If the subject in (142b) is truly an internal recipient argument, we predict that er-nominalization of the presumed undative verb krijgen is excluded, since this process requires an external argument; cf. the generalization in (60a) in Section 2.1.2, sub IIG. Example (145a) shows that this prediction is correct. The verb krijgen differs in this respect from the verb ontvangento receive, which, although semantically close, has a subject that is more agent-like, in the sense that it is more actively involved in the event.

We can therefore expect that the main verbs in (145) will also behave differently in imperatives, and the examples in (146) show that this expectation is borne out: only the more agent-like verb ontvangen can occur as an imperative.

According to generalization (60d) in Section 2.1.2, sub IIG, the presence of an external argument is a necessary condition for passivization; this would correctly predict that passivization of (142b) with the presumed undative verb krijgen is excluded. The examples in (147) show that krijgen is again different from the verb ontvangen, which allows passivization; this verb must therefore be analyzed a regular transitive verb.

Note that Haeseryn et al. (1997: §2.2.3, sub 3) claims that ontvangen does not allow passivization; this is clearly incorrect, as can be seen from the fact that a Google search (February 27, 2024) for the string [werd te laat gekregen/ontvangen] was received too late resulted in 46 hits for ontvangen, but none for gekregen. The claim that ontvangen must be classified as a regular transitive verb therefore stands. However, the examples in (145) and (147) are not conclusive for classifying krijgen as an undative verb, since we know that not all verbs with an external argument allow er-nominalization and that there are several additional restrictions on passivization; cf. Section 3.2.1. However, there is additional support for the hypothesis that the subject of krijgen is a derived recipient-subject.

The hypothesis that the subject of krijgen is a derived recipient-subject can explain the fact that the more or less idiomatic double object construction iemand de koude rillingen bezorgento give someone the creeps in (148a) has a counterpart in (148b) with krijgen. This would be entirely coincidental if Jan were an external argument of the verb krijgen, but it follows immediately if it originates in the same position as the indirect object in (148a); it also explains why the transitive verb ontvangen cannot be used in this context.

Another argument for the assumption that krijgen has an recipient-subject is that krijgen can be used in inalienable possession constructions with a locative PP such as op de vingers. The nominal part of the PP of such constructions, which typically refers to a body part, functions as a possessee and its possessor is usually expressed by a dative noun phrase. This means that (149a) expresses the same meaning as (149b), in which the possessive relation is made explicit by the possessive pronoun haarher (which makes the dative possessor optional). Cf. Section 3.3.1.4 for further discussion.

Subjects of active constructions do not normally function as inalienable possessors of the nominal part of such locative PPs: an example such as (150a) cannot express an inalienable possession relation between the (underlying) subject Jan and the nominal part of the PP, which makes it pragmatically odd (at least when the context provides no information about the possessor of the body part). To express inalienable possession between the subject and the body part, the simplex reflexive object pronoun zich must be added, as in (150b).

Note that the reflexive pronoun in (150b) is most likely dative (rather than accusative). Of course, this cannot be seen by examining the form of the invariant weak reflexive in (150b), but it can be made plausible by examining the structurally parallel German examples in (151), in which the possessor appears as a dative pronoun; cf. Broekhuis et al. (1996) for a detailed discussion.

The crucial observation for our present discussion is that the subject of the verb krijgen is an exception to the general rule that subjects of active constructions do not function as inalienable possessors of a body part embedded in a locative PP: it can be understood from the fact that the subject Marie is interpreted as the inalienable possessor of the noun phrase de vingers in example (152a). This follows immediately if (i) inalienable possessors must be internal recipient arguments, and (ii) the subject Marie (152a) is not an external argument but a derived recipient-subject. Example (152b) is added to show that, as in (149), the inalienable possession relation can be made explicit by the possessive pronoun haarher.

Note that the inalienable possessive constructions in (149) and (152) all allow an idiomatic reading comparable to English to give someone/to get a rap on the knuckles, i.e. “to reprimand/be reprimanded”. This would be entirely coincidental if Marie were an external argument of the verb krijgen, but it follows immediately if it originates in the same position as the indirect object in (149); cf. the discussion of example (148).

For the sake of completeness, we performed a Google search (February 24, 2024) on the string [kreeg/ontving een tik op de vingers], which showed that the verb krijgen is once more different from the transitive verb ontvangen: the search returned 146 unique hits with kreeg, but none with ontving.

The hypothesis that krijgen is an undative verb is particularly interesting in view of the fact that it is also used as an auxiliary in the so-called krijgen-passive, in which not the direct but the indirect object is promoted to subject. This can be seen in the following examples: (153b) is the regular passive counterpart of (153a), in which the direct object is promoted to subject; example (153c) is the krijgen-passive counterpart of (153a), which involves the promotion of the indirect object to subject.

The obvious question raised by the passive constructions in (153b&c) is what determines which of the two internal arguments is promoted to subject. Since worden is clearly an unaccusative verb (e.g. it takes the auxiliary zijn in the perfect), the hypothesis that krijgen is an undative verb suggests that it is the auxiliary verb that is responsible: if the auxiliary is unaccusative, the direct object of the corresponding active sentence cannot be assigned accusative case and must therefore be promoted to subject; on the other hand, if the auxiliary is undative, the indirect object cannot be assigned dative case and must therefore be promoted to subject. Assuming that passive participles cannot assign case (cf. Section 3.2.1), the case assignment in the two types of passive construction is as shown in (154)

The contrast between (155a) and (155b-c) shows that krijgen, but not hebben and houden, can take a van-PP that seems to express an agent. The percentage sign in (155b) is added to express that some speakers accept this example with the van-PP, although this shifts the meaning of hebben towards that of krijgen; a more or less idiomatic example of this type is Marie heeft dat trekje van haar vaderMarie has/got that feature from her father.

The contrasts in (155) can be related to the meanings expressed by the three verbs: the construction with krijgen in (155a) expresses that the theme het boek has changed position, with the referent of the complement of the van-PP referring to its original and the subject of the clause referring to its new location. This suggests that the van-PPs express a source rather than an agent (as previously suggested). If this is the case, the fact that the van-PP in sentence (155b) is not possible with hebben is due to the fact that this verb does not denote transfer, but expresses possession. Something similar applies to sentence (155c) with houdento keep in (155c), which explicitly expresses that the transfer of the theme is not in order.

This subsection discusses data suggesting that hebben is an undative verb, on a par with krijgen. The first thing to note is that hebben does not allow er-nominalization, which distinguishes it from the verb bezittento possess, which is semantically very close to it. The contrast between (156a) and (156b) may again be related to the fact that the subject of the latter is more agent-like. This is supported by the fact that the verb hebben can be used with individual-level predicates such as grijs haar hebbento have gray hair or with non-control predicates such as de griep hebbento have the flu, whereas the verb bezitten cannot: Jan heeft/*bezit grijs haarJan has gray hair; Jan heeft/*bezit de griepJan has the flu.

For completeness’ sake, note that there is a noun hebberd which is used to refer to a greedy person. This noun is probably lexicalized, as is clear not only from the specialization in meaning, but also from the fact that it is derived with the unproductive nominal suffix –erd, and that it does not inherit the theme argument of the input verb: een hebberd (*van boeken).

Second, hebben is like krijgen in that it cannot be passivized. Note that this is also true for the verb bezitten, which was shown to be a regular transitive verb in (156b). This again shows that passivization is not a necessary condition for a verb to be transitive.

For the sake of completeness, note that examples such as Marie werd bezeten door een kwade geestMarie was possessed by an evil spirit are quite common, but probably irrelevant, since such cases usually have no active counterpart; cf. Een kwade geest bezat MarieAn evil spirit possessed Marie. Probably bezetenpossessed functions here as an adjectival predicate, which is supported by the fact that it can also occur with the copula-like verb raken: cf. Marie raakte bezeten door een kwade geestMarie became possessed by an evil spirit.

Third, besides the idiomatic example in (148a), repeated as (158a), we find (158b) with a similar meaning. This would be entirely coincidental if the subject were an external argument of the verb hebben, but expected if it is an recipient-subject.

Finally, like the subject of krijgen, the subject of hebben can be used as an inalienable possessor of the nominal part of a locative PP. This would again follow if (i) inalienable possessors must be recipient arguments and (ii) the subject Peter in (159b) is not an external argument but an recipient-subject.

The verb houdento keep in (160a) seems to belong to the same semantic field as hebbento have and krijgento get, but expresses that there is no transfer of the theme argument. The claim that houden is an undative verb is supported by the examples in (160b-d), which show that er-nominalization and passivization are excluded, and that the subject of this verb can act as an inalienable possessor.

However, several possible problems with the assumption that houden is an undative verb raise their head. First, there are cases of er-nominalization such as (161b). However, these cases are special, because the corresponding verbal construction does not occur with the intended meaning, and we can therefore conclude that we are dealing with an idiomatic noun (constructed for marketing purposes) .

Second, the (a)-examples in (162) show that there are constructions with houden that do allow passivization; the deviant behavior of these examples may be due to the fact that we are dealing with an idiomatic expression with more or less the same meaning as the transitive verb bespiedento spy on, which also allows passivization. Somewhat surprisingly, the corresponding construction with krijgen behaves as expected in that it does not allow passivization; this is illustrated in the (b)-examples.

Third, er-nominalization and passivization are possible with the verb houden if this verb is used in reference to livestock, as in (163). However, the fact that the object in (163a) can be a bare plural (or a bare mass noun) suggests that we are dealing with a semantic (i.e. syntactically separable) compound verb, comparable to a particle verb (although it should be noted that the bare noun can be replaced by a quantified indefinite noun phrase such as veel schapenmany sheep).

We will leave the special cases in (161) to (163) for future research and rely on the more general findings in (160), which suggest that houden is an undative verb.