Section 1.5.1 has shown that the characteristic property of perfect-tense constructions is that the eventuality denoted by the main verb is presented as completed. This is illustrated for the transitive verb lezento read in the primeless example in (133): whereas the imperfect construction in (133a) presents the eventuality of reading a book as an ongoing event, the perfect-tense construction in (133b) presents it as completed: Marie has finished the book. The primed examples in (133) illustrate the same for the unaccusative verb vallento fall: (all) the leaves are on the floor.

The question we will raise now is whether the perfective meaning is introduced by the past participle of the main verb or by the accompanying perfect auxiliary. The latter would imply that the expression of the perfective meaning requires the presence of an auxiliary, but this is not true. The examples in (134), for example, show that the past participle can also express the perfect meaning by itself as an attributive modifier; in this function it stands in opposition to the present participle, which is used to express the imperfect meaning.

Note in passing that in the case of transitive verbs, the modified nouns are also different in the two attributive constructions; while the past participle modifies a noun corresponding to the internal (theme) argument, the present participle modifies a noun corresponding to the external (agent) argument of the verb lezento read. The internal argument of the unaccusative verb vallento fall, on the other hand, can be modified either by the past or the present participle; cf. Section 2.1.2, sub IID.

That perfective meaning is expressed by past participles can also be shown by non-finite constructions such as (135), which are normally used to express the speaker’s surprise about some presupposition apparently held by his interlocutor; it often functions as an emphatic denial of that presupposition. Example (135a) presents the eventuality of Peter taking his PhD as ongoing: the (presumed) completion of this eventuality is after speech time. The default interpretation of example (135b), on the other hand, is similar to that of the corresponding present perfect sentence Jan is gisteren gepromoveerdJan took his PhD yesterday in that it locates the completion of this eventuality before speech time.

The above discussion suggests that perfective aspect is a meaning contribution of the past participle. It seems that we can ascribe a similar meaning contribution to the passive participle. The reason for this claim is that it is not a priori clear whether the participle in (134a') is a past or a passive participle: the fact that we can easily add an agentive door-PP to this example suggests that the latter is at least a possibility. Of course, similar examples cannot be given for unaccusative verbs such as vallento fall in (134b), simply because such verbs do not allow passivization.

A viable working hypothesis therefore seems to be that past and passive participles are similar in that they both present the eventuality as completed; cf. Duinhoven (1985) for a similar conclusion. Section 1.5.1 further argued that the perfect tenses do not locate the eventuality of the event as a whole in a specific temporal domain, but only its endpoint. In future examples such as (137), for example, it is not the complete event of reading of the book/falling of the leaves that is situated after speech time, but only the reaching of the endpoint of that event; Marie may have started reading the book long ago, and the same may be true for the falling of the leaves.

The past/passive participles in the attributive constructions in (134) and (136) are like set-denoting adjectives in that they denote a property of the modified noun, namely that it has reached some endpoint of the eventuality denoted by the input verb of the past/passive participle. In this context, it seems interesting to note that passive participles are like set-denoting adjectives in that they do not take an agentive argument; indeed, it is one of the characteristic properties of passivization that the agent of the input verb is demoted to adjunct status. Let us assume, for the sake of argument, the far from obvious position that past participles are also incapable of taking an agentive argument, and see what this can teach us about the function of the perfect/passive auxiliaries.

A standard hypothesis is that past and passive participles differ in their case-assigning abilities: past participles can assign accusative case to their internal theme argument, which therefore appears as direct object, as in (138a); passive participles cannot assign accusative case to their internal theme argument, with the result that (i) the internal theme argument must appear as the (nominative) subject of the clause and (ii) the original subject must be demoted to adjunct, as in (138b); cf. Jaeggli (1986) and Roberts (1987) for a possible rationale for this difference in case assignment.

This subsection adopts the more controversial hypothesis, put forward in Subsection I, that perfect and passive participles constitute a single category; cf. Hoekstra (1984a) and Wegner (2019a/2019b) for similar proposals. Let us now consider the auxiliaries in more detail. Our first observation is that the perfect auxiliary zijnto be and the passive auxiliary wordento be are homophonous with the copulas zijnto be and wordento become. Of course, this could be entirely accidental, but the more interesting assumption would be that it means that the auxiliaries zijn/worden have one or more crucial properties in common with the copulas zijn/worden. Let us therefore take a closer look at the function of the latter category.

It seems that the copula zijn does not play any semantic role in the sense of traditional calculus logic: the adjective is predicated of the noun phrase Marie, and this makes the well-formed proposition ziek(Marie). One reason to assume that the copula must be present is that it is needed to express present or past tense. It could be argued that the same is true for zijn in the perfect-tense construction in (140); Subsection I has shown that the perfect auxiliary is not needed to express the perfective meaning aspect, but it is needed to express present/past tense, i.e. to locate the perfect eventuality within the present/past-tense interval.

The copula worden in (139b) also has no semantic function in the sense of calculus logic, but it nevertheless makes a semantic contribution of its own by indicating that the logical subject of the adjective is involved in a polarity transition: Marie is undergoing a change from a state in which she is healthy (not ill) to a state in which she is ill. Interestingly, the passive auxiliary worden makes a similar contribution; an example such as (141) also expresses that the book is undergoing a change from a state in which it has not been read (by Marie) to a state in which it has been read.

Now that we have seen that the auxiliaries zijn and worden have properties in common with the copulas zijn and worden, let us consider the auxiliary hebbento have. If we compare the passive construction in (141) with the perfect-tense construction in (142), we notice two conspicuous differences between the two constructions: (i) the internal argument of the main verb appears as the nominative subject of the clause in the passive construction, but it is assigned accusative case in the perfect-tense construction; (ii) the external argument (agent) of the main verb cannot be expressed as a nominal argument in the passive construction, whereas it can in the perfect-tense construction.

If past and passive participles are indeed of the same category, these two differences must be attributed to the copular verb and the auxiliary. This seems possible if we assume that participles are not able to assign accusative case; this is, of course, a standard assumption for the passive participle, which is needed to account for the promotion of the direct object to subject, but not for the past participle. On this assumption, the fact that the internal argument of the verb lezen in the perfect-tense example in (142) can be assigned accusative case should be accounted for by assuming that hebben can assign accusative case not only as a main verb, as in examples such as (143), but also as an auxiliary.

That hebben can also assign accusative case in functions other than that of main verb can be independently supported by the (semi-)copular constructions in (144), which show that the nominative subject of the copular construction with zijn can appear as an accusative object in the semi-copular construction in (144b) with hebben. This follows directly from the fact that zijn and hebben differ in that only the latter is able to assign accusative case to the noun phrase het raamthe window, which functions as the subject of the set-denoting adjective open/dicht: with the copula zijn the noun phrase het raam must appear as the subject of the sentence to receive nominative case, whereas with the semi-copula hebben it can appear with accusative case.

The examples in (144) also show another important property of the semi-copula hebben, namely that it can introduce an additional nominal argument, here Jan. Assuming that this property of hebben is also found with the auxiliary, we can explain the second difference between the passive and the perfect-tense construction; if passive and past participles are indeed of the same type, the past participle cannot be held responsible for the presence of the subject Jan in (142), which must therefore be attributed to the auxiliary hebben.

If we continue with this line of reasoning, the fact that the auxiliary wordento become triggers passivization in (141) can be attributed to its unaccusative status (which is clear from the fact that it takes the auxiliary zijn in the perfect tense): since neither the passive participle nor the auxiliary worden can assign accusative case to the internal (theme) argument of lezen, the latter must appear as the nominative subject of the clause.

The use of the perfect auxiliary hebben in perfect-tense constructions with intransitive verbs such as lachento laugh cannot be motivated by the need to assign accusative case because intransitive verbs do not take an internal argument that requires that case. Nevertheless, hebben may be needed in (145a) to (re-)introduce the agent of the main verb; the auxiliary worden can be used in the impersonal passive in (145b) because it is neither needed to assign accusative case nor to (re-)introduce the agent of the verb lachen.

This account of auxiliary selection in the passive/perfect-tense constructions in (141), (142) and (145) may also explain the fact that the perfect auxiliary zijn is often used in perfect-tense constructions with unaccusative verbs such as vallento fall, as shown in (140). Since the internal arguments of such verbs already appear as the nominative subject in simple present/past-tense constructions, it is not necessary to use the verb hebben in the corresponding perfect-tense constructions; there is no need to assign accusative case or to introduce an additional agentive argument, and consequently the use of the unaccusative verb zijn suffices for the expression of present/past tense.

Above we argued from a synchronic point of view that the difference between perfect-tense and passive constructions is not due to the participles but to the auxiliaries used in these constructions. We will reinforce this point by discussing some diachronic and dialectal evidence that supports this proposal. In reconstructing the development of the various types of participles, Duinhoven (1985) argues that diachronically, participles have a nonverbal stem: the suffixes -end and -t/d/en, which derive present and past/passive participles, originated as postpositions with a simultaneity and completeness meaning, respectively. At some point, the internal structure of these adpositional phrases became obscure, and they were reinterpreted as adjectival. At yet another stage, the use of these adjectival past participles in predicative position led to a verbal interpretation; cf. Kern (1912). More precisely, the semi-copular construction in (146a), which expresses that Jan has a letter in a finished (i.e. written) state, was reinterpreted as in (146b), which added a dynamic meaning aspect to the construction. Duinhoven claims that once this reinterpretation had taken place for dyadic verbs, the clausal structure was also applied to monadic verbs, giving rise to the current productive perfect-tense construction; cf. Broekhuis (2021). A similar reinterpretation derived the perfect-tense and passive constructions with zijn from copular constructions with zijn; cf. Van der Wal (1986).

Duinhoven’s reconstruction is fully compatible with the above proposal. First, it takes into account that past and passive participles have the adjectival properties (i) that they cannot assign accusative case and (ii) that they do not take an external (agentive) argument. Second, if we assume that the case-assigning and thematic properties of the verb hebben and the participle geschreven are the same in the two constructions in (146), the reinterpretation involves only one feature, namely the categorial status of the participle: in (146a) the participle is adjectival and denotes a stative property, whereas in (146b) the participle is verbal and denotes a (completed) dynamic eventuality. In non-standard varieties of Dutch that have a productive semi-copular construction, the ambiguity in (146) still arises; example (147) is ambiguous between a perfect-tense and a semi-copular reading in such non-standard varieties, while it has only a perfect-tense interpretation in standard Dutch. We refer the reader to Section A31.3.1, sub IA2, for further discussion.

The previous subsections have argued that past and passive participles constitute a single category, and that it is the choice of auxiliary that determines whether we are dealing with a passive or a perfect-tense construction. Such an analysis presupposes that the properties normally ascribed to passive participles also hold for past participles: participles have the adjectival properties that they (i) cannot assign accusative case and (ii) do not take an external (agentive) argument. That participles have these adjectival properties should not be surprising, since past and passive participles are diachronically derived from adjectives. Note, however, that past and passive participles do not have the categorial status of adjectives, since they exhibit different syntactic behavior (e.g. with respect to verb clustering).

In short, we have found that the perfect auxiliary hebben is used in perfect-tense constructions of transitive verbs because it can assign accusative case to the internal argument of the participle and/or (re-)introduce the agentive argument of its input verb. Since the auxiliary worden does not have these properties, the internal argument (if present) of the participle is promoted to subject; this gives us the passive construction. The auxiliary zijn is often used as a perfect auxiliary with unaccusative verbs, because in such cases there is no need to assign accusative case or to (re-)introduce an argument of the input verb.