Jump to ContentJump to Main Navigation
Show Summary Details
More options …


An Interdisciplinary Journal of the Language Sciences

Editor-in-Chief: van der Auwera, Johan

6 Issues per year

IMPACT FACTOR 2017: 0.644
5-year IMPACT FACTOR: 0.878

CiteScore 2017: 0.79

SCImago Journal Rank (SJR) 2017: 0.418
Source Normalized Impact per Paper (SNIP) 2017: 1.386

See all formats and pricing
More options …
Volume 56, Issue 1


Reproachatives and imperatives

Daniël Van Olmen
  • Corresponding author
  • Department of Linguistics and English Language, Lancaster University, Bailrigg, Lancaster LA1 4YW, UK
  • Email
  • Other articles by this author:
  • De Gruyter OnlineGoogle Scholar
Published Online: 2017-12-14 | DOI: https://doi.org/10.1515/ling-2017-0033


This paper studies constructions dedicated to the expression of an after the fact reprimand to a second person in the languages of Europe. Taking a usage-based perspective, it argues against earlier analyses of these reproachatives as imperatives, optatives or conditionals, which fail to capture their idiosyncrasies and overpredict both their cross-linguistic frequency and the grammaticality of types of imperative in a language. Based on a closer examination of Dutch, the paper assumes a middle position between the existing views in that it argues for an account of the Dutch reproachative as the hybrid outcome of the interaction of the aforementioned constructions and of processes such as analogy, conventionalization and insubordination. It explores to what extent such an analysis applies to the other European languages featuring a reproachative and what its implications are for our understanding of imperative semantics.

Keywords: reproachative; imperative; Dutch; languages of Europe

1 Introduction

Although directivity and past tense do not seem a natural fit, it has been argued in the literature that there exist languages with “past imperatives”. The classic example is Syrian Arabic (e.g., Palmer 1986: 112; Aikhenvald 2010: 132; Malchukov 2011: 242). In the Palestinian variety too, as in (1), the morphological imperative can combine with a second person past form of ‘be’ or with the non-agreeing past form of the verb. 1 With this construction, the speaker appears to reprimand the addressee 2 for not following a course of action that she has established as the right one only after the facts and to urge him to own up to his poor choice.


Palestinian Arabic (Afro-Asiatic, Semitic) 3


‘You should have eaten when you were at home!’

(Karawani p.c.)

It is constructions with this meaning that are the topic of the present article. In earlier research on Spanish and Dutch, they have been called, among other things, retrospective or irrealis imperatives (see Bosque [1980] and Duinhoven [1995] respectively). Following Kaufmann (2012: 102–104), however, we will refer to them as reproachative constructions.

The term “reproachative” has been chosen because it only alludes to the constructions’ function. The fact that it does not make any assumptions about their form is useful in light of the debate about their structural status: some linguists regard the reproachative constructions in Spanish and Dutch, for instance, as imperatives (e.g., Mastop 2005: 72–74; Vicente 2013) but others explicitly or implicitly reject such a view (e.g., Alcazár and Saltarelli 2014: 140; Jary and Kissine 2014: 100) and propose to analyze them as, for example, optatives or conditionals (e.g., Bennis 2007: 124; Biezma 2010). 4 In the present article, we will argue for a middle position between an imperative analysis and a non-imperative one in the languages under consideration. Both analyses can be said to be reductionist in that they seek to capture the complexity of reproachatives in terms of one existing construction. The former will be shown, inter alia, to necessitate an unwelcome expansion of imperative semantics and to wrongly predict constructions such as (1) to be frequent cross-linguistically. The latter will be argued to ignore “the tight semantic relation between imperatives and reproachatives” (Kaufmann 2012: 103) and the formal features that they share in certain languages. Our account, by contrast, will try and do justice to the hybrid nature of the reproachative constructions in the languages under examination.

The analysis proposed in this article will appeal to the principles of the usage-based model and of construction grammar. Put differently, language structure is regarded here as being shaped by language usage: it is influenced by specific instances of use and it can change through the repetition of such instances (e.g., Bybee 2006: 715–718). Language is also seen as consisting not of discrete categories but of a network of conventional form-meaning pairings or constructions varying in levels of abstraction and complexity and interconnected formally and/or functionally in diverse ways (e.g., Goldberg 1995: 1–23; Bybee 2010: 76–104, 136–150). The present study will make little or no use of quantitative corpus evidence, though. The cross-linguistic perspective that it takes means that we inevitably depend on earlier descriptions and information from native speakers and scholars of the languages under investigation, if available. In fact, for lack of detailed data about languages from other continents, 5 the focus will be on Europe and the Palestinian Arabic example in (1) will not be examined in any more detail. Another reason for the non-quantitative nature of the study is that, in the languages for which corpora may be used, such as Dutch, the reproachative construction cannot be searched for in a straightforward manner and appears to be fairly infrequent anyway (e.g., Van Olmen [2011: 30] mentions only one attestation in a 370,000-word corpus of spoken language and drama texts). In short, we will not undertake a systematic analysis of usage data as such. Rather, we will draw on insights form the usage-based model and from construction grammar (e.g., the potential conventionalization of a pragmatic use as part of the meaning of a construction) to explain the reproachatives in the languages of Europe.

The rest of this article is structured as follows. In Section 2, we will first define the notion of reproachative and then give an overview of the constructions in the languages of Europe that match our definition. Section 3 will look at existing accounts of reproachatives and discuss the problems with their analysis as either an imperative or a non-imperative. In Section 4, we will put forward our own analysis of the phenomenon. It will take Dutch as its starting point and then move on to the other languages. Section 5, finally, is the conclusion, which will briefly consider the implications of our account for the theoretical characterization of the imperative.

2 Reproachatives in the languages of Europe

2.1 Definition

In light of Bosque’s (1980: 416–418) and Duinhoven’s (1995: 346–348) descriptions of the meaning of their retrospective and irrealis imperatives, we define the reproachative in the following three-part way: it is (i) a construction of grammar (ii) which is essentially dedicated to the expression of (iii) a reprimand of the addressee by the speaker for not having exhibited some behavior that she has determined to be the appropriate one retroactively and of an appeal to him to acknowledge the faultiness of his course of action and to face its consequences. The meaning of a reproachative in (iii) can roughly be paraphrased as ‘you should have + past participle!’. It differs from that of the comparative concepts of imperative and optative: the former typically refers to the non-past (see Takahashi 2012: 71–72; Jary and Kissine 2016: 132) and, as Mastop (2005: 72) points out, the latter does not normally have the directive-like communicative function of attempting to commit – in the sense of the aforementioned stimulus – the addressee to an action that he did not carry out (see also Dobrushina et al. 2005).

Not all formal manifestations of the function described in the preceding paragraph count as reproachatives. Our characterization of the phenomenon in terms of grammar or, put differently, syntax and/or morphology in (i) excludes more lexical ways of conveying this type of reprimand from the discussion. In addition, as (ii) states, the construction of grammar needs to be dedicated to the expression of reproachative meaning. In European Portuguese, for instance, the past perfect subjunctive is used, though infrequently, “to express a contrary-to-fact wish directed to the interlocutor” (Becker 2010: 182), as in (2). The conditional protasis can occur without an apodosis and has clear reproachative overtones here but, in fact, the construction is not limited to the second person and the basic meaning is optative (Almeida p.c.; Becker p.c.). Its function as a reprimand is not part of the semantics but an implicature, which an unfulfilled and now unfulfillable wish can convey when it contains an action verb and a second person subject (see Haberland [2010: 484] for a comparable example in Greek).


Portuguese (Indo-European, Romance)


estava pronto!

be.ipfv.3sg ready

‘Had you peeled the potatoes, lunch would already be ready!’

(Hundertmark-Santos Martins 1982: 218)

Dedication is a matter of degree, however. The extent to which a construction is conventionalized with a particular meaning may vary (see Bybee 2010: 28–31) and, as we will discuss in more detail later, there may be language-specific reasons for relaxing, for instance, the criterion of addressee-orientedness. Our definition of reproachatives therefore reads “essentially dedicated” rather than simply “dedicated”.

2.2 Survey

To our knowledge, which is based on reference grammars and more specialized literature as well as on information from native speakers and other linguists, only six European languages have a reproachative construction: Spanish, Breton, Estonian, Latvian, Dutch and Hungarian. In the first four languages, it takes the form of a non-finite past participle. In the last two, a finite verb form is used. The two types are discussed in Sections 2.2.1 and 2.2.2 respectively.

2.2.1 Non-finite past participle constructions

Probably the most well-known language with a reproachative is Spanish, as in (3a). Bosque (1980: 415–416) calls the construction a retrospective imperative and argues that it has numerous properties in common with the regular Spanish imperative, the most important of which at this point is that it cannot be directed at a first or a third person, as (3b) and (3c) show.


Spanish (Indo-European, Romance)


‘You should have said it earlier!’

(Ríos García p.c.)


‘I should have got up earlier.’

(Bosque 1980: 416)


‘John should have come.’

(Bosque 1980: 416)

The Breton example in (4) closely resembles the Spanish one in (3a), but with the perfect requiring the auxiliary ‘be’ instead of ‘have’. Hewitt (2010: 305) characterizes it as “expressing a wishful imperative or suggestion” and Gros (1984: 318) as conveying a moral obligation that has not been abided by and as having the meaning and intonation of a reprimand. It should be noted, though, that this use of the construction has been described specifically for the Trégor variety of the language and does not appear to be acceptable for all native speakers of Breton (Le Pennec p.c.).


Breton (Indo-European, Celtic)


‘You should have told me.’

(Hewitt 2010: 305)

The third language that features a non-finite past participle construction as a reproachative is Estonian (see also Aikhenvald 2010: 132–133). The example in (5) comes from a corpus of online language and exhibits all the constructional traits that this use of the past participle typically has in the data (Metslang p.c.): there is no overt subject, the verb is in sentence-initial position and it is often followed by the particle siis ‘then’, which is indicative of the relation of a reprimand with the preceding context. 6


Estonian (Uralic, Finnic)


‘You should have told this before the elections!’

(etTenTen Corpus of Estonian)

However, what Muižniece et al. (1999: 147) describe as the past participle’s use as “an imperative expressing a command that is directed to the past (and thus cannot be carried out anymore)” is just one of the many cases of this verb form’s “finitization” in the language. It may serve, among other things, as a quotative, as in (6).




‘You are said to have visited the theater yesterday.’

(Metslang and Sepper 2010: 542)

Still, most of the other finite uses of the past participle all involve slightly different structures. In (6), for instance, the subject cannot be omitted. Muižniece et al.’s (1999) overview of past participle finitization in Estonian suggests that only the use as a so-called jussive past form, of which (7) is an example, is identical in form to the reproachative. But, as we will argue in Section 4, the fact that the construction appears to be vague with respect to the person to whom the post hoc obligation applies may not be so surprising if the imperative paradigm of the language is taken into consideration.




‘He overslept this morning.’


‘He should have come home in time last night!’

(Muižniece et al. 1999: 148)

Latvian, finally, is very similar to Estonian. Like its neighbor, it has a wide range of finite uses of the past participle, including a reproachative one. Moreover, as (8) shows, the construction with the verb in sentence-initial position and no overt subject is ambiguous with regard to the person at whom it is directed as well, though the number of possible interpretations is smaller as past participles are marked for number and gender in this language.


Latvian (Indo-European, Baltic)


‘You/he should have come home!’

(Muižniece et al. 1999: 134)

According to Muižniece et al. (1999: 147), a second minor difference with Estonian is that the use in (8) really requires the presence of “a word that expresses some emotionality”. Reiz ‘once’ and the reinforcer tak appear to fulfill that function here. Example (5), in comparison, contains no such particles.

2.2.2 Finite verb constructions

One of the two European languages in which the reproachative involves a finite verb is Dutch. The reference grammar of the language, Haeseryn et al. (1997: 106), more or less describes the construction in (9a) as conveying a counterfactual wish. But the strangeness of (9b), whose predicate would be perfectly acceptable in an optative construction with the meaning ‘if only you had been taller’, and the ungrammaticality of (9c) make clear that this characterization misses the fact that (9a) appeals to the addressee in the way typical of a reproachative (see Mastop [2005: 72] and Section 3.2.1, in which we will briefly look at the structural differences with the Dutch optative as well). Note also that the construction evokes a strong sense of conditionality. In (9a), for instance, an apodosis of some kind is felt to be present (e.g., ‘if you had kept quiet, none of this would have happened’). It calls to mind the positive consequences of the behavior that the speaker would have preferred to have seen in the addressee and may even be expressed by means of, say, a clause headed by dan ‘then’.


Dutch (Indo-European, Germanic)


‘You should have kept quiet!’


‘You should have been taller!’


‘She should have kept quiet!’


‘Keep quiet!’

The use of the past participle in (9a) is, of course, reminiscent of the reproachative constructions discussed earlier. The main difference with the Spanish and Breton ones in (3a) and (4) is that the perfect auxiliary is finite here in that it carries tense (and number) marking. What is more, the fact that this past singular form occurs in sentence-initial position and does not have an overt subject makes the construction look not unlike the imperative in Dutch, as in (9d) (zwijg ‘keep quiet!’ is the verb stem, which is also employed for the present first person singular) (see Boogaart and Janssen 2010: 124). 7 This similarity is the reason why it has been called an irrealis or pluperfect imperative in Dutch linguistics (see, respectively, Duinhoven [1995: 346] and Haeseryn et al. [1997: 106]). 8

The other language that has a reproachative with a finite verb is Hungarian. É. Kiss (2011: 85) mentions the “modal construction with a counterfactual, reproaching reading” in (10a) and points out that it has a very idiosyncratic combination of linguistic features. These are: (i) an implicit apodosis (as suggested by the original, unidiomatic translation), which can be made explicit; (ii) a verb in the past conditional; (iii) negation by modal ne rather than declarative nem; (iv) a tendency for the verb to occur in sentence-initial position; and (v) the absence of complementizers such as ha ‘if’ or bárcsak ‘if only’. What is interesting for our analysis in Section 4 is that features (i) and (ii) are shared by the conditional construction in Hungarian, (ii) to (iv) by the optative one – though verb-first is stronger in reproachatives – and (iii) to (v) by the imperative one – in which verb-first is compulsory (see É. Kiss 2011: 86–92; Kormos p.c.).


Hungarian (Uralic, Ugric)


‘Had you [only] come home in time!’

(É. Kiss 2011: 86)


‘They shouldn’t have been so late!’

(É. Kiss 2011: 86)

Like the Estonian and Latvian reproachatives in (5) and (8), the construction in (10a) need not be addressee-oriented, as evidenced by (10b). But the typical subject does appear to be second person and first person subjects in particular are judged unusual, perhaps because actually uttering a self-reprimand like ‘I/we should have + past participle!’ is not very common (Kormos p.c.). Still, taking the imperative paradigm of the language into account may again shed some light on the fact that the construction, in principle, allows all person and number combinations (see Section 4).

2.3 Interim conclusion

The reproachative seems to occur in a fairly diverse set of European languages, both in genealogical terms (two Uralic ones of different branches and four Indo-European ones all of different branches as well) and in geographical terms (though contact probably does play a role for the two languages spoken in the Baltics). The constructions fall into two types. The reproachatives in Spanish, Breton, Estonian and Latvian, on the one hand, are all non-finite past participle constructions. In the first two languages, the perfect auxiliary verb is expressed in its infinitival form and the interpretation is restricted to the second person, which may be singular or plural. The constructions can, in other words, exclusively serve as a reprimand of the addressee(s). In the last two languages, only the past participle is expressed and non-second person readings are possible as well. The reproachatives in Dutch and Hungarian, on the other hand, feature a finite verb form. The Dutch construction combines the past singular form of the perfect auxiliary with a past participle and, like the Spanish and Breton reproachatives, is limited to the second person singular and plural. The Hungarian construction contains a verb in the past conditional and resembles the Estonian and Latvian reproachatives in that it allows non-second person interpretations too. Importantly, the above discussion suggests that the reproachative constructions in the six languages may be linked not only to the imperative but also to the optative and the conditional. In fact, what most existent analyses of reproachatives do is reduce them to one of those constructions.

3 Existing accounts

Of the six languages discussed in the previous section, only Spanish and Dutch have received any significant attention in the literature, the former mostly from a formalist perspective (e.g., Bosque 1980; Biezma 2010; Vicente 2013) and the latter from both a formalist (e.g., Mastop 2005: 71–78; Bennis 2007: 124; Schwager 2011: 49–52) and a functionalist perspective (e.g., Proeme 1984; Duinhoven 1995). The present section will, for that reason, primarily be concerned with the reproachative in those two languages. We will first look at imperative accounts of the construction and then at non-imperative ones.

3.1 Imperative analyses

3.1.1 Spanish

According to Bosque (1980: 415), the Spanish reproachative “behaves like a real imperative”. He rightly points out that: (i) it has no truth value, i.e., ‘that is true/false’ is not a felicitous reply to (11a); (ii) it cannot be embedded, as (11b) shows; (iii) it sanctions overt subjects as long as they fulfill a discriminatory function, as in (11c); (iv) it is always addressee-oriented, as (11d) and (11e) make clear. 9 However, of these features, only (iv) seems to be shared by just the imperative: (i) and (ii) apply to, for instance, exclamatives too while (iii) is a general property of Spanish subject pronouns (see Van der Wurff 2007: 47).




‘You should have said it earlier!’

(Ríos García p.c.)


‘I think that you should have come.’

(Bosque 1980: 415)


‘You – not he or she – should have got out.’

(Bosque 1980: 416)


‘I should have got up earlier.’

(Bosque 1980: 416)


‘John should have come.’

(Bosque 1980: 416)

To the above list, Vicente (2013: 10–15) adds, among other things, a certain speech act variability and an agentivity requirement. On the one hand, the construction under examination is said to allow a wide range of illocutionary uses such as invitations, suggestions and threats. Yet, the post hoc character of the obligation makes that, unlike the imperative, it always has reprimanding overtones. On the other hand, the reproachative normally combines with action verbs. When it does not, as in (12), it coerces an agentive interpretation.




‘You should have caused/allowed yourself to be bored!’

(Vicente 2013: 12)

Bosque’s (1980: 416) last argument for an imperative account of the Spanish reproachative is that, if more than one person is addressed, the auxiliary ‘have’ can appear not only as an infinitive but also as a second person plural imperative. Few native speakers seem to accept habed ‘(you all) have!’ here, though (see Den Dikken and Blasco [2007: 144] and Biezma [2010: 3], who considers it a case of hypercorrection). In fact, the use of the infinitive in the construction raises the rather obvious question why one would analyze it as an imperative in the first place – unless one regards the directive infinitive in Spanish as an imperative as well. As Biezma (2010: 5–6) indicates, the reproachative also differs from the imperative in that it can only be used as a reply. 10 A speaker who knows that an expert baker was going to help the addressee in a few days cannot utter (13) when just walking into the kitchen and seeing his failed soufflé. The addressee needs to have said something like ‘oh my, the soufflé was a disaster’ for (13) to be suitable.




‘You should have made your soufflé next Tuesday!’

(Biezma 2010: 6)

The imperative, by contrast, does not require licensing by a preceding utterance.

Despite the dissimilarities, the reproachative is analyzed as an imperative by Vicente (2013). His argument is two-fold. First, if imperatives involve a performative modal operator or, in other words, a function over possible worlds (see Kaufmann 2012: 86–87), “there is no a priori reason why past counterfactual imperatives should not exist” (Vicente 2013: 24). They are simply the result of the interaction of this operator with a silent past morpheme (not unlike, for instance, the English modal operator if combining with a perfect marked for the past). Second, the felicity of imperatives is said to depend on the presence of a salient world-time in the common ground in relation to which they can be interpreted (see Portner 2007: 356–357). For counterfactual imperatives, this means that they can only be acceptable when one of the countless points in time before the moment of speaking has become salient in the common ground. Put differently, the discourse needs to contain something about a specific past event to which the speaker can react.

In essence, in Vicente’s (2013) view, the construction at issue here is quite a normal extension of the imperative: the former just makes full use of opportunities provided by the very general nature or, rather, definition of the latter as a function over possible worlds. However, in addition to fairly straightforward objections concerning the lack of imperative morphology in the reproachative and the need to postulate an unrealized past morpheme in Spanish, the problem is that his analysis makes an inaccurate prediction about cross-linguistic frequency. As Van der Wurff (2007: 50) points out, accounts that “suggest that past imperatives fit more or less naturally into the general conception of imperatives … [can]not explain … their undoubted rarity”. Admittedly, Vicente (2013: 46–47) too is aware of this issue but leaves it for further research.

3.1.2 Dutch

The constructional similarities to the imperative and the shared meaning of an appeal to the addressee (see Section 2.2.2) have led some scholars to regard the Dutch reproachative as an imperative as well. One of them is Proeme (1984: 245, our translation), 11 whose account involves an even broader characterization of imperative semantics than the one described in the preceding paragraph: “The speaker urges the addressee(s) to consider himself (/themselves) as fulfilling the role normally fulfilled by the subject referent in what is mentioned.” He distinguishes a “perform” and an “imagine” interpretation. In (14a), for instance, the addressee is told to keep quiet or, put differently, to perform the state of affairs. In (14b), by contrast, he is invited to imagine the state of affairs and, more specifically, to think of traveling to a particular country alone so as to make him aware of the possible consequences (which can be spelled out in a clause starting with dan ‘then’ or in the second part of a conditional coordination construction containing this type of imperative).




‘Keep quiet!’


‘Imagine traveling to that country on your own!’

The reproachative thus amounts to the sum of imperative, modal past tense, moment of speaking as “narrated period” or the period for which “the property expressed by the predicate is attributed to the referent of the subject” (Proeme 1984: 249, our translation) and perfect. The case in (15), for instance, can be paraphrased as follows: the addressee is urged (=imperative) to put himself in another reality than that of the speaker (=modal past tense) and to view himself at the time of speaking as the person to whom the property ‘keeping quiet’ (=moment of speaking as narrated period) was ascribed at an earlier point in time (=perfect) – which, of course, gives rise to a clash between the state of affairs desired by the speaker and the fact that the right time for its realization has passed (see Proeme 1984: 254).




‘You should have kept quiet!’

Importantly, this compositional approach presupposes that the parts of the aforementioned sum can occur separately and be combined in various ways. The perfect in (16) can serve as an example: 12 the speaker is said to prompt the addressee to regard himself as the person to whom the property ‘reading that book’ can be attributed at a time before the upcoming exam. The strangeness of (16) is partially due to the fact that, out of context, the reasons for the aspectual focus on the resultative state are unclear. But the crucial point here is that the construction is, in principle, possible in Proeme’s (1984) view.




‘Preferably have read that book before your exam [tomorrow]!’

(Proeme 1984: 248)

The same holds for the examples in (17), which are particularly interesting because they are structurally comparable to the reproachative. The cautious encouragement in (17a) is argued to differ from (15) only in that the time for the realization of the state affairs lies in the future and the addressee can still perform it. Like in (16), the perfect entails emphasis on a resultative state in the future. The past tense has a modal distancing effect (see footnote 8) and mainly makes (17a) a less insistent version of (16). In (17b), in comparison, had is said to receive a temporal interpretation: the addressee is invited to imagine himself in the past situation of having broken one’s leg at the beginning of a ski trip.




‘Preferably have read that book before your exam [tomorrow]!’

(Proeme 1984: 254)


‘Great! Imagine that you had broken your leg on the first day of your ski holiday!’

(Proeme 1984: 254)

Proeme’s (1984) definition of imperative meaning, which recognizes virtually no constraints relating to (modal) tense and grammatical and lexical aspect, does not account for the facts, though. On the one hand, few native speakers accept the combinations of features in (17) and, to a lesser extent, (16) and it is hard to find non-fabricated examples of these uses (see Boogaart and Janssen [2010: 125–126], who do mention one online attestation of 17b). On the other hand, even if one does not have a problem with these sentences, it remains to be explained why the reproachative is the standard interpretation of the construction in (15) and (17), i.e., the one which it tends to receive out of context – as Proeme (1984: 254) himself acknowledges – and the only one of which authentic instantiations can fairly easily be obtained. In Section 4.1, this reading is analyzed as the conventionalization of a specific pragmatic use.

In short, not unlike Vicente (2013), whose account overpredicts the cross-linguistic occurrence of so-called past imperatives, Proeme’s (1984) analysis of the Dutch reproachative as an imperative requires such a general characterization of the latter that the language is expected to have more “imperative” options than it actually appears to have. 13

3.2 Non-imperative analyses

3.2.1 Dutch

The traditional view in Dutch as well as Spanish linguistics is that the reproachative is some kind of imperative. But other analyses have been put forward. Bennis (2007: 124), for one, suggests that the Dutch reproachative is an optative. In Section 2.2.2, we have established that, from a semantic perspective, this claim does not do justice to the reproachative’s intrinstic appeal to the addressee to acknowledge the faultiness of his course of action and to face its consequences, as evidenced by (18).




‘You should have been taller!’


‘She should have kept quiet!’

Yet, it does not seem so unreasonable at first sight. Both the reproachative in (19a) and the optative in (19b) have a past perfect with the auxiliary in clause-initial position and a counterfactual meaning. They also share a sense of conditionality: (19b) too evokes the positive consequences that the realization of the state of affairs at an earlier point in time would have had. Moreover, it is not hard to think of contexts in which, pragmatically, the optative serves as a reprimand, just like the English construction with if only in the translation (see Section 2.1 as well).




‘You should have kept quiet!’


‘If only you had kept quiet.’

Upon closer inspection, though, it is evident that we are dealing with two different constructions. As the ungrammaticality of the “zero modal particle” in (19b) indicates, maar is needed in the optative. According to Mastop (2005: 73), this particle fits nicely with the meaning of a counterfactual wish in that it is typically used to “downplay the material” in a sentence and “is compatible with an added comment like ‘but how could [you] have known in advance?’”. It can also occur in the reproachative, which then acquires optative overtones, but is not required in this construction, as the brackets in (19a) make clear. A modal particle like toch, by contrast, brings in some type of presupposition and “seems to suggest that the addressee could have known, at the time of action, that the action was a bad choice and would lead to the unfortunate circumstances” (Mastop 2005: 73). It goes well with the appeal characteristic of reproachatives, in other words, but not with the completely non-directive-like semantics of optatives, as (19) shows (see Mastop 2005: 74).

Some more differences between the two constructions are exemplified in (20).




‘If only you had been taller.’


‘If only she had kept quiet.’


‘If only you were silent for once.’

As Van der Wurff (2007: 46–47) points out, optatives – unlike the reproachatives in (18) – allow non-agentive predicates as well as non-second person subjects, as in (20a) and (20b) respectively. What is more, the subject is also always overtly expressed here. Example (20c), finally, demonstrates that the optative is not even restricted to past counterfactuality and past perfect verb forms.

3.2.2 Spanish

For Spanish, Biezma (2010: 7) argues that, in an analysis of the reproachative as an imperative, many of the former’s idiosyncratic features such as the use as a reaction only (but see footnote 10 for an assessment of their alleged specificity) would have to be considered “mere conventionalizations”. She views the reproachative as a conditional clause of which only the protasis is expressed and which is marked by inversion, as in (21a), rather than by the conjunction (solo) si ‘(only) if’, as in (21b).




‘Had you left earlier, you would have arrived on time.’

(Biezma 2010: 9)


‘If (only) you had left earlier, you would have arrived on time.’

(Biezma 2010: 9)

The ellipsis, on the one hand, would be the reason for which the reproachative is always a reply: “[It] can only be spelled out when the context is rich enough to retrieve the consequent” (Biezma 2010: 8). Conditional inversion, on the other hand, would generally indicate that the proposition in the protasis is part of the common ground and therefore partly explain the construction’s function as a reprimand: the speaker presents the alternative course of action as having been “epistemically available at the time” (Biezma 2010: 10) when the addressee nevertheless chose to do something else. 14

However, this analysis, like the optative one of the Dutch reproachative, fails to account for some of the more “imperative-like” characteristics of the construction. Conditional clauses exhibit no preference for agentive predicates, for instance. Another obvious difference with the reproachative is that the protasis in (21a) contains a past subjunctive form of the verb ‘have’ instead of the infinitive. Biezma’s (2010: 8) solution is to posit a modal element that is just not spelled out “since in Spanish the modal is realized in the verb morphology of the consequent” (i.e., the conditional form habrías ‘you would have’), which is left unexpressed here. From a usage-based point of view, this explanation is, of course, not particularly attractive. But even if one is open to the idea of such a silent modal, one faces the challenge that the conditional inversion construction of which the reproachative is said to be a special case appears to be unacceptable for most native speakers, who do not find the reproachative itself problematic in the least (see Vicente 2013: 43; Ríos García p.c.).

3.3 Interim conclusion

What the imperative and non-imperative analyses in Sections 3.1 and 3.2 have in common is that they try and reduce the reproachative to an established construction with which it shares a number of properties. Indicatively, Vicente (2013: 4) writes that it “cannot be assimilated [emphasis ours] to many (if not all) of the other clause types that lack a truth value”.

The similarities to which the various accounts draw attention – such as a formal resemblance to the imperative, an undoubtedly not unrelated directive-like meaning and a sense of conditionality – need to be taken seriously but should not be seen as evidence that the reproachative is the same as an imperative, an optative or a conditional protasis. Such views lead to a characterization of the imperative unable to explain the apparent scarcity of past imperatives in the world’s languages and the unacceptability in a specific language of types of imperative that are definitionally possible or to a disregard of the morphosyntactic and semantic differences between the reproachative and the construction of which it is argued to be an instance (e.g., infinitival rather than imperatival or past subjunctive ‘have’ in Spanish, the non-optative-like appeal to the addressee).

Our way out is to consider the reproachatives in the languages under investigation as constructions in their own right but with strong connections to other constructions or, in other words, to acknowledge their hybrid nature. In actual fact, implicitly and despite the fact that he calls the Dutch reproachative an irrealis imperative, Duinhoven (1995) adopts a similar approach. His account of the construction as the result of the interaction of a number of constructions is the point of departure for our own analysis.

4 Hybrid account

4.1 Dutch

4.1.1 Duinhoven (1995)

Duinhoven’s (1995) analysis starts from the conditional inversion construction and, more precisely, its counterfactual instantiation, as in (22). It often has plaintive overtones and, when the subject in the protasis is second person and the predicate is agentive, i.e., something that the addressee could have actively done, it tends to serve as a reproach.




‘Had you kept quiet, none of this would have happened!’

This pragmatic function is argued to carry an “adhortative connotation” (Duinhoven 1995: 352, our translation): the speaker signals what, in her opinion, the addressee should have done in order to prevent the coming about of the negative consequences of his actual behavior. This deontic appeal to the addressee is shared with the imperative construction, which is said to have resulted in the analogous deletion of the subject in protasis too.

Duinhoven (1995) adds that the variation between the conditional inversion construction in (23a) and the conditional imperative construction in (23b) has probably played a role as well.




‘If you keep quiet, all will be fine.’


‘Keep quiet and all will be fine.’

The idea is that if speakers can omit the second person pronoun in the former and get the latter “without any substantial difference in meaning” (Duinhoven 1995: 350, our translation), they may find it easier to leave out the subject in an inverted counterfactual conditional that is used as a reproach and is hence not dissimilar to an imperative.

The final step concerns the disappearance of the apodosis. As Duinhoven (1995: 355) points out, the main clause of a conditional construction is frequently elided if the consequences are vague or, conversely, if they are self-evident. In such a case, the conditional subclause normally contains one or more modal particles, which evoke the implicit part of the construction (see, for instance, Section 3.2.1 on toch) and, as such, indicate that the clause is still felt to be incomplete. This type of ellipsis is claimed to have happened to the counterfactual conditional and, eventually, to have given rise to the autonomous construction in (24).




‘You should have kept quiet!’

This reproachative construction does not require modal particles anymore: “Without inversion [due to the loss of the subject] or a modal particle, the clause no longer has the character of a subordinate clause” (Duinhoven 1995: 356, our translation).

4.1.2 Evaluation

The account sketched in the previous paragraph seems plausible from a typological perspective in that it appeals to two commonly attested processes of change. On the one hand, conventionalized non-imperative directive strategies often lose their subject marking (see Aikhenvald 2010: 343). This change is, in all probability, due to analogy with the imperative, which tends to be non-finite cross-linguistically (see Nikolaeva 2007: 139). In many a Bantu language, for instance, the subjunctive, which (25a) shows ends in -e, loses its subject marker when used directively, as in (25b). It resembles the imperative in (25c), which ends in -a, in this respect (and, in Shangaci, also in its tonal pattern – as pointed out by Devos and Van Olmen [2013: 18–19]).


Shangaci (Niger-Congo, Bantu)


… that you hold me’

(Devos and Van Olmen 2013: 19)


‘Hold me!’

(Devos and Van Olmen 2013: 19)



(Devos and Van Olmen 2013: 19)

On the other hand, the established use of an initially subordinate clause as a matrix clause, i.e., a protasis as an irrealis imperative in Duinhoven’s (1995) analysis, is a cross-linguistically common phenomenon called insubordination (or desubordination). Moreover, it is most frequently “found in various types of clause concerned with interpersonal control – primarily imperatives and their milder forms such as hints and requests, but also permissives, warnings and threats” (Evans 2007: 387). Conditional subclauses in particular often develop into independent directives (e.g., if you (dare) touch me! or if you could give me hand) (see Evans 2007: 393).

The above account also appears to be compatible with the usage-based model (see Section 1). Repeated instances of use of the counterfactual conditional inversion construction that features a second person subject and serves as a reprimand may be assumed to have reinforced each other. This reinforcement may be expected to have led to a separate exemplar with a reproachative function that is stored “near” the original exemplar and, in time, to the establishment of a new construction. What Biezma (2010: 7) would dismiss as a mere conventionalization is thus an integral part of the usage-based model. It also offers a relatively straightforward explanation for a number of properties of the construction under investigation: its inherited sense of conditionality, its (imperative-like) restriction to agentive predicates and its use just as a reply. For one thing, the counterfactual conditional inversion construction is only interpretable as a reprimand if the addressee was able to (but did not) realize the state of affairs or, in other words, if the protasis contains an action rather than, say, a state, as in (26a) versus (26b). This pragmatic constraint could be argued to have conventionalized in the reproachative.




‘Had you kept quiet, none of this would have happened!’


‘Had you been taller, none of this would have happened.’

In addition, the apodosis can only be elided if its content is somehow retrievable. As the consequences here are somehow the reverse of the actual results of the addressee’s past behavior, they cannot just be recovered from the immediate context of the moment of speaking. The link between his earlier conduct and the subsequent circumstances, which are judged to be undesirable, needs to have been activated in the common ground before the speaker can provide her alternative course of action and leave its opposite/desirable outcome implicit. Sentence (27a), for instance, cannot be uttered to someone who is late to a meeting and has just walked into the room. But it is acceptable to leave out the apodosis when the addressee establishes a connection between his behavior and its result first, by saying something like ‘sorry I’m late but my tram broke down’ on entering the room (see also Section 3.1.1). The reproachative in (27b) can also only be used in the second situation.



Hadjeeentaxigenomen, …[danwasje

‘Had you taken a cab, … [you would have been on time].


‘You should have taken a cab!’

The ellipsis of the apodosis is thus possible only when the speaker is reacting to something in the preceding discourse, which may be said to have conventionalized in the use of reproachatives as replies only.

Another part of Duinhoven’s (1995) argument that goes well with the usage-based model is that, by functional analogy with the imperative, i.e., the appeal to the addressee, the reproachative has also undergone a formal, imperative-like change, i.e., the loss of the second person subject pronoun. As Fischer (2008: 352–368) among others argues, language users can indeed mentally connect constructions (of different levels of abstraction and complexity) to other constructions based on functional and/or formal similarities and even model the former after the latter in time.

The analysis is not unproblematic, though. The analogy with the imperative resulting in the deletion of the subject, for instance, requires a more substantial amount of cognitive flexibility than Duinhoven (1995) acknowledges. In an imperative, the state of affairs assigned to the addressee is typically realizable in the present or future whereas, in a reproachative, it can no longer be carried out. This point of divergence is probably why the variation between the conditional inversion construction in (28a) and the conditional imperative construction in (28b) is brought in: it is essentially argued to have facilitated the omission of the subject.




‘If you keep quiet, all will be fine.’


‘Keep quiet and all will be fine.’

However, many a difference between the two constructions is glossed over. First, (28b), which is said to have enabled the loss of je in (26a) used as a reproach, involves coordination instead of subordination. Second, semantically, the construction in (28a) is purely conditional. The conditional imperative construction, by contrast, can receive a directive reading, as is likely to be the case in (28b), where the addressee is urged to keep quiet, as well as an entirely conditional one, as in (29). In this example, nobody is encouraged to give the woman the floor. Rather, it is an assertion that one will not be able to say anything anymore if one allows her to speak. 15




‘Give her the floor once and you’ll never be able to say anything yourself.’

(Fortuin and Boogaart 2009: 653)

Third, the conditional imperative construction has the so-called restrictive reading that “only X [i.e., the protasis] is needed for Y [i.e., the apodosis] to occur” (Fortuin and Boogaart 2009: 656). In (28b), for instance, it is suggested that keeping quiet will almost automatically be followed by everything being fine. This feature has been attributed to the coordination in the construction and is thus absent from both the conditional inversion construction in (28a) and its counterfactual instantiation used as a reprimand in (26a), which adds to the doubt over the impact of the pair in (28).

Finally, the claim that the “variation” between (28a) and (28b) plays a role in the development of the Dutch reproachative assumes that the loss of the subject must have preceded the ellipsis of the apodosis: omission of the subject of the counterfactual protasis by analogy with these two constructions would be impossible if the counterfactual apodosis was elided earlier and the developing reproachative consisted of one clause instead of two. Duinhoven (1995: 360) subscribes to this chronology but, as we will see in Section 4.1.3, the data do not appear to support it.

4.1.3 Modified analysis

This section presents a modified version of Duinhoven’s (1995) account. It start from an exploration of the constructions central to his analysis in a corpus of Dutch plays and fictional prose from 1500 to 2000. The corpus is made up of texts from the Digitale Bibliotheek voor de Nederlandse Letteren. 16 The sixteenth-century component contains 239,932 words of plays and 1,947,614 words of fictional prose. The two genres account for, respectively, 858,632 and 704,518 words in the seventeenth-century component; 269,677 and 678,717 words in the eighteenth-century component; 226,917 and 2,279,896 words in the nineteenth-century component; and 232,133 and 2,328,004 words in the twentieth-century component. A few comments are in order. First, these numbers are given for the sake of completeness. Because of the infrequency of the constructions under investigation, a quantitative analysis makes little sense. Second, the choice to exclude Old and Middle Dutch from the corpus is motivated by the fact that the reproachative is a recent innovation in the language, as suggested by Duinhoven’s (1997: 48, our translation) observation that “before 1900, the construction is not mentioned in any grammar”. Third, the relevant cases were manually selected from all WordSmith 6.0 hits for the past indicative and subjunctive forms of the perfect auxiliaries hebben ‘have’ and zijn ‘be’ (including spelling variants) which start with a capital letter and/or are preceded by a punctuation mark or conjunction or, put differently, which might be in clause-initial position. 17

In the sixteenth-century data, we only find full-fledged examples of the counterfactual conditional inversion construction. Some of these feature a second person subject in the protasis and possibly have a reprimanding connotation. Sentence (30) is a case in point. 18




Doort volghen van u lust, volght u jammer en noodt. Kondy dit (goede man) noch van zelfs niet verstaan? ‘T verzoecken leeret u, wat behoevet mijn vermaan?

‘If you had followed my advice, you would still be free and eminent. When you go along with your lust, disaster and paucity follow you. Could you, good man, not see this yourself? Tempting fate will teach you. What use is my reprimand?’

(Coornhert 1590, Comedie van Israel)

Insubordinated instances of the counterfactual conditional inversion construction appear in the seventeenth- and eighteenth-century data. The cases in (31) and (32), where an apodosis such as ‘we would not have to deal with her’ and ‘this would not have happened’ respectively is left implicit, can serve as examples.




Krelis is dood! Teeuwis is nu de Haan in de Korf: en geen mensch zal ‘er hem uit werpen…… Ha, daar is uw Moeder.

‘Krelis is dead! Teeuwis [the speaker] is number one now and nobody will take his place… Ah, there comes your mother.’


Ga heen Antje, anders krijgt gij ‘er wat van. – Nu zal de kwelling eerst beginnen.

‘‘If only we had gone away!’ Go away, Antje. You’ll regret it if you don’t. – Now, the torment will start.’


Neen, neen; nu mijn Broeder dood is, zal ik uwe Moeder vast wellekom zijn…

‘No, no. Now my brother is dead, I, your mother, should be welcome.’

(Brender à Brandis 1791, De Gouden Bruiloft van Kloris en Roosje)



‘t Is mijn eygen schult.


‘It’s my own fault. If only I hadn’t opened the door.’

(De Mol 1671, Bedrogen Lichtmis)

Like (31), they often contain the modal particle maar, which makes them very similar to the optative in Present-day Dutch (see Section 3.2.1). However, like (32), which is not really acceptable anymore now, they can also occur without it and feature another modal particle such as doch (or even no modal particle at all).

It seems, in other words, that the Early Modern Dutch construction to express a counterfactual wish was not yet as fixed as its present-day descendant. Examples (33) and (34) therefore need to be regarded simply as second person instantiations of this construction which have reproachative overtones in context.




GOeden dagh vrinden. Maar, wat ik van Tys hoor, is ‘er uw man zo kwaâlyk aan?

‘Good day, my friends. I’ve heard from Tys that your husband is in bad condition?’


‘If only you had come a little earlier.’


‘k Héb alles om jou laaten staan.

‘I did drop everything for you.’

(Van der Goes 1682, De Gelyke Twélingen)




Men houdt u … voor een Meisje, dat zich, in alle hare daden, geen ander oogmerk voorstelt dan Divertissement. Kan ik, die u als myne eigen Dochter bemin, dit zien, zien, zonder op u misnoegt, en over uw gedrag zelf bedroeft te zyn? …

‘People regard you as a girl that, in all her actions, has no other purpose than enjoyment. Can I, who loves you like my own daughter, witness this without being annoyed with you and unhappy about your behavior myself?’


‘If only you said something to me!’


Hoeneer moest ik u dit ééne woord gezegt hebben? Vóór de zaak gebeurde?

‘How early should I have said something to you? Before the affair happened?’

(Wolff and Deken 1782, Historie van Mejuffrouw Sara Burgerhart)

This interpretation is supported by, respectively, Dr Polifémus’s indignant reaction that he actually dropped everything to get to Jórden’s house as fast as possible in (33) and Mss Buigzaam’s – Sara’s guardian – response in (34), in which she takes up the implied post hoc obligation.

The examples in (31) to (34) suggest that, pace Duinhoven (1995), the loss of the apodosis preceded that of the subject and resulted in an insubordinated construction of a more general, optative nature. This insubordinated apodosis could then serve as a reprimand pragmatically. Yet, the first and only attestation of a reproachative construction in the nineteenth-century data – which is earlier than claimed by Duinhoven (1997: 48) – appears to be part of a complex conditional construction in which the apodosis dan hadt jij de premie binnen ‘you would have got the bounty’ is spelled out, as (35) shows.



‘Wat heb je er nu aan?’ vroeg vrouw Grerowitz, toen haar man eindelijk opgestaan was en naast haar voortliep. ‘Immers net zooveel, zal ik maar zeggen, als niemendal.


terwijl je nu daarentegen weer, zooals gewoonlijk, niets thuis brengt.

‘‘What good is it for you now?’ Mrs Grerowitz asked, when her husband had finally got to his feet again and was walking next to her. ‘About as much as nothing, I’d say. If only, instead of showing the soldiers the way, you had put them on the wrong track and had gone and searched yourself with one or two of our people, then you would have got the bounty, while now, as usual, you don’t bring home anything.’

(De Goeje 1861, De Dochter van den Kozak)

The protasis consists of two coordinated clauses. The first one is optative, as the presence of je and maar indicates and as our translation by means of if only reflects (though it is clearly used as a reprimand). The second one, by contrast, has no subject and is thus the new, imperative-like formal embodiment of the function of an after the fact reproach (because the clause is linked to an optative by en ‘and’, it is hard to translate it as ‘you should have gone and searched yourself!’, though). 19

Example (35) supports the view that both the optative and the reproachative stem from the protasis of the counterfactual conditional inversion construction. It could also be argued to suggest that, in spite of the fact that ellipsis seems to predate subject omission, the relative mutual substitutability between the conditional imperative and inversion constructions (see Sections 4.1.1 and 4.1.2) played a role after all in the functional analogy of the reprimand-exemplar of the optative construction in (33) and (34) with the imperative, leading to the formal analogy of deleting the subject. 20

Another perspective on (35) is offered by the usage-based idea that “as the new construction becomes established in the grammar, it gradually [emphasis ours] loses its associations with the constructions from which it arose” (Bybee 2010: 31). The occurrence of the optative and the reproachative in a conditional construction here, together with their conditional overtones even when they are used on their own, may be due to a persisting link with the conditional inversion construction. Postma et al.’s (2013: 290–292) discussion of (36) is revealing.




You should have worked harder so that you wouldn’t have had to beg later!’

(Postma et al. 2013: 290)

For some speakers of Dutch, this link is still so strong that they do not accept the addition of a ‘so that’-clause instead of an apodosis headed by dan. For others, the reproachative is independent enough to find (36) unproblematic. Put differently, the autonomy of a new construction is a matter of degree in the usage-based approach and may very well involve intersubjective variation.

In the same vein, it is interesting to note that most reproachatives in our twentieth-century data are like (37). This example closely resembles the original optative construction in that it contains the modal particle maar and is more (jokingly, here) plaintive than scolding, as evidenced by the aunt’s calm and elaborate reply.



‘Je bent en blijft een verwend kreng’ zei haar tante. ‘Ik had je heel heel anders aan moeten pakken toen je bij ons logeerde.’ Henriët voelde zich bijna gelukkig. De gebeurtenissen kregen door die opmerking hun juiste maat.


‘Je was zo grappig’ zei haar tante. ‘Je dacht dat ik je niet doorhad wanneer je jokte, en ik kon het niet over mijn hart verkrijgen om je straf te geven.

‘You continue to be a spoilt brat,’ her aunt said. ‘I should have dealt with you very, very differently when you were staying with us.’ Henriët felt almost happy. The events were put in the right perspective by that remark. ‘You should have done it,’ she said.’ ‘You were so funny,’ her aunt said. ‘You thought that I didn’t see through you when you were lying and I couldn’t bring myself to give you a punishment.’’

(Kossmann 1980, Geur der Droefnis)

Furthermore, the reproachative in (38) is the only one in the corpus which occurs with no modal particle or no clearly conditional context and which, following Duinhoven’s (1995: 356) point about the role of modal particles, can be said to exemplify full independence. 21



Een zwaar neerdrukkend schuldgevoel woog op haar.


schreeuwde het steeds in haar hoofd.

‘A heavy, depressing feeling of guilt was weighing on her mind. ‘You should have been bold enough! You should have been bold enough!’ was always screaming through her head.’

(Albert Helman 1931, De Stille Plantage)

Examples (35) to (37) and the infrequency of (38) could be taken to suggest that, to varying extents, the reproachative is still associated with the optative construction as well as with the counterfactual conditional inversion construction. Such a view is accommodated for by the usage-based approach. Bybee (2006: 719), for one, argues that “low levels of repetition lead to conventionalization only” and (very) high frequency is required for the establishment (and, finally, grammaticalization) of a new construction. Given that, with the exception of the imperative, the constructions involved here do not occur very often, as mentioned earlier (see Section 1 and the beginning of the present section), the reproachative is perhaps simply not (yet?) established as a wholly autonomous construction. (Certain) speakers nowadays may still “actively” see it against the background of the counterfactual conditional inversion construction and of the imperative and the optative. In other words, they may still understand the reproachative more as a protasis with an elided apodosis and a directive-like variant of a counterfactual wish rather than as a completely insubordinated and independent construction in its own right.

Another possible analysis of the typical presence of maar in the reproachative is, of course, that its meaning is just very compatible with the construction (like with the optative, see Section 3.2.1). 22 According to Vandeweghe (1994: 149, our translation), it presupposes that the addressee “is focused on/benefits from/has nothing against the action/state referred to” and that “there may be objections/impediments/barriers/hesitations/contraindications to/about the realization”. The latter aspect fits well with the reproachative’s function of an after the fact reprimand. The former probably depends on the situation in which the reproachative is used (e.g., on the question whether the consequences would have been beneficial to the addressee and/or the speaker).

4.1.4 Summary

In our view, the Dutch reproachative is not a simple imperative or optative (see Section 3) but a hybrid construction sharing formal properties (e.g., the lack of a subject, the typical occurrence of maar, a preference to be “completed” by a clause introduced by dan) and functional properties (e.g., the appeal to the addressee, the use as a reply only, the conditional overtones) with (exemplars of) the constructions that bring it about through ellipsis/insubordination, conventionalization and analogy: the counterfactual conditional inversion construction, the optative construction and the imperative construction and, perhaps, the conditional imperative construction too. The relations can be summarized as follows:

  1. the counterfactual conditional inversion construction, which can be used with reprimanding overtones in the second person, gives rise to the counterfactual optative construction through the ellipsis of its apodosis and the gradual insubordination of its protasis;

  2. the pragmatic use as a reprimand of this counterfactual optative construction with agentive verbs in the second person becomes more and more conventional through usage;

  3. the functional similarity between this conventionalizing use and the imperative construction, i.e., the appeal to the addressee, makes the analogical formal change of leaving out the subject possible;

  4. this formal change may be facilitated by the variation between the conditional inversion construction and the conditional imperative construction.

One could, of course, argue that, because of (iii) and “despite” its origins, the Dutch reproachative construction is now part of an imperative whose formal and functional scope has been expanded. 23 Formally, this new imperative construction would, for instance, allow not only verb stems but also preterit verb forms in clause-initial position. Yet, the fact that, for many speakers (see footnote 8), only the past perfect auxiliaries had and was are acceptable would need to be accounted for somehow and an analysis stipulating that the imperative only permits certain preterit forms in the reproachative amounts to saying that there is something special and different about the latter construction after all. Functionally, the new construction would have semantics that cover the imperative’s normal uses as well as an after the fact reprimand. As pointed out in Section 3.1.2, Proeme’s (1984) attempt fails in that it overpredicts the possible meanings that the imperative actually has. Moreover, any proposal would have to explain why, unlike the ordinary imperative, the reproachative always has conditional overtones and can only serve as a reply (see also Section 3.1.1). These differences are easily accounted for by viewing the reproachative as a hybrid, though imperative-like, construction in its own right. In the rest of Section 4, we will examine whether and to what extent such a hybrid analysis also applies to the other reproachatives in the languages of Europe.

4.2 Other languages

4.2.1 Estonian (and Latvian)

The present section will focus on Estonian but, as Muižniece et al. (1999) points out, Latvian has undergone essentially the same developments. As discussed in Section 2.2.1, the reproachative is just one of a range of “finite” functions of the past participle in the language. The use of this verb form as the predicate of a clause is a very old feature of Finno-Ugric (and Indo-European). Originally, it “expressed the completion of an event and thus its past character” in a “narrative that focuses on the events under discussion and does not actualize the speech situation” (Muižniece et al. 1999: 138), appeared in clause-initial position and was restricted to third person readings. This function has survived, for instance, in folk tales, as (39) shows.




‘Once there lived a king and …’

(Metslang and Sepper 2010: 543)

Over time, the lone past participle was extended from monologic contexts to dialogic ones, where it became desirable to link the event to the speech situation in temporal and/or modal terms and first and second person interpretations – the interlocutors – developed too. In addition to indicative and evidential uses (see Section 2.2.1 on the quotative), the narrative past has been argued to have yielded a so-called relative use. Sentence (40) can serve as an example. The coordinated lone past participles lasknud ‘left’ and pugend ‘climbed’, in clause-initial position, convey “a preferred or recommended situation, i.e., a situation that would have given rise to a positive result” (Muižniece et al. 1999: 146). In view of their meaning, which Metslang (p.c.) stresses does not involve a directive-like post hoc obligation (yet), it is not so remarkable that they are only found in the protasis when they occur in a complex conditional construction like (40): the apodosis, which contains the conditional marker oleks here, mentions the “positive” result. 24




‘You could have left us in the fire, you two could have climbed out of the window of the window, then you could have cursed in peace after all.’

(Muižniece et al. 1999: 146)

This relative use is said to have evolved into, on the one hand, a purely conditional one and, on the other hand, an imperative and jussive or, in other words, reproachative one: “When applying it to the second person, the previous link [i.e., (40)] would provide a development of an imperative expressing a command that it is directed to the past” (Muižniece et al. 1999: 147). It is not specified how this change came about but it is cross-linguistically not uncommon for possibility markers to turn into necessity markers, especially in the deontic domain (e.g., Traugott and Dasher 2002: 120). One can easily imagine that, in particular contexts, a speaker employs a suggestion about preferable past behavior pragmatically as an indirect way of signaling the required course of action earlier on and of reprimanding someone for not having followed it and that, in the end, this usage may conventionalize into (41).




‘You should have come home in time last night!’

(Muižniece et al. 1999: 147)

The reproachative may thus not derive from the imperative but the two constructions do have a few properties in common. As the comparison between (41) and (42) shows, they both: (i) convey some type of addressee-oriented obligation; (ii) have the verb in clause-initial position, which may be a dedicated negative auxiliary in the imperative; and (iii) do not overtly express the subject.




‘Water the flowers!’

(Metslang and Sepper 2010: 535)


‘Don’t be a bore!’

(Metslang and Sepper 2010: 535)

Evidence for the fact that these similarities seem to have resulted in analogy of the reproachative with the imperative comes from the former’s person and number constraints and from its negation. First, as discussed in Section 2.2.1, (41) is not limited to a second person reading. In fact, the only interpretation of the subject that is excluded is ‘I’ (Metslang p.c.). Analyzing this construction as a reproachative nonetheless might be considered too much of a departure from our definition in Section 2.1 but it is probably not a coincidence that the imperative paradigm in Estonian has forms for all person and number combinations apart from the first person singular (see Metslang and Sepper 2010: 534). Second, generally, the finite uses of the lone past participle have no negative forms but example (43), from a corpus of online language, suggests that the reproachative is an exception.




‘Well, you/he/… should not have brought it, no one told you/him/…!’

(etTenTen Corpus of Estonian)

Interestingly, it is not the standard negative particle ei that appears here but a novel past participle form of the auxiliary ära (see also Erelt 2012: 34–35). This verb is normally only used to negate imperatives, as in (42b), and makes clear that the reproachative has been linked to the imperative construction, at least for some speakers of the language. 25

4.2.2 Hungarian

In Hungarian, there are good indications that the reproachative is of conditional origin. Both the first sentence of (44a), an instance of the construction under examination, and the protasis of (44b), headed by the complementizer ha ‘if’, contain a verb in the past conditional: the combination of the conditional marker volna and the second person singular past form of the particle verb ‘finish’ befejezted. These two examples also show that the reproachative, like a conditional subordinate clause, evokes an apodosis, which can be spelled out but tends to remain implicit.




az állást.)

the job

‘Had you [only] finished your dissertation! (You would have got the job.)’

(É. Kiss 2011: 89)


‘If you had finished your dissertation, you would have got the job.’

(É. Kiss 2011: 89)

However, unlike the conditional construction in (44b), in which the order of protasis and apodosis is not set, as exemplified by (45a), the reproachative cannot be preceded by its consequence, as in (45b). This is what “happens in the case of independent sentences describing events with a fixed temporal order” (É. Kiss 2011: 89).






‘You would have got the job, if you had finished your dissertation.’

(É. Kiss 2011: 89)




‘You would have got the job. Had you [only] finished your dissertation!’

(É. Kiss 2011: 89)

In other words, the ungrammaticality of (45b) could be taken as evidence for the fact that the two clauses in (44a) are autonomous and, more specifically, that the reproachative has been insubordinated.

Insubordination does not seem to be the whole story, though. As briefly mentioned in Section 2.2.2, the construction in (44a) also exhibits a number of non-conditional properties. For instance, unlike the protasis in (44b), the reproachative has no complementizer in clause-initial position. Instead, it starts with the verb fejezted while the particle be is separated from this verb and moved to the right. Furthermore, it is negated not by the standard negative marker nem, as in (46a), but by the modal variant ne, as in (46b).



Hanemhagytadvolnafélbeadisszertációdat, …

‘If you hadn’t abandoned your dissertation, …’

(É. Kiss 2011: 89)


‘You shouldn’t have abandoned your dissertation!’

(É. Kiss 2011: 89)

Crucially, as É. Kiss (2011: 89–90) points out, all these other properties are shared with the imperative. In the positive one in (47a), the verb occupies clause-initial position – as expected given the strong cross-linguistic tendency for verb-first in imperatives (see Aikhenvald 2010: 96) – and the detached particle follows it. In the negative one in (47b), ne rather than nem is used.




‘Finish your dissertation!’

(É. Kiss 2011: 90)


‘Don’t abandon your dissertation!’

(É. Kiss 2011: 89)

In light of our account for Dutch in Section 4.1, it does not seem unlikely that, also in Hungarian, this resemblance between the reproachative and imperative constructions is the result of functional and then formal analogy of the former with the latter.

Two final comments are in order. First, the optative too constitutes an insubordinated protasis with lingering conditional overtones, is negated by means of ne and allows both a preposed and a postposed particle, as in (48). These facts suggest that the construction may have played an intermediary role here as well but more diachronic research is needed to check this hypothesis.




‘If only you hadn’t abandoned your dissertation.’

(É. Kiss 2011: 89)


‘If only you had finished your dissertation.’

(É. Kiss 2011: 90)

Second, as indicated in Section 2.2.2, the Hungarian reproachative accepts not only second person but also third and first person subjects – though the latter are regarded as unusual possibly because of the exceptional nature of self-reprimands. The fact that a construction akin to the imperative has no restrictions on the subject is not so remarkable for this language, though: its imperative paradigm has forms – which may occur in a few other contexts but exhibit the syntactic behavior also found in the reproachative (e.g., the positions of the verb and its particle) when used directively – for all person and number combinations (see Aikhenvald 2010: 49–50, 206).

4.2.3 Breton and Spanish

For the variety of Breton spoken in Trégor, Gros (1984: 318, our translation) suggests that the past participle construction in (49a) results from the “ellipsis of a conjugated modal of the type dleout ‘should’ [i.e., ‘(you should) have made her work!’]”. Hewitt (2010: 305), by contrast, seems to assume the insubordination of an optative-like protasis and the ellipsis of its conditional marking ma karjes ‘if you had only wanted to’, as in (49b).




‘You should have made her work!’

(Gros 1984: 318)


‘You should have told me!’

(Hewitt 2010: 305)

At this stage, we are not in a position to evaluate these proposals properly. Still, it may be worth pointing out that the structure ‘infinitival perfect auxiliary + past participle’ can be employed as a counterfactual protasis on its own, without a finite verb in the conditional (Jouitteau p.c.). Example (50) is a case in point.




‘Had I known, I would have brought meat.’

(Jouitteau p.c.; Le Pennec p.c.)

Considering that this conditional clause can occur with any subject in the apodosis and the protasis will be interpreted as having the same one, one could imagine a second person instantiation of the construction used as a reprimand giving rise to the reproachative through conventionalization and insubordination. Obviously, this scenario does not go beyond mere speculation.

For Spanish too, perhaps somewhat surprisingly, it is not immediately clear to us whether the reproachative is the hybrid outcome of the interaction of other constructions and how it would have come about. To our knowledge, the language possesses no conditional or optative construction that just involves the structure ‘infinitival perfect auxiliary + past participle’. In the vein of Gros (1984), one could hypothesize that the reproachative results from the ellipsis of the conjugated modal verb deberías ‘you should’ but one would not be able to account for its conditional overtones (see Section 3.2.2 and Biezma 2010) in a straightforward way. Another construction that is potentially relevant in view of our above description of other reproachatives is (51).




If you had come earlier, you would have won.’

(Ríos García p.c.)

The combination of the preposition ‘of’ and an infinitive can be used as the protasis of a conditional clause and, if that infinitive is a perfect auxiliary followed by a past participle, as a counterfactual one, like in (51). For this construction to be a/the source of the reproachative, one would still need to explain the loss of de, though. In short, an in-depth study of the history of the structure ‘infinitival perfect auxiliary + past participle’ and its occurrence in various constructions seems necessary. Let it suffice for now to say that the form of the perfect auxiliary in the reproachative is probably not completely unmotivated as the infinitive “is very frequently used [directively] as [an alternative to] an imperative” (Bosque 1980: 416).

4.2.4 Summary

The reproachative constructions in Section 4.2 for which we have sufficient information resemble the Dutch one in that they result from the interaction of other constructions. In Estonian, the use of the non-finite past participle construction as a reprimand appears to have emerged in the following way:

  1. the original narrative function of this construction, which was restricted to third person interpretations, gave rise to, inter alia, a “relative” function and was extended to other persons in dialogic contexts;

  2. in this relative function, the construction serves as a suggestion about a preferable past situation which would have had positive results and can occur in the protasis but not the apodosis of a conditional;

  3. the pragmatic use of this expression of possibility for the necessity function of reprimanding becomes more and more conventional through usage;

  4. the functional similarity between this conventionalizing use and the imperative construction, i.e., their deontic and directive-like meaning, makes the analogical formal change of negating the now reproachative construction by way of the prohibitive auxiliary possible and accounts for its imperative-like person and number options.

The Hungarian reproachative construction, with its verb in the past conditional, may be explained as follows (the unclear role of the optative, which also derives from the conditional construction, is not considered here):

  1. the counterfactual conditional construction, which can itself be used with reprimanding overtones, undergoes a process of eliding the apodosis, which gradually gives rise to an insubordinated protasis;

  2. through usage, the pragmatic function of this counterfactual protasis as a reprimand becomes increasingly conventional;

  3. the functional resemblance of this conventionalizing use to the imperative is reflected in its negation by the modal rather than the standard negative marker, the clause-initial position of the verb and the postposition of verbal particles as well as in its person and number possibilities.

4.3 Interim conclusion

The reproachative constructions in the languages of Europe have essentially been argued to be the motivated product of a number of interacting constructions – like the conditional, the optative and the imperative – and a number of typologically and cognitively plausible processes – like subject deletion in directive(-like) strategies, insubordination, conventionalization and analogy. There is, of course, no a priori reason why the reproachatives in languages from other parts of the world (see Section 1 for a Palestinian Arabic example) should all necessarily be linked to and derived from other constructions. 26 A reproachative like the Dutch one can, at any rate, not be assumed to emerge across the world. Language-specific factors have to be taken into consideration, which Estonian and Hungarian and their respective imperative paradigms show may even compel us to expand our view of what counts as a reproachative. More generally, in the theoretical approach adopted here, change is regarded as coming about in discourse and being based on usage patterns, which may but need not be conventionalized by speakers and make them link certain constructions with each other (e.g., Fischer [2008: 369] on the unpredictable nature of analogy). Explaining in this way, for instance, the absence of something like *had kept quiet! ‘you should have kept quiet!’ in English is, of course, not entirely satisfactory, since the language seems to have the same constructional ingredients as Dutch (e.g., Van den Nest 2010) and comparable insubordination patterns (e.g., D’Hertefelt 2015). So why has a similar construction not arisen in English then? One possible additional reason is that the conditional inversion construction is very rare in (the history of) the language (see Leuschner and Van den Nest 2015, as compared to German), probably with “fixing of SVO order as the major motivation for the reduced productivity of English verb-first structures” (Hawkins 1986: 211). Note, in this respect, that its optative, unlike the Dutch one, is not VSO either but if only + SVO. Another reason might be that the variation between the conditional imperative (e.g., sing and …) and the conditional inversion construction (e.g., *sing you, then …), which has possibly facilitated the loss of the subject in Dutch, is not found in English.

5 Conclusion

We have argued here that the reproachative constructions in the languages under investigation are best analyzed not as “simple” imperatives, optatives or conditionals but as hybrids resulting from established processes of change and sharing particular properties with all these constructions. What is more, our usage-based approach can quite straightforwardly account for differences between speakers of the same language (e.g., the fact that the autonomy of a new construction is a matter of degree means that the Dutch reproachative may still be so closely linked to the conditional for some people that they only accept a following dan-clause while it may be independent enough for others to find a zodat-clause unproblematic) and for differences between languages (e.g., analogy with the imperative does not entail that the reproachative is limited to the second person in Estonian).

The present analysis also has a number of implications for our understanding of the imperative. We have pointed out, inter alia, that including past tense in its semantics and expanding its meaning to (entertaining a) potentiality in order to somehow rope in imperative-like reproachatives cannot explain the cross-linguistic scarcity of imperatives with past tense marking and semantics or the ungrammaticality of patterns which should then be theoretically possible within a language (see Section 3.1.2). Comparable proposals have been made to account for the occurrence of the imperative in conditional clauses. Dancygier and Sweetser (2005: 243), for one, write that the English imperative “sets up a potential space, which serves as a background to the prediction” (see Davies 1986 too). But such a general definition does not capture the construction’s precise import in a conditional, which Fortuin and Boogaart (2009: 653) describe as a call upon the addressee for his involvement and which they regard as a trace of the directivity of the original construction in the complex construction. In our analysis, which does not overpredict cross-linguistically or language-specifically, it is the similarity to this type of appeal as well that may cause a construction with a reproachative function to start resembling an imperative.

In other words, this article endorses a view of the imperative as primarily directive – not necessarily as its core meaning, as often assumed in speech act theory (e.g., Vanderveken 2002: 46), but perhaps as the prototypical outcome of a particular bundle of semantic features, as argued in recent work by Takahashi (2012: 76–88) and Jary and Kissine (2014: 292–293).


Earlier versions of this paper were presented at the 5th UK Cognitive Linguistics Conference in Lancaster and the 6th International Syntax of the World’s Languages Conference in Pavia in 2014. The parts about Dutch elaborate on Van Olmen (2013). Thanks are due to João Almeida, Alberto Arceri, Martin Becker, Frank Brisard, Denies Du Mon, Claire Nance, Hartmut Haberland, Karolien Janssens, Mélanie Jouitteau, Hadil Karawani, Judit Kormos, Brenda Laca, Ronan le Pennec, Helle Metslang, Carmen Ríos García, Nin Sauleda-Brossa, Lauren Van Alsenoy, Johan van der Auwera, Marleen Van Peteghem, two anonymous reviewers and the editorial board of Linguistics.


  • Aikhenvald, Alexandra Y. 2010. Imperatives and commands. Oxford: Oxford University Press. Google Scholar

  • Alcazár, Asier & Mario Saltarelli. 2014. The syntax of imperatives. Cambridge: Cambridge University Press. Google Scholar

  • Becker, Martin. 2010. Mood in Portuguese. In Björn Rothstein & Rolf Thieroff (eds.), Mood in the languages of Europe, 179–197. Amsterdam & Philadelphia: John Benjamins. Google Scholar

  • Bennis, Hans. 2007. Featuring the subject in Dutch imperatives. In Wim van der Wurff (ed.). Imperative clauses in generative grammar: Studies in honour of Frits Beukema, 111–134. Amsterdam & Philadelphia: John Benjamins. Google Scholar

  • Biezma, María 2010. Inverted antecedents in hidden conditionals. In Seda Kan, Claire Moore-Cantwell & Robert Staub (eds.), Proceedings of North East Linguistics Society 40. Amherst: Graduate Linguistics Student Association. https://mariabiezma.files.wordpress.com/2012/10/biezmapaperrevised.pdf (accessed 27 October 2015). 

  • Boogaart, Ronny & Theo Janssen. 2010. Mood in Dutch. In Björn Rothstein & Rolf Thieroff (eds.), Mood in the languages of Europe, 117–132. Amsterdam & Philadelphia: John Benjamins. Google Scholar

  • Bosque, Ignacio. 1980. Retrospective imperatives. Linguistic Inquiry 11(2). 415–419. Google Scholar

  • Bybee, Joan L. 2006. From usage to grammar: The mind’s response to repetition. Language 82(4). 711–733. CrossrefGoogle Scholar

  • Bybee, Joan L. 2010. Language, usage and cognition. Cambridge: Cambridge University Press. Google Scholar

  • D’Hertefelt, Sarah. 2015. Insubordination in Germanic: A typology of complement and conditional constructions. Leuven: University of Leuven dissertation. Google Scholar

  • Dancygier, Barbara & Eve Sweetser. 2005. Mental spaces in grammar: Conditional constructions. Cambridge: Cambridge University Press. Google Scholar

  • Davies, Eirlys. 1986. The English imperative. London: Croom Helm. Google Scholar

  • Den Dikken, Marcel & Mariví Blasco. 2007. Clitic climbing in Spanish imperatives. In Wim van der Wurff (ed.), Imperative clauses in generative grammar: Studies in honour of Frits Beukema, 135–152. Amsterdam & Philadelphia: John Benjamins. Google Scholar

  • Devos, Maud & Daniël Van Olmen. 2013. Describing and explaining the variation of Bantu imperatives and prohibitives. Studies in Language 37(1). 1–57. CrossrefGoogle Scholar

  • Dobrushina, Nina, Johan van der Auwera & Valentin Goussev. 2005. The optative. In Martin Haspelmath, Matthew S. Dryer, David Gil & Bernard Comrie (eds.), The world atlas of language structures, 298–301. Oxford: Oxford University Press. Google Scholar

  • Duinhoven, A.M. 1995. Had gebeld! De irreële imperatief [Had gebeld! The irrealis imperative]. Tijdschrift voor Nederlandse Taal- en Letterkunde 111(4). 346–364. Google Scholar

  • Duinhoven, A.M. 1997. Stopte dan ook! De imperatief in ontwikkeling [Stopte dan ook! The evolving imperative]. In Els H. C. Elffers-van Ketel, Joop M. van der Horst & Wim G. Klooster (eds.), Grammaticaal spektakel: Artikelen aangeboden aan Ina Schermer-Vermeer [Grammatical spectacle: Articles offered to Ina Schermer-Vermeer], 47–58. Amsterdam: Vakgroep Nederlanse Taalkunde. Google Scholar

  • É. Kiss, Katalin. 2011. On a type of counterfactual construction. In Tibor Laczkó & Catherine O. Ringen (eds.), Approaches to Hungarian: Volume 12: Papers from the 2009 Debrecen conference, 85–108. Amsterdam & Philadelphia: John Benjamins. Google Scholar

  • Erelt, Mati. 2012. Keeldkõne vormidest [Concerning the forms of prohibitive speech]. Emakeele Seltsi aastaraamat 58. 30–38. Google Scholar

  • Evans, Nicholas. 2007. Insubordination and its uses. In Irina Nikolaeva (ed.), Finiteness: Theoretical and empirical foundations, 366–431. Oxford: Oxford University Press. Google Scholar

  • Fischer, Olga. 2008. On analogy as the motivation for grammaticalization. Studies in Language 32(2). 336–382. CrossrefGoogle Scholar

  • Fortuin, Egbert. 2003. De directive infinitief en de imperatief in het Nederlands [The directive infinitive and the imperative in Dutch]. Nederlandse Taalkunde 8(1). 14–43. Google Scholar

  • Fortuin, Egbert. 2004. De syntaxis van imperatiefsubjecten en modale particles: Een pragma-semantische benadering [The syntax of imperative subjects and modal particles: A pragma-semantic approach]. Nederlandse Taalkunde 9(4). 355–375. Google Scholar

  • Fortuin, Egbert & Ronny Boogaart. 2009. Imperative as conditional: From construction to compositional semantics. Cognitive Linguistics 20(4). 641–673. Google Scholar

  • Goldberg, Adele E. 1995. A construction grammar approach to argument structure. Chicago: University of Chicago Press. Google Scholar

  • Gros, Jules. 1984. Le trésor du breton parlé III: Le style populaire. Brest: Emgleo Breiz-Nevez. Google Scholar

  • Haberland, Hartmut. 2010. Mood in Greek. In Björn Rothstein & Rolf Thieroff (eds.), Mood in the languages of Europe, 473–491. Amsterdam & Philadelphia: John Benjamins. Google Scholar

  • Haeseryn, Walter, Kirsten Romijn, Guido Geerts, Jaap de Rooij & Maarten C. van den Toorn. 1997. Algemene Nederlandse Spraakkunst [Comprehensive Dutch Grammar]. Groningen & Deurne: Nijhoff & Wolters Plantyn. Google Scholar

  • Haspelmath, Martin. 2010. Comparative concepts and descriptive categories in cross-linguistic studies. Language 86(3). 663–687. CrossrefGoogle Scholar

  • Hawkins, John A. 1986. A comparative typology of English and German: Unifying the contrasts. London: Routledge. Google Scholar

  • Hewitt, Steve. 2010. Mood in Breton. In Björn Rothstein & Rolf Thieroff (eds.), Mood in the languages of Europe, 292–308. Amsterdam & Philadelphia: John Benjamins. Google Scholar

  • Hundertmark-Santos Martins, Maria Teresa. 1982. Portugiesische Grammatik. Tübingen: Max Niemeyer. Google Scholar

  • Jary, Mark & Mikhail Kissine. 2014. Imperatives. Cambridge: Cambridge University Press. Google Scholar

  • Jary, Mark & Mikhail Kissine. 2016. When terminology matters: The imperative as a comparative concept. Linguistics 54(1). 119–148. Google Scholar

  • Karawani, Hadil & Hedde Zeijlstra. 2013. The semantic contribution of the past tense morpheme kaan in Palestinian counterfactuals. Journal of Portuguese Linguistics 12(1). 105–119. CrossrefGoogle Scholar

  • Kaufmann, Magdalena. 2012. Interpreting imperatives. Dordrecht: Springer. Google Scholar

  • Leuschner, Torsten & Daan Van den Nest. 2015. Asynchronous grammaticalization: V1-conditionals in Present-Day English and German. Languages in Contrast 15(1). 34–64. CrossrefGoogle Scholar

  • Malchukov, Andrej. 2011. Interaction of verbal categories: Resolution of infelicitous grammeme combinations. Linguistics 49(1). 229–282. Google Scholar

  • Mastop, Rochus J. 2005. What can you do? Imperative mood in semantic theory. Amsterdam: University of Amsterdam dissertation. Google Scholar

  • Metslang, Helle & Maria-Maren Sepper. 2010. Mood in Estonian. In Björn Rothstein & Rolf Thieroff (eds.), Mood in the languages of Europe, 528–550. Amsterdam: John Benjamins. Google Scholar

  • Muižniece, Liena, Helle Metslang & Karl Pajusalu. 1999. Past participle finitization in Estonian and Latvian. In Mati Erelt (ed.), Estonian: Typological studies III, 128–157. Tartu: University of Tartu. Google Scholar

  • Nikolaeva, Irina. 2007. Constructional economy and nonfinite independent clauses. In Irina Nikolaeva (ed.), Finiteness: Theoretical and empirical foundations, 138–180. Oxford: Oxford University Press. Google Scholar

  • Paardekooper, P.C. 1951. De imperatief als grammatische categorie in het ABN [The imperative as a grammatical category in Standard Dutch]. De Nieuwe Taalgids 44. 97–107. Google Scholar

  • Palmer, Frank R. 1986. Mood and modality. Cambridge: Cambridge University Press. Google Scholar

  • Platzack, Christer & Inger Rosengren. 1998. On the subject of imperatives: A minimalist account of the imperative clause. Journal of Comparative Germanic Linguistics 1(3). 177–224. Google Scholar

  • Portner, Paul. 2007. Imperatives and modals. Natural Language Semantics 15(4). 351–383. CrossrefGoogle Scholar

  • Postma, Gertjan, Daniël Van Olmen & Evie Coussé. 2013. Discussie [Discussion]. Nederlandse Taalkunde 18(3). 289–305. CrossrefGoogle Scholar

  • Proeme, Henk. 1984. Over de Nederlandse imperativus [About the Dutch imperative]. Forum der Letteren 25(4). 241–258. Google Scholar

  • Schwager, Magdalena. 2011. Imperatives and tense. In Renata Musan & Monika Rathert (eds.), Tense across languages, 37–58. Berlin: De Gruyter. Google Scholar

  • Takahashi, Hidemitsu. 2012. A cognitive linguistic analysis of the English imperative: With special reference to Japanese imperatives. Amsterdam: John Benjamins. Google Scholar

  • Traugott, Elizabeth C. & Richard B. Dasher. 2002. Regularity in semantic change. Cambridge: Cambridge University Press. Google Scholar

  • Van den Nest, Daan. 2010. Emergenz und Grammatikalisierung von V1-Konditionalen: Ein Rekonstruktionsversuch am Beispiel des Deutschen und Englischen. Ghent: Ghent University dissertation. Google Scholar

  • Van der Wurff, Wim. 2007. Imperative clauses in generative grammar: An introduction. In Wim van der Wurff (ed.), Imperative clauses in generative grammar: Studies in honour of Frits Beukema, 1–94. Amsterdam & Philadelphia: John Benjamins. Google Scholar

  • Van Olmen, Daniël. 2011. The imperative in English and Dutch: A functional analysis in comparable and parallel corpora. Antwerp: University of Antwerp dissertation. Google Scholar

  • Van Olmen, Daniël. 2013. De imperatief in de verleden tijd [The imperative in the past tense]. Nederlandse Taalkunde 18(3). 253–267. CrossrefGoogle Scholar

  • Vanderveken, Daniel. 2002. Universal grammar and speech act theory. In Daniel Vanderveken & Susumo Kubo (eds.), Essays in speech act theory, 25–62. Amsterdam: John Benjamins. Google Scholar

  • Vandeweghe, Willy. 1994. Modale partikels en taalhandelingstypes [Modal particles and speech act types]. In Fons Maes, Paul van Hauwermeiren & Luuk van Waes (eds.), Perspectieven in taalbeheersingsonderzoek [Perspectives in language proficieny research], 144–151. Dordrecht: ICG Publications. Google Scholar

  • Vicente, Luis. 2013. Past counterfactuality in Spanish imperatives and its implications for the analysis of imperatives. http://semanticsarchive.net/Archive/DM5ZGY1N/vicente—past-counterfactuality-in-spanish-imperatives.pdf (accessed 27 October 2015). 

  • Wolf, Henk. 2003. Imperatieven in de verleden tijd [Imperatives in the past tense]. Taal en tongval 55(2). 168–187. Google Scholar


  • 1

    Karawani and Zeijlstra (2013) examine the semantics of kaan in declaratives and interrogatives and argue that it conveys tense when the clause does not contain any other tense marker and counterfactuality when another tense marker is present. This analysis may be extendable to (1): if one assumes that imperatives like kol ‘eat!’ express tense (e.g., non-past according to Takahashi [2012: 71–72]) or just rule it out (e.g., Platzack and Rosengren 1998), one expects kaan in imperatives to have a counterfactual meaning as well – which is exactly what ‘should have + past participle’ is. This issue is, however, beyond the scope of the present article, which will concentrate on the languages of Europe. 

  • 2

    The speaker will be referred to as feminine and the addressee as masculine here. 

  • 3

    The following abbreviations will be used here: 1,2,3 first, second and third person; acc accusative; ade adessive; cmpr comparative; comp complementizer; cond conditional; emp emphatic; f feminine; fv final vowel; gen genitive; ill illative; imp imperative; ine inessive; inf infinitive; ipfv imperfective; m masculine; mp modal particle; neg negation; nom nominative; obj object; pl plural; poss possessive; proh prohibitive; prs present; prtv partitive; prtc particle; pst past; ptcp participle; refl reflexive; rel relative; sbj subject; sbjv subjunctive; sg singular. 

  • 4

    The notions of imperative, optative and conditional, like that of reproachative, are considered comparative concepts à la Haspelmath (2010) here. The imperative, for instance, can be characterized as a “sentence-type whose only prototypical function is to provide the addressee(s) with a reason to act, that is suitable for the performance of the full range of directive speech acts, and whose manifestations are all morphologically and syntactically homogeneous with the second person” (Jary and Kissine 2016: 132). This functional definition allows us to compare constructions across languages but does not presume that imperatives are identical in form and function cross-linguistically or, for that matter, that every language has a dedicated imperative. So when, in the rest of the present article, the reproachative construction is related to, say, the imperative or the optative in some language, we are referring to the language-specific instantiations of these comparative concepts. Still, one of our main claims is that, despite the fact that the reproachative in one language may differ from that in another and the same holds for the imperative, the optative and the conditional, similar relations appear to exist between them in the languages under investigation. 

  • 5

    Van der Wurff (2007: 50) rightly points out: “If features like this [i.e., past tense marking in imperatives] can escape notice in a language as well studied and documented as Dutch, establishing whether other languages have them may not be a simple matter of checking reference grammars.” 

  • 6

    One of the reviewers also points to a special morphological verb form for the expression of reproaches in the South Estonian dialect of Mulgi. For more (though very limited) information, see Muižniece et al. (1999: 147). 

  • 7

    As one of the reviewers points out, it is possible to insert the strong form of a second person subject pronoun (e.g., jij rather than je) in the Dutch reproachative construction. One could, for instance, add it to (9a), i.e., had jij toch gezwegen!, to emphasize that it is the addressee who should have kept quiet and not someone else. Importantly, in this respect too, the reproachative resembles the imperative (see Fortuin 2004 on imperative subjects in Dutch). 

  • 8

    Dutch has been argued to have a so-called preterit imperative too (e.g., Duinhoven 1997). The attested example in (ia) is a case in point: it has a preterit verb form in clause-initial position and no overt subject. It relates not to the past but to the present/future, though. The difference with the imperative in (ib) is that the directive is presented as less firm and as unlikely to be complied with. The preterit thus appears to have a distancing effect (see Proeme 1984: 251–52).




    ‘Man, you should get angry for once!’

    (Boogaart and Janssen 2010: 125)


    ‘Man, get angry!’

    According to one of the reviewers, (ia) can also have a past reading, which can be triggered by the addition of a clause like dan kreeg je meteen op je donder ‘you immediately got a scolding’ and can be paraphrased as ‘imagine that you got angry (back then)!’. This use is not generally accepted, however (see Wolf 2003: 169; Van Olmen 2013: 255). Neither is the construction as such, for that matter: the reference grammar of Dutch discusses a pluperfect imperative but does not mention a preterit one. For that reason, little attention will be paid to the latter in the rest of this article. Let the following two remarks suffice. First, its conditional overtones suggest that, like the reproachative (see Section 4.1), it derives from a conditional inversion construction (e.g., werd je boos, dan kreeg je meteen op je donder ‘if you got angry, you immediately got a scolding’). Second, as the reviewer hypothesizes, its potential occurrence in a conditional coordination construction with no real sense of directivity (e.g., ?werd boos en je kreeg meteen op je donder literally ‘got angry and you immediately got a scolding’) would then be a further development: as the insubordination of the protasis and the omission of the subject conventionalize, which may be facilitated by the pre-existence of the reproachative, the preterit construction becomes more autonomous and, like the imperative, may be used in the conditional coordination construction. 

  • 9

    Biezma (2010: 5), however, gives an example of a reproachative that is directed at a third person. Vicente (2013: 5–9) too argues that such an interpretation is possible and that this should not actually come as a surprise since ordinary directive infinitives in Spanish accept it as well. But he admits that a third person reading of the construction in (11a) is “somewhat uncommon” (Vicente 2013: 7). Confirmation of this observation (Ríos García p.c.) suggests to us that it may be of a pragmatic nature: the speaker reprimands a third person as if they are present. 

  • 10

    She discusses a number of other “non-imperative” properties but it is unclear how specific to the reproachative they really are. The claim that reproachatives are only acceptable if they mention the weakest alternative that would have made the state of affairs desired by the speaker and/or the addressee come true (see Biezma 2010: 6–7) can serve as an example. As compared to ‘you should have taken a cab!’, ‘you should have taken a red cab!’ indeed sounds strange as a reply to ‘sorry I’m late’ but, as a reaction to ‘I think I’ll be late’, the imperative advice ‘take a red cab!’ also makes sense only if red cabs are faster than regular ones. 

  • 11

    Schwager’s (2011: 49–52) analysis of Dutch closely resembles Vicente’s (2013) account of Spanish and will therefore not be discussed. Let if suffice to say that, at least, she does not have to resort to assuming the presence of some silent morpheme as the perfect auxiliary in the Dutch reproachative carries past tense marking. 

  • 12

    The judgments in (16) and (17) are not Proeme’s (1984). They are based on our own intuitions and on consultations with other native speakers (e.g., Du Mon p.c.; Janssens p.c.; Van Alsenoy p.c.). 

  • 13

    More generally, the belief that a definition of the imperative should accommodate the imagine reading is debatable (e.g., Fortuin and Boogaart 2009: 648–655). Not only does corpus research indicate that 99 % of Dutch imperatives have a directive or perform interpretation (see Van Olmen 2011) and can we thus assume from a usage-based point of view that this reading is much more entrenched, but the appeal to imagine some state of affairs also arises only in implicit or explicit conditional contexts, which we may want to attribute it to (see Fortuin 2003: 26–27). 

  • 14

    The judgment in (21a) is not Biezma’s (2010). 

  • 15

    At first glance, this vagueness between directivity and assertion could be argued to make analogy of the counterfactual conditional inversion construction as a reprimand with the conditional imperative construction somewhat easier in that the former actually involves a post hoc – thus more purely conditional – appeal to the addressee – like a directive. But a closer look at the conditional imperative construction reveals a further difference with the future reproachative: when part of a conditional assertion, as in (24), imperatives typically have a (covert) subject that is interpreted generically and refer to an undesirable rather than a desirable state of affairs (see Fortuin and Boogaart 2009: 644, 650). 

  • 16

    The Digital Library of Dutch Literature is available online at<http://www.dbnl.org>and was accessed in July 2015. 

  • 17

    As one of the reviewers points out, it is relatively easy to look for examples of the constructions under examination that contain modal particles. One could, for instance, search a sizable corpus for combinations such as had/was (+ je) + maar/toch + past participle in an automatic way and extract the potentially numerous relevant attestations manually. Such a study would, however, be somewhat biased in that the cases without modal particles are not taken into account. Those cases can, of course, be searched for in the manner sketched in the present section. For a large corpus, this would be a very time-consuming enterprise, though, given the huge amount of irrelevant hits. A systematic diachronic corpus study of the Dutch reproachative seems interesting but is beyond the typological scope of this article. 

  • 18

    For reasons of space, the linguistic context of the constructions under discussion will only be translated here, not glossed. For the sake of clarity, the core elements of the constructions will be highlighted in bold in the longer examples. 

  • 19

    One of the reviewers suggests that the second clause in the protasis could also be regarded as a continuation of the initial optative clause in which the subject is not expressed. Such an analysis, though perhaps not entirely impossible, would mean that the constraints on contraction at the level of the sentence in Dutch (e.g., Haesery et al. 1997: Section 27.5) are violated here. In a coordination construction, a second clause subject that is coreferential with the first clause subject and follows the finite verb cannot, in principle, be left out. 

  • 20

    Additional supporting evidence for the path from counterfactual optative used as a reprimand to reproachative comes from one of the reviewer’s observations. The earliest attestations of the reproachative that they found online, one of which dates back to 1819, all contain liever ‘preferably’. This adverb also frequently occurs in older examples of the optative. 

  • 21

    Paardekooper’s (1951: 99) intuition about the prototypical presence of maar and the atypical absence of any modal particles in the construction appears to be accurate. 

  • 22

    This point was raised by one of the reviewers. They also put forward the hypothesis that, synchronically, the optative and the reproachative are instantiations of a more abstract verb-first construction with maar. The idea is not unappealing but it remains to be seen what this construction’s more general semantics would be and why maar would be compulsory in its optative instantiation and optional in its reproachative one. 

  • 23

    This argument is made by one of the reviewers. 

  • 24

    Special thanks are due to Helle Metslang for her help with the glossing of (40). 

  • 25

    Two Estonian participants of the 6th International Syntax of the World’s Languages Conference in Pavia in 2014 told us that they recognized the form ärnud but would never use it themselves. 

  • 26

    This was pointed out to us by one of the reviewers. 

About the article

Published Online: 2017-12-14

Published in Print: 2018-01-26

Citation Information: Linguistics, Volume 56, Issue 1, Pages 115–162, ISSN (Online) 1613-396X, ISSN (Print) 0024-3949, DOI: https://doi.org/10.1515/ling-2017-0033.

Export Citation

© 2018 Walter de Gruyter GmbH, Berlin/Boston.Get Permission

Citing Articles

Here you can find all Crossref-listed publications in which this article is cited. If you would like to receive automatic email messages as soon as this article is cited in other publications, simply activate the “Citation Alert” on the top of this page.

Daniël Van Olmen
Journal of Pragmatics, 2019, Volume 139, Page 146

Comments (0)

Please log in or register to comment.
Log in