This chapter deals with R-pronominalization and R-extraction. The first term refers to the pronominalization of the nominal complement of a preposition by a so-called R-word, which must precede the preposition. In (1) we give two examples with the R-word er, which will be glossed as there for reasons that will become clear when we discuss the examples in (6).

Pronominal R-words such as er will generally be referred to as R-pronouns, although we will also use the notation [+R] pronouns in contrast to [-R] pronouns like hij/hemhe/him. The combination of an R-pronoun and its associated adposition will be labeled pronominal PP.

Table 1, which is based on the classification of the pronouns developed in Section N18.2, shows that all non-anaphoric third person [-R] pronouns which can be used as independent arguments have a [+R] counterpart. The fact that these [+R] pronouns all contain an /r/ accounts for their name.

subgroup A		[–neuter]	[+neuter]	R-pronoun
referential		hij/zij ‘he/she’	het ‘it’	er ‘there’
demonstrative	proximate	deze ‘that’	dit ‘this’	hier ‘here’
	distal	die ‘this’	dat ‘that’	daar ‘there’
relative		die ‘who’	dat ‘which’	waar ‘where’
Subgroup B		[+human]	[–human]	R-pronoun
interrogative		wie ‘who’	wat ‘what’	waar ‘where’
quantificational	existential (positive)	iemand ‘someone’	iets ‘something’	ergens ‘somewhere’
	existential (negative)	niemand ‘nobody’	niets ‘nothing’	nergens ‘nowhere’
	universal	iedereen ‘everyone’	alles ‘everything’	overal ‘everywhere’

Table 1 shows that the [-R] pronouns can be divided into four main groups on the basis of the features [±neuter] and [±human]. For the referential, demonstrative, and relative pronouns (subgroup A), the grammatical gender distinction is the most prominent one, although it should be noted that most nouns referring to [+human] entities are also [-neuter]. Nouns with the feature constellation [+neuter] and [+human] are typically diminutives; some exceptions are the noun meisjegirl, which has the formal but not the semantic characteristics of a diminutive, the noun kindchild and a number of nouns with negative connotations like wijfbitch and mensperson. For the interrogative and quantificational pronouns (group B), only the distinction between [+human] and [-human] is relevant. Distinguishing between the groups A and B is useful because Section 36.1 will show that [+R] pronouns are typically, but not exclusively, used as alternative realizations of [-human] pronouns; cf. (2) for examples and Table 2 in Section 36.1, sub VII, for an overview.

In contrast to English, preposition stranding in Dutch cannot arise by extracting a complete noun phrase or a regular pronoun from a prepositional phrase. This means that English constructions like (3a&b) are usually considered ungrammatical in Dutch (although they are easily interpretable), as shown by the unacceptability of the primed examples in (3).

Dutch does not have so-called pseudo-passives either: English (4a) cannot be translated into Dutch as in (4b), but must be rendered by the impersonal passive construction in (4b').

However, preposition stranding is often possible with pronominal PPs. As mentioned above, we will refer to this type of preposition stranding as R-extraction; this form of extraction is illustrated in the primed examples in (5) by wh-movement of the R-word. To indicate that the R-word and the stranded preposition form a semantic unit, we will often italicize the discontinuous pronominal PP in this chapter.

Example (6) illustrates R-extraction for all the R-pronouns in Table 1. We will follow the custom in the linguistic literature to gloss the R-pronouns with an English spatial pro-form. This custom is due to the fact that all R-words can also be used as spatial pro-forms, as shown in the primed examples.

We conclude this brief introduction to R-extraction by discussing a possible problem for our earlier conclusion based on the examples in (3) to (5) that preposition stranding is possible only with R-pronouns. This pertains to the fact that examples such as (7a) are more or less acceptable for some (but not all) speakers in informal speech. There are at least two possible analyses of this construction. The first analysis is given in (7b) and assumes that we are dealing with preposition stranding of the English type in the sense that it is the noun phrase that has been extracted from the PP. The problem with this analysis is that it forces us to develop an independent account of the unacceptability of the Dutch examples in (3) and (4b). Moreover, on this analysis it is rather surprising that the noun phrase bananen cannot be scrambled leftward in the middle field of the clause: *Ik ben (de) bananen dol op; cf. Van der Horst & Van der Horst (1999:269). The second analysis is given in (7c) and assumes that the movement involves an R-pronoun which is subsequently deleted; in essence, we are dealing with a left-dislocation construction with a deleted resumptive pronoun.

The analysis in (7c) has at least four advantages. First, it is compatible with our conclusion, based on the examples in (3) and (4b), that preposition stranding is possible only with R-pronouns. Second, it is supported by the fact that the resumptive pronoun can also be overt: Bananen, daar ben ik dol op. Third, the (b)-examples in (8) show that the preposition met must appear in its stranded form mee, both in the construction with and without daar; cf. Klooster (2001a:324). Note that the percentage sign is used in (8b') for the same reason as in (7a'); there is no R-pronoun.

Fourth, the analysis in (7c) does not predict the acceptability of constructions such as \`1Ik ben (de) bananen dol op, and correctly predicts that PPs that do not allow R-extraction do not license the form of preposition stranding in (7a); cf. Klooster (2001a:324).

Therefore, we adopt the analysis in (7c) and assume that the markedness of (7a) is due to the fact that deletion of the R-pronoun is apparently not a preferred option for many speakers of Dutch. We refer the reader to Klooster (1989/2001a), Van der Horst & Van der Horst (1999:268ff), and Van der Horst (2008) for further discussion.

Before we start our more comprehensive discussion of R-pronominalization and R-extraction, we also want to point out that, besides its use as a spatial pro-form, the R-word er can also function as an expletive in impersonal passives and existential/presentational sentences like (10a&b) and as a licenser of the nominal gap [e] in so-called quantitative er constructions such as (10c).

The fact that er can perform more than one function is important because a single occurrence of er can occasionally express more than one of these functions at the same time. In (11b), for example, er simultaneously performs the function of an expletive, the licenser of the nominal gap in the noun phrase remnant corresponding to drie sigaren in (11a), and the complement of the stranded preposition in. We will discuss examples like these in more detail in Section 36.5.3.

This chapter is organized as follows. Sections 36.1 to 36.3 discuss the restrictions on R-pronominalization and R-extraction: Section 36.1 considers the semantic restriction that, generally speaking, the R-pronoun refers to a [-human] entity; Section 36.2 looks at the lexical restrictions on the adpositions, e.g. that locational but not temporal prepositions can easily be part of a pronominal PP; Section 36.3 concludes with a discussion of some syntactic conditions on R-extraction. Section 36.4 provides examples of idiomatic constructions with pronominal PPs. Finally, Section 36.5 explores an empirical domain that slightly extends the discussion of R-pronominalization and R-extraction, namely the co-occurrence restrictions on the R-words in (6) and (10), as well as erʼs ability to perform more than one function simultaneously, as illustrated in (11b).