PART 1: DIACHRONIC EVIDENCE

 

Russian constructions with 'take' expressing an unexpected event: Their historical origin and development in the 19th century

 

 

Daniel Weiss

Slavisches Seminar der Universität Zürich, Switzerland E-mail: dawe@slav.uzh.ch

 

 

---

ABSTRACT

This study aims at elucidating the rise and expansion of four competing Russian constructions expressing the unexpectedness of an event. Three of them are built according to the pseudo-coordinative model 'take and do' and one follows the serial model 'take do'. The historical data stemming from the Russian national corpus and covering the whole 19^th century reveals striking differences between these constructions in terms of frequency and grammaticalisation, the most peripheral being the serial model. Evidence for ongoing grammaticalisation is mainly based on the rise of non-canonical second verbs denoting an uncontrollable event and inanimate subjects. Special attention will be given to the meanings of the imperative and contextually bound pragmatic effects.

Keywords: "Surprise" constructions; pseudo-coordination; verb serialisation; syndetic linking; grammaticalisation of 'take'; imperativus dramaticus; pragmatic effects; Russian colloquial syntax

---

 

 

1. Introduction

The present study is devoted to a family of Russian constructions that may be loosely described as 'take and V2' with a varying linking element in between them and the approximate meaning 'suddenly do V2'. These constructions have been discussed in several studies including Kuznetsova 2006; Kor-Saxin 2007; Weiss 2007; Stojnova 2007; and Weiss 2008. In its overall conception, the present study continues the latter article but unlike all abovementioned studies, it examines the said constructions not in contemporary Russian but traces them back to their origin and historical development throughout the 19^th century. Thus, it provides the East Slavic counterpart for Andrason, Gebka-Wolak and Moroz' analogous scrutiny of the Polish construction with wziqc (this volume). Moreover, since the construction with the imperative shows marked effects of an ongoing pragmaticalisation, this study also relates to the investigation of similar phenomena in Weiss forthcoming. The areal affiliations of such 'take' constructions as illustrated in Coseriu 1966; Ekberg 1993; and Nau et al. 2019¹ as well as the riddle of the semantic change which brought about the transition from the literal understanding of 'take' to the meaning 'suddenly, unexpectedly' will only briefly be commented on in section 5.

The essence of the constructions to be discussed (at least in their non-imperative uses) has already been captured in Usakov's 1935 dictionary, which provided the basis for all subsequent Soviet dictionaries of the Russian language:

Взять 12. Неперех. Разг. «Употр. в сочетании с союзами и, да и следующим глаголом употр. для выражения внезапного или неожиданного действия». Ušakov 1935

Vzjat' 12. Intrans. Coll. «ís used in combination with the conjunctions i, da and the following verb to express a sudden or unexpected action».

Similar wordings have been chosen by all subsequent authors who examined the Russian 'take and do' constructions, notably Isacenko 1960; Prokopovic 1969; Fortuin 2000:149-161; and Kor-Saxin 2007. The component 'sudden, unexpected' is indeed crucial for the semantic change of the first verb vzjat', originally meaning 'take'. The term 'semantic bleaching', which is often used in descriptions of grammaticalisation processes, seems ill-fitted here since the meaning 'take' is not bleached but replaced by a completely different meaning that still has to be specified: 'unexpected' to whom? The current speaker, a fictive observer or a protagonist of the ongoing narrative? At any rate, we are dealing with a specific type of modality. The examples illustrated below will show that all three cases are met. Be this as it may, 'take' constructions come now to compete and/or combine with adverbs expressing surprise in Russian. ín Balkan languages, the admirative mood serves a similar purpose.

The second component of Usakov's definition that calls for clarification is the term dejstvie referring to V2: in metalinguistic comments, dejstvie denotes much more than simple actions, it also covers events and even states. This will in particular be illustrated by those examples that allow for non-human actors. Moreover, the loss of the action meaning of 'take' is corroborated by the main syntactic change undergone by it: most authors claim that it can no longer govern a direct object - hence Usakov's indication "intransitive". This feature is, however, questionable, as has already been shown in Weiss 2008:488-489. What remains unchallenged is another, semantic invariant: all these 'take' constructions denote one single event (probably the most appropriate metalinguistic equivalent of Russian dejstvie), which runs counter to any interpretation in terms of a coordinative linking of two clauses; hence the term 'pseudo-coordinative' intended to design such structures². V2 usually denotes an achievement or an accomplishment, seldom an activity but never a state in Vendlerian terms. Vzjat' may be said to function as a kind of adverbial modifier of V2.

The further characteristics of said constructions may be summarised as follows:

(i) the morphological outfit of both verbs is identical regarding ±finiteness and the grammemes of person, number, gender, mood, and voice, but not necessarily tense and aspect.

(ii) both verbs share at least the subject valence and may also have a common object slot and/or a common adjunct (adverbial) position.

So far, both the single event constraint and the restrictions (i)-(ii) are reminiscent of those that characterise the so-called double verbs in Russian as described in Weiss 2012:613-615. By contrast, the ban on the repetition of auxiliary morphemes found with double verbs does not hold without exceptions, and the inversion of the two verbs, which is frequent with double verbs, seems impossible in the case of 'take' constructions. Moreover, unlike double verbs, which allow for both the negation of V2 and, though less often, for the negation of both verbs, in the 'take' construction only V2 can be negated. The whole relationship between the two construction families is even more intricate since the 'take' family also features a serialized member (see below).

The abovementioned characteristics hold also for chains with the imperative voz'mi/voz'mite, although they do not carry the same meaning: as already noted by Usakov 1935, their principal meaning is not related to unexpectedness but on the contrary, it marks an urgent request, or in Fortuin's words (2000:156): "... the vz/'at'-construction is used to eliminate the addressee's possible hesitation to do the action." On the other hand, the 'take' imperative participates in several semantically dedicated constructions, among which the so-called 'imperativus dramaticus' is the most frequent one: not surprisingly, it marks an unexpected turn in narration, thus re-activating the meaning of the non-imperative constructions. It is, however, not available in the serial construction. In all these special uses, the imperative loses its directive meaning.

A final remark concerns the characterization of the 'take and V2' model as colloquial. Most examples in my corpus are indeed quoted from direct speech and free reported speech, which allows for more freedom in the choice of colloquialisms. The same holds for the numerous examples from Internet communication (including social networks) in today's language quoted in Weiss 2008.

2.1 A first overview: the main properties

In what follows, I will attempt to summarise the principal findings in Weiss 2008. The following overview presents all available members of the whole construction family:

(a) vzjat' i X-ovat'

(b) vzjat' da X-ovat' a)-c) ^ pseudo-coordination

(c) vzjat' da i X-ovat'

(d) vzjat' X-ovat' ^ serial verb

(e) vzjat', X-ovat' ^ asyndetic linking

The first three constructions (a) - (c) contain coordinative conjunctions with the basic meaning 'and'. Yet, since there are no two separate events that could be conjoined, the whole constructions can only be subsumed under the pseudo-coordinative category. The construction with the complex conjunction da i is in Russian lexicography treated as an expression of surprise, hence the variant c) is inherently redundant. Since it also co-occurs with the imperativus dramaticus, such examples exhibit a threefold marking of the same semantic element, i.e. 'take' + da i + imperativus dramaticus:

'Die' as V2 is very frequent, see below, section 2.4. As if this were not enough, sometimes an adverb such as vdrug provides the fourth marker:

Construction d) represents a case of true serialisation, more precisely: an asymmetrical SVC according to the typology sketched out in Aikhenvald 2006 and Aikhenvald 2018. Thus, the 'take' model overlaps with the VSC construction type, and we end up with a bundle of three pseudo-coordinative constructions and one serial counterpart. It should be added that the asymmetrical relation characterises the whole 'take' construction family. To quote from Aikhenvald 2018:55:

"An asymmetrical serial verb construction consists of a verb from a relatively large, open, or otherwise unrestricted class, and a verb from a semantically restricted (or closed) class. Asymmetrical SVCs will tend to denote a single event described by the verb from the non-restricted class. The verb from a closed class provides an additional specification: ..."

In our case, all this pertains also to the four constructions involved: V2 belongs to a practically unrestricted class excluding only states, whereas vzjat' (the "minor verb" according to a different terminology) is the unique member of a restricted class and serves as an adverbial modifier of V2. The only difference is that besides the serial pattern, the asymmetrical relation encompasses also three syndetic pseudo-coordinative constructions. To complete this picture it should be emphasised that the asymmetrical nature of the constructions under scrutiny distinguishes it from Russian SVC constructions in general, which may also belong to the symmetrical type (Weiss 2012:615-617).

Case e) requires a special comment. Contemporary examples suggest that uses with a prosodic break between the two verbs or its graphic equivalent with comma or dash should indeed be distinguished from "pure" serial linking (Weiss 2008:475). However, historical examples seem less reliable in this respect. Moreover, the comma variant often occurs in triplets of the type vzjat', V2 i V3 like the following:

"as soon as I joined some feast assembly of Russians I always got the impression that they only [pretend] but out of a sudden would get up and go on the rampage, as if they were at home."

This constellation leads to an incongruency between the prosodic and the semantic articulation: semantically, the verb form vstanut belongs to the 'take' construction, whereas prosodically it enters the subsequent coordinative chain. Yet, such triplets never render one single event as required for the 'take' construction but the last verb enters a temporal sequence with the two preceding ones (Weiss 2008:486). As syndetic uses are sometimes also separated by a comma, I decided to disregard interpunction altogether in this study and count all chains of the type vzjat'(,) V2 as serialised.

2.2 The historical data

All characteristics described above may also be found in my historical corpus collected from the Russian national corpus (NKRJa). A caveat concerns the composition of this corpus: its unbalanced structure with a marked preference for literary texts at the expense of almost all other genres makes itself even more felt in the 19^th century. Not surprisingly, my historical data comprise almost only literary sources and a few examples of the press (Russkaja mysl'). If we wanted to obtain a more balanced sample we would have to systematically mine personal letters, diaries, and the like, which is beyond the scope of this study. The search was limited to immediately juxtaposed components of the string except for negation, e.g. vzjat' da ne V2. Taking into account strings with one or two intermediate components before V2 might substantially alter the proportions, as was shown in Weiss (2013:329-332) with double imperatives. It will be the object of a later investigation.

The chronology of the above-listed constructions does not vary substantially: the first attestations range from 1827 to 1844. However, the divergence in terms of numeric representation is remarkable: it varies between 41 (serial construction) and 416 instances (da i construction). The following table presents the numerical distribution of the different constructions.

 

Table 1

 

The preponderance of the inherently redundant type with da i becomes even more striking considering that 11% of these uses are additionally marked by the imperativus dramaticus (the first occurrence dates from 1843). Taking into account cases with an additional adverb such as vdrug as in (3) with a cluster of four different markers, it seems obvious that the 'take' constructions are prone to the maximal expression of the intended meaning right from the beginning. This impression is corroborated by the figures in the last column: they represent additional adverbial markers with the meaning "suddenly", such as vdrug, ni s togo ni s sego or the combination of these two. Another synonymous adverb would be vnezapno, which is not attested in my corpus;³ neozidanno occurs only once (Starik vzjal i podpisal neozidanno dlja menja, A. Cexov 1894). On restrictions of vdrug cf. Kor-Saxin 2007:237, more on adverbial markers Weiss 2008:476. Interjections like bac! or bux! replacing V2 serve a similar purpose in contemporary language (Weiss, ibidem) but do not occur in the historical corpus.

In terms of grammatical categories, all persons, genders, tenses and moods are represented; for aspect, see section 1.4. As for voice (of V2), only one passive form was noted (example 10), but the quotation does not reflect standard use anyway. Among non-finite forms, only infinitives were found.

2.3 Ambiguity

As the rubric 'ambiguous cases' suggests, the counts reflected in the table are not fully waterproof. Two sources of ambiguity may be singled out (Weiss 2008:488f.): the existence of a direct object of vzjat' and the exact meaning of vzjat' itself. Theoretically, an existing overt object may precede vzjat' , follow V2 or occupy the intermediate position. Instead, the object may be (and in colloquial Russian very often is) covert (elliptical), which increases the possible variants to four. Moreover, V2 may itself be transitive; if we take into account that an object may depend on vzjat' , V2 or both, we end up with twelve different possibilities. Note that position and dependency are not interrelated due to the free word order within such coordinative strings, cf. Ja tebja sizu i ozidaju lit. 'I you sit wait' / I am sitting here and waiting for you', where the object is preposed to an intransitive verb.

In my sample most overt objects follow V2 , while pre- and interposition practically never occur. Things get more complicated when it comes to the more frequent elliptical objects: a 0 governed by vzjat' is hard to distinguish from the non-existence of an object in the new meaning of unexpectedness. In this situation, determining the meaning of the first verb becomes decisive. Several examples do not allow the literal reading because this would create a misleading temporal sequence. Such is the case in

Obviously, you cannot first 'take' a journal and then establish it, therefore vzjat' does not function in any literal sense but lends itself only to the pseudo-coordinative reading. The next fragment refers to an already established referent with an anaphoric pronoun:

The adverb 'suddenly' points to an unexpected event but is compatible with the idea that the parents seized their daughter physically to marry her, all the more so as the adverbial 'by force' may also be related to physical violence. Yet, the sequence 'seized and married her' is not well-formed for temporal reasons, since the second action usually presupposes a lengthy and careful preparation. Therefore, this is also an unequivocal instance of the pseudo-coordinative vzjat'. The following context allows for both the literal and the derived meaning:

The physical action of turning a table upside down is of course possible but also pretty senseless and hence unexpected, which is elaborated in the subsequent two rhetorical questions. Thus, this is an example of true ambiguity. Interestingly enough, the first seven attestations of the pseudo-coordinative construction with da i including the famous quotation from Gogol's "Nose" (see below, example 16) all represent the same type of ambiguity. This might be interpreted as evidence for the initial stage of the grammaticalisation process. On the whole, one has to admit that most ambiguous examples call for a lengthy discussion and often do not yield a satisfactory solution. Whether they provide the turning point for the emergence of the meaning of unexpectedness remains an open question.

2.4 Further evidence of continuous grammaticalisation

So far, only examples with 'take' in the perfective aspect have been examined. There is, however, also an alternative with the imperfective counterpart brat'. In Weiss (2008:480-481), I discussed a few contemporary uses with iterative meaning in the present and past tense and with the historical present. Predictably, the progressive meaning was not represented, as this meaning runs counter to the punctual character of the event characterised as sudden. My historical sample from the NKRJa contains only two instances of the construction brat' i V2 in the iterative present and one with the imperative (beri i sadi); all three stem from the end of the 19^th century.

The constructions with brat' da and brat' da i in the NKRJa are attested as late as 1991 and 1967, respectively. The contemporary examples quoted in Weiss 2008 belong to all three types of pseudo-coordinative linking (with i, da and da i). The late appearance of the imperfective counterparts suggests that this criterion should likewise be taken into consideration for a comprehensive account of the grammaticalisation process.

Another criterion that might be significant is the non-concordant marking (Aikhenvald 2018:101) of aspect, i.e. the combination of both grammatical aspects in the same chain. This usually also triggers a change of tense as in vzjal i pisn 'I took.PST and write.PRES' (Weiss 2008:479). The same combination was found in a quotation from 1877 (Vzjal i pisu: «Syn! Kogda tty menja nazyvaes' tjaten'koju, to ja tebja synom nazyvaju» 'I took and write: My son! When you call me daddy, I call you son'), and a similar example (Vzjal i molen 'I took.PST and am silent.PRES' dates from 1865. In such cases, the imperfective V2 serves as a historical present.

A combination of two different aspects in the same future tense appears in the formula voz'mu i ne budu, which today serves as a pragma-phraseme, expressing a note of humoristic defiance, cf. Voz'mu i ne budu vzroslet' lit. 'I'll take and not grow up'. The phraseme obviously emerged only in the 20^th century; the first attestation in the NKRJa dates from 1950, the only earlier unequivocal instance being remarkable in that it denotes a state (ne budu ljubit'), which does not fit in the overall characteristics of the 'take' pseudo-coordinative constructions. Nowadays, this phraseme has become very frequent in Internet communication and appears in all grammatical persons. However, the historical corpus already yields the synonymous variant with stann:

The non-concordant combination of both aspects is illustrated in a few cases with V2 denoting an activity:

The speaker's command of standard Russian in (10) seems, however, not impeccable. On the whole, this historical data is too scarce to serve as evidence for an ongoing grammaticalisation of the pseudo-coordinative constructions under discussion. The same holds even more for negation (recall that only V2 may be negated): example (8) is the only unambiguous attestation. By contrast, in the 20^th century the NKRJa yields dozens of hits with vzjat' i ne V2.

A more reliable criterion is also illustrated by example (10): the detransitivisation of vzjat' is often accompanied by the loss of the agent role of its subject. This makes possible the use of non-human subjects as in mesto 'place', which already marks an advanced stage of deagentivisation. All preceding stages are attested in my material. To begin with, animals may be assigned an agentive role, for instance, Sobaka-to ... vzjala da i smygnula vperednjuju 'the dog suddenly slipped to the front (Turgenev 1847). The da i-construction features ten animals as subjects, among them a bear, a fox, a snake, a cock, a seagull, and a pike, the construction with i has four animals (a dog, a hare, a he-goat, a cock).

Close to the canonical subjects of vzjat' in its principal, physical meaning are human beings that involuntarily undergo a process such as 'die', 'forget', 'remember', 'fall asleep', 'get angry', 'mix up', 'break down', etc., expressed by V2. Such examples are quite numerous in all syndetic constructions: they amount to a total of 51, among them 29 belonging to the lexical field of dying, including such verbs as 'croak', 'get poisoned', or 'drown'. Yet, they are very unequally distributed: the construction with da i features 39 occurrences, among them 19 with the meaning 'die', while the construction with da has only 5 instances (among them 3 uses of dying but also 2 instances of vzjat' da udavit'sja 'suddenly hang oneself), and the i-construction features 6 cases of dying. Even more striking is the complete lack of involuntary events within the serial construction.⁴ All this allows for a first conclusion: being the most innovative and advanced of the three pseudo-coordinative competitors, the construction with da i also allows for most diversity.

Inanimate entities do not lend themselves to an agentive role and hence represent a stage more remote from the starting point of the grammaticalisation. The da i-construction features twelve subjects referring to physical objects (the sun, a roof beam, a boat, a place) and abstract entities (rumours, disputes, the nature, illness, a feeling, society, the general conversation, and England), while the i-construction has only two such subjects (an idea and strawberry flowers). The following excerpt in the imperativus dramaticus illustrates the use in question:

Some uses, such as spletni 'rumours' and Anglija 'England', are based on metonymy and may thus be said to be closer to the starting point.

Table 2 summarises the results of the last two criteria examined:

Interestingly enough, the serialized construction vzjat' V2 does not manifest any of the properties indicating ongoing grammaticalisation discussed here and was therefore not included in the table. Its only particularity consists in its proneness to enter coordinated triplets (30 out of 34 non-imperative uses) as illustrated above in example 3.

Against the state of affairs in contemporary language (Weiss 2008:490-495), the cases discussed look rather modest: today, one even finds uses with an impersonal (subjectless) weather predicate, cf. kak vdrug voz 'mi daprimoroz' 'when it froze out of a sudden' or voz 'mi da i zamerzni (same meaning).

That the grammaticalisation is still underway is corroborated by another observation. Some authors seem to feel the need to mark the unusual character of the new construction metacommunicatively. Such is the case of Turgenev, who in the next example not only puts the form vzjal in brackets but adds the metacommunicative marker 'as they say':

Note the adverb ni s togo ni s sego, which underlines the unexpectedness of the whole event. Thus, not only the contents are marked as unexpected but also the wording. The second example dates much later but may be conditioned by the non-canonical V2 combined with an abstract subject:

The unexpected character of the event is again enhanced by vdrug and the additional marker 'particularly amazing' in the subsequent string.

All in all, we may conclude that the different criteria for grammaticalisation examined in this section are represented very unequally. The imperfective aspect of the minor verb is attested only three times at the end of the period under scrutiny and negation of V2 only once. By contrast, inanimate subjects and uncontrollable events denoted by V2 begin to appear in the second half of the 19^th century. Both processes are mainly reflected within the most frequent construction with da i. Strikingly enough, more than half of all instances of involuntary events are centred around the general meaning 'die'; other meanings begin to appear in the construction with da i. All these findings call for a comparison with a sample of similar size from the 20^th century to grasp the full grid of different grammaticalisation paths. As has already been mentioned, the later period brings a rise of negated V2 constructions and uses of the imperfective brat' but it also remains to be clarified whether the parameters already productive in the 19^th century continue to gain momentum in the subsequent period.

A look at the general historical development of the most frequent construction with da i in the NKRJa shows a sharp increase between 1840 and 1856, which brought about a quadruplication of hits, and another peak in 1882, followed by a smooth decrease until 1900 ending up with about half of the hits against the peak. The amount of noise during the whole period is negligible but recall that this statistic also contains 26 ambiguous cases. The majority of inanimate subjects and involuntary events denoted by V2 are located within the same period. The count starts with seven ambiguous instances, which makes the graph even steeper. The 20th century sees the da i- construction in decline: the last hit dates from 2004. A similar fate affects the much less frequent construction with da: it reaches its peak in the early 1850s, its last occurrence in the NKRJa being from 1987. Due to the scarce data and the unbalanced composition of the Russian national corpus, these numbers can, however, not be considered statistically significant. By contrast, the third construction with i still flourishes and has become the most frequent of all three syndetic constructions.

 

3. Imperatives and conditionals

As mentioned in the beginning, imperatives do not fit in the semantic characteristics of the take-constructions if they do not represent an imperativus dramaticus. Yet, it turns out that the latter constitutes the majority of all uses: the 63 cases by far outnumber all other uses taken together. To begin with, the directive meaning ascribed to voz'mi(te) da/da i/i V2-uj(te) by Russian lexicographers ('don't hesitate, go on') underlies only four examples within the i-construction, among them three by Cexov and one by Gor'kij. The following example illustrates the addressee's hesitation, if not unwillingness to perform the requested act:

The da i-construction features five directive uses, the da-construction two other ones.⁵ All other uses are not directed at a concrete addressee but represent modal uses of the imperative. The latter may function as a marker of a conditional or concessive clause:

Most probably, this utterance is no longer directed at a concrete addressee but a generic 'you'. The same holds for three uses introduced by xot'. They render a hypothetical consecutive meaning as described in Fortuin 2000:112 f:

Cf. the paraphrase 'so wet that one just could.' This dedicated construction still combines with 'take' - pseudo-coordinative constructions in contemporary language, cf. Weiss 2008:481484. The particle xot' may also co-occur with or be replaced byprosto 'just'. Such is the case of a famous quotation from Gogol's Nose:

Note that this example is not fully waterproof since voz 'mi could also be interpreted literally, the elided object referring back to the human being without a nose.

All in all, we may conclude that the overwhelming majority (63 = 80%) of the 78 attested pseudo-coordinative imperative forms occur in the construction of the imperativus dramaticus, whereas the prototypical directive function of the imperative is represented only in 11 (= 14%) cases (and the remaining modal uses yield 5 (= 6.4%) cases. This amazingly small portion of prototypical (directive) uses is all the more striking as the serial construction has only imperatives with directive meaning (seven instances). Thus, serial imperatives with voz'mi/voz'mite behave more like serial imperatives in general, which seldom lose their directive function even if they take new pragmatic meanings (Weiss 2013:334-336). Recall that

the serial voz 'mi/-te construction also never functions in the imperativus dramaticus. Example 17 illustrates the directive use:

In sum, imperatives reveal two main tendencies: a marked divergence between pseudo-coordination and serialisation and a drift away from the core meaning of the imperative in the case of the pseudo-coordinative constructions.

Forms of the conditional, including both the l-form and the infinitive (see example 9), occur 20 times in the historical corpus. As was mentioned in the beginning, unlike with serial verbs, the auxiliary morpheme by may be repeated:

Otherwise, the morpheme by may be placed between the minor verb and V2 or else after V2. As for their semantic interpretations, conditionals do not offer any new insights but appear in most typical contexts, such as wishes (18) and entire or abbreviated conditional periods (cf. inoj by vzjal da uexal 'somebody else would [take and] leave').

 

4. Pragmatic effects

Some pragmatic effects that were illustrated in Weiss 2008:495-497 with contemporary data can already be found in the historical corpus. They may be divided into deviations from the lexical characteristics of V2 in the given context and a seemingly irregular structure of the whole discourse fragment. The first type occurs in jocular reports or announcements of somebody's death: 'dying' as the most frequent non-canonical meaning of V2 (see above, section 2.4) lends itself to a pragmatic reinterpretation where the act of dying is conceived of as controllable:

The implicit meaning 'against sb.'s will' present in (19) comes to the fore in explicit markers such as naperekor, na zlo:

In terms of speech act theory, this may be called a mocking threat. Intentional death can also be interpreted ironically:

The adverbial 'as a reward' obviously renders the protagonist's perspective, thus providing a clear illustration of Wilson and Sperber's (2017) echo theory of irony. Similarly, the speaker takes his distance toward another's utterance in (22):

Note that this example is unique in violating the selectional restriction of V2 formulated in section 1: ljubit' denotes a state, not an activity, achievement, or accomplishment. This semantic deviation additionally marks the awkwardness of the echoed attitude.

Another effect arises in replies to questions which express a doubt about the feasibility of the act to be performed:

The simple resumption of the interlocutor's wording contradicts the meaning of unexpectedness. In other words, this example belongs to the second of the aforementioned types (unusual structure of the discourse fragment). What should, however, be explained first is the use of the past tense together with the zero subject: it serves as another marker of a directive act, much as in, for example,poselvon! 'go away!' According to this interpretation, the seeming contradiction turns out to fit neatly in the meaning 'don't hesitate' described for directive imperatives above, cf. (14). Yet, there seems to be an additional critical note which is not inherent to directive imperatives and which may be paraphrased as 'Look, it's so easy!', cf. 'in one word'. Example (23) explicitly states that it is generally known how to act in the way requested:

Weiss 2008:499 discusses similar examples in contemporary texts that likewise point to an ultimate assessment of the type 'no need of further explanation'. The prerequisite of such uses seems to be the mention of V2 in the preceding utterance by the interlocutor. However, for the time being the data is too scarce for any general conclusions. This holds also for the next case in which the speaker himself resumes a past tense form as V2 in a pseudo-coordinative construction:

The sarcasm conveyed first by the initial sentence ('not ashamed to visit') and then by the triple marker of surprise (vdrug, vzjal da i) is boosted by the subsequent act of an alleged willingly remembering.

 

5. Conclusions

This study has examined the emergence and expansion of four competing Russian constructions expressing the unexpectedness of an event, three of them being built according to the pseudo-coordinative model 'vzjat' and V2' and one following the serial model vzjat' V2. The historical data (683 unambiguous examples) taken from the Russian national corpus and beginning in 1827 shows a marked divergence between the serial type and the three coordinative types with the conjunctions da, i, and da i in terms of frequency and further grammaticalisation. Most striking is the weak representation of the serial model: it reaches only 5,7% of the whole bulk⁶. Within the pseudo-coordinative model, the da i-construction constitutes 65% of all instances; it exceeds the da-construction five times in frequency. The corpus contained a considerable amount of ambiguous cases due to the competing literal vs. new reading of vzjat' in the given string and the presence or absence of overt or covert direct objects. The exact number is hard to determine but may well amount to 10% of the whole bulk. Ambiguity is more widespread at the beginning of the period under discussion: the first seven hits of the most productive construction vzjat' da i V2 are all ambiguous.

The three pseudo-coordinative constructions tend to boost the surprise semantics by the so-called imperativus dramaticus, which constitutes almost 10% of all hits, and additional adverbials with the meaning 'suddenly'. The directive function of the imperative 'take and do! = don't hesitate to do' runs counter to the overall surprise semantics but is far less frequent than the other uses of the imperative, especially the imperativus dramaticus. Evidence of ongoing grammaticalisation involves mainly inanimate subjects and V2s denoting uncontrollable events (notably related to the semantics of dying), whereas the imperfective aspect brat' of the first verb and the negation of V2 appear only at the end of the period and are negligible in number. By contrast, the serial construction vzjat' V2 remains behind all this both in terms of frequency and grammaticalisation. In the historical corpus, it most often provides the first two parts of coordinative triplets of the type vzjat' V2 i V3. Additional pragmatic efforts, notably those resulting in irony or sarcasm, originate either from lexical-semantic anomalies or unusual repetitive patterns; they are again limited to the pseudo-coordinative model.

A final, although speculative word concerns the presumable origin of the semantic change undergone by vzjat'. The meaning 'do suddenly', ignored by Heine and Kuteva 2002, is far less widespread than the completely unrelated grammaticalisation resulting in an instrumental meaning ('take the knife and cut' - 'cut with a knife'). Less frequent, but semantically closer to unexpectedness is the new meaning 'begin', attested in many European languages, cf. Coseriu 1966 and Ekberg 1993. Nevertheless, the step from 'begin' to 'do suddenly' still remains to be explained, all the more so as in the history of Russian nothing indicates that vzjat' i V2 or vzjat' alone previously had come to mean 'begin'. The contemporary Russian constructions have striking areal affiliations in Polish (Andrason, this volume and 2018) and even more so in Baltic (Lithuanian and Latvian, Nau et al. 2019). Much as in Russian, in Baltic the pseudo-coordinative model prevails. The Estonian (Tragel 2017) and Finnish equivalents are both geographically and structurally at the periphery of the areal convergence. Whatever the mutual relations between these neighbouring languages may be, it deserves our attention that there exists a close parallel in Brazilian Portuguese (Aikhenvald 2018:125): pegou falou lit. '(he) took (he) spoke' means 'he spoke all of a sudden' along with pegou e falou 'he took and spoke'. All in all, an updated and comprehensive typological overview of the possible grammaticalisation processes of 'take' seems highly desirable.

 

References

Aikhenvald, A. 2006. Serial verb constructions in typological perspective. In A. Aikhenvald and R. M. W. Dixon (eds.) Serial Verb Constructions: A Cross-linguistic Typology. Oxford: Oxford University Press. pp. 1-68. https://doi.org/10.1093/oso/9780198791263.003.0010        [ Links ]

Aikhenvald, A. 2012. The essence of mirativity. Linguistic Typology 16(3): 435-485. https://doi.org/10.1515/lity-2012-0017        [ Links ]

Aikhenvald, A. 2018. Serial Verbs. Oxford: Oxford University Press.         [ Links ]

Andrason, A. 2018. The wziqc gram in Polish. A serial verb construction, or not? STUF -Language Typology and Universals 71(4): 577-629. https://doi:10.1515/stuf-2018-0022.         [ Links ]

Coseriu, E. 1966. Tomo y me voy. Ein Problem vergleichender europäischer Syntax. Vox Romanica 25: 13-55.         [ Links ]

Ekberg, L. 1993. The cognitive basis of the meaning and function of cross-linguistic take and. Belgian Journal of linguistics 8: 21-41. https://doi.org/10.1075/bjl.8.03ekb        [ Links ]

Fortuin, E.L.J. 2000. Polysemy or Monosemy: Interpretation of the Imperative and the Dative-Infinitive Construction in Russian. Amsterdam: Institute for Logic, Language and Computation, Universiteit van Amsterdam.         [ Links ]

Giusti, G., V. Di Caro and D. Ross (eds.). 2022. Pseudo-Coordination and Multiple Agreement Constructions. Amsterdam: John Benjamins.         [ Links ]

Heine, B. and T. Kuteva 2002. World Lexicon of Grammaticalization. Cambridge: Cambridge University Press.         [ Links ]

Isacenko A.V. 1975. Die russische Sprache der Gegenwart. Formenlehre. München: Max Hueber.         [ Links ]

Kor-Saxin, I. 2007. O vozmoznom puti grammatikalizacii russkogo vzjat'. Russian Linguistics 31: 231-248.         [ Links ]

Kuznetsova, J. 2006. The first verb of pseudocoordination as an auxiliary. Bloomington: Slavic Linguistic Society.         [ Links ]

Nau, N., K. Kozhanov, L. Lindström, A. Laugalienè and P. Brudzynski. 2019. Pseudocoordination with "take" in Baltic and its neighbours. Baltic Linguistics 10: 237-306. https://doi:10.32798/bl.365        [ Links ]

Nau, N. this volume. Constructions with 'take' in Latgalian: The limits of diachrony.

Prokopovic, E.N. 1969. Stilistika castej reci. Moskva: Nauka.         [ Links ]

Ross, D. 2016. Between coordination and subordination: typological, structural and diachronic perspectives on pseudocoordination. In F. Pratas, S. Pereira and C. Pinto (eds.) Coordination and Subordination: Form and Meaning - Selected Papers from CSI Lisbon 2014. Newcastle upon Tyne: Cambridge Scholars Publishing. pp. 209-243.         [ Links ]

Smelev, A. 2002. Russkajajazykovajamodel'mira. Moskva: Jazyki slavjanskoj kul'tury.         [ Links ]

Stojnova, N. M. 2007. Konstrukcija vzjat' i sdelat' v russkom jazyke. In F.I.Dudcuk, N.V. Ivlieva and A.V. Podobrjaev (eds.) Struktury i interpretacii: raboty molodyx issledovatelej po teoreticeskoj i prikladnoj lingvistike. Moskva: Izdatel'stvo Moskovskogo universiteta. pp. 144171.         [ Links ]

Tragel, I. 2017. Serial verb constructions in Estonian. In B. Nolan and E. Diedrichsen (eds.) Argument Realisation in Complex Predicates and Complex Events: Verb-Verb Constructions at the Syntax-Semantics Interface. Amsterdam-Philadelphia: John Benjamins. pp. 169-189. https://doi.org/10.1075/slcs.180.06tra        [ Links ]

Usakov, N. 1935. Tolkovyj slovar' russkogo jazyka. T. 1. Moskva: Nauka.         [ Links ]

Weiss, D. 2007. The grammar of surprise: the Russian construction of the type Koska vzjala da umerla 'Suddenly, the cat died'. In T. Reuther, L. Wanner and K. Gerdes (eds.) Proceedings of the 3rd International Conference on Meaning - Text - Theory, Klagenfurt (Wiener Slawistischer Almanach, Linguistische Reihe, Sonderband 69). München & Wien. pp. 427436.

Weiss, D. 2008. Voz'mu i ne budu! Zum Inexspektativ im modernen Russischen. In P. Kosta and D. Weiss (eds.) Slavistische Linguistik 2006-7. München: Otto Sagner. pp. 474-504.         [ Links ]

Weiss, D. 2012. Verb serialization in North East Europe: the case of Russian and its Finno-Ugric neighbours. In B. Hansen, B. Wiemer and B. Wälchli (eds.) Grammatical Replication and Grammatical Borrowing in Language Contact. Berlin & Boston: Walter de Gruyter Mouton. pp. 619-655. https://doi.org/10.1515/9783110271973.611        [ Links ]

Weiss, D. 2013. Dvojnye glagoly v russkoj razgovornoj reci v zerkale Nacional'nogo korpusa: formy imperativa mn. c. In: E. Vel'mezova (ed.), Schweizerische Beiträge zum XV. Internat. Slavistenkongress in Minsk. Bern-Frankfurt a.M.: Peter Lang. pp. 319-341.         [ Links ]

Weiss, D. forthcoming. Imperatives as a source for the emergence of new particles in Russian. In E. Graf and U. Schweier (eds.) Pragmaticalization: Language Change between Text and Grammar. Berlin: De Gruyter Mouton.

Wilson, D. and D. Sperber. 1992. On verbal irony. Lingua 87(1/2): 53-76. https://doi.org/10.1016/0024-3841(92)90025-e        [ Links ]

Wilson, D. 2017. Irony, hyperbole and banter. In J. Blochowiak, C. Grisot, S. Durriemann and C. Laenzlinger (eds) Formal Models in the Study of Language: Applications in Interdisciplinary Contexts. Cham: Springer. pp. 201-220. https://doi.org/10.1007/978-3-319-48832-511        [ Links ]

 

 

1 The Irish English parallel construction of the type 'He took and died' will here be exploited in the translations of some Russian examples for the sake of more transparency.
2 For a cross-linguistic investigation of such constructions see Ross 2016 and Giusti et al. 2022.
3 The reason seems to be that vnezapno (unlike neozidanno) does not open a valence for an observer, as is pointed out in Smelev 2002:149-164.
4 For pragmatic effects based on playing with one's own death, see section 4.
5 Besides "true" imperatives, the directive function also characterises two uses of the preterit vzjal, see below, examples 23 and 24.
6 As evidence from Latgalian (Nau, this volume) shows, the divergence between the SVC model and the pseudo-coordinative model may also be due to a difference of registers: SVC dominates in older oral folklore texts, whereas it is completely absent in modern Latgalian and also in the first written Latgalian texts.