Impersonal verbs are verbs that can be assumed to take no nominal argument at all, which is why they are also known as avalent verbs. Weather verbs like regenento rain and sneeuwento snow in (9) are typical instantiations of this type.

The subject pronoun het in these examples is not referential and therefore should not be considered an argument of the weather verb; it seems to be present only to satisfy the syntactic requirement that the verb has a (nominative) subject. Section 2.2.3, sub IB, will support this view by showing that het is obligatorily suppressed if some other element in the clause introduces a nominal argument that can function as a subject. This is illustrated here by the resultative construction in (10), in which the noun phrase Jan is licensed by the complementive natwet.

Dutch differs from German in that it has no impersonal verbs with a (dative or accusative) experiencer object; German examples such as (11a), in which both accusative and dative are found, are rendered in Dutch by a semi-copular hebben-construction, as in (11b); cf. Section A28.2.1, sub IB, for discussion.

Since impersonal verbs in Dutch do not take other nominal arguments, there is not much to say about them in the present context. Therefore, we limit ourselves here to a small sample of these verbs in (12): the (a)-examples are truly impersonal in the sense that they are not normally used with an argument, while the (b)-examples are verbs that can also be used as monadic or dyadic verbs; cf. Het gietIt is pouring versus Jan giet water over de vloerJan is pouring water on the floor.

Before concluding this section, we need to point out two things. First, the examples in (13) show that there are a number of exceptional, probably idiomatic cases in which weather verbs of the type in (12a) do seem to take an internal argument.

Second, we should mention that Bennis (1986: §2.2) has argued against the above claim that weather het is non-referential by showing that it is able to control the implicit PRO-subject of an infinitival clause in examples such as (14a). A problem with this argument is that the pronoun het in the main clause is not the subject of a weather verb, but of a copular construction with a nominal predicate, similar to the one we find in examples such as (14b); since the pronoun het in (14a) is clearly non-referential, the same may be true for the PRO subject it controls.

Of course, it is possible to construct examples such as (15) in which PRO is controlled by weather het, but since PRO can be controlled by the non-referential pronoun het in (14a), this can no longer be taken as evidence for the referential status of weather het.

Bennis is more convincing by pointing out that weather verbs may sometimes take a referential subject as in (16b), in which weather het can be replaced by the referential noun phrase de windthe wind.

However, example (16) does not show that weather het is also referential. A serious problem for such a view is the earlier observation that it is not possible to realize het in resultative constructions such as (10). This is unexpected if het were referential, because example (17a) shows that the referential noun phrase de wind must be realized in such resultative constructions. Example (17a) thus contrasts sharply with the (b)-examples in (17), which again show that weather het is obligatorily omitted in resultative constructions (on the intended natural-force reading); cf. Section 2.2.3, sub I, for further discussion.