-initial position: die in (528b) can therefore only be replaced by the strong pronoun haar. Replacing the weak pronoun ’r in (528b') with the distal demonstrative die again seems to trigger a contrastive reading (although the effect is less clear in this case).

The distal demonstrative die and the referential personal pronouns exhibit different anaphoric behavior when they occur unstressed in the clause-initial position; cf. Haeseryn (1997:307-8) and Van Kampen (2010a). In the examples in (529), for instance, unstressed die cannot refer to the subject of the preceding clause, but must refer to the direct object, whereas referential pronouns are not restricted in this way.

Before discussing the anaphoric behavior of the distal demonstrative pronoun die, we want to emphasize that we are dealing with a normal distal demonstrative, as is clear from the fact that its form is sensitive to the number and gender features of its antecedent. This is illustrated in (530) by the neuter noun boek(en)book(s). For completeness’ sake, note that dat in (530a) can also refer to the predicate het boek lezenreading the book; cf. Section 19.2.3.1, sub III.

However, the distal demonstrative die in (529) is special in that it cannot easily be replaced by its proximate counterpart dezethis/these in informal speech, although this is less clear than in the cases discussed above, since the proximate demonstrative dezethis/these can be found in more formal texts; cf. Subsection III for examples.

The anaphoric behavior of the distal demonstrative die differs from that of the referential personal pronouns hij/zijhe/she, as illustrated in (529). This difference is not related to the syntactic function of the antecedent, but rather depends on the role of the antecedent in the information structure of its clause. Consider the following discourse chunk, where the (b)-sentences are possible continuations of (531a). The two (b)-examples differ in that in the first, but not in the second the distal demonstrative in the second conjunct can refer to the subject de leraarthe teacher of the first conjunct.

The contrast seems to be related to the fact that de leraar is preferably interpreted as part of the focus (new information) of the clause in the first conjunct in (531b), but as part of the presupposition in (531b'). Van Kampen (2010a) accounts for this by claiming that unstressed distal demonstratives are used as a topic-shift devices: they take an element from the focus of the previous sentence and turn it into a new topic for the ongoing discourse. This would explain why the unstressed distal -new topic- pronouns often occur as subjects, because discourse topics typically occur as subjects of their clauses. However, it is difficult to determine whether distal demonstratives with other syntactic functions can also be used as topic-shifting devices. Possible cases are the discourse chunks in (532), where the (b)-examples are again intended as continuations of (532).

The fact that the discourse chunk can be followed by a longer story about the antecedent of the distal demonstratives suggests an affirmative answer to the question whether distal demonstratives with other syntactic functions can also be used as topic-shifting devices. However, we cannot support this with phonological evidence, for the simple reason that topicalized elements are normally accented (as is clear from the fact that daar in (532b') cannot be replaced by unstressed er). This means that examples of the type in (532b&b') can, in principle, be ambiguous between a deictic and a topic-shift reading of the distal demonstrative. Since we currently have no means of distinguishing between the two readings, we leave this as a topic for future research.

That the information structure is crucial for the question whether unstressed die can be used is also clear from the fact that it cannot take a referential personal pronoun as antecedent, since such pronouns are never part of the focus of the clause and are therefore unsuitable as antecedents for the topic-shifting pronoun die. This is shown by the two examples in (533), which should be compared with (529b) and (531b).

Information structure may also be the key to the apparent free variation of the distal demonstrative and referential personal pronouns in (529b), since scrambling of intonationally unmarked noun phrases is restricted to noun phrases that are part of the presupposition of the clause (cf. Section 22.1.3). Van Kampen (2010a) claims that for this reason such scrambled noun phrases cannot function as antecedents for the distal demonstrative die. If scrambling can be applied string-vacuously in (529b), this implies that the free variation is only apparent and due to the co-occurrence of the two structures in (534): die is used when the antecedent Jan belongs to the focus of the clause and occupies its base position, as in (534a), while hij is used when it belongs to the presupposition of the clause and is scrambled, as in (534b).

However, the examples in (535) show that the contrast is not as sharp as Van Kampen claims: it seems that die is indeed preferred when the antecedent Jan has not been scrambled, as in (535a), but some of our informants also accept the distal pronoun when the antecedent has been scrambled, as in (535b).

The unexpected finding in (535b) need not imply that Van Kampen’s claim is wrong, because the judgments on (535b) may be somewhat blurred by the fact that this example becomes fully acceptable with die when either the antecedent Jan or the distal demonstrative die is assigned (contrastive) accent; in the former case, the proper noun Jan receives a contrastive/emphatic focus reading, which cancels the implication that it belongs to the presupposition of the clause; in the latter case, the distal demonstrative does not function as a topic-shift device, and consequently the requirement that its antecedent be part of the focus of its clause does not hold. Given this complication, it is not clear whether examples like as (535b) should be considered counterexamples to Van Kampen’s claim; for this reason, we will also leave this issue to future research.

The examples in (536) show that the antecedent of the distal pronoun die cannot be embedded in a potential antecedent of the distal demonstrative: in (536a), for instance, the antecedent of the distal demonstrative is Peter’s mother, not Peter. That this is indeed due to the fact that the proper noun is embedded in another noun phrase is clear from the unacceptability of (536b): since the noun phrase embedding the proper noun Peter cannot function as the antecedent of the demonstrative for reasons related to our knowledge of the world, it seems unlikely to give a pragmatic account of the judgments in (536a) by saying that de moeder van Peter is simply a more prominent antecedent for die than Peter.

De Vries (2010) has noted that in some cases contrastive accent on the embedded noun phrase can make examples like (536b) perfectly acceptable, but this may again be related to the fact that Peter then receives a contrastive/emphatic focus reading that cancels the implication that it belongs to the presupposition of the clause; cf. Van Kampen (2010b).

The discussion in this subsection supports the claim that unstressed die functions as a topic-shift device in the sense that it takes (part of) the focus of the preceding clause as its antecedent and presents it as the new discourse topic. Referential personal pronouns, on the other hand, signal that the current discourse topic is maintained. See Van Kampen (2010a) for further discussion.

Left-dislocation constructions such as (537) resemble topicalization constructions, but differ from them in that they do not involve movement of the left-dislocated phrase. This phrase is external to the clause and is associated with a resumptive pronoun: when the resumptive pronoun is moved into the clause-initial position, as in (537a), it preferably takes the form of a distal demonstrative; when it occupies the middle field of the clause, as in (537b), the referential personal pronoun gives the best result. The demonstrative in constructions such as (537) is usually die, unless the antecedent is clearly neuter, as in Dat meisje[+neuter], dat ken ik nietThat girl, I do not know her.

However, there is no agreement between the left-dislocated element and the demonstrative when the demonstrative functions as the logical subject of a nominal predicate; whereas the demonstrative must agree in gender with the dislocated element in the primeless examples in (538) with an adjectival predicate, it cannot agree with it in the primed examples with a nominal predicate.

The fact that the demonstrative always has the [+neuter] form dat in copular constructions with a nominal predicate is clearly related to the fact that dat can also occur in such constructions when there is no left-dislocated antecedent, as in the examples in (539).

The examples in (540) show that similar facts can be found in the vinden-construction; (540a) shows that the pronoun die is used in the case of an adjectival predicate, and the (b)-examples show that the pronoun dat is used in the case of a nominal predicate.

The examples in (541) show that the demonstrative dat would also be used if instead of the subject the predicate was left-dislocated. This is illustrated for the nominal predicate of a copular construction in the primeless examples, and for the nominal predicate of the vinden-construction in the primed examples.

But of course we cannot appeal to this fact to account for the contrast between the primeless and primed examples in (538), since the demonstrative dat is also used with the left-dislocated adjectival predicate aardig in (542).

The distal demonstrative pronoun die has a pejorative connotation when used deictically to refer to a [+human] entity. This subsection has shown that this is not the case when it is used anaphorically, i.e. when it has an antecedent in the immediately preceding sentence. We discussed three cases: cases in which die simply refers to an antecedent in the preceding sentence, cases in which it functions as a topic-shift device, and cases in which it functions as a resumptive pronoun in a left-dislocation construction.