In German, inalienable possessors are typically dative phrases; this holds for both locational and non-locational inalienable possession constructions. Although Dutch does not show morphological case distinctions, the same is arguably true for locational constructions such as (185a). To see this, consider the structurally similar example in (186a). That the possessor is not accusative but dative is clear from passive formation: example (186b) shows that in the regular passive it is not the possessor Jan that is promoted to subject, but the noun phrase de boekenthe books; (186c) shows that in the so-called krijgen-passive it is the possessor that is promoted to subject. This is sufficient to conclude that the possessor is an indirect object, and is therefore assigned (abstract) dative case; cf. Section V3.2.1.

Note that example (186c) shows that inalienable possessors can function as subjects if they correspond to an “underlying” indirect object, i.e. an internal argument, which we will call goal for convenience. We will adopt this as the null hypothesis for our discussions in Subsections B and C.

Subsection A has shown that a subject can function as an inalienable possessor if it corresponds to an internal goal argument. Now consider the examples in (187) with the verb gevento give. Example (187b) is excluded because krijgen-passivization of the verb geven is generally impossible. Broekhuis & Cornips (1997) has claimed that this is due to the fact that the intended meaning can also be expressed by the verb krijgen alone, as illustrated by example (187b').

One could argue that (187b') is derived from (187b) by omitting the semantically light participle gegevengiven. If so, we can conclude that this example is fully compatible with the null hypothesis that inalienable possessive subjects are goal arguments. A problem for this analysis is that the dynamic verb krijgen has a more stative counterpart, hebben, which also allows a subject with the function of inalienable possessor; in the examples in (188) it seems impossible to assume the omission of a semantically light participle like gegevengiven, since this would lead to a wrong (dynamic/transfer) meaning.

If we want to maintain the hypothesis that the subjects in the examples in (187b) and (188) are goal arguments, we are led to the conclusion that besides the so-called unaccusative verbs, there is a set of undative verbs which take a goal argument; this goal argument has to be realized as nominative subject, because the verb cannot assign dative case to it. An argument for this view is that verbs like hebben and krijgen cannot be passivized (regardless of the presence of the locational PP): this may be due to the fact that they have no (external) agentive argument.

If the proposed analysis of the examples with krijgen and hebben is on the right track, we can expect there to be more undative verbs: good candidates are the verbs nemento take and houdento keep; these verbs also seem to take subjects with a kind of goal role, exhibit a similar aspectual difference as krijgen and hebben, and resist passivization.

Furthermore, the examples in (191) show that the subjects of these verbs can indeed function as inalienable possessors. However, there is a small hitch: it seems that the passive counterparts of these examples are better than those in (190), which is also clear from the fact that they can be easily found on the internet.

This means that we can only maintain our hypothesis if we assume that the verb nemen with a PP-complementive is ambiguous between an agentive transitive and a less agentive undative form. Perhaps this can be supported by the fact that the particle verb meenemento take away in (192) is clearly agentive and easily allows passivization. we will leave this issue to future research.

It is not expected that the direct object of a locational construction can act as an inalienable possessor, since it functions as a theme (and not as a goal), i.e. it refers to an entity that is located with respect to the possessum: the direct object een handa hand in (193), for example, is the entity that is given a certain location with respect to the nominal complement of the locational PP, de schouderthe shoulder.

Nevertheless, there is a class of verbs that systematically allow what appears to be their direct object to act as an inalienable possessor: the common denominator of these verbs is that they imply some form of bodily contact between the subject and the direct object of the clause; cf. Broekhuis et al. (1996). Two examples are given in (194a) and (195a); the (b) and (c)-examples show that these examples allow regular passivization, but not semi-passivization, which should be sufficient to show that we are dealing with a direct and not an indirect object in the (a)-examples.

Nevertheless, there are reasons to doubt the conclusion that we are dealing with true direct objects in (194a) and (195a). First, it should be noted that the verbs that enter such inalienable possession constructions are generally denominal and can be paraphrased by the semantically light verb geven followed by an indefinite noun phrase. Some examples are given in (196); cf. Section 18.1.4.1, sub B, for a discussion of the semantic difference between the denominal and the light verb construction.

This implies that the examples in (194a) and (195a) are more or less equivalent to those in (197), where the inalienable possessor functions as indirect object. If we now assume that the light verb geven has a phonetically empty counterpart, which triggers so-called incorporation of the direct object, by which the denominal verbs in (196) are derived, we can maintain that the inalienable possessors in (194a) and (195a) actually have the same thematic role as the indirect objects in (197).

The proposed analysis for the problematic examples in (194a) and (195a) makes it possible to maintain the null hypothesis that the inalienable possessor must be a goal argument. Of course, we still have to solve the problem that regular passivization is possible, but semi-passivization is not. Broekhuis et al. (1996) claim that this is due to the fact that these examples are actually ambiguous between a structure with a dative and a structure with an accusative object; they substantiate this by referring to the German examples in (198), where the possessor can be either dative or accusative.

This still does not solve the whole problem, since there seems to be no discernible difference in meaning between the two alternatives. There may be several ways to solve this problem, but we will leave this to future research and refer the reader to Broekhuis et al. (1996: fn.3) for a suggestion.

The PPs in the locational constructions discussed above are predicative in the sense that they take an argument and assign it a location with respect to its nominal complement, the possessum. This located argument is usually the direct object, but we have seen that it can also appear as the subject in regular passive constructions. The relevant example is repeated here as (199b).

Since unaccusative verbs also involve a derived subject with the thematic role of theme, we expect that the subjects of these verbs can also function as located arguments. The examples in (200) show that this prediction is indeed borne out.

The definite articles found in the possessum of the standard Dutch locational inalienable possession constructions, as well as in the other dialectical inalienable possession constructions, exhibit various deviant semantico-syntactic properties similar to those of the definite articles found in weak definites; cf. Scholten (2018).

First, the possessum het beenthe leg in (201) does not have to express familiar information; it can also be used when its referent has not been mentioned earlier in the discourse. This is clearly related to the fact that the possessum is construed as an inalienable part of the dative possessor: the possessor must be indefinite, as in (201a), when the leg in question belongs to someone not previously mentioned, but definite, as in (201b), when it belongs to someone who is part of the domain of discourse (domain D). These examples also show that the same holds when the possessum contains a possessive pronoun. That the possessor is crucial is clear from the fact that example (201c) with the definite article can only be used if the leg has been mentioned earlier in the discourse (which we have indicated by means of a number sign); note that example (201c) with zijn is possible in out-of-the-blue contexts, provided that the pronoun refers to a familiar person.

Second, the possessum het been (as well as its alternate zijn been) behaves like a weak definite in that it has no unique reference; it is singular despite the fact that the dog’s victim will normally have two legs. However, it differs from weak definites in that it can be the antecedent of a deictic pronoun, which is clear from the fact that (201a&b) can easily be followed by the sentence in (202).

Third, although the possessum can refer to an entity, it differs from regular definite noun phrases in that its reference is not rigid, which is evident from the fact that it allows a sloppy reading in elliptical contexts. This is illustrated by example (203a), which expresses that the dog bit both Jan and Marie. Example (203b) shows that the same applies to a possessum with a possessive pronoun; note in passing that this example is quite natural, despite the fact that the feminine possessor in the second conjunct is construed with the possessum zijn been, which has a masculine possessive pronoun.

Finally, the possessum in (204a) has a de dicto reading in the sense that it co-varies with each specific choice from the contextually defined set of persons; again, this also holds when the possessum contains a possessive pronoun, as in (204b). Both sentences imply that more than one person’s leg was bitten, even though the noun been is singular.

The main finding of this section is that definite noun phrases used as possessum in an inalienable possession construction do not have the characteristic properties of regular definite noun phrases: they do not refer to a unique entity in domain D. Rather, they resemble weak definites, although they differ from them in that they can refer to specific entities. As far as the phenomena discussed here are concerned, possessums with a possessive pronoun behave similarly to those with a definite article.