Ditransitive verbs select an external argument, which is realized as the subject of the clause, and two internal arguments, which are realized as an indirect object (the goal/source argument) and a direct object (the theme argument), respectively. Examples of such ditransitive verbs are aanbiedento offer and afpakkento take away in (82).

Example (83) provides a small sample of such double object verbs. Note that the goal/source arguments of these verbs can often also be expressed with an aan/van-PP; this is not relevant here and will therefore be discussed in Section 3.3.

Although Dutch has no morphologically realized cases for non-pronominal noun phrases, it is generally assumed, on the basis of comparable constructions in German, that the two objects are assigned different cases: the indirect object is assigned dative case, while the direct object is assigned accusative case.

The indirect object often does not need to be realized overtly, but is usually semantically implied: for example, if we omit the dative noun phrases in the examples in (85), the goal of the event is assumed to be a salient entity in the discourse domain.

Monadic unaccusative verbs are characterized by having an internal theme argument, which appears as the theme-subject of the clause. We therefore expect there to be a class of unaccusative verbs with two internal arguments, one of which appears as a derived subject. Den Besten (1985) has argued that such dyadic unaccusative verbs do exist, and that they are instantiated by the so-called nom-dat verbs. The name of these verbs expresses that they take a theme argument that is assigned nominative case, as well as an experiencer argument that is assigned dative case. This is not directly observable in Dutch, because it does not morphologically express the difference between dative and accusative case, but it is in German examples such as (86a), of which (86b) provides the Dutch translation.

The experiencer argument (indirect object) is usually obligatory, or at least semantically implied. In the latter case, the implicit experiencer often refers to the speaker, but it can also be interpreted generically.

Subsection II will show that subjects of nom-dat verbs differ from subjects of transitive verbs in that they are internal arguments, i.e. they behave in several respects like the theme-subjects of monadic unaccusative verbs discussed in Section 2.1.2; they also show similar behavior to the derived subjects of the passivized ditransitive verbs in (88).

Section 2.1.2, sub III, has argued that there are two classes of monadic unaccusative verbs, one with the auxiliary zijn and another with the auxiliary hebben in the perfect tense, and Subsection IIC will support this claim by showing that the same holds for nom-dat (i.e. dyadic unaccusative) verbs. The order nominative-dative in the two examples in (89) clearly indicates that we are dealing with nom-dat verbs; cf. Subsection IIF.

Example (90) provides samples of both types of verbs, taken from the more general list in Den Besten (1985:fn.7). Since Dutch does not express case morphologically, it cannot be straightforwardly established that the verbs in (90) are indeed nom-dat verbs, but this is possible for the German counterparts of these verbs; cf. Drosdowski (1984: 608-10) for an extensive list and Lenerz (1977) for a more detailed discussion of the behavior of such German verbs.

Native speakers sometimes have different judgments about auxiliary selection; for some speakers, the verb bevallento please is (also) compatible with the auxiliary hebben, as can be seen from the fact that such cases can be easily found on the internet. To our knowledge, it has not been investigated whether this shift in auxiliary selection affects the other properties of the verb that will be discussed in Subsection II.

Note that it is sometimes difficult to give satisfactory English renderings of the verbs in (90), because English usually expresses the same meaning by using completely different syntactic frames; in English, the experiencer is often realized as the subject rather than the object of the clause (which is perhaps not surprising, in view of the fact that passivization of English ditransitive constructions usually requires the goal argument, not the theme, to be promoted to subject).

In German, objects of nom-dat verbs are assigned dative case, as are indirect objects of double object constructions. This might lead us to expect that the objects in these constructions show similar syntactic behavior. However, there is at least one striking difference between them; the examples in (92) show that, while dative objects of ditransitive verbs often alternate with prepositional phrases, objects of nom-dat verbs do not have that option. This fact could be related to a difference in the thematic roles carried by the respective dative objects; the dative/PP alternation prototypically involves recipient/goal arguments and not experiencers. The reason may be that experiencers, but not goals, are inherently affected by the eventuality in some sense: cf. Section 3.3.1 on the dative alternation illustrated in the (a)-examples for a discussion of the claim that affectedness is a typical property of dative objects but not of prepositional indirect objects. A more trivial solution might be to appeal to the fact that Dutch has so-called role prepositions for expressing goals and sources, but not for experiencers; cf. Section P32.3.3, sub IIA.

Some nom-dat verbs seem to be undergoing a process of reanalysis in the direction of regular transitive verbs. This is clearly the case with the verb passento fit in (93); besides (93a), in which the experiencer is realized as a dative object, the construction in (93b) is considered acceptable by many (but not all) speakers; cf. https://onzetaal.nl/taalloket/ik-pas-mij-past-die-broek-niet-meer.

Perhaps this reanalysis goes hand in hand with a change of meaning; although example (93b) can be used in the same sense as (93a), with the subject as experiencer, it can also be used to express that someone is trying on the shoes, in which case the subject is construed as an agent. However, an alternative would be to assume that the second reading is related to the particle verb aanpassento fit on, which is not used as a nom-dat verb.

A closer look at the individual nom-dat verbs in (90) reveals that many of these verbs are either morphologically complex, in the sense that they are prefixed by the morphemes be- or ont-, or obligatorily accompanied by a verbal particle. Although this has been noted before, it has not been thoroughly investigated whether this is theoretically significant, as might be expected since prefixes like be- and ont- as well as verbal particles can both be considered secondary predicates; cf. Section 2.2.3, sub IIIB. We leave this question for future research.

Section 2.1.2, sub IIIA, has shown that intransitive and transitive verbs usually denote activities; the subject of the clause normally functions as agent and therefore typically refers to a [+animate] entity. Examples (94a&b) show that the same is true for ditransitive verbs; the subject of the double object construction is usually the agent performing the activity denoted by the verb, and for this reason it is typically a [+animate] participant (or an institution, which is then seen as a collection of individuals). Although there are some exceptional cases, such as (94c), the overall pattern seems consistent with the idea that subjects of double object constructions are external arguments.

Nom-dat verbs, on the other hand, denote processes or states. The subject of such verbs functions as a theme, i.e. it undergoes the process or is in the state denoted by the verb. The fact that the subject is not an agent explains why the subjects of nom-dat verbs are often [-animate]; it is also consistent with the hypothesis that the subject of a nom-dat verb is an internal argument, just like the subject of the unaccusative verbs discussed in 2.1.2. Two examples are presented in (95).

External arguments are usually noun phrases; cf. the introduction to Chapter 2. The fact, illustrated in (96), that the subject of a nom-dat verb can be a clause therefore also suggests that it is an internal argument. Note in passing that the subject clause can occur either in sentence-initial or sentence-final position; if it is in final position, the regular subject position is occupied by the anticipatory pronoun hetit.

Section 2.1.2, sub IIB, has shown that agentive er-nouns refer to entities that perform the activity denoted by the input verb. Since ditransitive verbs have an external argument, it is not surprising that they can also be the input of er-nominalization. Some examples are given in (97).

However, as in the case of transitive verbs, er-nominalization of ditransitive verbs sometimes leads to a marginal or even impossible result for unclear reasons; some examples are given in (98).

Since the nom-dat verbs have no external argument, it is predicted that they cannot be input for the formation of agentive er-nouns. As shown in (99), this seems to be confirmed. The examples in (99a) and (99b) correspond to some of the nom-dat verbs in (90a) and (90b), respectively. The form opvaller is marked with a percentage sign since it does occur occasionally on the internet; overkomer is marked with a number sign because it can be used as a lexicalized form referring to a large wave washing over a ship.

Note that, as in the case of the monadic unaccusative verbs, there seem to be a number of lexicalized exceptions. That these forms are not the result of the productive process of er-nominalization is clear from the fact that e.g. the derived form in (100b) cannot be used to refer to the referent of the (underlined) subject in an example such as Dat boek/Jan viel me tegen that book/Jan disappointed me.

Section 2.1.2, sub IIC, has argued that all verbs with an external argument take the auxiliary hebben in the perfect tense. The examples in (101) show that ditransitive verbs also take this auxiliary.

Section 2.1.2, sub III, on the other hand, has argued that monadic unaccusative verbs can take either hebben or zijn in the perfect tense, depending on their aspectual properties. The same is true for dyadic unaccusative verbs: the cases in (102) and (103) give examples of nom-dat verbs taking the auxiliary zijn and the auxiliary hebben, respectively.

The fact that the verbs in (102) take the auxiliary zijn is sufficient to conclude that they are unaccusative and, consequently, that the subject is a theme-subject. The fact that the verbs in (103) do not use zijn but hebben is due to the fact that they are atelic; they denote a state of affairs without an inherently implied endpoint.

Section 2.1.2, sub IID, has shown that past/passive participles of transitive verbs can be used attributively to modify nouns corresponding to the direct object of the corresponding active verbs. As shown in (104a&b), the same holds for the past/passive participles of ditransitive verbs. The indirect object usually remains implicit in such cases, but can be expressed overtly if it is a pronoun. The question marks indicate that if it is a non-pronominal noun phrase, the result may be slightly marked for some speakers; however, it is clearly grammatical as it is fully acceptable to other speakers.

As in the case of transitive verbs, the past/passive participles of ditransitive verbs cannot be used to modify a noun corresponding to the subject of the corresponding active verb; this is shown in (105).

The use of the past/passive participle to modify the indirect object is unacceptable for many speakers, but marginally acceptable to some others. Note that the theme argument must be expressed overtly in these cases: if it is omitted, the examples in (106) become completely unacceptable to all speakers in the intended goal reading of the modified noun (which could then be read as a theme).

Section 2.1.2, sub III, has shown that past/passive participles of monadic unaccusative verbs selecting zijn can be used attributively to modify a noun corresponding to the subject of the corresponding active verb, while past/passive participles of monadic unaccusative verbs selecting hebben cannot. The same distinction holds for dyadic unaccusative verbs. The examples in (107) are cases with past participles of nom-dat verbs selecting zijn; the fact that the past participles can be used to modify the nouns matching the subjects of the corresponding active verbs is sufficient to conclude that the base verb is unaccusative.

The examples in (108) are cases with past participles of nom-dat verbs selecting hebben. The fact that the past participles cannot modify the noun matching the subjects of the corresponding active verbs is due to the fact that these verbs are atelic; they denote a state of affairs without an implied endpoint.

Note that, as with the ditransitive verbs in (105), the use of a non-pronominal experiencer is somewhat awkward but possible in (107); omitting it in (107a) would actually worsen the result, although (107b) is perfectly fine without it. As expected, omitting the experiencer in (108) does not improve the result.

Section 2.1.2, sub IIE, has shown that while intransitive and transitive verbs can be passivized, unaccusative verbs such as arriverento arrive cannot. From this we concluded that the presence of an external argument is a necessary condition for passivization. This correctly implies that ditransitive verbs can normally be passivized, as shown in (109); note that the agent can optionally be expressed by an agentive door-PP.

If nom-dat verbs are analyzed as dyadic unaccusative verbs, we would expect that they cannot be passivized. The examples in (110) and (111) show that this expectation is borne out: impersonal passivization is impossible.

The examples in (112) show that the dative object of an active sentence cannot function as the subject of a passive sentence either. This provides additional evidence for the claim that nom-dat verbs cannot be considered regular transitive verbs.

Note that we have used examples with human subjects, because it is often claimed that there is an animacy restriction on passivization. As expected, clauses with a [-animate] subject cannot be passivized either.

Although the word order in the middle field is relatively free in Dutch, the relative order of the arguments of the verb is more or less fixed. As shown in (113), the subject of a transitive verb must normally precede the direct object.

The same is true for the arguments of a ditransitive verb. With a neutral intonation of the clause, the subject must precede the indirect object, which in turn must precede the direct object. All other orders are excluded; the addition of contrastive accents sometimes improves the result slightly, but it is always highly marked (unlike in German, which allows the inversion of indirect and direct objects under the right intonation pattern).

However, nom-dat verbs differ from (di)transitive verbs in this respect. The examples in (115) and (116) show that two orders are possible; the subject can either precede or follow the dative object. This again provides direct evidence for the claim that these verbs are not regular transitive verbs.

Interestingly, the examples in (117) show that the same freedom of word order can be seen in passive constructions with ditransitive verbs. This similarity between nom-dat and passivized ditransitive verbs supports the claim that the subject of a nom-dat verb is an internal argument comparable to the direct object of a ditransitive verb.

The data in (115) to (117) provides evidence for the claim that the base position of the theme-subject of a nom-dat verb is the same as the direct object of a (di)transitive verb. These positions follow the canonical position of the indirect object, i.e. the primed examples of the nom-dat and passive constructions in (115) to (117) are derived by moving the derived subject into the canonical subject position of the clause. In other words, the structure of the primeless examples in (115) to (117) is as shown schematically in (118a), where e represents the empty canonical subject position, and the structure of the primed examples is as shown in (118b), in which the nominative noun phrase has been moved into this subject position.

The difference in word order between the structures in (118a) and (118b) is not arbitrary, but seems to be related to the information structure of the clause. If the nominative argument occupies the position in (118a), it is interpreted as belonging to the focus (discourse-new information) of the clause. If it occupies the position in (118b), it belongs to the presupposition (discourse-old information) of the clause. This is also clear from the fact that existentially quantified subject pronouns, which typically belong to the focus of the clause, must follow the dative noun phrase.

The same is shown by the fact that definite subject pronouns, which typically belong to the presupposition of the clause, must be placed in the regular subject position. See Sections 13.2 and N21.1.4 for further discussion of the relationship between word order and information structure.

This subsection has shown that the word order of the nominal arguments in the clause is quite strict in Dutch, with one exception. An internal theme argument can either follow or precede an internal goal argument if there is no external agent argument (as in passive and nom-dat constructions). Depending on the information structure of the clause, the theme can then either follow the goal (i.e. remain in its base position) or precede it (i.e. move into the canonical subject position). This freedom of word order can be explained by the hypothesis that subjects must move into the canonical subject position of the clause (as in English), unless this conflicts with the generally assumed tendency that discourse-old information of the clause precedes discourse-new information (unlike in English). This generalization will play a crucial role in Subsection III, where we will discuss an analysis of the nom-dat construction that competes with the one in (118).

Although Section 2.1.2, sub IIIF, has shown that the wat-voor split is not a very reliable test for distinguishing between external and internal arguments, we will demonstrate that in the case of the nom-dat verbs it can be used to show that the subject is a theme-subject. Let us first look at some data. Example (121) indicates that the wat-voor split seems to be possible with all arguments of ditransitive verbs, although some speakers may have some difficulty in extracting wat from the subject and indirect object. As with intransitive and transitive verbs, a wat-voor split of the subject usually requires the presence of the expletive er; if it is omitted, as in (121a), the sentence becomes marked for all speakers.

As shown in (122a), a wat-voor split is also possible from the derived subject in a passive construction headed by a ditransitive verb; the expletive er is optional, probably due to the fact that the indirect object Marie can be interpreted as belonging to the presupposition of the clause; cf. Section N21.1.2 for a discussion of the restrictions on the occurrence of the expletive er. However, example (122b) shows that the wat-voor split requires the indirect object to precede the derived subject: if the subject is moved into the canonical subject position, the wat-voor split is unacceptable regardless of whether the expletive er is present or not.

The ungrammaticality of (122b) can be made to follow from the assumption that the movement of the theme-subject into the regular subject position causes a freezing effect; a moved phrase is generally assumed to be an island for wh-extraction, i.e. one cannot move an element from a phrase that itself has moved. This supports the hypothesis that the examples in (122) have the structures in (118) in Subsection F.

Since we have argued that clauses with a nom-dat verb also have the structures in (118), we expect a similar contrast as in (122) to arise with these verbs: if the nominative noun phrase follows the dative noun phrase, a wat-voor split is expected to be possible, whereas it is expected to be excluded if it precedes the dative noun phrase. The examples in (123) bear these expectations out for nom-dat verbs selecting zijn.

Nom-dat verbs taking hebben, on the other hand, do not qualify; in (124) the wat-voor split leads to a degraded result in both orders.

The above data has shown that the wat-voor split provides evidence for the derived status of the subject of nom-dat verbs taking zijn; since the split is only possible if the nominative noun phrase follows the dative noun phrase, the subject must be generated in the same position as the direct object of a transitive verb. The wat-voor split is inconclusive in the case of nom-dat verbs taking hebben, since it seems impossible in either order (for reasons that are unclear).

For completeness’ sake, we conclude this subsection with a brief (and somewhat theory-dependent) discussion of the wat-voor split of dative noun phrases in passive ditransitive and nom-dat constructions. Consider the examples in (125). Example (125a) shows that a wat-voor split from an indirect object seems possible, although native speakers’ judgments on the exact status of these examples may differ. To allow the split, the subject must be indefinite: if it is definite, as in (125b), the acceptability of the construction decreases. The split is completely forbidden if the subject is moved into the regular subject position, as in (125c).

The ungrammaticality of (125c) can be explained as follows. In order to license the wat-voor split, the indirect object must occupy its base position in order to avoid a freezing effect. However, it has been argued that movement of a theme argument (a direct object or a theme-subject) across an indirect object in its base position is blocked. In order to move the theme argument, the indirect object must be scrambled to a more leftward position; cf. Haegeman (1991) and Den Dikken (1995). This is easy to show for ditransitive verbs. The examples in (126b&c) show that indirect and direct objects can be scrambled to a position before the clause adverbial zekercertainly. However, while the indirect object can be scrambled on its own, as in (126b), scrambling of the direct object is only possible if the indirect object has also been scrambled, as is clear from the ungrammaticality of (126d). Note that the judgments hold only under neutral intonation; example (126d) improves when the indirect object receives contrastive focus; cf. Section 13.2 for further discussion.

The examples in (127) show that something similar holds in the passive construction; moving the theme-subject into the canonical subject position to the immediate right of the complementizer datthat requires scrambling of the indirect object: this movement is blocked when the indirect object remains in its base position after the clause adverbial zekercertainly. Again, this only holds under neutral intonation; example (127c) improves with a contrastive focus accent (e.g. on the indirect object or the adverb/particle zeker).

The discussion of (126) and (127) strongly suggests that in (125c) the indirect object must be scrambled, and that the impossibility of the wat-voor split is therefore due to a freezing effect. The intermediate status of (125b) may also be due to a freezing effect, since the definite noun phrase de boekenthe books is more likely to be scrambled than the indefinite noun phrase boekenbooks.

A pattern similar to that in (125) can also be seen in the case of the nom-dat verbs. This again provides evidence for the claim that the base position of the theme-subject is to the right of the indirect object and that its placement in the regular subject position is the result of movement, as shown in example (118b).

Note that this evidence holds only to the extent speakers agree that there is a clear contrast between the completely unacceptable (c)-examples and the marked but acceptable (a) and (b)-examples; judgments on the latter may vary.

The previous subsections have discussed ditransitive and dyadic unaccusative (nom-dat) verbs. We have seen that the latter, like the monadic unaccusative verbs, come in two types: the first type selects the auxiliary zijn in the perfect tense, while the second type takes hebben. Ditransitive verbs are easy to distinguish from transitive and nom-dat verbs because they take three nominal arguments instead of two. Transitive and nom-dat verbs are more difficult to distinguish because they take the same number of arguments and, unlike in German, the objects are not morphologically marked for case. However, they still differ in several ways: transitive verbs can undergo er-nominalization and be passivized, while nom-dat verbs cannot; furthermore, transitive verbs require the word order subject > object, while nom-dat verbs also allow the order object > subject under the right information-structural conditions. The properties of transitive and nom-dat verbs discussed in this section are summarized in Table 5. Columns 1-6 should be read in the same way as in Table 3; column 7 is added and indicates whether it is possible for the dative object to precede the subject (i.e. the nominative argument).