The final type of dative alternation involves benefactives. The examples in (435) show that benefactives are normally expressed by means of a voor-PP in Dutch.

There is, however, a very small subset of verbs denoting activities relating to the serving of food and drinks that also allow a dative object: typical examples are the verbs schenken'to pour' and opscheppen'to dish up' in (436).

The examples in (437) show that benefactive constructions like the (b)-example are special in that the direct object can be left implicit. This might be related to the fact that the direct object must refer to some entity in a restricted semantic field: it must refer to something that can be consumed. We may therefore be dealing with implicit cognate objects of the type we also find in pseudo-intransitive verbs like eten'to eat', drinken'to drink', roken'to smoke', etc.

The two alternants in the examples in (436) clearly differ in meaning. The double object constructions express that the entity denoted by the direct object is intended for the referent of the indirect object: Marie is also the recipient of the drink/potatoes. The periphrastic indirect object constructions, on the other hand, express that the subject performs the activity on behalf of the referent of the indirect object: Marie may be the recipient of the drink/potatoes, but it may also be the case that Jan is performing the activity of pouring out a drink/dishing up potatoes to help Marie in her task of serving some guests; cf. Van Hout (1996:47). This is in keeping with the meaning representations in Table (335), according to which the nominal but not the periphrastic indirect object is affected by the event denoted by the verb.
      The Standard Dutch alternation is much more restricted than the corresponding one in English. For example, verbs of food preparation like bereiden'to prepare' and bakken'to bake' in the primeless examples in (438) do not readily allow it. It should be noted, however, that Dutch still has the idiomatic expressions in the primed examples in (438) and that the double object constructions are very common (in fact: pervasive) in various eastern and southern dialects of Dutch with a wide variety of verbs; cf. Van Bree (1981) and Cornips (1994).

      It seems controversial to analyze the voor-phrase as an indirect object. This can be illustrated by the fact that it is only in the second edition of the Algemene Nederlandse Spraakkunst that it is unambiguously treated as an indirect object (Haeseryn et al. 1997:1160ff.); the first edition (Geerts et al. 1984:882ff.) treated it primarily as an adverbial phrase. One reason for treating the voor-phrase as an adverbial phrase is that this PP differs from objects in that it can always be omitted without it being semantically implied. Another reason is that the voor-phrase satisfies the adverb-test in (439). Note that (439b) allows the same range of interpretations as (439a); Peter may intend the drink for Marie or he may perform the activity of pouring drinks for the benefit of Marie.

Another reason is that benefactives may appear in the form of a simplex reflexive, which normally cannot be bound by a co-argument; the acceptability of the examples in (440) with a reflexive pronoun would fall out if benefactives are actually not arguments of the verb.

For completeness' sake, it should be noted that a potential argument against adjunct status is that R-extraction from the voor-PP is possible, as is shown in (441a). We know, however, that this is not a reliable test for establishing complement status given that R-extraction is also possible from other PPs that are normally assumed to be adjuncts such as the instrumental met-PP in (441b).

The discussion above shows that it is not a priori clear that the supposed dative alternation with voor-phrases should be treated on a par with the dative alternations discussed in the previous sections. We leave this as a question for future research.