Since agents are typically [+animate] entities, the regular passive in most cases involves the demotion of an animate subject, as in the (a)-examples in (68). However, the (b)-examples show that passivization is also possible when an inanimate entity is construed as agentive; cf. Section 3.2.1.1, sub III, for further discussion.

A causer can be considered a special kind of agent, so it is not surprising that verbs with a causer subject can also be passivized. This is illustrated by the transitive verb brekento break in the (a)-examples in (69). The demoted subject of the causative verb can also be inanimate, as long as it is interpreted as the causer of the event, as shown in the (b)-examples.

The primed examples in (70) suggest that causative object-experiencer psych-verbs like irriterento irritate and overtuigento convince (cf. Section 2.5.1.3, sub II) can also be passivized. However, this is only possible if the met-PP referring to the cause (i.e. the means by which the causer brings about the mental state of the experiencer) is not overtly realized.

A typical property of the object-experiencer psych-verbs in (70) is that the cause can also be realized as the subject of the active construction, as in the primeless examples of (71). The primed examples again suggest that passivization of such causative psych-constructions is possible.

The claim that we are dealing with passives in the primed examples in (70) and (71) presupposes that the door-PPs are agentive phrases, i.e. similar to those found in unambiguous passive examples. However, this seems to be contradicted by the fact that the door-PPs in (71) contain inanimate, non-agentive noun phrases. Moreover, there is an alternative analysis according to which the door-PPs function as causative adjuncts comparable to those we find in unaccusative constructions such as De ruit brak door de harde windThe window broke because of the strong wind; cf. also Smeman (2023). Another reason to doubt the passive analysis of the primed examples in (70) and (71) is that the verb worden can be replaced by rakento get, which typically has a copular-like function when it occurs with a past/passive participle.

The examples in (72) strongly suggest that the verb worden in (70) and (71) is also used as a copular verb meaning “to become”. If so, we can explain the observation in Pollmann (1975:26-27) that pseudo-intransitive constructions such as (73a) differ from other pseudo-intransitive verbs such as rokento smoke in that they have no impersonal passive counterpart: the unacceptability of (73b) would follow from the fact that adjectival complementives require the overt presence of a logical subject; cf. *Er is/wordt boos (lit. there is/becomes angry).

We would also expect that in embedded clauses the participle must precede the finite verb, but the judgments on the primed examples in (74) show that this expectation is not fully borne out, since at least some speakers consider the order worden-participle to be substantially better than the order raken-participle.

Despite the possible problem posed by the primed examples in (74), we believe that we have shown that there is good reason to conclude that we are dealing with passive or copular (adjectival passive) constructions in the primed examples in (70) and (71); cf. Section 2.5.1.3, sub IID, for further relevant discussion.

There is a small set of causative non-experiencer verbs that behave more or less like the object-experiencer psych-verb irriterento irritate, cf. Section 2.5.1.3, sub V. A typical example is verduidelijkento clarify in (75), which is like irriteren in that it allows the subject of the active construction to be either a causer or a cause.

Again, it is not immediately clear that the primed examples are passive counterparts of the primeless examples, since the door-PP need not be agentive in nature. This is especially obvious with a causative verb such as reddento save, which can also take a causative door-PP in the active voice, as shown in (76a). Consequently, there is no doubt that the door-PP in (76c) can be interpreted as causative.

If (76c) were a passive construction, and if the door-PP in this example were of the same type of phrase as the causative door-PP in (76a), we would expect it to be possible to add an agentive door-PP, as in (76c). Our judgments in (77) show that this leads to a marginal result at best, and thus seem to contradict a passive analysis of (76c).

Although a more precise examination is needed before we can draw a firm conclusion, the discussion of the examples in (76) and (77) suggests that we may also be dealing with a causative door-PP in (75b'), and thus with a copular (i.e. adjectival passive) construction.

There are several types of non-agentive/non-causative verbs with inanimate subjects that nevertheless allow passivization. Some examples are given in (78); other verbs of this type are begrenzento bound, omcirkelento encircle, omlijstento frame, omringento surround, overdekkento cover and overwoekerento overgrow. Note that the passive counterparts of the stative primeless examples in (78) require the door-PP to be present; without it, the passive verbs receive an activity reading.

Other non-agentive verbs that can be found in regular passives are verbs that can take an object with propositional content, like aantonento demonstrate, bewijzento prove, demonstrerento show/demonstrate, bepalento determine, implicerento imply, as well as the verb vormento make up. The examples in (79) show that also in these cases the passive constructions must contain a door-PP.

The (a)-examples in (80) show that measure verbs like durento last, kostento cost, tellento count and wegento weigh with a non-agentive subject cannot be passivized. Note that the verbs tellento count and wegento weigh can also be used as activity verbs, as shown for wegen in the (b)-examples, in which case passivization is possible.

The difference between the two primed examples could in principle be attributed to the non-agentive nature of the subject in (80a), but another assumption holds that it is the nature of the nominal complement (here: 100 pond) that is relevant; it is not a direct object but an adverbial phrase comparable to the adjective zwaar in Jan weegt zwaarJan weighs heavy; cf. Section 2.4.

Since regular passivization results in the promotion of the theme argument of the active verb to subject, it is sometimes claimed that an important function of regular passivization is the “externalization” of the internal argument of the active verb. Section 3.2.1.1, sub IV, has already shown that this cannot be the case; the obligatoriness of the complementives van de tafel affrom the table and kapotbroken in the primeless examples in (81) shows that the accusative noun phrases are logical subjects (external arguments) of these predicates, not internal arguments of the verb vegen.

Section 3.2.1.1, sub IV, concluded from this that passive participles differ from the corresponding active verbs in that they are not able to assign accusative case to the noun phrase de kruimels/de bezem, which must therefore be promoted to subject in order to receive nominative case. That we are not dealing with externalization of an internal argument is also clear from the fact that arguments which are not assigned accusative case but occur in the form of a PP cannot be promoted to subject; intransitive PO-verbs only give rise to impersonal passivization; cf. Subsection IVB.

The (a)-examples in (83) show the same for complement clauses. Note that the expletive er in (83a') does not have the syntactic function of subject, i.e. it is not an anticipatory pronoun introducing the embedded clause; this function is restricted to the pronoun het in the (b)-examples. The passive examples in (83) thus differ in that the passive construction in (83a') is an impersonal passive, whereas the one in (83b') is a regular passive.

In English, the derived subject is not only assigned nominative case, but is also obligatorily placed in the regular subject position of the clause. This is not the case in Dutch: the derived subject can remain in its original position, i.e. the position normally occupied by the direct object of the active verb. This is easily demonstrated by the passive counterparts of the active ditransitive construction in (84a); the derived object can either follow or precede the indirect object, an option not available to the subject of active constructions (such as Jan in (84a)).

The difference between the two (b)-examples in (84) is related to the information structure of the clause: if the derived subject appears in its original position, as in (84b), it typically belongs to the focus (“new” information) of the clause, whereas it is presented as part of the presupposition (“old” information) of the clause if it is placed in the canonical subject position, as in (84b'). That this is the case is supported by the distribution of (non-specific) indefinite noun phrases like een mooi boeka beautiful book, which typically belong to the focus, and referential personal pronouns like hetit, which typically belong to the presupposition of the clause; the examples in (85) show that the former usually follow the indirect object, while the latter must precede it.

The examples in (84) and (85) show that the placement of the derived subject in the regular subject position is subject to conditions similar to scrambling of nominal objects; cf. Sections 13.2 and N21.1.4. This is not really surprising, since the placement of subjects of active clauses is also subject to similar conditions. This is illustrated in example (86a), where the position of the adverbial phrase gisterenyesterday shows that the subject does not have to occupy the canonical subject position right-adjacent to the complementizer. The (b) and (c)-examples show that the information structure of the clause is also involved in this case. Note in passing that the presence of er in (86b) depends on whether gisterenyesterday is presented as part of the focus or the presupposition of the clause; cf. Section N21.1.2. Note that we assume a neutral (non-contrastive) intonation pattern; (86b') is more or less acceptable when the subject een student is assigned contrastive focus.

In the passive counterparts of the idiomatic expressions in (87a&b), the derived subjects een stokje and de draak cannot be placed in the regular subject position either This is shown by the (optional) presence of the expletive erthere and especially by the position of the subject after the modal verb, which suggests that the noun phrase is still within its underlying position within the verb phrase (cf. the discussion of A-scrambling in Section 13.2). This may be because the derived subject is not referential and therefore cannot be part of the presupposition of the clause.

The derived subject in regular passives usually corresponds to the accusative phrase in the corresponding active clause. In some cases, however, it seems that dative phrases can also be promoted to subject in the regular passive.

The interpretation of certain adverbs is sensitive to grammatical function and thereby to passivization: the subject-oriented adverb graaggladly is related to the agent in the active sentence in (102a), but to the theme in the passive sentence in (102b).

The examples in (103) show that passivization can affect the binding possibilities of pronouns. The possessive pronoun haarher in the active example in (103a) can either be bound (i.e. interpreted as coreferential with) Marie or as referring to another person previously mentioned in the discourse, e.g. Els. In the passive sentence in (103b), on the other hand, the possessive pronoun is preferably interpreted as referring to a previously mentioned person, e.g. Els.

More or less the same is shown by the examples in (104a&b): while in (104a) the reciprocal pronoun elkaareach other can be bound by the indirect object, this is not possible in (104b). However, the example in (104c) shows that this depends not only on passivization, but also on word order; if the subject is not moved into the canonical subject position, but remains in its underlying position following the indirect object, binding by the indirect object remains possible; cf. Section N22.2.

The examples in (105) also show that it is a combination of passivization and word order that determines the interpretation of the sentence. In the active sentence in (105a), the possessive pronoun zijnhis can be interpreted as bound by the quantifier iedereeneveryone or it can refer to some entity previously mentioned in the discourse. The first interpretation gives rise to the so-called bound-variable reading, in which the pronoun functions as a variable in the semantic representation of the sentence: ∀x (x kissed x’s brother). The second interpretation is called the independent reading, because the interpretation of the pronoun is independent of the quantifier: ∀x (x kissed his brother). The passive sentence in (105b) does not allow a bound-variable reading of the pronoun, i.e. it can only be interpreted as referring to a previously mentioned person: ∀x (his brother was kissed by x). But again, word order seems to play a role; if the subject is not moved into the canonical subject position, but remains in its underlying position following the door-PP, as in (105c), the bound-variable reading of the pronoun is easier to get (though it is not fully felicitous due to the fact that the quantifier is the complement of a PP and that pronominal binding from such a position is generally somewhat marginal).

The examples in (103) and (105) have shown that binding is bled by passivization when the derived subject moves into the canonical subject position. However, binding can also be fed by passivization. Example (106a) shows that a possessive pronoun embedded in a subject cannot be bound by the direct object: this example can only be construed with an independent reading of the pronoun zijn. In the corresponding passive construction in (106b), on the other hand, both the independent and the bound-variable reading are available. Note that the derived subject must be moved into the canonical subject position in order to make the bound reading available: example (106c) only licenses the independent reading of the pronoun.

Passivization can affect the relative scope of quantified phrases. Consider the examples in (107). In (107a) the universal quantifier iedereen has scope over the indefinite noun phrase twee talen; the languages spoken may differ from person to person. In the passive construction in (107b), on the other hand, the scope relations are reversed; the sentence expresses that there are two languages which are spoken by all the persons in question. This reversal of scope requires that the derived subject be moved into the regular subject position: example (107c), in which the derived subject remains in its base position, has the same scope relation as (107a).

If an active clause contains a modal verb, passivization is usually possible. The modal verb remains the finite verb of the clause, and the passive auxiliary is realized as an infinitive, but seems to be optional.

Examples such as (108b) without the auxiliary are often assumed to contain an empty counterpart of the passive auxiliary. However, there is reason to believe that such an analysis is on the wrong track. Since passive constructions can normally contain an agentive door-PP, the assumption of an empty passive auxiliary would incorrectly predict that this adjunct phrase can also appear in examples such as (109) in the absence of worden.

Example (109) thus suggests that the participle in the construction without worden is not a passive participle but a predicatively used adjective; cf. De muur moet geelthe wall must be made yellow. That we are dealing with an adjectival participle can perhaps also be supported by the examples in (110); while the unambiguous verbal participle in (110a) can either precede or follow the clause-final verbs, the participle in (110b) prefers the preverbal position, which is a hallmark for nonverbal status.

Further support for the assumption that the participle functions as a predicative adjective in the absence of worden is that it can be coordinated with other predicative phrases, such as the PP in de lak in (111).

Constructions with modal verbs that take an adjective as their complement are discussed in more detail in Section A28.2.3, sub II.

Passivization of intransitive PO- verbs such as rekenen opto count on in (112a) always involves the impersonal passive in (112b); passive constructions such as (112c), in which the complement of a preposition is promoted to subject, are unacceptable (but see the discussion of (114) below).

Quirk et al. (1985/1991: §3.69) and Huddleston & Pullum (2002:1433) show that English often allows passivization of the kind in (112c). It is tempting to relate this to the fact that English does not allow impersonal passivization of the kind in (112b); English passivization of the kind in the primed examples in (113) could then be given a functional explanation, assuming that it compensates for the unavailability of impersonal passivization.

Some Dutch speakers (marginally) accept the string in (112c). This does not mean, however, that such speakers allow the complement of a preposition to be promoted to subject. Instead, they interpret example (112c) as a case of left dislocation; the structure is as given in (114), with the pronominal part of the PP daar ... opon him optionally omitted (indicated by strikethrough). That the noun phrase zijn vader in (114a) is not a subject in the absence of the resumptive pronoun daar is clear from at least two facts. First, example (114b) shows that replacing this noun phrase with a subject pronoun severely degrades the result. Second, since left dislocation occurs only in main clauses, the unacceptability of (114c) shows that we correctly predict that the surface string Zijn vader werd op gerekend in (114a) has no embedded counterpart.

That the noun phrase in sentence-initial position does not function as a subject in the absence of daar is also clear from the fact that this noun phrase does not agree in number with the finite verb; cf. Klooster (2001a:324). For all Dutch speakers, example (115b') is unacceptable both with and without the pronominal part of the PP; this contrasts sharply with the following English example from Huddleston & Pullum: These problems weren’t faced up to by the committee.

Section 3.2.1.1, sub IV, has shown that accusative noun phrases can be promoted to subject in the regular passive, regardless of whether they are internal arguments of the passivized verbs or not. However, there are specific additional restrictions on the accusative noun phrase. Consider example (116a), in which the accusative noun phrase de vaas is not an argument of the verb slaanto hit but of the predicative adjective kapotbroken. However, since this verb assigns case to this noun phrase, passivization still results in its promotion to subject, as in (116b).

Given this analysis of (116b), we would expect something similar to happen if we were to passivize the causative/permissive verb latento make/let in (117a); since laten is generally taken to assign accusative case to the external argument of the verb dansen, we expect the latter to appear as the nominative subject of the clause after passivization. However, example (117a') shows that regular passivization is impossible. The (b)-examples in (117) provide similar examples with the perception verb horen.

It is not immediately clear what the unacceptability of the primed examples in (117) shows. For example, it might be taken to show that the earlier assumption that laten assigns accusative case to the subject of the infinitival clause is wrong, but then we would wrongly predict that impersonal passivization is possible in (118).

This means that some independent reason must be found for the impossibility of passivization. The constructions with the participles gelaten and gehoord could be excluded by the fact that they are part of the verbal complex; as in the perfect-tense constructions in (119), the verb is expected to appear as an infinitive (the so-called infinitivus-pro-participio effect).

Evidence for this analysis is that German, which does allow participles in such verb sequences, also allows passivization of the type in the primed examples in (117); cf. Reis (1976) and Rutten (1991:121).

This leaves us with the constructions with the infinitival forms laten and horen; the impossibility of these constructions could be explained by appealing to the hypothesis discussed in Section 3.2.1.1, sub I/II, that passive morphology is needed in order to demote the external argument of the verb or to absorb case; cf. Bennis & Hoekstra (1989b).

The examples in (121) pose, in a sense, the opposite problem to the one in the previous subsection. The primeless examples in (121) seem to be cases in which the subject of the infinitival clause is promoted to subject of the matrix clause. However, the expected active counterparts of these constructions in the primed examples are unacceptable; cf. Section 5.2.2.2, sub III, for a more detailed discussion of this construction.

The unacceptability of the primed examples would follow if the verbs achten and veronderstellen are unable to assign accusative case to the subject of the infinitival clause (which may be related to the fact that these verbs differ from the causative and perception verbs in that they do not trigger verb clustering but extraposition of the infinitival clause). If so, the primed examples can be used to support the claim that the core property of passivization is the demotion of the external argument of the verb, not the absorption of accusative case; cf. Section 3.2.1.1.

For completeness’ sake, note that the idiomatic expression in (122a) is an exception to the rule that the active verb achten, which in this case alternates with menen, cannot assign accusative case to the subject of its infinitival complement. However, more cases of this sort can be found in transparent free relatives with an infinitival clause headed by the copular te zijn, such as (122b); cf. Woordenboek der Nederlandsche taal (Meenen, #13), Zajijeck (1970) and Van de Velde (2015: §3.3) for discussion.