An indefinite subject in an expletive construction can be either specific or non-specific. The most plausible reading of (9a) is the non-specific one, in which the speaker is unable to identify the person in question, while the most plausible reading of (9b) is the specific one, in which at least the speaker is able to identify the person in question in the discourse. These examples also show that the non-specific indefinite subject in (9a) must follow the adverbial phrase, i.e. it cannot be scrambled. The specific indefinite subject in (9b), on the other hand, can more easily be placed before the adverbial, indicating that it can at least marginally be scrambled. In the case of a quantifier such as iemandsomeone in (9c), A-scrambling is even the normal means of distinguishing between the two interpretations: if the quantifier follows the adverbial, it is preferably interpreted as non-specific, whereas if it precedes the adverbial, it is usually interpreted as specific.

The examples in (10) show that non-specific indefinite subjects are not commonly used without the expletive, whereas the specific ones can be used without the expletive er. Note that we have put aside the fact that in some varieties of Dutch, examples such as (10a) are also acceptable without the expletive; we are only discussing the varieties that do not allow this.

The distribution of the expletive er is sensitive not only on the nature of its subject, but also on that of other material in the clause. For example, Bennis (1986: §3.5.3), Koeneman (2000: §4.5) and Wesseling (2018: §1.5.5) argue that the presence of other presuppositional material is relevant: every sentence must have a presupposition, and the expletive er is a dummy presupposition marker that must be used in its absence. Consider the (a)-examples in (11). In (11a) the spatial adverbial phrase voor mijn huis follows the indefinite subject and is interpreted as part of the focus of the clause: since there is no other element functioning as presupposition, the expletive er must be realized. However, if the adverbial precedes the subject, as in (11b), it can (but does not have to) function as a presupposition. Therefore, the expletive can be optionally omitted, depending on the intended reading of the subject; cf. Grondelaers (2009) for a discussion of similar examples with the expletive in sentence-initial position.

Note in passing that the locational proforms hierhere and daarthere in (12) are usually presuppositional in nature, which means that the use of the expletive er yields a marked result. Wesseling (2018: §3) takes the acceptability of (12b) without er as evidence that the locational proforms hier and daar can also be used as expletives in standard Dutch. This conclusion is based on the assumption that the canonical subject position to the immediate right of the finite verb in main clauses (or the complementizer in embedded clauses) must be filled. However, this claim runs into problems with examples such as (11b), where er is also optional, as well as the fact that nominative subjects can follow dative objects in nom-dat and passivized ditransitive constructions without the need to insert an expletive in the canonical subject position; for discussion and examples, see Section V2.13, sub IIF. We therefore adhere to the traditional assumption that only the weak R-word er and its allomorphs dʼr and ʼr can be used as an expletive in standard Dutch, while acknowledging that the strong form daar (but not hier) can be used as an expletive in other Dutch and German varieties; cf. Barbiers et al. (2005: §2.4) and Hartmann (2008: §4). We believe that the results from the acceptability judgment task reported by Wesseling are consistent with Bennis’ proposal that expletive er is a dummy presupposition marker; we leave this for further research.

Grondelaers & Brysbaert (1996) demonstrate, based on a limited corpus (a Dutch and a Flemish novel), that the expletive is often practically impossible with a spatial PP in clause-initial position. Example (13a) illustrates this with a constructed example; such constructions with an expletive are marked and are not common in the corpus. This might also follow from the proposal in Bennis (1986), since topicalized constituents are often discourse topics, and can be supported by the finding in Grondelaers & Brysbaert that the expletive is possible (and in fact much preferred) when the topicalized PP is preceded by a focus marker such as ook also, since this blocks the topic reading. Note that the diacritics in (13) represent our own judgments.

Grondelaers & Brysbaert (1996) does not explain the distribution of expletive er in terms of information structure (i.e. the presence of a presupposition): they claim that the pragmatic function of er is to announce that the sentence “introduces a still unknown but communicatively important entity”. Spatial (and, to a lesser extent, temporal) adverbial PPs may have a similar function when used as domain adverbials (cf. Section V8.2.2, sub XI): they provide an anchoring point that helps the hearer single out the intended referent. However, not all domain adverbials are equal in this respect: the less specific the spatial PP is, the more likely it is that the expletive er will be used; cf. (14a). The main verb also seems to play a role in that the expletive er seems to be obligatory if the verb is the copular zijnto be; cf. (14b). The attested examples from NRC Handelsblad given in the article also suggest that the nature of the subject may also be relevant, since the subjects of these cases generally refer to abstract entities. Again, the diacritics in (14) represent our own judgments.

From the above discussion we can conclude that the distribution of expletive er is not a purely syntactic matter, but partly depends on pragmatic considerations; cf. see Grondelaers & Brysbaert (1996), Grondelaers et al. (2002/2009), Grondelaers (2009) and the references cited there for further discussion.

The examples in (15) show that in transitive constructions the nature of the direct object is also relevant. Although, for as yet unclear reasons, (15a) may be somewhat marked on a non-specific interpretation of the subject, it seems that this reading requires the presence of the expletive, which is consistent with the fact that the indefinite object een boek is preferably interpreted as non-specific. In (15b), which again may be somewhat marked on a non-specific interpretation of the subject, the expletive can easily be left out; this seems to be related to the fact that the definite object het boek can (but need not) be interpreted as part of the presupposition of the clause. In (15c), which easily allows for a non-specific interpretation of the subject, the expletive cannot be used; this is due to the fact that the object pronoun het must be interpreted as part of the presupposition of the clause; cf. Koeneman (2000), Koster & Zwart (2000), and Broekhuis (2007/2008) for more discussion of these so-called transitive expletive constructions.

The data in (10) to (15) have shown that in the absence of certain material (e.g. a presupposition or a domain adverbial), the expletive must be realized when the subject is non-specific. This does not hold, however, for non-specific indefinite subjects modified by certain attributive adjectives or restrictive relative clauses. The primed examples in (16) show that they can be placed in the regular subject position, i.e. the position occupied by the expletive in the primeless examples; we have added the particle nog to these examples to make it visible that the subjects in the primeless and primed examples do indeed occupy different positions. Apparently, the attributive adjective/relative clause makes the noun phrase sufficiently specific to occupy the canonical subject position.

The examples in (17) illustrate that modified noun phrases show a clear preference for being placed in the regular subject position when the clause contains sentence negation. Note that these examples should not be confused with examples such as Er is een tachtig jaar oude man niet goed gewordenAn eighty-year-old man has become unwell, where the negative adverb is construed with the adjectival predicate. In these cases we are probably dealing with constituent negation (niet goednot well ≈ onwelill).

A second exception concerns examples in which the head of the indefinite subject is contrastively stressed. Thus, while an indefinite subject such as een man in (18a) does not normally occur without the expletive, it may occur without the expletive if the noun man is contrastively stressed, as in (18a'). If the noun phrase contains a numeral or quantifier, as in (18b), the expletive may also be omitted if the numeral/quantifier is contrastively stressed, although in this case the noun phrase is likely to receive a partitive reading; cf. De Hoop (1992).

Finally, it should be noted that there are regional differences in the licensing of er in expletive constructions with an adjunct in sentence-initial position, like those in (10), (11), and (16); cf. Grondelaers et al. (2008), briefly summarized in Grondelaers et al. (2020:75-6), for some differences between Dutch and Belgian speakers.

That non-specific indefinite subjects usually show a preference for being introduced by an expletive is also clear from the fact that indefinite subjects induce special semantics when they occur in the regular subject position. Consider the two (a)-examples in (19). Example (19a), in which the determiner is normally pronounced as [ən], simply asserts that some student has been arrested. When the indefinite subject is placed in the regular subject position, the indefinite article is preferably stressed (i.e. pronounced as [e:n]), so that we cannot directly distinguish whether we are dealing with the article or the numeral éénone. The preferred reading of the primed example is a partitive one: it is claimed that a certain student from a contextually determined set of students has been arrested — the interpretation of the indefinite subject comes rather close to één van de studentenone of the students; cf. Section 19.4, sub I. The (b)-examples in (19) show that the same phenomenon can be found in cases that unambiguously involve a numeral or a quantifier.

As noted above, it is not immediately clear whether we are dealing with the indefinite article or the numeral éénone in (19a'). The fact illustrated in (20) that the indefinite plural subject studenten cannot occur in the regular subject position suggests the latter.

This supports our earlier conclusion in Subsection I that unmodified non-specific indefinite noun phrases do not normally occur in the regular subject position, but remain in their VP-internal base position.

Another difference between the expletive construction and the construction with the indefinite subject in the regular subject position is that in the former the noun phrase can never be interpreted generically. Consider the examples in (21): the indefinite subject in the expletive construction in (21a) cannot be interpreted generically, whereas example (21b) must be interpreted generically. The difference can be made clearer by putting the examples in the past tense: (21a') is perfectly acceptable and expresses that it used to be the case that a hippo was lying in the water; (21b'), on the other hand, is strange because it suggests that hippos in general have changed their habit of lying in the water. Note that (21b') becomes acceptable on a specific or partitive interpretation if we stress een: it used to be the case that a certain hippo or one of the hippos was lying in the water.

The examples in (22) show that the same pattern arises in the case of plural indefinite subjects; ∅ indicates the indefinite null article. Again, the primed (b)-example is not entirely acceptable for non-syntactic reasons, since it suggests that hippos in general have changed their habit of lying in the water.

This section has shown that (unmodified) non-specific indefinite subjects introduced by the indefinite article een/∅ must normally be part of an expletive construction. Specific indefinite subjects, on the other hand, can either be part of an expletive construction or occupy the regular subject position. Finally, indefinite subjects with a partitive or generic interpretation cannot occur in an expletive construction but must occupy the regular subject position.

	in canonical subject position	expletive construction
non-specific indefinite subject	—	+
specific indefinite subject	+	+
partitive/generic indefinite subject	+	—

It should be noted that the general rule that non-specific indefinite subjects headed by an indefinite article do not occur in the regular subject position can be waived in narratives. A story may well begin as in (24), where the function of the noun phrase een man is clearly to introduce a new discourse entity, without the implication that the speaker can uniquely identify the intended referent. The sentence in (24) is only acceptable if the discourse continues with a story about the person sitting in the waiting room.