Restrictions on past-tense passives in Late Modern Danish

2.1 Periphrastic and inflectional passives

Finite verbs in contemporary standard Danish are inflected for tense (present vs past), mood (indicative vs imperative), and voice (active vs passive). Additional tense and modality distinctions are expressed by various periphrastic constructions, most of them consisting of a finite auxiliary plus an infinitive or past participle. Historically, the verb was also inflected for person and number, and a morphological subjunctive existed, but these forms are obsolete in the modern language (see e.g. Haugen 1976, 376–9; a brief sketch of verbal morphology in the present-day language may also be found in Herslund 2002, 63–5).

Like in the other Mainland Scandinavian languages, both an inflectional and a periphrastic passive are available. The inflectional passive is formed by adding -(e)s either to the stem or to the past-tense active form and is hence usually referred to as the ‘s-passive’ (on non-passive verbs with the -(e)s suffix, see below). The periphrastic passive consists of the auxiliary blive ‘become’ plus a past participle.^[1] The latter construction is thus a parallel to the periphrastic passives found in most of the other Germanic languages (see Askedal 2010 for a comparative survey). Two examples with the verb fælde ‘fell, log, cut down’ are given in (3) and (4), one with the s-passive and the other with the periphrastic passive:^[2]

Unlike in English – but like most of the other Germanic languages (Askedal 2010, 91) – intransitive verbs in Danish may also be passivized (on passives from intransitive verbs, see also Keenan and Dryer 2007, 345–8). In this construction, the subject position is filled by one of the obligatory expletives der ‘there’ or her ‘here’. Both the s-passive and the periphrastic passive occur in this construction, as shown in (5) and (6) with the intransitive verb danse ‘dance’:

The two passive constructions are not available in the same tense forms. Because past participles do not occur with the s-suffix, periphrastic tense forms using the past participle, such as the perfect, cannot form the s-passive, as shown by the starred forms ∗ har hørtes and ∗ har skrevets in Table 1. Hence, in the perfect (and pluperfect), one can only use the periphrastic passive: er blevet hørt ‘has been heard’; er blevet skrevet ‘has been written’. In the past tense, on the other hand, the s-passive is possible, but only with some verbs. Most importantly, weak verbs like høre ‘hear’, which form their past tense with a suffix (-ede or -te), may also form a past-tense s-passive, as shown in Table 1 and in (1). Weak past-tense passive forms like hørtes ‘was heard’ are perfectly possible (even if they have a rather low discourse frequency, as discussed below). In this, they differ from the great majority of strong verbs, i.e. verbs with suffixless past-tense forms. These do not usually form a past-tense s-passive (but see Section 2.2). The hypothetical strong past-tense passive ∗ skreves ‘was written’ is impossible in modern standard Danish. In other words, the past-tense s-passive is lexically restricted in the terms of van Lier and Messerschmidt (2022): one set of lexical items (namely, weak and a few strong verbs) may form it, while another (the majority of strong verbs) may not. Note, however, that the restrictions on the Danish s-passive differ from the cases mentioned by van Lier and Messerschmidt (2022, 7–8), as they are limited to a particular tense form – the lexical restrictions concern strong verbs in the past tense, not strong verbs in general.^[3]

The s-passive and the periphrastic passive also differ with respect to semantics, but the exact nature of this difference has been the object of some discussion in the literature, and for many speakers, it does not appear to be categorical. According to the traditional account, the distinction is primarily one of genericity vs specificity: the s-passive is preferred for habitual and generic situations, in normative statements, and with certain stative verbs, whereas the periphrastic passive is mainly used for specific (including future) situations; see e.g. Diderichsen (1962, 117, 136) or Lundskær-Nielsen and Holmes (2010, 355–61). Compare the use of the s-passive in the generic statements in (3) and (5) with the use of the periphrastic construction in (4), which concerns a specific plan for the future, and (6), where one particular situation is described.

An interpretation of the distinction in terms of mood has been proposed by Heltoft and Jacobsen (1996) and Heltoft (2006) (see also Hansen and Heltoft 2011, 741–51). According to this view, the s-passive can be described as expressing a kind of objective mood, the periphrastic passive a subjective mood. By using a periphrastic passive, the speaker makes a simple declarative statement which is ‘anchored’ in their own consciousness. By contrast, the s-passive ‘indicates the presence of another consciousness or intention located in the propositional layer’ of the clause (Heltoft and Jacobsen 1996, 208). This, the authors maintain, explains why the s-passive is often used in generic and normative statements – these do not reflect the speaker’s point of view alone, but depend on some other consciousness as well, e.g. people in general in the case of norms. According to Heltoft and Jacobsen (1996), this analysis also explains a well-known fact about the two passives, namely, that they lead to different interpretations when combined with modal verbs. In such cases, the periphrastic passive usually has a more subjective meaning than the s-passive, e.g. in the promise (‘subjective guarantee’) in (8) as opposed to the instruction in (7) (examples from Heltoft and Jacobsen 1996, 210):

The alternation between the two passives has recently been investigated in detail by Laanemets (2009, 2012, 2013), who compares the use of passive constructions in four genres, two spoken and two written.^[4] I will only report some of her findings from the newspaper and conversation corpora here, since these are by far her largest written and spoken corpora, respectively. Laanemets finds that the s-passive is frequently used in newspaper texts, whereas it is much less frequent than the periphrastic passive in the conversation corpus; as she notes (Laanemets 2012, 96), similar observations have been made in the earlier literature (e.g. Hansen 1967, iii, 54, Rehling 1934), so the lower frequency of the s-passive in the colloquial spoken language is not an entirely new phenomenon. However, Laanemets’s figures reveal an interesting detail when broken down according to the form of the passive verb (infinitive, present, or past): the difference between the newspaper texts and the conversations is mainly due to a difference in the use of the present tense. In the infinitive – e.g. in contexts with a modal verbs, as in (7)–(8) – the s-passive is more frequent than the periphrastic passive in both corpora (71.7% s-passive in the newspaper corpus, 76.4% in the conversation corpus). In the past tense, the periphrastic passive is much more frequent in both corpora (3.2% s-passive in the newspaper corpus, 0.6% in the conversation corpus). Compare the distributions of the two passive constructions in the two corpora in Table 2.

The figures in Table 2 suggest that no matter how the semantic difference between the s-passive and the periphrastic passive is characterized, both genre and linguistic context (the tense of the clause) appear to be factors influencing their distribution. Laanemets’s findings show that the s-passive is infrequent in the present tense in the spoken language, but frequently used in the written newspaper corpus. In the past tense, on the other hand, the s-passive is infrequent in both speech and writing. It should thus be kept in mind that the morphological restrictions discussed in the following section concern forms which are quite rare in contemporary Danish speech and writing. As the findings presented later in this article will show, this was not the case in all Late Modern Danish texts.

2.2 Restrictions on the s-passive

In this section, the restrictions on past-tense s-passives on strong verbs are described in more detail, and two attempts at explaining them are discussed. As already mentioned, most strong verbs can only form the periphrastic passive in the past tense, not the s-passive. This fact is generally acknowledged in descriptions of the language (see e.g. Herslund 2002, 72, Lundskær-Nielsen and Holmes 2010, 357, Hansen and Heltoft 2011, 629, 752–4), and a number of different explanations for it have been proposed in the literature. I return to these below.

As noted earlier, the s-passive is infrequent in the past tense in both the written and spoken language, but past-tense s-passives do occur. With weak verbs, i.e. verbs forming their past tense with one of the suffixes -ede and -te, the form is grammatical and productive, even if mainly encountered in the written language. The examples in (9) and (10) are from KorpusDK (Det Danske Sprog- og Litteraturselskab 2019), a large corpus of contemporary written Danish (c. 107 million words).

A few weak verbs with the suffix -te have stem modifications in addition to the suffix, e.g. bringe ‘bring’ (pst  bragte) and sælge ‘sell’ (pst  solgte) (Hansen and Heltoft 2011, 652). This does not appear to be a problem for the formation of an s-passive in the past tense, as indicated by (11) and (12):

The majority of strong verbs, i.e. verbs with a suffixless (usually ablauted) past tense, cannot form the past-tense s-passive. The ungrammatical example in (13) is from Hansen and Heltoft (2011, 755):

While descriptions of contemporary Danish agree that most strong verbs cannot form the s-passive in the past tense, the accounts differ in their level of detail. Some authors mention quite a number of exceptions to the generalization, while others mention only a few possible forms or none at all. Lundskær-Nielsen and Holmes (2010, 357) mention the four strong past-tense passives sås ‘was seen’, gaves ‘was given’ (said to have “an old-fashioned ring”), afholdtes ‘was held/organized’, and vandtes ‘was won’; according to the authors, the last two verbs allow an s-passive because their active past-tense forms end in -t. Herslund (2002, 72) only writes that “[m]ost strong verbs” have no past-tense s-passive, but gives no examples of strong verbs that do. In the reference grammar by Hansen and Heltoft (2011, 752–4), the exceptions to the generalization are discussed at greater length. According to the authors, the main exceptions are strong verbs historically belonging to the so-called sixth ablaut class, most of which have the stem vowel /o:/ in the past tense, e.g. joges ‘was chased’, toges ‘was taken’, and lodes ‘was let’. They also mention a few additional verbs, including some ‘prefixed strong verbs in the written language’ (“Sammensatte skriftsproglige stærke verber;” Hansen and Heltoft 2011, 752) belonging to other ablaut classes, e.g. frigaves ‘was released’ (from give ‘give’, fifth ablaut class). (See also Heltoft 2006, 276–8 for a brief overview in English.)

Three corpus examples of strong past-tense s-passives are given in (14)–(16): a verb belonging to the sixth ablaut class (toges), one of the verbs with -t mentioned by Lundskær-Nielsen and Holmes (2010) (vandtes), and a prefixed verb mentioned by Hansen and Heltoft (2011) (frigaves).

In Table 3, I have attempted to compile a comprehensive list of the strong past-tense s-passives attested in contemporary Danish. The list is organized according to the historical ablaut classes, like the list compiled by Heltoft (2006, 277–8), but gives more forms than Heltoft’s overview. Note that I only list a small selection of the hypothetical ungrammatical s-passives; the total number of strong verbs is much larger (about a hundred; Hansen and Heltoft 2011, 656–9 list 104 underived strong verbs in the present-day language). The overview in Table 3 was compiled from the grammatical descriptions discussed earlier and supplemented with forms found in KorpusDK; a question mark before a form indicates that the verb is mentioned in the literature but that the simplex form was not found in the corpus. A number of verbs, such as -fandtes, -lodes, and -stodes, are attested only with prefixes, as pointed out by Hansen and Heltoft (2011). Some examples of such forms are listed in the last column. As the table suggests, the past-tense s-passive is indeed more frequent with verbs of the sixth ablaut class, but is also found with a few other strong verbs.

Note that some strong verbs which do not allow the past-tense s-passive are found as past-tense deponents (middles). As discussed by Heltoft (2006, 274–5), the s-passive historically developed from an earlier reflexive construction, following the well-attested pathway reflexive  →  anticausative (‘intransitive’ in Heltoft’s terms) →  passive (see Haspelmath 1990; Heltoft and Nielsen 2019, 212–5). Such reflexive and anticausative verbs are still relatively frequent in the language, such as enes ‘agree’, skændes ‘argue’, slås ‘fight’, længes ‘long, miss’, or findes ‘be, exist’. These occur in past-tense and participial forms without any of the restrictions applying to passives. Hence, a strong deponent verb like slås ‘fight’ (from slå ‘beat’) may form the past tense sloges, as shown in (17). This would also hypothetically be the past-tense s-passive of slå ‘beat’, but the form is not available with this meaning – the past-tense passive of slå has to be formed with the auxiliary blive, as in (18):

Such examples show that the restrictions on the s-passive cannot be due to phonotactic or other purely formal constraints: a form like sloges is perfectly possible in Present-Day Danish, it is just not available as a past-tense passive form, only as a deponent (as also noted by Heltoft 2006, 278).

A number of different explanations for these restrictions have been proposed. I will only discuss the two most recent ones here; for a concise discussion (and criticism) of some earlier proposals see Laanemets (2012, 29–32). In both of the most recent accounts, syncretism within the verbal paradigm plays a crucial role, but the accounts differ in important respects. Whereas Heltoft (2006) proposes a historical-functional explanation, the account in Lundquist (2016) only makes reference to the synchronic grammatical system.

According to Heltoft (2006) (see also Hansen and Heltoft 2011, 753–7), the ‘gaps’ in the past-tense paradigms in Present-Day Danish are to be explained by the historical development of the system and the meaning of the s-passive. Very briefly stated, according to the author, only verbs which had identical past indicative and past subjunctive stems in Middle Danish (c. 1100–1500) can form a past-tense s-passive in the contemporary language. This includes all weak verbs as well as strong verbs of the sixth ablaut class. The reason for this, according to Heltoft, can be found in the modal character of the s-passive, which precluded it from being combined with indicative stems in Middle Danish. Since most of the strong past-tense forms in Present-Day Danish go back to Middle Danish indicative forms, these never allowed a past-tense s-passive. However, because weak verb and strong verbs of the sixth ablaut class had indicative–subjunctive syncretism in the past tense (i.e. no formal mood opposition), the past-tense s-passive could be formed in these conjugation classes without any problems. Hence, under this view, the present-day situation is a historical accident, as it were, owing to the patterns of syncretism in the Middle Danish verbal paradigm.

Lundquist (2016) also accounts for the restrictions by referring to syncretism, but in the paradigms of the present-day language. According to this explanation, the past-tense s-passive in contemporary Danish (and Norwegian Bokmål^[5]) is licensed by the partial syncretism between the past active forms and the participle. The reason for this, in Lundquist’s view, is that the morphological passives in the two languages are underlyingly tenseless; finite passives in a sense ought not to be grammatical at all, but are acquired by the language learner because of ‘form-parasitism’ on other forms in the paradigm. In other words, a present-tense passive form like bruges ‘is used’ is possible because it is syncretic with the passive infinitive, and a past-tense passive like brugtes ‘was used’ is only possible because the participle brugt with the same stem exists. If there is no syncretism between the simple past (e.g. sang ‘sang’) and participial stem (sunget ‘sung’), no past-tense s-passive is possible (see in particular Lundquist 2016, 206–9).

Although my material from Late Modern Danish will not be able to confirm or disprove either of these accounts unequivocally, I note here that they both have a number of potential weaknesses – and that my Late Modern Danish findings may at least cast some doubt on both of them. As for Heltoft’s historical explanation, it should be noted that it depends on a particular ‘modal’ analysis of the Danish passive alternation (outlined in Section 2.1) and further assumes that the same semantic distinction between the s-passive and the periphrastic passive already applied in Middle Danish; while this might turn out to be the case, it is an empirical question which remains to be investigated in detail.^[6] In addition, Heltoft’s account presupposes that there has also been morphological stability between the late Middle Ages and the present day: the past-tense s-passives which are possible in the contemporary language, according to Heltoft, are the ones which were possible in Middle Danish. As the findings presented in Section 4 will show, however, quite a number of past-tense s-passives are attested in Late Modern Danish texts which are not expected according to Heltoft’s account, e.g. of verbs belonging to the first and second ablaut classes.

As for Lundquist’s synchronic explanation, it should be stressed that it only attempts to explain why weak past-tense s-passives are possible – it does not account for the existence of strong past-tense forms like toges ‘was taken’ or frigaves ‘was released’, which have no syncretism with the corresponding past participles (taget, frigivet) and are therefore unexpected in Lundquist’s account. The author acknowledges this in a footnote (Lundquist 2016, 208) and suggests that these might just be lexical exceptions to the general rule “or part of a more general pattern,” though he does not comment on the fact that strong past-tense s-passives appear to be more frequent in the sixth ablaut class.^[7] In any event, if such forms can be shown to have been more frequent in the past – in this case in Late Modern Danish – this suggests that Lundquist’s syncretism analysis cannot be extended to earlier stages of the language, but only works (with several exceptions) for Present-Day Danish.

As this overview has shown, one of the Danish passive formations is severely restricted in the past tense. While the overall discourse frequency of the s-passive is low in the past tense both in speech and writing (see Laanemets’s findings), some forms are clearly felt to be grammatical and are used at least in the written language. A past-tense s-passive can be formed from weak verbs (see [9]–[12]) and a minority of strong verbs (e.g. [14]–[16]). There is no agreement on why this would be the case, and there are few empirical investigations of the restrictions. In the following section, I will discuss the reasons for looking at their historical development and present the methods used in this study.

4.1 Restrictions according to early grammars

The grammars surveyed for this part of the study are listed in Table 4. The authors, titles, and years of publication are given, and in the final column, I indicate whether the grammars mention any restrictions on past-tense passives of strong verbs. For the ones that do, the forms said to be ungrammatical are listed.^[10] Note that the grammars included in the survey were all written either by native speakers of Danish or by people living and writing in Denmark, who we can assume must have had a good knowledge of the language; a few grammars written by non-native speakers living elsewhere were excluded. It should also be noted that all of the grammars must to some extent be considered prescriptive by today’s standards. The target audience was of course not modern linguists, but contemporaries of the authors who needed to learn (standard) Danish, such as foreign language learners (e.g. the grammars by Tode, Schneider, Otté) or students in the Danish school system (e.g. Høyer, Jensen, Bojesen). It cannot be ruled out that some judgements in the grammars reflect the author’s ideas of ‘proper’ language rather than actual usage at the time. With these caveats in mind, let us now look more closely at the restrictions mentioned in the grammars.

As Table 4 shows, about half of the surveyed grammars mention restrictions on past-tense passives. The two (short) seventeenth-century descriptions contain no information on any restrictions. Of the eighteenth-century grammars, Høysgaard (Methodisk Forsøg, 1752) and Baden (Forelæsninger, 1785) mention restrictions, but Baden’s list of ungrammatical past-tense passives is clearly copied almost verbatim from Høysgaard’s. Seven of the nineteenth-century grammars mention restrictions. In the grammars that mention restrictions, the level of detail and the number of examples differ greatly. For instance, in Rask’s Grammar of the Danish Language (1830), there is only a very oblique reference to any restrictions. In the verbal paradigms, various forms are given, including the strong past-tense passives gaves ‘was given’, droges ‘was drawn’, fandtes ‘was found’, dreves ‘was driven’, and strøges ‘was stroked’. However, in the paradigm for faa ‘get’, the past-tense passive form fikkes is given between brackets (p. 51). While Rask uses brackets for somewhat different purposes, e.g. to indicate obsolete forms or innovations not belonging to the ‘cultivated’ language, the common denominator is that he did not consider the bracketed forms part of the standard language. The brackets around fikkes most likely indicate that the form is hypothetical and was not actually used in Rask’s day.

Other grammarians discuss passive morphology at greater length, including the restrictions in the past tense. Høysgaard (1752) writes that ‘in particular, there are many passives whose simple imperfect [past] is never used’ (“I sær ere der mange Passive, hvis Enkelte Imperfectum aldrig bruges,” § 1649) and gives several examples of past-tense s-passives ‘which no one uses’. For instance, the periphrastic forms in (20) are said to be used instead of the s-passives sködes ‘was shot’ and stakkes ‘was stabbed’:

Similarly, Matzen’s Modersmaalets Sproglære (1893) gives a number of examples of periphrastic passives in the past tense, noting explicitly that the s-passive form is not used, e.g. in (21):

The material surveyed here is obviously too limited for any statistical analysis, and it should also be kept in mind that the observations in the grammars may not be entirely independent – some of the grammarians certainly read each other’s works and may have borrowed examples from each other (cf. the comment above on Baden borrowing his examples from Høysgaard). Hence, the fact that the verbs drikke ‘drink’ ( ∗ drakkes) and synge ‘sing’ ( ∗ sanges) are often singled out as having restrictions on the s-passive may just be coincidental. On the other hand, it indicates that many grammarians at least agreed on the unacceptability of these s-passives. Among the forms listed in Table 4, the third and fourth ablaut classes seem to be especially frequent (e.g. drikke and synge, but also hjælpe ‘help’, tvinge ‘force’, stjæle ‘steal’). From the first and fifth classes, on the other hand, only a single example of each is mentioned in the surveyed grammars (bide ‘bite’ and æde ‘eat’, respectively), whereas no verbs from the sixth class are mentioned as having restrictions. This could be coincidental, but it might also indicate that past-tense s-passives of these ablaut classes were generally more acceptable. Unfortunately, only few of the grammars explicitly mention past-tense s-passives that do occur. Among these, skreves ‘was written’ from the first ablaut class is the most commonly cited form (Werfel, Tode, Høyer, Matzen). The first-class form dreves ‘was driven’ is mentioned in two grammars (Rask, Jensen), as is grebes ‘was seized’ (Jensen, Jessen). None of these forms are used in Present-Day Danish, but as the corpus study in Section 4.2 will show, at least two of them are attested in other nineteenth-century texts.

It is also worth noting here that the grammars which mention restrictions do not generally venture any explanations for them.^[11] The discussions are often rather short, and as Table 4 shows, many of the grammars give only a few examples of ungrammatical past-tense forms. One grammarian (Heckscher, p. 158) explicitly states that there are no rules for the choice between s-passive and periphrastic passive (“darüber lassen sich keine Regeln aufstellen”), as this depends entirely on euphony (“vom Wohnklange”). Others advise that a past-tense s-passive should be avoided if the form could be confused with another verb, such as a present-tense s-passive or a deponent verb. This is mentioned by Høysgaard (Methodisk Forsøg), Baden, Høegh-Guldberg, Jensen, and Matzen. For instance, a number of these grammarians point out that the potential past-tense passive form sloges ‘was beaten’ is identical to the deponent verb (‘fought’, cf. Section 2.2), and that the periphrastic passive with blive is thus preferable. Another example, mentioned by Høegh-Guldberg, Jensen, and Matzen, is the potential past-tense s-passive saaes/saas ‘was seen’ (modern spelling sås), which according to these authors might be confused with the present-tense s-passive of the verb så ‘sow’; Matzen (p. 302) gives the example Frøet saas i August, which he states would be ambiguous (‘The seed is sown in August’ or ‘The seed was seen in August’) if the past-tense s-passive of se ‘see’ was allowed. However, I think these comments must be interpreted as prescriptive advice rather than serious attempts to explain the restrictions. As the grammarians in question must have realized, most of the ungrammatical past-tense s-passives (e.g. ∗ drakkes, ∗ hjalpes, ∗ sanges, ∗ aades, and so on) are never subject to any ambiguity, so the majority of the verbs would still be unaccounted for. It is probably also rather doubtful that a form like sås ‘was seen’/‘is sown’ ever resulted in ambiguity in actual linguistic usage, and in any event, this is actually one of the strong past-tense s-passives which is attested – both in Present-Day Danish (see Table 3) and in my Late Modern Danish material.

Before turning to the primary sources, I will briefly discuss an important finding concerning the sociolinguistic status of the s-passive in the period. Although many of the grammars surveyed here mention s-passive forms which they judge to be ungrammatical, some of them also explicitly advise readers to use the s-passive as much as possible. Two different reasons are given for why the s-passive is preferable to the periphrastic passive. In Jensen’s Forsøg til en dansk Sproglære (1833), the argument is that the s-passive is more ‘characteristic’ of Danish because many other languages have only periphrastic forms:

[…] the two expressions [s-passive and periphrastic passive] are usually employed without any distinction; but it should be noted in this connection that, where the simple form is possible, it would seem that it, as more characteristic of our language, is to be preferred^[12] (Jensen, p. 338; my transl.)

In Tode’s Danish grammar for speakers of German (1797), the argument is that the synthetic formation gives the language more ‘brevity and strength’ compared to a periphrastic construction. Tode also observes that the periphrastic passive is more frequent than the s-passive in the spoken language, in his view clearly a regrettable state of affairs:

Such a simple and unparallelled most important transformation, i.e. that of an active into a passive, through the addition of a single letter [i.e. -s], gives the language an extraordinary brevity and strength, and I therefore hold it against the Danes that they do not make use of this peculiar passive every day, but usually, at least in the spoken language, and sometimes even in formal speech, make use of an auxiliary.^[13] (Tode, § 543; my transl.)

Finally, Høegh-Guldberg uses a combination of these two arguments: most other modern languages lack such a convenient formation as a synthetic passive, and hence, Danish has an advantage over these languages. He explicitly advises his readers to use the s-passive as much as possible:

Because our language, unlike many other modern ones, can usually form a passive in the present and imperfect without an auxiliary, we should not let this advantage and convenience pass into oblivion, by too often using the also permissible formation with the auxiliary blive. We should therefore preferably say: ledes and lededes, mættes and mættedes, frelses and frelsedes, rather than: bliver and blev ledet, bliver and blev mættet, bliver and blev frelset!^[14] (Høegh-Guldberg, p. 224; my transl.)

At the same time, Høegh-Guldberg goes on to concede that one must ‘resort to the auxiliary’ in the case of some irregular (i.e. strong) verbs. There is thus a clear tension in the advice given here. On the one hand, the s-passive is considered the preferable Danish form which one should use whenever possible; on the other hand, Høegh-Guldberg agrees with several of his contemporaries that not all potential s-passive forms are actually grammatical.

It is of course impossible to know exactly what the impact of such prescriptive statements may have been. Perhaps some readers of these grammars actually followed the advice and deliberately attempted to increase their use of the s-passive; for others, such advice may have had little or no effect. The statements are interesting, however, because they suggest that a particular linguistic ideology was at work in the Late Modern Danish period: the synthetic s-passive seems to have been valued over its periphrastic alternative, even though some of the grammars also mention that the periphrastic passive was the more usual construction in the spoken language. (I will discuss a very similar example from the history of German in Section 5.) The findings presented in the following section will indeed suggest that such prescriptive attitudes may have had an effect on linguistic behaviour.

4.2 Past-tense passives in primary sources

As mentioned in Section 3, I carried out a search for past-tense passives in the writings of 11 Late Modern Danish individuals. Some of these are well-known writers, others are more obscure historical figures. The texts may be divided roughly into three types, namely, academic texts, fiction of various kinds, and autobiographical texts and letters (ego-documents). The texts are listed in Table 5. Together they run to more than 700,000 words, but the length of the individual texts varies greatly. The shortest texts are the autobiography (“Levnetsbeskrivelse”) of the soldier Hans Pedersen and a collection of letters by the socialite and writer Kamma Rahbek. By far the longest text is the autobiography of the criminal Ole Kollerød.

In order to have a degree of control over regional background as a possible variable, I only included writers from Zealand or Funen, i.e. the central Danish islands.^[15] Standard Danish originated in this dialect area, and most of the writers included here had a strong command of the standard written language. The academic and fictional texts all appear to conform closely to the standard written language of the period, as does the published autobiography of the naval officer Carl van Dockum. The other autobiographies (S. Pedersen, H. Pedersen, and Kollerød) were not written to be published, however, and the authors had a varying degree of proficiency in the standard written language. The findings presented later suggest that this correlates with the use of past-tense s-passives. One author, Jens Baggesen (1764–1826), is represented by two texts in the corpus. One is a translation of Ludvig Holberg’s Nicolai Klimii Iter Subterraneum (‘Niels Klim’s Subterranean Journey’), a fantasy novel originally published in Latin (1741); the other is Baggesen’s own Labyrinten (‘The Labyrinth’), a travelogue describing the author’s journey from Copenhagen to Basel in the year 1789. The differences between these two texts will be discussed briefly in the following section.

Table 6 gives an overview of the use of the two past-tense passive constructions in the texts. The three columns in the middle give the number of past-tense s-passives, past-tense periphrastic passives, and past-tense passives in total. The shares of the two constructions are given between brackets. So, for instance, in Baggesen’s translation of Niels Klim, there are 62 (=25.4%) s-passives and 182 (=74.6%) periphrastic passives in the past tense, giving a total of 244. As the table shows, in the letters by Rahbek, only four past-tense passives were found (not too surprisingly, as the letters are mainly written in the present tense); because of this low number, I will not consider these letters further here.

In the final column of Table 6, I have listed the strong s-passives attested in the texts, including the number of attestations of each form. A hyphen before a form indicates that it is only attested with a prefix; for instance, in Kierkegaard’s Forførerens Dagbog a prefixed form (oplodes ‘was opened’) is used once, but not the corresponding simplex verb (lodes ‘was let’). A hyphen between brackets indicates that both simplex and prefixed s-passives are attested; for instance, in the writings of Baggesen, simplex fandtes ‘was found’ is found alongside a prefixed form (befandtes ‘was found, judged’).

From the perspective of Present-Day Danish, some of the strong s-passives found in the texts are surprising. Compare the forms in Table 6 with the ones listed for the contemporary language in Table 3. As I discussed earlier, in Present-Day Danish strong past-tense s-passives mainly seem to occur with verbs belonging to the sixth ablaut class, e.g. droges, toges and prefixed forms like overlodes and fastsloges. These verbs also occur in the Late Modern Danish texts, as do (-)saaes ‘was seen’ (modern spelling sås), -gaves (from give ‘give’), and vandtes ‘was won’. But the historical material also contains forms which are no longer current in the language. These include grebes ‘was seized’ and skreves ‘was written’ (both ablaut class 1), (-)bødes ‘was bidden’ and brødes ‘was broken’ (class 2), and the prefixed form ombares ‘was carried around’ (class 4, from bære ‘carry’). The simplex forms fandtes ‘was found’ (class 3), gaves ‘was given’ (class 5), and sloges ‘was beaten’ are also found, which in the present-day language are only attested with prefixes. In other words, while some forms found in the historical material still occur in the present-day language, a number of forms in the eighteenth- and nineteenth-century texts have since become obsolete. The situation in the Late Modern Danish material is thus clearly not identical to the present-day state of the language.

It should be noted that in a small minority of cases, a past-tense form may be interpreted either as an s-passive or a deponent (‘middle’) verb, as the suffix -(e)s is shared by the two constructions (cf. also the languages discussed by Keenan and Dryer 2007, 352–3). As mentioned in Section 2.2, the distinction between s-passives and deponents is important in Present-Day Danish because past-tense deponent verbs are not subject to the same restrictions as s-passives. For instance, the deponents slås ‘fight’ and findes ‘be, exist’ may readily be used in the past tense (sloges and fandtes). Two possible deponents from the corpus are fødtes ‘was born’ (22) and kaldtes ‘was called’ (23), from the weak verbs føde ‘bear, give birth’ and kalde ‘call’. These could be interpreted as deponent one- and two-place predicates, respectively, rather than true passives with an implied agent. In (22), the focus is not on the process of being born, but on the fact that some people in ancient Albania, according to the author, had grey hair from early childhood. The verb in this context could perhaps simply be paraphrased ‘existed’. In (23), kaldtes means ‘was called’ in the sense ‘went by the name of’, i.e. the focus is on the name of this fairy-tale character rather than on the act of calling her by this name.

However, since both of these forms are still transparently related to their active counterparts (and may in fact be substituted by periphrastic passives, at least in the present-day language), they were counted as past-tense s-passives and included in the count in Table 6.

A tricky case is the use of fandtes in Ole Kollerød’s autobiography (spelt fandes or fantes in his non-standard orthography). Interpreted as a passive, this form means ‘was found’; as a deponent, it means ‘was, existed’. Only the latter meaning is found in Present-Day Danish, and the verb frequently occurs with this sense in Kollerød’s text, but there are a few examples where a passive interpretation seems quite likely. In (24), Kollerød relates an episode where he was arrested and searched because he had been accused of theft. Because none of the stolen goods were in his possession any longer, he of course denied this:

This might also be interpreted as a deponent, with fandes meaning ‘were (in a certain place)’, but the situation described here makes a passive interpretation seem quite likely to me. I have counted (24) and one other example (den fantes hos mig ‘it was [found] in my possession’) as passives, but it must be noted that they could also be interpreted as deponents; hence the question mark next to fandes/fantes in Table 6. In any event, as the table shows, the number of past-tense s-passives in this text is very low no matter how these individual examples are analysed.

Turning to the overall frequency of past-tense passives, the analysis reveals great differences in the shares of s-passives versus periphrastic constructions. For the sake of clarity, the relative frequencies from Table 6 are shown on a bar chart in Figure 1. As the chart suggests, the texts cluster in three groups. In the first group, consisting of the autobiographies by Hans Pedersen and Ole Kollerød, the s-passive is never or hardly ever used in the past tense. Pedersen has no past-tense s-passives at all (i.e. 0%), Kollerød only very few (2.8%). The second group consists of all the fictional texts and two autobiographies, by the copyholder Søren Pedersen and the naval officer Carl van Dockum. In these, the s-passive is clearly in the minority in the past tense, but cannot be described as marginal: its share ranges from 14.9% (Gyllembourg) to 25.4% (Baggesen NKl.). In the third and final group, the s-passive is in the majority in the past tense compared to the periphrastic construction. These are Baggesen’s travelogue Labyrinten (60.5% s-passives) and the two academic texts, Rask’s treatise on the origins of the Old Norse language (79.7% s-passives) and an article by the historian F. Schiern (66.7% s-passives). The implications of these frequency differences will be discussed in the following section.

Figure 1

Shares of past-tense passives in the texts.