Like the German Konjunktiv I, the morphologically marked subjunctive in Dutch is a relic of older stages of the language. It is mainly found in the formal/archaic register; clear examples can be found in the first five lines of the 1951 translation of het Onzevader (The Lord’s Prayer) by the Nederlands Bijbelgenootschap in (246a). In informal speech, the subjunctive occurs only in formulaic expressions such as (246b).

The examples in (246) show that the subjunctive is normally formed by adding the suffix -e to the stem of the verb, but there are also some irregular forms, such as the conjunctive form zij of the verb zijn in (247a). The Dutch subjunctive is normally used to form clauses that are not declarative or interrogative. It can express incitements and wishes, as in the examples in (246), but also acquiescence, as in (247a). Example (247b) shows that the subjunctive usually takes the first or second position in the main clause and must therefore be considered a finite verb form.

The fact that the morphologically marked subjunctive is no longer part of everyday speech is evident from the fact that wishes, incitements, etc. are generally expressed by other means, like modal (ad)verbs and periphrases. A clear example of this can be found in the 2004 Bible translation by the Nederlands Bijbelgenootschap, where the subjunctives in het Onzevader in (246a) are replaced by a construction with the verb latento make; cf. Section 5.2.3.4, sub VI, for more discussion of the laten-construction in (248).

We will not discuss the present subjunctive in more detail here, but refer the reader to Haeseryn et al. (1997:103ff) for further discussion of the relics of this category in modern Dutch. Finally, note that the present subjunctive can still be recognized in certain lexical items, such as the preposition dankzij thanks to and fixed lexical expressions like koste wat het kostat all costs and godzijdankthank God.

The German past subjunctive (Konjunktiv II) is much more productive than the present subjunctive (Konjunktiv I) and is normally used to refer to non-actualized eventualities or (in literary German) in contexts of reported speech to express the speaker’s lack of commitment to the truth of the proposition; cf. e.g. Drosdowski (1995:156ff) and Palmer (2001). Dutch differs from German in that it does not have a special morphological verb form to express the past subjunctive; one case mentioned in Haeseryn et al. (1997) that can still occasionally be found in writing is ware, but it seems that most speakers use this form only in the fixed expression als het wareso to speak.

It seems that in many cases German past subjunctive constructions can simply be translated into Dutch by using a regular past-tense form. To get an impression of the semantic difference between the simple past and the past subjunctive in German, consider the examples in (250), taken from Erb (2001:69).

Placing the simple past verb in the first position of the sentence, as in (250a), leads to a regular question interpretation, whereas placing the past subjunctive in the first position, as in (250b), leads to an irrealis interpretation: the speaker is expressing a wish. The German examples in (250) can be easily translated by the examples in (251), both of which use the past-tense form waswas, but have a different (interrogative versus exclamative) intonation contour of the sentence.

The difference in meaning between the two Dutch examples is exactly parallel to the difference between the two German examples in (250). This suggests that Dutch is like German in that it also has a past subjunctive, although the form of the Dutch past subjunctive happens to be identical to that of the simple past. One argument in favor of this proposal is that the interrogative construction can easily occur in the present, whereas the irrealis construction cannot.

The use of the past tense in irrealis contexts is very common in Dutch, and the examples in (253) show that the past tense can be expressed on both main and non-main verbs.

Note, however, that the irrealis meaning only arises in examples like (251b) and (253) when a modal particle such as maar is present; the examples in (254) show that without such a particle the irrealis reading becomes impossible. The unacceptability of these examples therefore suggests that the irrealis reading arises as a result of combining the past tense with modal particles of this type; cf. Foolen (1993:179).

It is also very common to express irrealis without a modal particle by using a past-tense form of an epistemic modal. Such verbs are used to indicate the speaker’s judgment about the likelihood that a particular proposition is true: for example, by using the modal verb zullen in Jan zal komen morgenJan will come tomorrow, the speaker indicates that he has sufficient evidence to support the claim that the proposition morgen komen (Jan) is/will be true; cf. Section 1.5.2, sub II, for a more detailed discussion. The irrealis reading arises as a result of contextual information: the counterfactual reading of the first conjunct in (255), for example, arises from the fact that the second conjunct indicates that the assessment of the speaker-in-the-past was incorrect; cf. Section 1.5.4 for a more detailed and careful discussion.

The discussion above suggests that the irrealis reading arises as a result of temporal, modal and contextual information. The syntactic construction as a whole may also provide clues that an irrealis reading is intended. Conditional constructions in the past tense like those in (256), for example, are often interpreted with a counterfactual reading of the embedded conditional clause. Section 1.5.4 will show that this counterfactual reading is again triggered by contextual information. The primed examples show that conditional clauses can also appear with the past-tense form of zullen without any noticeable change in meaning.

A special case is the past-tense form of the verb hebben. The finite verb had in (253b) above can be interpreted as the regular perfect auxiliary hebben, but it seems that this is not always the case. Consider the examples in (257a&b) with the deontic modal verb moetento be obliged; it seems that the perfect-tense counterpart of the simple present example in (257a) is as given in (257b), where the auxiliary has an infinitival form and must therefore be in the scope of the finite modal verb. The crucial example is (257c), where we find a second instance of hebben: it must occur as a finite verb in the past tense and triggers a counterfactual reading. The fact that there is a perfect auxiliary in the scope of the finite verb had makes it quite implausible that the latter would also be a perfect auxiliary.

The examples in (257) show that the past subjunctive had is much higher in the structure than the perfect auxiliary hebbento have. In fact, the examples in (258) strongly suggest that past subjunctives such as had are always the highest verb in their clause: first, (258a) shows that the modal verb zullen is like English will in that it usually cannot be embedded as an infinitive under another verb (including the perfect auxiliary) and therefore usually occurs as a finite verb; second, example (258b) shows that zullen can still be embedded easily as an infinitive under the past subjunctive had, which must therefore be even higher in the structure.

The examples in (257) and (258) may suggest that in certain cases the past-tense form had should be considered a genuine past subjunctive form. The earlier examples in this subsection, on the other hand, strongly suggest that for other verbs it is not only the past tense that triggers the irrealis meaning, but that certain modal and contextual information is also relevant: Section 1.5.4 will argue that in many cases pragmatic considerations can indeed be used to account for such readings, which suggests that Dutch does not have an abstract past subjunctive that is morphologically identical to the past.