Zurich Open Repository and Archive University of Zurich Main Library Winterthurerstrasse 190 CH-8057 Zurich www.zora.uzh.ch

Year: 2012

Verb serialization in North East Europe: the case of Russian and its Finno-Ugric neighbours

Daniel Weiss

Posted at the Zurich Open Repository and Archive, University of Zurich

http://dx.doi.org/10.5167/uzh-72092 Originally published at: Weiss, Daniel (2012). Verb serialization in North East Europe: the case of Russian and its Finno-Ugric neighbours. In: Wiemer, Björn; Wälchli, Bernhard; Hansen, Björn. Grammatical replication and grammatical borrowing in language contact. Berlin/Boston, 611-646. ISBN 978-3-11-027009-9.

Verb serialization in northeast Europe: the case of Russian and its Finno-Ugric neighbours Daniel Weiss

The topic of the present paper needs a special justification since the existence of the Serial Verb Construction (SVC) in Finno-Ugric has simply escaped the attention of SVC typologists up to now, while the SVC status of the Russian double verb construction of the type sidit-pla3et ‘s/he sits-is crying’, sjadem-podumaem ‘We’ll sit down-think’ or pridi-poznakom’sja ‘come-introduce yourself’ was even explicitly rejected by Aikhenvald 2006 (see section 1.1 below). This unfortunate omission of the North East European area is clearly due the conviction that there “appear to be none (SVCs) in Europe or north or central Asia” (Dixon 2006: 338). Therefore, the whole first section of my paper will be devoted to the illustration of the claim that the constructions to be examined in Russian and its Finno-Ugric neighboursi illustrate rather nicely all the major properties that have been described as typical of SVCs. If the “scalar, or continuum approach to SVC” proposed by Aikhenvald (2006: 3) is to be taken seriously, the said languages are sure to occupy a position closer to prototypical SVCs on this scale, even if their frequency is very low (in this connection it may be mentioned that according to Dixon (2006: 338) in Khwe SVCs constitute less than 1 per cent of all clauses). Second, it will be argued that Russian SVCs originated from contact with Finno-Ugric languages. This is plausible since, as will be shown in section 2, they share a large number of common properties with today’s Komi counterparts, the latter representing on the whole a more advanced stage of serialization. This convergence offers, however, only indirect evidence for my claim, since the most important Finno-Ugric languages located in Central Russia that may have provided the source of the Russian innovation have been extinct since the late middle ages; all our parallels are

based on data from still existing Permian and Volga Finnic languages located in the North and East of European Russia, where the influx of Russian immigrants started later and did not reach the same levels as in the Podmoskov’e. All this implies that the very assumption that the RussianFinno-Ugric parallels are contact-induced requires a separate justification, all the more so since in recent time, the direction was reversed, i.e. Russian exerted a strong influence on the structure and the lexical stock of the surviving Finno-Ugric languages. To make things even more complicated, we are unable to determine the exact age of Russian SVC: since in earlier times written sources represent almost exclusively formal registers, a colloquial construction such as SVC may well have existed for centuries without being attested in written documents. The problem is aggravated by the unreliable punctuation of that time which does not allow us to unequivocally distinguish SVC from asyndetic verb coordination (see, section 1.4 below): thus, the lack of a comma between two grammatically identical verb forms in a manuscript is not conclusive; moreover, sometimes the comma was added later by the editor. All we can state for the time being is that the first examples are to be found in the epistolary genre and in the so-called merchant stories of the 17th century, with further evidence being provided by foreigners’ notes and grammatical descriptions. On the other hand, to my knowledge the famous Novgorod birch-bark letters (11-15th centuries) do not contain one single instance of SVC; this does not, however, come as a surprise since, in the Fennic (Balto-Finnic) languages spoken in the Novgorod-Pskov, this type of construction is no longer productive, either.

1. Russian double verbs

1.1. A first approximation The present study continues a series of previous papers devoted to the analysis of double verbs (DV) in Russian: in Weiss (1993) the main accent was laid on the description of the different subtypes and their conditions of use in modern colloquial speech and the language of traditional folklore; in Weiss (2000) I tackled the question of how this construction should be modelled in the framework of a dependency syntax; Weiss (2003) examined new corpus data from the language of the contemporary press and fiction and established a parallel to similar constructions in FinnoUgric languages; and Weiss (2008) focuses on the interpretation of DVs with perfective aspect. The present study will add but a few new observations on these topics; it mainly aims at a more systematic overview of what are already well-established facts by highlighting those features that invite a comparison with the general picture of verb serialization. Early modern Russian has developed a type of verbal juxtaposition that represents much more than the germs of verb serialization: it meets not only the minimal definition of SVC established by Bisang (1995: 137) but fits fairly well into the overall picture of SVC drawn by Aikhenvald (2006), notwithstanding the author’s own objections.ii In addition, it has to satisfy the following language-specific requirements formulated in Weiss (2000): a) the morphological outfit of both verbs should be identical; hence, person, number, gender, mode, voice and representation (the opposition finite : nonfinite) of both verbs have to coincide. As for tense and aspect, the exceptions to the identity requirement are predictable and well-defined, cf. Weiss (2008); b) both verbs share the same subject and optionally a second or third argument or an adjunct, cf. Pojdem k sponsoru sxodim lit. ‘Let’s go to the sponsor drop by = Let’s go and visit the sponsor’;

c) auxiliary morphemes such as the conditional marker by and the auxiliary of ipf. future forms (budu, bude9’, etc.) cannot be repeated, cf. Po9el by proveril ‘You should go [and] find out’; d) the two verbs may be reversed, cf. Sadis’ rasskázyvaj > Rasskázyvaj sadis’ ‘tell [us the story] sit down!’ The main stress shifts together with the component it falls on, cf. XY > YX. From a typological point of view, the last-mentioned feature is most striking: in the overwhelming majority of languages, the verb order within a SVC cannot be reversed; even in such exceptional cases as Mandarin, inversion is only possible with symmetric predicates, hence without infringing on the principle of iconic ordering (Matthews 2006: 78). Russian, however, allows anti-iconic sequences as in the example quoted under d) or in ty by spala legla ‘You should sleep-lie down’, where the natural chronology of the sub-events involved does not coincide with syntactic order. Contrary to what is stated in Aikhenvald (2006: 47), Russian DVs are not “restricted in their mood, polarity, tense, and aspect choices”, but may take all values of the grammatical categories listed under a) above. The same holds for the aspectual meanings traditionally distinguished in Russian grammars: with the doubtful exception of the so-called general factive meaning of the imperfective aspect, all possible semantic values of both aspects are represented in my corpus. It should be added, though, that some types occur only with general or habitual time reference. Moreover, DVs may take inanimate subjects and may occur even in ‘impersonal’ sentences without overt subjects. As for the double (‘concordant’) inflection of both verbs, it is not uncommon in SVC languages; for morphological analysis this means that we are dealing with two separate words forming a syntactic entity, which is, at any rate, indicated by the term ‘construction’. As the term ‘juxtaposition’ suggests, the two verbs usually are immediately conjoined without a prosodic break. If the latter condition is not satisfied, we obtain a biclausal construction that is not subject to any of the restrictions a)-d) formulated above, viz. asyndetic linking. Nominal or

adverbial elements may, however, be inserted into the DV construction (see above, Pojdem k sponsoru sxodim), but the more elements that divide the two verbs, the more the unity of the whole is endangered. The overwhelming majority of examples in my corpus have contiguous structure, i.e. no other constituents go between the verbs; exceptions occur only within prototypical uses (see below). As a consequence of the syntactic merger of the two verbs, a complement of the second verb may be located before the first verb, cf. Ona na nas stoit smotrit lit. ‘She at us stands-looks = She is standing there and looking at us’. The term ‘double verb’ is somewhat misleading since we also find verb ‘triplets’ and even ‘quadruplets’. Such examples mainly belong to two types: either their first component is semantically redundant (most often the verb in question is ‘go’ and / or the form is an imperative, see section 1.6), or they are the result of spontaneous, but ephemeral creation. The impact of the latter factor should not be underestimated: such data constitute a considerable part of my own corpus, which contains several hundred different Russian DVs (406 were already cited in my articles),iii and Internet blogs, discussion forums, etc. simply abound in new material; one single story of four pages may contain 45 examples (cf. footnote 15 below). On the whole, DVs are, however, not frequent (the percentage of textual clauses that include an SVC is certainly less than 1 per cent, which, however, is also characteristic of other SVC languages, cf. Dixon 2006: 338), and their colloquial flavour prevents their use in more formal text genres, such as scientific or bureaucratic discourse, press agency information and the like. The distribution among the remaining genres (colloquial speech, Internet communication, fiction, press reports and comments, folklore, etc.) depends partly on the distinction between prototypical and other uses to be discussed in the next section. The majority of DVs are based on a common semantic denominator, viz. the logical conjunction. Therefore the conclusion seems tempting that DVs represent a specific kind of coordinative structure where the syndetic link (the conjunction ‘and’) or the asyndetic prosodic break is omitted. There are, however, several objections against this reductionist analysis; above all, not all DVs can be replaced by ‘and’-linking (see section 1.4).

Therefore, corresponding ‘and’-links are treated as mere paraphrases of DVs in this study. The main communicative feature that distinguishes DVs from coordinative clause linking is an iconic effect: the two events denoted undergo a specific kind of merger or even a morphing process, which eventually leads to the rise of one single complex event; this holds true at least for prototypical DVs, see below. 1.2. Prosody and meaning: prototypes and twins The features listed so far are shared by all occurrences of the DV construction. The picture changes, however, when we turn to prosody and meaning: here we must, as was pointed out in Weiss (2000), distinguish prototypical and non-prototypical uses. The former constitute a prosodic unity, i.e. they are not only pronounced without a pause between the two verbal components, but also with one single main stress (usually on the second verb, in the case of inversion on the first one); the latter are characterized by two equal peaks (prosodic twins, so to speak). In the same vein, the semantic prototype of a DV denotes one single event (process, activity, action) with the second verb modifying the first one, whereas nonprototypical DVs denote two separate, though closely interrelated events (semantic twins). In other words, the above mentioned iconic fusing effect is stronger with the prototype than the twin structures. Note that the prosodic and the semantic distinction may, but need not necessarily, coincide: in particular, synonymous pairs represent a semantic but not a prosodic unity. In terms of descriptive analysis, a DV may constitute one single phonological and / or semantic word, but never one single morphological word. From a typological point of view, the conjunction of prosodic and semantic prototype of DVs meets the general definition of SVC, formulated by Aikhenvald (2006:1), cf.: “Serial verb constructions describe what is conceptualized as a single event. They are monoclausal; their intonational properties are the same as those of a monoverbal clause…”. It should be emphasized that the prosodic merger does not lead to phonetic attrition of either of the verbal components of a Russian DV.

Similarly, the semantic fusion does not produce notable effects of semantic bleaching; when desemantization occurs, it affects the first component exclusively. One exception is constituted by certain imperatives as first members conveying additional pragmatic meanings (section 1.6.1), but this is not a property of DV constructions, but reflects a general tendency typical of the Russian imperative. Prototypical uses prevail in colloquial Russian, while folklore and dialects admit both types. The same holds for contemporary poetic and journalistic language and especially for Internet communication; in all these varieties, however, a more creative, playful approach makes itself felt that produces a host of freshly coined combinations, as will be shown below (section 1.7). On the other hand, colloquial and folkloristic uses tend to be stereotyped, if not phraseologized (see section 1.5.). For instance, the first component usually is a verb of position (sit, stand, lie, hang), motion (go, walk, run, etc.) or a combination of both (sit down, lie down, but rarely: stand up)iv , and less often a verb of intentional perception (watch, listen), whereas the second component may be chosen more or less freely if it is compatible with one of the above-mentioned verbs, cf. sidit-mol"it ‘s/he sits [and] is silent’, begi i9"i ‘run [and] look for it’, xodit #aluetsja ‘s/he goes around [and] complains’, pri9li poznakomilis’ ‘they came [and] introduced themselves’, slu9aju o#idaju ‘I am listening [and] waiting’, etc. Because of these semantic restrictions of the first verb, such DVs belong to the asymmetrical type of serialization, singled out in Aikhenvald (2006). Besides this, there are two important subtypes that need not satisfy the said scale of preferences. The first consists of verb pairs whose second component expresses the lack of any obstacle or interference that could affect the process described by the first component. It may be illustrated by examples such as bit’ ne #alet’ lit. ‘beat up [and] not have pity’, radovat’sja ne naradovat’sja ‘be glad and not have enough joy’, #it’ ne ssorit’sja ‘live [and] not quarrel’, etc. The crucial role of negation in such pairs is obvious (see next section). Note that the DV construction is not the only technique that can express the concept of an easy, smooth and untroubled process in colloquial Russian: in particular, the same idea may be rendered by the particle sebe as in #ivet / spit / sidit / rabotaet / xodit

sebe ‘lives / sleeps / sits / works / walks around without trouble’. Both devices may combine, e.g. #ivu sebe ne tu#u ‘I live [and] do not mourn’. The temporal equivalent of the lacking obstacle is uninterrupted duration; its markers are the first verbs sidet’, stojat’ and xodit’ (section 1.6), moreover, the reduplication of one verb as in idem-idem ‘we go and go’ iconically serves the same purpose. The idea of an undisturbed process is further developed by the second subtype, in which the second component serves as an intensifier of the first one, cf. kri"at’-razdirat’sja ‘shout very loudly’. Such pairs are often phraseologically bound (section 1.5) and / or based on synonymous variation (section 1. 7) In non-prototypical DVs neither of the components (the semantic ‘twins’) is subject to specific lexical restrictions; according to the terminology proposed by Aikhenvald (2006), these are symmetrical SVCs. What is salient here are certain fundamental semantic relations between them, viz. cohyponymy, antonymy, conversion or a semantic script (see section 1.7). In a certain way, even non-prototypical DVs may be said to constitute a kind of (higher order) semantic unity, which becomes particularly palpable in the case of a common script or frame. On the other hand, such DVs maintain their semantically composite structure, i.e. they do not manifest the same tendency to develop new lexicalized meanings, as is the case in Finno-Ugric (see section 2 below). As for time reference, the semantic prototype may be said to refer to one single event in time (the time intervals denoted by both verbs are identical). There is, however, one exception not to be neglected: with verbs of motion as first members, the DV may require a sequential reading with the first verb referring to the initial stage or sub-event of the process denoted by the second verb. Most of these cases are trivial in that the first verb is redundant and could as well be omitted, cf. pojdu proverju ‘I’ll go [and] verify’, sjadem podumaem ‘we’ll sit down and think of it’, ty by legla spala ‘You should lie down [and] sleep’. Non-prototypical uses refer to two still distinguishable, though not entirely separate events; yet, they tend to cover the same time span, or else time reference becomes irrelevant, as is often the case with habitual or generic readings. The progressive

meaning of the imperfective aspect is found almost exclusively with the semantic prototype.

1.3. Negation The position of negation may vary: it can be placed before either the second or (due to inversion) the first verb, cf. e9’ ne bespokojsja! ‘Eat, don’t worry!’ and ne polenis’ pro"itaj ‘don’t be lazy, read it through’ or even before both of them, cf. ne peli ne pljasali ‘they neither sang nor danced’, ne projti ne proexat’ ‘one can neither pass by walking nor by driving’ (note that the normal double negation ‘neither…nor’ would be ni projti, ni proexat’, which is also possible). In the latter case the second occurrence can be omitted as well, cf. ne peli pljasali.v Thus, location before the first component has two sources: either it may be interpreted as the result of permutation, or it has wide scope and comprises both verbs. On the other hand, initial position never marks negation of the first verb alone. In view of the definition of the semantic prototype (only one situation is referred to), one would not expect there to be any partial negation (with narrow scope). The same prediction is made by Aikhenvald (2006: 1): according to her, SVC should have the same polarity value. This assumption, however, turns out to be incorrect: my corpus contains numerous examples where only the second verb is negated.vi Obviously, this is due to the above-mentioned semantic constant: what is negated is the presence of an obstacle that could somehow affect (slow down, inhibit, impair or even prevent) the process denoted by the first verb. Besides the imperatives ne polenis’ ‘don’t be lazy’ and ne bespokojsja ‘don’t worry’ quoted above, this pattern is represented by such collocations as idet ne spe9it ‘is walking without hurry’, slu9aet ne perebivaet ‘is listening without interrupting’, le#it ne 9evelitsja / ne podnimetsja, ‘lies without moving / will not stand up’, #ivut ne ssorjatsja ‘they live without quarelling’, #ivet ne tu#it ‘lives without mourning’, vypil ne pomor9"ilsja ‘he drank [it] out without frowning’, bili ne #aleli ‘they hit mercilessly’, l’et ne perestaet

‘it’s raining incessantly’, radujus’ ne naradujus’ ‘I am glad without end’, etc. This peculiarity helps to explain why partial negation of the first component is not attested in non-inverted double verbs: the lack of the obstacle is always expressed by the second component.

1.4. The closest paraphrases Russian offers several finite and non-finite constructions that can compete with verb serialization. Above all, the following three types of paraphrases should be distinguished: instead of choosing a DV, the second component can be linked with the first by a) syndetic coordination, b) adjoined as a converb (adverbial participle), or c) transformed into an infinitive. These are the different options: sidit mol"it 1 ‘sits is silent’ sidit mol"it 1 ‘sits is silent’ pridi poznakom’sja! 1 ‘come introduce yourself’

sidit i mol"it syndetic coordination ‘sits and is silent’ sidit mol"a converb ‘sits being silent’ pridi poznakomit’sja! infinitive ‘come to introduce yourself’

None of these variants is a full equivalent of serialization, nor do their conditions of use coincide. To begin with, syndetic linking is by far the closest variant, and one might even conjecture that the DV construction goes back to a simple omission of the conjunction ‘and’. This is, however, not very probable: as will be shown in section 2, the roots of Russian verbal serialization should not be traced backed to Russian itself, but to contact with Finno-Ugric languages. Besides this, contrary to Aikhenvald (2006: 46), there is a whole range of DVs that block the insertion of ‘and’: first, set phrases of the intensifying type kri"at’-nadryvat’sja to be discussed in the next section; second, half-synonymous pairs describing one single event of the type poxlopal-pogladil moju ruku (see section 1.7); third, DVs with perfective aspect that do not denote a sequence of actions,

cf. Weiss (2008); and, fourth, DVs in the imperative whose first component is pragmatically specialized and functions much like a particle, cf. davaj pej ‘go on [lit. give] drink’, smotri ne prostudis’ ‘watch [out] don’t catch cold’, podi pover’ ‘go believe’, etc. On the whole, one would expect that prototypical DVs in general constitute an obstacle to syndetic paraphrasing since they denote only one single event, while coordination typically combines two more or less autonomous events; in view of the numerous “exceptions” where ‘and’ combines not two separate propositions, but rather two different aspects of the same situation (Weiss 2008), one has, however, to abandon such a simplistic hypothesis. The converb paraphrase is neutral as to the opposition , but changes the equal status of both components into an internal hierarchy: the aspect rendered by the converb is either backgrounded, or else the converse phrase is prosodically focused. Besides the complications arising with such natural pairs as ‘eat and drink’ (cf. *Pil edja ‘he drank [while] eating’), this triggers a host of semantic complications, e.g. with regard to modality and time reference (see Weiss 2000: 374). For instance, the time span denoted by the converb typically includes that of the main verb, whereas in a prototypical DV both intervals coincide. Besides this, there are several obstacles hampering the transformation of a DV into a converbial construction; among them one notes morphology (whole classes of verbs do not form converbs at all) and different stylistic values (DVs belong to colloquial language, converbs to the formal register, therefore they usually do not occur in the same text).vii The infinitive differs from DV with respect to modality, since the proposition denoted is not conceived of as real. Thus, in the example quoted above, DV and infinitival variants are synonymous since the action denoted by the second component is located in the future. However, if we transfer this example into the past, the DV pri9el poznakomilsja ‘came introduced himself’ refers to two events that really took place, whereas in pri9el poznakomit’sja ‘came to introduce himself’ the truth value of the proposition poznakomilsja ‘he introduced himself’ remains open. Moreover, imperatives with pragmatically specialized first components (see section 1.6.1) do not allow for the infinitive transformation, cf. Podi

pover’ / *poverit ‘Go believe / *to believe’, Smotri ne proboltajsja / *ne proboltat’sja ‘Watch [out] don’t let out [the secret] / *not to let out’. To sum up, syndetic coordination shares the equal status of both conjuncts with the DV construction, whereas converb and infinitive constructions are based on hierarchical relations. The latter two types of constructions share with the DV construction the requirement of a common subject for both verbal components, whereas coordination admits different subjects.

1.5. Phraseological boundness DVs with an intensifying second component tend to become phraseologized. Such is the case of kri"at’-nadryvat’sja ‘scream with all one’s strength’, xoxotat’-zalivat’sja ‘laugh uproariously’, l’et-ne perestaet ‘it rains incessantly’: on the basis of the second verb we are able to predict the first one with high probability. For instance, the second verb nadryvat’sja “forecasts” a first verb denoting a sound activitiy, cf. Ona oret-nadryvaetsja ‘she is shouting with all her strength’, mladenec kri"alnadryvalsja ‘the child was screaming very loudly’, but also sobaka laetnadryvaetsja ‘the dog is barking very loudely’, telefon zvonil-nadryvalsja ‘the phone was ringing very loudely’ etc. It should be noted that nadryvat’sja alone can already mean ‘to shout with all one’s strength’; thus, in the case of a human agent the second verb of the DV becomes redundant, whereas with animals and inanimate sources such as phones we observe a metaphorical transfer. A similar situation obtains with xoxotat’zalivat’sja, since zalivat’sja alone may also mean ‘laugh loudly’. In this way, the examples mentioned belong to the same semantic subtype as the series with identical roots of the kind #dat’-o#idat’ or zvonit-nazvanivat’, which also serve the purpose of intensifying (Weiss 2003: 42); moreover, as we will see in the next section, the same meaning is also expressed by a host of “free” DV combinations. Note that the concept of intensity reinforces the semantic invariant described in the previous section, where the DV construction denoted an uninterrupted and unhindered process.

This semantic type also shows a certain tendency towards frozenness: for instance, one finds #it’ ne ssorit’sja ‘live without quarrelling’, but rather not ?brigada rabotaet ne ssoritsja ‘the brigade works without quarrelling’, ?deti igrajut ne ssorjatsja ‘the children are playing without quarrelling’. The frozenness of the examples from this series is best illustrated by the following restrictions: (i) they do not allow for inversion, cf. *nadryvaetsja-oret, and (ii) they cannot be transformed into a coordinative string with syndetic linking, cf. *zvonit i nadryvaetsja, nor into a converb, cf. *oret nadryvajas’. In both respects, they differ from regular (“free”) DV constructions. The following examples represent the highest degree of lexicalization in that they do not allow for any grammatical variation besides that triggered by subject agreement, i.e. number (last four examples) and gender in the preterite (last two examples): po9lo-poexalo lit. ‘went off-drove off = And this was it!’, podaj-prinesi lit. ‘hand over-bring = a servant for the most inferior services’, (ego) ne razberi-pojme9’ ‘you won’t understand (him)’, #il-byl lit. ‘lived-was’ = there once was’, Po#ivem-uvidim lit. ‘We’ll live-see = We’ll wait and see’,viii Uvidim-posmotrim lit. ‘We’ll see-watch = We’ll wait and see’, Kak #ivete-mo#ete? lit. ‘How do you live-be able = How is life?’ Besides their fixed morphological arrangement, these set phrases share the above-mentioned restrictions with the examples discussed earlier, viz. they lend themselves neither to inversion nor to syndetic linking, cf. *byl-#il ‘was-lived’, *po9lo da poexalo‘went off and drove off’, etc. On the other hand, their frozen character often provokes playful modifications. For example, Internet bloggers and forum discussants may merge two double verbs into triples, cf. Po#ivem-uvidim, posmotrim ‘we’ll live-see, watch’, or enlarge a given DV by adding “free” pairs, cf. uvidim posmotrim, glazkami pogljadim ‘we‘ll see-watch, we’ll watch with our little eyes’ or Po#ivu uvi#u, do#ivu uznaju, vy#ivu u"tu.

1.6. Grammaticalization phenomena As is predicted by Aikhenvald (2006: 30), grammaticalization phenomena exclusively affect the asymmetrical subtype of SVC, i.e. prototypical DVs with restricted first verbs in my terminology. However, first components of Russian double verbs do not change into real grammatical markers such as TAM morphemes or new prepositions. The possible candidates for such a transformation may be counted on one hand: the verbs sidet’ ‘sit’, xodit’ ‘go around’, pojti / idti ‘go’, vzjat’ ‘take’. The last case is, however, not primarily related to the DV construction, but to coordination: strings of the type ‘he took and died’ in the meaning of ‘all of a sudden, he up and died’ illustrate a widespread semantic evolution, which according to Coseriu (1966) is attested in not less than 24 European languages, covering whole compact areas in north, southeast and especially east Europe.ix As for Russian, the construction vzjal i umer / vzjal da umer ‘took and died’ has been focused upon in several studies, see Fortuin (2000: 149-161), Kor Chahine (2007) and Weiss (2008). Although it allows for a serialized variant of the type vzjal umer instead of syndetic linking vzjal i umer, the former is subject to far more restrictions than the latter (Weiss 2007). Sidet’ ‘sit’ as first verb of a DV emphasizes the uninterrupted performance of the activity or the ongoing duration of the process in question and occurs most often with progressive meaning. Such uses attest to a slight semantic bleaching since the sitting position itself is not conceived of as uninterrupted, but the selectional restrictions imposed on the full verb remain valid: only animate subjects are acceptable, and the action denoted by the second verb should in principle be compatible with the sitting position of the agent. Xodit’ ‘to walk’ is the dynamic counterpart of sidet’ ‘to sit’: the activity denoted by the second verb is performed at different places, but not necessarily by going around, cf. xodit #aluetsja ‘keeps complaining’. Thus, xodit’ takes the adverbial meaning ‘everywhere’, and the whole DV has habitual or iterative meaning. In this way, both sidet’ and xodit’ convey the idea of uninterrupted duration; we

are back to the semantic invariant evoked in section 1.3, viz. the undisturbed process. Contrary to this, the verbs idti and pojti as first components of a DV are associated with the idea of beginning, which does not come as a surprise in view of the fact that in combination with event nouns as in Po9la novaja #izn’ ‘a new life began’ and infinitives as in Kak po9el rugat’sja! ‘And then did he ever start cursing!’ the verb pojti also denotes a starting point. This does not mean, however, that idti or pojti function as real ingressive markers in the DV construction, because, as is shown in Weiss (2007b), they are subject to too many aspectual and selectional restrictions. Instead of this, they tend to become simply redundant, since the activity in question implies a physical movement anyway, cf. Po9la vymylas’ ‘She wentwashed herself’ or po9li sobralis’ ‘they went-gathered’; this type of redundancy most often occurs in the imperative and hortative where it gives rise to numerous triplets as in idi begaj (zanimaj) ‘go run (borrow some money)’, pojdem sxodim posmotrim ‘Let’s go-go there-have a look’. As is shown by the latter examples, idti or pojti may be followed by another verb of motion. On the whole, our Russian list does not look very impressive, since basic verbs of motion such as ‘go’ and of posture such as ‘sit’, ‘stand’ or even ‘lie’ are not only the most liable to take grammatical meanings in SVC (Aikhenvald 2006: 47 f.), but belong generally to the most often grammaticalized verbs in the languages of the world (cf. Majsak 2005). Therefore, we do not even have to search for parallels in such exotic languages as creoles. For instance, the verb ‘sit’, as first component of a coordinative construction, has become a durative marker in a whole range of European languages (Hungarian, Bulgarian, Swedish, and Dutch, cf. Kuteva 1999, Ebert 2000, Csató 2001, and Raxilina 2003). Unlike these languages, Russian admits the verb ‘sit’ not only in the coordinative, but also in the serialized model, but the degree of grammaticalization of both variants is lower than in the other languages (Weiss 2003: 50 f.). As for the verb ‘go’ as component of coordinative structures, it is subject to multifarious types of semantic bleaching, including the surprise meaning noted for Russian ‘take’ above, cf. English And then he went and died or

German Und da geht er doch hin und stirbt. As we have seen, in the Russian serialized examples with ‘go’ we find no similar semantic bleaching, but rather redundancy. Against the background of data from other languages, the almost complete lack of grammaticalization in Russian serialization is, of course, striking. It finds its natural explanation in the fact that unlike in many SVCs (e.g. creoles), serialization is not a construction Russian simply cannot do without, but rather a kind of syntactical toy designed for special effects such as increasing iconicity, morphing meanings (see below) and the like. Why should such a marginal device come to mark such grammatically basic relations as case, aspect, modality and the like? Russian has plenty of morphological and lexical means that are more appropriate for this purpose.

1.6.1. Pragmatic effects related to the imperative There is, however, one instance of the DV construction that requires special treatment: with some imperatives, we observe a peculiar kind of desemanticization whose function is pragmatic rather than grammatical. In particular, this holds true for po(j)di ‘go!’ In its literal reading, it simply serves as an additional marker of the directive illocution expressed by the imperative of the second verb, which is, by the way, a very widespread function among the languages of the world (Majsak 2005: 178). But when combined with a second imperative denoting a mental activity, it can no longer be interpreted literally. Such is the case of podi pover’ (takomu alibi)! lit. ‘go believe such an alibi’, podite razberites’ ‘go find out’ and the like. Their derived meaning (obviously based on a frozen Gricean implicature) is a negative statement, cf. podi pover’ = ‘one cannot believe’ or podi razberis’ = ‘you can’t find out’. When the second verb denotes a physical activity that may be preceded by moving, the DV construction may become ambiguous between literal and rhetorical meaning, cf. Podi pojmaj vora! ‘Go catch the thief!’ or ‘You won’t be able to catch the thief.’ Interestingly enough, the grammatically hybrid set phrase (ego) ne razberi-

pojme9’ quoted in the previous section contains both the derived (ne pojme9’ ‘you won’t understand’) and the literal meaning (razberi ‘understand!’). Inversion remains still possible, cf. podi razberi = razberi podi ‘understand go’, whereas syndetic linking with ‘and’ works freely in the strings podi i razberi, podi da razberi ‘go and understand’, but rarely occurs in other combinations, cf. podi da pojmi-ka ‘go and catch’, ?podi i pover’! ‘go and believe!’ In this way, podi(te) has become a marker of rhetorical requests, whose derived (ultimate) meaning is the corresponding negative statement, much as the derived meaning of a rhetorical question is the corresponding negative statement. In this use, it functions as a synonym of the imperative poprobuj(te) ‘try!’, being also used as a marker of rhetoricity, cf. Takogo poprobuj ugrob’ lit. ‘Try kill such a guy!’ > ‘You can’t kill such a guy’. Not surprisingly, both imperatives may combine in triplets of the type Podi poprobuj pojmi ix ‘You can’t understand them’. Podi and poprobuj differ, however, with respect to their syntactical distribution: the former combines only with a second imperative in the DV construction, while the latter admits also infinitives, cf. podi *ponjat’ / *razobrat’ ‘go *to understand’, but: poprobuj poverit’ / ponjat’ / razobrat’ ‘try to belive / understand’. In other words, the desemanticization of podi is limited to serialization and (to a much lesser extent) coordination. All this said, it should be emphasized that the pragmatic specialization discussed is not due to any structural properties of the DV construction; it rather reflects an overall tendency to develop imperatives of the type go! and come! into particles with specific pragmatic load, cf. for example German (ach) geh! or Polish Id$ ty! which function as dissent markers.x In Russian, the rhetorical interpretation of the imperative holds as well for replies such as On skoro vernetsja. – Do#idajsja vernetsja! ‘He’ll come soon. – (lit.) Just wait he’ll come! = he won’t come at all!’, where there is no trace of any DV. Moreover, the tendency to pragmatically mark off particular imperatives also affects other verbs, cf. smotri ‘watch out!’, davaj and valjaj as hortative or suggestion markers. Although all these forms can appear as part of a DV of the type davaj pej! ‘have a drink!’, they are much more frequent in other syntactic contexts; moreover, they all

distinguish number (i.e. besides davaj (sg.) we have davajte (pl.)) and therefore cannot be considered real particles.

1.7. Non-prototypical DVs in contemporary press and fiction Traditional Russian folklore and dialects manifest a marked tendency to exploit both prosodic and semantic non-prototypical DVs combining accentual twins with such fundamental semantic relations as synonymy, antonymy, cohyponymy, etc. In the contemporary press (e.g. Literaturnaja Gazeta, Argumenty i fakty) as well as in modern fiction (e.g. with such diverse writers as Sol3enicyn, Ale9kovskij, Sorokin, Petru9evskaja, Makanin or Ratu9inskaja) and in the Internet (blogs, forums), we witness a remarkable revival of this tendency, which has lead to the rise of numerous spontaneous, fresh creations. The lack of lexical restrictions characteristic of these “symmetrical” uses favours the rise of new formations. Of course, mere quotation of an already existing DV formula also remains possible; in such cases, the tension between the popular idiom and the modern content will evoke a humorous or ironic effect, cf.: (1)

...blagopolu"no ["inovni"estvo] pere#ilo v takom successfully survive-PST-3PL in such-LOC.SG ka"estve sovetskie gody i vot #ivetquality-LOC.SG Sovet-ACC.PL year-ACC.PL and live-PRS.3SG zdravstvuet nyne. -is healthy-PRS.3SG now ‘[the officials] successfully survived the Soviet period and are now healthy and sprightly.’ (LG 42/02)

But more often than not, the lexical filling of the pairs is far beyond the scope of traditional folklore or dialect speech. To begin with, let me illustrate this by pairs whose first and second components are more or less synonymous. They constitute a hybrid case in that prosodically they belong to the non-prototypical subtype (both verbs have equal stress), whereas

semantically they behave like prototypical DVs (they denote one single event). Besides traditional examples such as dumat’-gadat’ ‘think-speculate’, grustit’-pe"alit’sja ‘be sad-mourn’, rydat‘-gorevat’‘weep-be sad’, spat’-po"ivat’ ‘sleep-rest’, xodit’-brodit’ ‘walk-stroll around’, ne znat’-ne vedat’ ‘not know-not be acquainted with’ we find the following pairs (all cited in the infinitive; the aspect of these pairs varies): prosit’-umoljat’ ‘beg-beseech’, perexitrit’-pereigrat’ ‘deceive-cheat’, propadat’-pogibat’ ‘be lost-die’, oskorbil-oklevetal ‘offend-slander’, odolet’-peresilit’ ‘overcome-defeat’, iskat'-vyiskivat' ‘look for-search’, ezdit’-katat’sja ‘drive-drive around (e.g., for fun)’, zagolosit’zaskandirovat’ ‘strike up-begin to scan’, nabrosat’-narisovat’ ‘sketchdraw’, poguljat’-poveselit’sja ‘have a good time-have a party’, ruku po#at’-potiskat’ ‘shake-press hands’, etc. This type is especially productive with perfective verb pairs, cf. Weiss (2008). Often the second verb serves as an intensifier of the first one. This seems to be a widespread feature: it is also attested in French-based Creoles and may even be characteristic of serialization in general (cf. Bisang 1995: 148 on “narrow serialization” in Khmer). In other cases, the verbal pair is built up with a kind of cohyponym that attests rather the author’s search for the appropriate formulation; such is the case in probe#al-prokatilsja po trave poryv vetra ‘a gust of wind randrove over the grass’ or poxlopal-pogladil moju ruku ‘he tapped on my hand-gave it a stroke’. In the latter example, the movement in question might be something in between these two activities. Consequently, the DV construction functions as a kind of semantic merger or a morphing device, and it still represents the semantic prototype by referring to one single event. It should be noted that Russian has two specialized conjunctions at its disposal that convey the same meaning: instead of the bare logical connector ‘or’, the speaker may choose ne to (p), ne to (q) ‘may be (p),

may be (q)’ or to li (p), to li (q) ‘the same’, if he is unable or unwilling to commit himself to either of two mutually exclusive descriptions. Thus, the DV poxlopal-pogladil ‘tapped-gave a stroke’ is roughly equivalent to ne to poxlopal, ne to pogladil ‘maybe, he tapped, maybe, he gave a stroke’, the closest English equivalent being a combination of ‘or’ and an explicit marker of uncertainty such as ‘may be p, or may be q’ (this paraphrase does not, however, always work). An especially productive subtype of this approximative device is constituted by descriptions of sounds that cannot be defined unequivocally; in the press we find for instance such formations as voju-siplju ‘I howlcroak’, gudit-pyxtit ‘throbs-puffs’, 9umit-zvenit ‘makes noise-rings’, noetzavyvaet ‘weeps-starts howling’, and the writer Ale9kovskij creates raskudaxtalas’-raskukarekalas’ ‘she went cackling-cook-a-doodle-dooing’ and #akae9’-kvakae9’ ‘you quack’ (the first verb is meaningless). The ludic effect caused by such onomatopoetic pairs is evident. My corpus of spontaneous DV creations contains still more candidates for another type of cohyponymous pair: instead of being fused into one single meaning, the two meanings remain separate but are added up into a new, complex whole. Folkloristic language mainly offers examples related to the most fundamental human needs: Odet’sja-obut’sja ‘get dressed-put on shoes’, odet’-obut’ ‘dress-put shoes on’, razut’-razdet’ ‘take off shoes-undress’ est’-pit’ ‘eat-drink’, kormit’poit’ ‘feed-water’, pet’-pljasat’ ‘sing-dance’ Since we are dealing with two distinct activities, these examples no longer represent the semantic prototype of DV: they are twins with respect both to their prosody and meaning. The same holds for similar examples to be found in the modern press, which, however, look somewhat more up-todate: stri"’-brit’ ‘cut hair-shave’, ubit’-ranit’ ‘kill-wound’, pisat’-lepit’ ‘paintsculpture’, "itat’-pisat’ ‘read-write’, pooxotit’sja-poryba"it’ ‘hunt-fish’, posolit'-popere"it’‘salt-pepper’

But are these real cohyponyms? If so, what would be the corresponding hypernyms? More often than not, they do not exist or are at least doubtful (for instance, ‘consume’ is too large and ‘nourish’ too restricted temporally to serve as hypernyms of ‘eat’ / ‘drink’ and ‘feed / water’, respectively) and stylistically inadequate. Thus, it becomes evident that the DV construction in such cases replaces non-existing or inappropriate hypernyms by enumerating at least two of their potential cohyponyms. Whether these are meant to denote the whole supercategory or just give an exhaustive list is open to interpretation, cf. ‘read + write’ = ‘be literate’ on the one hand, ‘to salt + to pepper’ =? to spice’ on the other hand. Some of the verbal pairs have nominal dvandva correlates, cf. ode#da-obuv’ ‘dressshoes’, sol’-perec ‘salt and pepper’. Moreover, the missing hypernym may even be represented by verbal triplets, cf. Smo#ete takuju nevestu prokormit’, odet’-obut’? AiF 42/02 ‘Will you be able to feed, dressprovide shoes for the bride = to support the bride?’, Xotite vse %to usly9at’uvidet’-po"uvstvovat’? LG 43/02 ‘Do you want to hear-see-feel all this?’, tam duri skol’ko vlezet, e9’-pej-no"uj (S.Sokolov) ‘How many crazy things will enter there, eat-drink-spend (imper.) a night!’. For some of the DVs given above, there exist, however, exact monolexical synonyms. For instance, ljudi, umeju9"ie "itat’-pisat’ ‘people who can write and read’ are simply gramotnye ljudi ‘literate people’. Note that the DV is subject to the same temporal restriction as its equivalent in that it denotes only a permanent property (a generic state in Vendlerian terms) and cannot be transformed into a predication with specific object as in *Deti v 9kole "itajut-pi9ut o Putine ‘At school, children read and write about Putin’.xi On the whole, most (if not all) cohyponymical DVs do not allow for all types of temporal reference: in particular, the progressive use of the imperfective aspect seems to be blocked. My next example is rather exceptional in that it illustrates the metaphorical use of a DV composed of two participles: (2)

Vy znaete, srazu priexali 2PL.NOM know-PRS.2PL at once arrive-PST-3PL

razutyrazdety, odelis’unshoed-PTCP-NOM.PL undress-PTCP-NOM.PL dress-PST-3PL-REFL obulis’ i den’#onok skopili. got shoed-PST-3PL-REFL and money-GEN.PL save-PST-3PL ‘You know, they arrived without suitable shoes and clothes and got dressed and shoed immediately and saved some money.’ (Russkaja razgovornaja re"’ evropejskogo severo-vostoka Rossii) In this case, the semantic bleaching of both participles is already lexicalized, the verb pair razut’-razdet’ itself being of considerable age. Note that on the continuum between the two poles , we have moved closer to the prototypical one-event reading. An even more evident case of lexicalization is constituted by the pair kormit’poit’ ‘feed-water > to provide sb. with sufficient financial means’. Let us add that there are also cohyponymous DVs whose components allow only for an alternative reading, for example Klintona pisali-lepili ‘They painted Clinton or made sculptures of him’. In view of the lack of lexical hypernyms, it is sometimes hard to determine if a given pair represents the type discussed so far or the next one: we often find pairs whose members are linked by some common frame or script (Weiss 2003: 46-48). They may then denote either two subsequent phases or two more or less simultaneously ongoing processes, cf.: vypit’-zakusit’ ‘drink [vodka]-have a bite’, provo#at’-plakat’ ‘escortlament’, ispovedat’-pri"astit’ ‘confess-receive Communion’, stirat’gladit’‘wash-iron’, ryt’-iskat’‘dig-search’, pet’-pljasat’ ‘sing-dance’, xodit‘-guljat’ ‘walk-have a good time’, pit’-prazdnovat’ ‘drink-have a party’, nalit’-popit’ ‘pour-drink’, ponyrjat’-poplavat’ ‘dive-swim’, prosve"ivat’-osmatrivat’ ‘X-ray-examine’, xranit’sja-pylit’sja ‘be storedbecome dusty’, ponosit’-privyknut’ ‘wear for some time-get used to’ Note again the selective character of such pairs: there may be “missing links”, i.e. activities belonging to the same script, but not indicated in the

given case, cf. su9it’ ‘dry’. On the other hand, an Internet search provides sufficient evidence for the existence of the variant stirat’- su9it’ ‘wash – dry’ (for instance when referring to diapers, which need not be ironed). The first three of the cited pairs denote culturally specific concepts: vodka consumption is obligatorily accompanied by eating gherkins, fish and the like to get rid of the bitter taste; young men leaving to join the army (or the dead at a funeral) were escorted by the whole village with public laments (the so called provody); and finally, in Orthodox rite Confession obligatorily precedes Communion. Such examples confirm Aikhenvald’s statement that “a function of verb serialization is then to represent complex events, which are – at least partly – a cultural construct” (2006: 11). Not surprisingly, verb order does not allow for inversion in such cases. In the remaining cases, inversion is possible even if it leads to anti-iconic order, cf. gladit’-stirat’ ‘iron-wash’. Moreover, such pairs may be enlarged, yielding triples of the type gladit’-stirat’-"istit’ ‘iron-washclean’ or stirat’-su9it’-gladit’ ‘wash-dry-iron’. On the whole, however, one has to admit that the same content is much more often expressed by means of asyndetic coordination, cf. stirat’, gladit’, ubirat’ etc. Other cases rather call for an analysis in terms of hypernymy vs. hyponymy, e.g. spasite-pomogite ‘save-help!’ (Gri9kovec), stali pit’-kutit’ (folklore) ‘they began to drink-tipple’, dvi#etsja-krutitsja ‘is movingrotating’, or pe"’-gotovit’ ‘bake-cook’. According to this analysis, such examples belong to the semantic prototype of DVs since they refer to one single event. The meaning of the whole may be the same as that of the hypernym alone; for example, the hyponym ‘bake’ in the last-mentioned pair could just as well be omitted. But in general, clear instances of hypernyms as first or second components are rare and at least in colloquial language mainly restricted to specifications of somebody’s way of walking (cf. idet-prixramyvaet ‘walks limping’, xodit-podprygivaet ‘walks hopping’) or talking (cf. govorit-zaikaetsja ‘speaks stuttering’). Note that in these pairs it is the hypernym that is omissible. Antonymous pairs may also be found in modern Russian DVs, cf.: gruzit’-razgru#at’ ‘load-unload’, uexat’-priexat’‘depart-arrive’, otveztiprivezti ‘drive away-drive back’,xii vklju"at’sja-vyklju"at’sja ‘be switched

on-switched off’, zapirat’sja-otpirat’sja ‘be locked-unlocked’, promerzat’ottaivat’‘freeze-thaw’, sognut’-razognut’ ‘bend-unbend’, razdevat’-odevat’ ‘undress-dress’, pripodnjat’-opustit’ ‘lift-lower’, #it’-umirat ‘live-die’ These are not typical for folkloristic use: all examples cited represent contemporary poetic or journalistic language. As for converse verb pairs (a relation that should strictly be kept apart from antonymy!), they occur rarely, the only two clear instances being pokupat’-prodavat’ ‘buy-sell’ and podat’-prinjat ‘serve-receive’. Note that he first one is backed by the corresponding binominal construction kupljaproda#a. The following examples shed some light on the creative potential provided by the DV construction; the first two denote particular subscripts of the mating season from a dog’s perspective and the Western approach to the adoption procedure, respectively: (3)

Pora vysmatrivat’vynjuxivat’ svoju Time look out-IMPF-INF sniff out- MPF-INF REFL.ADJ-ACC.SG.F vtoruju polovinu. Second-ACC.SG.F half-ACC.SG ‘It is time to seek (lit. ‘to look out-sniff out’) for one’s second half.’ (AiF 12, 2001)

(4)

kak inostrancy usynovljajut- pokupajut rossijskix how foreigner-NOM.PL adopt-PRS.3PL buy-PRS.3PL Russiandetejsirot. ACC.PL.ANIM child-ACC.PL orphan-ACC.PL ‘…how foreigners adopt (lit.: adopt-buy) Russian orphans.’ (AiF 24/03)

(5)

My – dvadcat' pi9u9"ixsnimaju9"ix 1PL.NOM twenty write-PTC-IPF-GEN.PL photographiz raznyx stran PTC-IPF-GEN.PL from different-GEN.PL country-GEN.PL i primknuv9ij k nim

and join-PTCP.PRF-NOM.SG to 3PL.DAT korrespondent MN – tol'ko "to podkatili k correspondent[NOM.SG] MN just arrive.-PST-3PL to Mozdoku Mozdok-DAT.SG ‘We – twenty reporters (lit. persons) writing-taking pictures) from different countries and an embedded MN correspondent – had just arrived at Mozdok.’ (Moskovskie novosti 49/02) Interestingly enough, even alternative subscripts are not excluded, as is shown by the next example, where three different methods of how to punish vandals are discussed: (6)

Ka#dyj desjatyj (11,5%) predlagaet «zapretit’Every-NOM.SG.M tenth-NOM.SG.M suggest-PRS.3SG forbid-INF udu9it’arestovat’». suffocate-INF arrest-INF ‘But it should be borne in mind that for some, vandalism is just a form of leisure time. Every tenth suggests to “forbid-suffocatearrest”’ (AiF 44/2002)

The last example from this series playfully alludes to a set phrase, viz. the DV imperative podaj-prinesi: (7)

Sama Nata9a sostojala v dol#nosti «podajSelf-NOM.SG.F Natasha exist-PST-SG.F in job-LOC.SG serveprimivymoj tarelki». 2SG.IMP receive-2SG.IMP wash-2SG.IMP dish-ACC.PL ‘Natasha herself was employed as a servant (lit.: ‘serve-receive-wash dishes’).’ (Ju. Daniel’)

On the whole, we may conclude that the non-canonical subtype of the DV construction provides a convenient tool for spontaneous descriptions of

complex activities, since it combines composite semantic structure with the idea of a higher-order unity conditioned by such relations as antonymy, cohyponymy, etc., and by strict syntactic contiguity. At least three pragmatic motivations for this technique can be singled out: first, either the author is unable to find an appropriate one-word description, or, second, he is too lazy (or else too reluctant) to bother with such a description. The latter situation often leads to the use of isolated imperatives or infinitives as ready-made quasi-nominal blocks as in ex. 32 or in den’gi na uexat’priexat’ ‘money for leave-come back’ that are simply quoted without being syntactically integrated into the remaining sentence. The following example taken from an Internet discussion under the title “Our tears” may illustrate this kind of use: (8)

$ena %to takoj pribor gladit’Wife-NOM.SG COP such-NOM.SG.M device[nom.sg] iron-INF stirat’, umeet #rat’ gotovit’ i vse wash-INF know-PRS.3SG eat-INF cook-INF and all-ACC.SG.N takoe! such-ACC.SG.N ‘A wife is a device [for] ironing-washing, she knows how to cook and all the rest of it!’ (another participant’s reply:) &to domrabotica, mal’"ik. ‘This is a maid, my boy’ (forum.antidate.org/lofiversion/index.php/t4461-150.html)

Finally, according to the third explanation, we are, in contrast, dealing with a conscious aesthetic device exploited by writers and journalists.xiii In all three cases this subtype of DV construction has moved far away from its formulaic origins in folkloristic language. On the other hand, the latter are still present in texts that systematically aim at imitating the folkloristic genre, as is often the case with modern religious legends.xiv

2. Equivalent constructions in Finno-Ugric languages As previously mentioned, the Permic and Volga Finnic languages possess the same type of verbal juxtaposition as the one represented in Russian DVs:xv with the exception of the last criterion (reversibility of word order), almost all requirements stated above in the definition of Russian DV (see section II.1) are also met by analogous constructions to be found in Komi (Zyryan), Udmurt (Votyak), Mari (Cheremis) and Mordvinian, in particular the double inflection.xvi There is only one exception: reversibility of word order is only given in Mordvinian, cf. andoms-s’imd’ems ‘feedwater; entertain (treat as a guest)’ = s’imd’ems-andoms (Honti 2005: 37). Moreover, the Finno-Ugric pairs are in general less marked stylistically than their Russian counterparts. Triple verbs also seem to occur frequently, cf. the Komi example kurgyny-dzurtny-ovny ‘breathe heavily-crack-live = live for a long time despite one’s sickness’. Some of them arise spontaneously in poetic language as they do in Russian; some others seem to be lexicalized already. Since the pertinent details were already presented in Weiss (2003: 52-57), I will here restrict myself to summarizing the main parallels and divergences and complement them with fresh material from Komi presented in Fedina (2005) and with some Mordvin parallels.xvii The distinction of prototypical and non-prototypical uses turns out to hold as well for the Finno-Ugric languages studied here as for Russian. As for double verbs referring to one single event (the prototype), we again find the subtype with intensifying meaning realized by quasi-synonymous pairs, reduplication or frozen combinations, which are reminiscent of the Russian situation; for examples from Komi folklore and fiction, see Ludykova (1995: 50 ff.). The hypernymous subtype offers a richer choice than in Russian. Above all, we find a multitude of combinations depicting a person’s manner of walking with very specific first verbs that add more lively colours to the visual or auditory perception evoked by the second verb; semantically, such pairs are vividly reminiscent of the so-called colourative construction in Finnish, see below. To quote just two particularly striking examples from Komi: tapikas’ö-munö ‘is going, clumsily stamping with his/her large feet (usually about bears)’ and

bad“#al’ö-vetledlö ‘is walking-waddling with heavy steps (usually about a hen)’. Note that the hypernym always occupies the second position. The same tendency to make the description as colourful as possible is manifested in spontaneous metaphorical uses of SVC, cf. the following quotation from Fedina (2005: 207; it contains a triple verb): (9)

Menam dzurtö-or “jas’ö-ojzö byd ly i byd kusyn, byd sön ‘At me are creaking-tearing-moaning every small bone every joint every vein.’

As for auditory perception, cf. rapiimunny-"egny ‘to break off with a crack’ or opöstny-gorödny ‘to begin to howl with a wild voice’ (both examples from Fedina 2005: 207). The hypernymic subtype allows also for other combinations that are not characteristic of Russian DVs, for example those that contain ‘make’ as its second component: kerny-vö"ny lit. ‘make-prepare = process, work on’. Similar pairs containing the verb ‘do, make’ are attested in Mordvinian, where the latter verb is desemantized, serving only to intensify the other component, cf. ojms’emst’ejems ‘rest-do’; interestingly, in such cases inversion is blocked (Honti 2005: 38 f.). According to Wälchli (2003: 148), such combinations express rather a delimitative meaning; the verb ‘do’ would thus function as an equivalent of the Russian verbal prefix po-. The remaining semantic relations that link the two verbs are identical to those to be found in Russian: besides synonymy, which characterizes prototypical DVs, we find antonymy, cohyponymy and / or common scripts or frames; besides Weiss (2003), see also Wälchli (2003: 116-128). Lexical equations with Russian are easy to find; to mention but a few: ‘plough-sow’, ‘eat-drink’, ‘feed-water’, ‘dress-put on shoes’, ‘cook-bake’, ‘live-be’, etc. The last-mentioned example clearly indicates the link to Russian folklore, cf. the opening formula of fairy tales #il-byl ‘there once was’ (see next section). On the other hand, it should be emphasized that unlike in Russian (but much as in many prototypical SVCs), semantic bleaching and blurring of composite meanings are frequent here and often give birth to new abstract meanings. They range from fairly predictable

results, for example ‘sing-dance > be merry, enjoy’, ‘plough-sow > work on the field’, ‘catch-kill > hunt’, ‘dry out-bare > eradicate’ to less expected ones, cf. ‘go in-go out > be restless / agitated’ (Mordvin) or ‘ask-pick > beg’ (Komi), and end up with totally unexpected new meanings, cf. ‘knitsew > take care of sb.’, ‘go in-go out > be friends with sb.’ (both from Komi). Since the composite reading usually remains also available, these are cases of polysemy;xviii they illustrate, at the same time, the transition from non-prototypical (composite interpretation) to prototypical (one event) meaning of SVC. Moreover, all four Finno-Ugric languages under discussion have verb pairs where one component is redundant and could be omitted, as was the case with Russian xoxotat’-zalivat’sja ‘laugh loudly’ etc.; it was, however, impossible to find exact semantic coincidences with Russian. In addition, there exist pairs, one of whose members does not occur as an autonomous single verb apart from the given combination; this feature is not typical of SVC since “each verb in an SVC may also occur as the sole verb in a clause” (Dixon 2006: 339). Such DVs often serve to intensify the meaning expressed by the other verb, which again is reminscent of Russian; the latter does not, however, have verbs that are bound exclusively to the DV construction. The following particularity is also alien to Russian: some Finno-Ugric DVs consist of linguistically mixed material, i.e. they may contain one Russian root as in Komi durytny-pör”jas’ny ‘cheat’, two verbs stemming from different dialects or from a dialect and the standard language (Fedina 2005: 207).xix Interestingly, Udmurt combines Russian first verbs with the vernacular verb ‘make’, cf. issledovat’ karyny ‘research-make’. A last difference concerns the rise of TAM markers. As might be recalled, grammaticalization of first components of DVs is at its initial stage and restricted to the meaning of uninterrupted duration in Russian; this language cannot be said to have already developed TAM markers within a SVC. This process seems to be more advanced in Komi, where we find munnö ‘go’ as second verb with ingressive and kerny ‘make’ with attenuative meaning. Besides this, Komi has transformed some verbal roots into grammatical suffixes (Sidorov 1992: 28). Some pairs are

restricted as to their time reference: in Mordvin for instance, synonymous pairs such as mus’kems-lovpatn’ems ‘wash’ appear only in generalizations or with negation (Wälchli 2003: 151). This is reminiscent of the behaviour of several Russian subtypes discussed in section 1.7. Grammaticalization is more salient in Mari and Udmurt, where it affects, however, a different type of verbal construction: in groups composed of a finite verb such as ‘go’, ‘come’, ‘take’ or ‘end’ and a converb the former undergoes auxiliarization. This development is clearly due to borrowing from the adjacent Turkish languages (Chuvash and Tatar), where this phenomenon is widespread (Tatar is reported to admit not less than 30 such auxiliaries in the said construction). The development of specific pragmatic meanings manifested by Russian imperatives in DVs is not attested in Finno-Ugric languages according to my sources. A few words may be added about the situation in the Balto-Finnic group. In Finnish, SVC is attested as a dialectal relic, competing with the so-called colourative infinitive construction (Jarva and Kytölä 2007).xx The two variants are considered synonymous, the serial variant being even acceptable in the standard language; it is, however, much less frequent than its infinitival counterpart. The following pair of sentences both meaning ‘I fell into the snow’ may illustrate this use: (10a) Minä sinne lumeen I there snow-ILL (10b) Minä sinne I there

lumeen snow-ILL

koatuat fall-INF

tupsahdin COL-PAST-1SG

kaaduin tupsahdin fall- PAST-1SG COL-PAST-1SG

The two-finite variant is also reported to exist in Sami languages. Since Ojutkangas (1998), this construction has been related to SVC. The same holds for Standard Estonian (cf. Tragel 2003). In this language, the set of first verbs is very restricted: apart from minema ‘go’, tulema ‘come’, käima ‘walk’ and võtma ‘take’ only a few other motion verbs such as jooksma and lippama are mentioned; this differs sharply from Russian, where many other initial components are allowed. Triplets do occur, but

usually only with minema and käima as the first two verbs (recall that in Russian, triplets also usually contain pojti/idti ‘go’ as first verbs). Grammaticalization is said to take place with minema ‘go’ and tulema ‘come’ marking a new immediate future-like meaning, which is not surprising, given the general tendency of these verbs to develop into future markers (Majsak 2005). On the whole, one gets the strong impression that these instances of SVC in modern Balto-Finnic are but the meagre relics of a construction type which has long been in decline. In conclusion, my little survey has shown a striking amount of structural and lexical convergence between Russian and Permic or Volga Finnic languages. Against this background, most divergences found are of minor importance, e.g. the existence of DVs with components that cannot function autonomously or the salience of the ‘colourative’ subtype. What deserves our attention, however, is the fixed word order within the DVs and the extensive grammaticalization and lexicalization, since these features bring the Finno-Ugric languages closer to the general characteristics of SVC. In this way, the not too numerous divergences between Russian and Finno-Ugric DVs all point to the same direction: SVCs were at the beginning (and still are nowadays in Permic and Volga groups) more deeply rooted and better integrated in the lexicon and grammar of Finno-Ugric than in East Slavic, where they play only a peripheral role.

2.1. Who borrowed from whom? On the basis of the descriptions given above, we can now attempt to determine whether the parallels detected may be contact-induced and, if so, which language was the target and which one(s) the source of the morphosyntactic calques. To begin with, SVC is by no means a common Slavic phenomenon: West and South Slavic languages either do not have it at all, or they allow only for lexically restricted serialization in the imperative and (like Polish) in the 1st person present (for details, see Weiss 2003: 49-51). Nor does our construction represent a relic from ancient

Indo-European times, since there is no other IE. language that would manifest such a construction (note that this lack contrasts sharply with double nouns, or else dvandva compounds,xxi which are not only common Slavic features, but represent a venerable Indo-European heritage). Consequently, we are dealing with an East Slavic innovation whose origins are still to be determined: when did it emerge? Was the decisive impulse triggered by contact with other languages, or should we assume that the DV construction originated on purely Slavic soil? The latter assumption seems by far less plausible since there simply was no need for such an innovation, given that Russian already had the full inventory of clause linking techniques mentioned at the beginning of the present paper. Therefore, we are forced to seek the impulse outside the Slavic medium. Among the possible sources of the borrowing there is still another group left, namely the Turkic languages, which indeed possess a structurally analogous construction (Csató 2001). Turkic influence must, however, be excluded for different reasons (Weiss 2003, 51-52): first, these languages prefer the structural model consisting of an auxiliarized finite verb and a converb (see previous section); second, among the remaining cases of finite SVC, there are no striking parallels with Russian DVs; and third, the temporally and locally restricted contact of Russian with, say, Tatar or Chuvash led to numerous lexical borrowings, but certainly did not favour the adoption of a whole syntactical construction. There remains, however, an isolated DV that has probably been borrowed from Turkic, namely the pair pokupat’-prodavat’ ‘buy-sell’ together with its nominalization kuplja-proda#a; this designation of trade seems to be known all over Eurasia.xxii In addition, Turkic provided the pattern for nominal twin formations whose second member is semantically empty and a mere phonological replica of the first with substitution of the initial consonant by an m-, cf. Russian xuxry-muxry ‘trifle’;xxiii this device may also be employed in creating new DV combinations, cf. 9irnu-myrnu, s "em vynyrnu‘ (Ju. Dani%l’).xxiv Thus, the most plausible explanation for Russian DVs is to trace them back to Finno-Ugric. This would be perfectly in line with the view that “languages with SVCs tend to form areal clusters” (Aikhenvald 2006: 52).

The main problem connected with this solution has already been indicated: the examples discussed in the previous section all stem from languages that entered into contact with Russian in a relatively late period and could exert but marginal adstratal influence on the dialects of Slavic settlers in the northeast and east of European Russia.xxv On the other hand, in view of its numerous loanwords borrowed from Russian in recent times one could even argue that Komi might have borrowed if not the whole DV construction, at least some subtypes from Russian. The scenario required by our innovation is quite different: we are talking of widespread and long-lasting Slavic-Finnic bilingualism which eventually led to language death, the former speakers of Finnic switching to Russian, but preserving some substratal relics of their ancient language. Such a scenario can only be met by languages that have long been extinct. The most reliable candidates for the role of source are the idioms spoken by two tribes mentioned already in the first Russian chronicles as tribute payers of the Kievan Empire, viz. Merja and Murom. Toponymy and historical sources indicate that they inhabited Central Russia including the Podmoskov’e area. This is exactly where Tka"enko in his carefully documented 1979 monograph locates the origin of the Russian fairy-tale formula #il-byl; dialectological and historical data suggest such a conclusion, and the author convincingly argues that no other source (vernacular evolution or borrowing from a third language) can be found that would explain the unique career of this formula than the Finno-Ugric double root *elä-wole- ‘be-live’ (for a more sceptical look at this pair, see Petruxin 2007). This combination is still attested in almost all Finno-Ugric languages including representatives of the Baltic group (Estonian, Karelian), most often not only as a frozen initial formula, but also in the middle or at the end of the tale, as politeness formula,xxvi and in whole finite and non-finite paradigms. Thus, we have detected a plausible pattern: the item in question is by far better developed (more productive, less restricted) in the Finno-Ugric source than in the Russian target language. The same pattern now obtains when we turn to the SVC model as a whole: as we have seen in the previous section, in Permic and Volga

Finnic this type of construction is more differentiated and less restricted in every respect than in Russian. The assumption that the SVC device already existed in Common Finno-Ugric is already shared by virtually all specialists in this field. It seems all the more plausible as there was no other paratactic linking mechanism: Finno-Ugric languages simply abounded in hypotactic constructions with converbs and infinitives, but they did not have paratactic conjunctions from the beginning; in particular, a word such as ‘and’ was lacking and had to be borrowed later by the individual branches either from Germanic or from East Slavic (cf. for example Bátori 1980). Thus, we may conclude that unlike in Russian, in Finno-Ugric verb serialization was by no means a syntactically superfluous phenomenon, but played a crucial role as a multiverb construction. Moreover, the cumulative semantics of the asymmetrical DV subtype fits perfectly in the picture of Finno-Ugric nominal derivation: as is well known, such a notion as ‘face’ is often composed by two designations of its most salient parts, such as ‘eyes-mouth’, ‘eyes-nose’ or ‘nose-mouth’ in these languages. On the other hand, if our assumption is correct, what results is a peculiar kind of cross-borrowing: Finno-Ugric owes its syndetic linking to East Slavic, whereas the latter developed SVC due to the substrate influence of the former. As a result, however, SVC is marginalized on both contacting sides: in Finno-Ugric it seems to be in decline (probably also under the pressure of the new conjunction ‘and’), and its introduction into Russian remains at the functional periphery of the clause-linking system. The common Finno-Ugric origin of the DV construction is corroborated by Hungarian folklore texts.xxvii In this language, DV are only attested in fairy tales and child lyrics (songs and verses); otherwise, they have practically ceased to exist, the last relics surviving in peripheral dialects in Transylvania and the Eastern Carpathians.xxviii The different types represented in Hungarian data are, however, very reminiscent of those discussed in the previous section. For instance, we find synonymous, antonymous and cohyponymous pairs, and their semantic fillings are often identical with Russian and Komi, to mention but a few: eszik-iszik ‘eatdrink’, etet-itat ‘feed-water’, ével3dik-vúcsálkodik ‘mourn-be sad’, sír-b3g

‘cry-weep’, jön-megy ‘come-go’, ad-vesz ‘buy-sell’, süt-f3z ‘bake-cook’, ír-olvas ‘write-read’, etc. New, non-composite meanings are not very frequent, but do occur, e.g. ás-kapál ‘dig-hack > nit-pick, intrigue, be jealous’. Besides this, as in Komi, there are verb pairs where one of the components is phraseologically bound to the DV construction (often it provides a paronymous replica of the other verb, cf. irul-pirul ‘blush’; irul does not occur apart from this DV). Other DVs are composed of the basic verb and its derivative, cf. megy-mendegél ‘go without haste’, which, as may be recalled, is a widespread type in Russian. Finally, Hungarian DVs simply abound in onomatopoetic pairs: the corpus analysed in Valko (2006) contains not less than 58 different pairs denoting sounds (to these one may add another 40 DVs describing different distorted modes of speech), practically all of which have a paronymic structure, cf. cseregbereg, hipeg-hupog, ripeg-ropog, etc. This preference is obviously due to the youthful audience. But, additionally, it constitutes the phonetic counterpart of semantically similar, i.e. more or less synonymous, verb pairs. On the whole, one may state that, notwithstanding its intense and long lasting contact with German and Slavic neighbours in Central Europe, Hungarian still preserves an astonishing number of parallels with DV constructions in the Finno-Ugric languages in the northeast of Russia. This archaic feature is all the more striking as Proto-Hungarian must have lost contact with the Finnic branch already at the beginning of our era. In the long run, however, Hungarian may be expected to join the languages with contact-induced SVC loss. My argumentation would be incomplete if no mention was made of another and more important syntactic calque that points to the same direction: although this issue is more controversial than the borrowing of the DV construction, the partial functional replacement of Common Slavic im&ti ‘have’ by existential sentences with byt’ ‘be’ and the adessive preposition u ‘at’ in Russian can hardly be explained otherwise than by Finno-Ugric substratal influence. The same holds true for the development of this preposition into a marker of “external” (raised) possessors,xxix which

in other Slavic languages are expressed by the dative. This is, however, another story, which cannot be discussed here. Notes 1. For the sake of convenience, I will focus the discussion on Komi, a language spoken in the northeast of European Russia to the west of the Urals. This language has the advantage, unlike its southern neighbours such as Udmurt or Mari, of not having been subject to significant Turkic influence. 2. In her otherwise inspiring paper, Aikhenvald (2006: 45-46) claims that Russian DVs behave the same way as similar phenomena in other “familiar European languages” (Swedish, Bulgarian, Hungarian, American English, Brazilian Portuguese) and “cannot be considered on a par with SVC”. This assumption is clearly based on insufficient Russian data. Counterevidence to Aikhenvald’s particular claims will be presented throughout this paper. 3. In view of these figures, the claim that Russian DVs “are limited to just a few verbs” (Aikhenvald 2006: 46) sounds somewhat bizarre. The results of a search of the Russian National Corpus (ruscorpora.ru), which already provided several thousand instances of DV imperatives, will be published in a separate paper. 4. A search for such examples with the Google search machine provided several instances of vstal vyprjamilsja (vo ves’ rost) ‘he stood up and stretched his body to full length’. 5. However, with ne pojme9’ ne razbere9’ ‘you don’t understand’, which consists of two synonymous verbs, neither the substitution of ni nor the omission of the second negation is possible. 6. Similarly, Aikhenvald (2006: 8 f.) quotes examples where only part of the whole SVC is in the scope of negation. 7. This does not mean that double converb forms are not attested in my corpus; in particular, one may find such examples in hybrid texts of the kind described in footnote 15. 8. To illustrate the grammatical frozenness of this example, using a large data collection, I quote the following figures provided by the Google search machine (04/12/2011): 1st pers.pl. 899,000 instances, 1st pers.sg. 284,000, 2nd pers.sg. 9820, 3rd pers.sg. 544 instances. The corresponding figure for uvidimposmotrim amounts to only 6140 instances.

9. Diachronically, this could well be a frozen instance of the former perfect with the auxiliary ‘be’. 10. Unfortunately, Coseriu does not sufficiently distinguish between the surprise meaning discussed here and the simple ingressive meaning, which seems to prevail, for instance, in Swedish, cf. Ekberg (1993). 11. It may be added that Russian podi (#e) functions also as a true particle expressing surprise. 12. On the other hand, Sol#enicyn has Lenin ina"e pisal-govoril! ‘Lenin wrote and spoke otherwise’, where a specific object (‘about this’) is at least presupposed. 13. In view of these examples, the claim that “unpredictable derivational restrictions may arise within each particular semantic group: for instance, in Russian, motion verbs containing the preverbs u- and ot(o)- cannot occur in double verb constructions, while verbs with other preverbs can” (Aikhenvald 2006: 46 ) turns out to be erroneous. 14. In modern fiction, the last two motivations may well overlap if the author seeks to imitate his character’s speech or way of thinking. 15. In a text entitled 'udotvornaja ikona i Anikitina Repnin’ (“The miraculous icon and A. Repnin”) comprising 1181 words, I found not less than 45 different DVs. Some of them sound rather artificial and add a kind of pseudofolkloristic flavour.[Fascninating evidence that this represents a folkloristic sound] 16. The term ‘double verb’ is not used in the Finno-Ugristic literature; the usual term there is ‘paired verb’, which, in view of the existence of triplet verbs (see below), is not more convenient than ‘double verb’. Curiously enough, the languages to be discussed here are not even mentioned in the chapter on double verbs in Aikhenvald (2006). 17. The term ‘Volga Finnic’ is used here as a mere geographic label because Mari and Mordvinian are no longer conceived as a separate genetic branch of FinnoUgric, cf. Abondolo (1998: 3). Due to the lack of space, the two main varieties of both Mari and Mordvinian will not be taken into account here, since they are irrelevant for our purpose; the same holds for the difference between the two standard languages of Komi (Zyryan vs. Permyak). For all pertinent details see Abondolo (1998). 18. For obvious reasons (lack of informants and corpora), information on FinnoUgric data is often too scarce for a direct comparison with Russian. 19. Contrary to this, their Russian counterparts usually have only compositional meaning.

20. For similarly mixed nominal constructions in Finno-Ugric and Turkic languages see Pacsai (1995: 90-91). 21. I owe this reference to Jussi Ylikoski (p.c.). The same holds for the reference to Tragel (2003) about Estonian (see below), an English paraphrase of which was kindly sent to me by the author. 22. For a full description of nominal pairs in modern Russian see Bergmann (2006). As for structural parallels and divergences between double nouns and double verbs, cf. Weiss (2000: 361-363). The Eurasian dimension of such binominal constructions is best described in Wälchli (2003); the author’s term ‘co-compounds’ also includes double verbs. 23. Verb pairs with this meaning are attested even in Vietnamese and Thai (Bisang 1992: 295, 334). Note that the main trade centre at the Chinese border located in front of the Russian town Kjaxta was named Maimachen ‘buy-sell-city’. 24. For the Turkic origins of this formation and its traces in modern Russian see Plähn (1987). 25. Komi seems to have the same type of formation, cf. 9arny-marny ‘slave away’ (Fedina 2005: 206). 26. That north Russian dialects were indeed affected by influence of neighbouring Komi dialects in recent times is shown by Leinonen (1998). 27. Note that the set expression Kak #ivete-mo#ete? ‘How is life?’, lit. ‘How do you live-be able?’ quoted in section i.5 also has an exact counterpart in many Finnic languages. As for Circumbaltic convergences in general, see Dahl and Koptjevskaja (2001). 28. The following observations are based on Pacsai (1995) and an MA thesis (Valko 2006) devoted to a systematic comparison of Hungarian and Russian double verbs that was defended at the Slavic Department of the University of Zurich. 29. For Hungarian, Aikhenvald’s (2006: 46) remark that “double verbs are often restricted to certain registers” is justified. However, to list Hungarian in the same row as Russian is highly misleading. 30. This case is illustrated by the example 9 from Komi. It should be emphasized that, unlike double verbs, the spatial model of possessive meaning is not restricted to Permian and Volga Finnic, but is also characteristic of Baltic Finnic including Finnish.

References Abondolo, D. 1998 The Uralic Languages. London-New York: Routledge. Aikhenvald, A.Y. 2006 Serial Verb Constructions in Typological Perspective. In: A.Y.Aikhenvald & R.M.W. Dixon (eds.): Serial Verb Constructions. A Cross-linguistic Typology. Oxford: University Press (Explorations in Linguistic Typology), 1-68. Bátori, I. 1980 Russen und Finnougrier. Kontakt der Völker und Kontakt der Sprachen. Wiesbaden: Harrassowitz. Bergmann, A. 2006 Binomina im Russischen als Kategorie der komplexen Benennung. Frankfurt a.M.: Peter Lang. Bisang, W. 1992 Das Verb im Chinesischen, Hmong, Vietnamesischen, Thai und Khmer (Vergleichnde Grammatik im Rahmen der Verbserialisierung, der Grammatikalisierung un der Attraktorpositionen). Tübingen: Gunter Narr. Coseriu, E. 1966 «Tomo I me voy». Ein Problem vergleichender europäischer Syntax. Vox Romanica 25: 13-55. Csató, E. 2001 Turkic double verbs in a typological perspective. In: K. H. Ebert and F. Zuniga (eds.), Aktionsart and aspectotemporality in nonEuropean languages, 176-187. Zürich: Arbeiten des Seminars für Allgemeine Sprachwissenschaft. Dahl, Ö. and M. Koptjevskaja (eds.) 2001 Circum-Baltic Languages. Vol. 1-2. Studies in Language Companion Series 54-55. Amsterdam and Philadelphia: Benjamins. Dixon, R.M.W. 2006 Serial Verb Construction: Conspectus and Coda. In: A.Y.Aikhenvald & R.M.W. Dixon (eds.): Serial Verb Constructions. A Crosslinguistic Typology. Oxford: University Press (Explorations in Linguistic Typology), 338-350.

Ebert, K. 2000

Ekberg L. 1993

Progressive markers in Germanic languages. In: Ö. Dahl (ed.), Tense and Aspect in the Languages of Europe, 605-654. Berlin: Mouton. The cognitive basis of the meaning and function of crosslinguistic take and V. Belgian Journal of Linguistics 8. Perspectives on Language and Conceptualization, 21-41.

Fedina, M. S. 2005 Obrazovanie parnyx glagolov v Komi jazyke. In Istorija, sovremennoe sostojanie, perspektivy razvitija jazykov i kul’tur finno-ugorskix narodov. Materialy III Vserossijskoj nau" noj konferencii finno-ugrovedov (1-4 ijulja 2004 g.), 206-208. Syktyvkar. Fortuin, E. L. 2000 Polysemy or monosemy: Interpretation of the imperative and the dative-infinitive construction in Russian. Amsterdam: Institute for Logic, Language and Computation. Honti, L. 2005 Uráli „páros igék”. Folia Uralica Debrecenensia 12: 33-46. Jarva V. and S. Kytölä 2007 The Finnish Colorative Construction and Expressivity. SKY Journal of Linguistics 20: 235-272. http://www.ling.helsinki.fi/sky/julkaisut/SKY2007/JARVA%20A ND%20KYTOLA.pdf Kokkonen, P. 2000 A comparison of Komi serial verbs with those in Udmurt and Volgaic. In: Congressus Nonus Internationalis Fenno-Ugristarum 7.-13.8.2000, 113-120. Tartu, Pars V, Red. Tonu Seilenthal, Tartu. Kor Chahine, I. 2007 O vozmo#nom puti gramatikalizacii russkogo vzjat’. Russian Linguistics 31: 231-248. Kuteva, T. 1999 On‘sit’ / ‘stand’ / ‘lie’ auxiliation. Linguistics 37-2, 191-213. Leinonen, M. 1998 The postpositive particle -to of Northern Russian dialects, compared with Permic languages (Komi Zyryan). In: Studia Slavica Finnlandensia, 74-90.

Ludykova, V. M. 1995 Sintaksis komi rasskaza 20–x godov. In Grammatika i leksikologija komi jazyka, 48-53. Syktyvkar. Majsak, T. A. 2005 Tipologija grammatikalizacii konstrukcij s glagolami dvi# enija i glagolami pozicii. Moskva: Jazyki slavjanskix kul’tur. Matthews, S. 2006 On Serial verb constructions in Cantonese. In: A.Y.Aikhenvald & R.M.W. Dixon (eds.): Serial Verb Constructions. A Crosslinguistic Typology. Oxford: University Press (Explorations in Linguistic Typology), 69-86. Ojutkangas, K. 1998 Asyndeettisistä verbi-ilmauksista suomalais-ugrilaisissa kielissä. In: A. Pajunen (ed.). Kieliopillistumisesta, analogiasta ja typologiasta, Suomi 185. Helsinki: Suomalaisen kirjallisuuden seura. Pacsai 1995 Areal’nye aspekty parnyx slov v russkom jazyke. Nyíregyháza. Petruxin, P. V. 2007 $ili-byli: vopros zakryt? Russkij jazyk v nau"nom osve9"enii, No 2 (14). Plähn, J. 1987 Xujnja-mujnja i tomu podobnoe. Russian Linguistics 11: 37-41. Pullum, G. K. 1990 Constraints on intransitive quasi-serial verb constructions in modern colloquial English. In: B.Joseph / A.M.Zwicky (eds.), When verbs collide. Papers from the 1990 Ohio State miniconference on serial verbs. Ohio State University , 218-239. Pütsch, H. 2001 Monosituationale und bisituationale Sachverhaltspräsentation am Beispiel des russischen Doppelverbs. In: V.Lehmann, J.Scharnberg (eds.), Slavistische Linguistik 2000. Referate des XXVI. Konstanzer Slavistischen Arbeitstreffens. München, 147-178. Raxilina, E. V. 2003 Rusistika i tipologija. Leksi" eskaja semantika glagolov so zna" eniem ‘sidet’ v russkom i niderlandskom. Russian Linguistics 27: 313-327. Sidorov, A.S.

1992 Izbrannye stat’i po komi jazyku. Syktyvkar. Tka"enko, O. B. 1979 Sopostavitel’no-istori"eskaja frazeologija slavjanskix i finno-ugorskix jazykov. Kiev. Tragel, I. 2003 On the syntactic function(s) of Estonian Serial Construction. In: Keel ja Kirjandus. www.eki.ee/keeljakirjandus/summarytragel.html Valko, P. 2006 Russische und ungarische Doppelverben im Vergleich. MA thesis, Zurich. Wälchli, B. 2003 Co-Compounds and Natural Coordination. Stockholm: Universitetsforlaget. Weiss, D. 1993 Dvojnye glagoly v sovremennom russkom jazyke. V: Kategorija skazuemogo v slavjanskix jazykax: modal’nost’ i aktualizacija. Akty me#dunarodnoj konferencii, Certosa di Pontignano (Siena), 26.-29.3.1992 (A cura di F. Fici Giusti e. S.Signorini), München: Otto Sagner, 67-97. Weiss, D. 2000 Russkie dvojnye glagoly: kto xozjain, a kto sluga? V: Slovo v tekste i v slovare. Sbornik statej k semidesjatiletiju akademika Ju. D. Apresjana, L.L.Iomdin, L.P.Krysin (red.), Moskva: Jazyki slavjanskix kul’tur, 356-378.. Weiss, D. 2003 Russkie dvojnye glagoly i ix sootvetstvija v finnougorskix jazykax. V: Russkij jazyk v nau"nom osve9"enii, vyp.2 (6), 37-59. Weiss, D. 2007 The grammar of surprise: the Russian construction of the type Ko9ka vzjala da umerla ‘Suddenly, the cat died’. In: T. Reuther, L. Wanner, K. Gerdes (ed.), (ed.), Proceedings of the 3rd International Conference on Meaning – Text – Theory, Klagenfurt, May 21 – 24. (Wiener Slawistischer Almanach, Linguistische Reihe, Sonderband 69), München – Wien: Verlag Otto Sagner, 2007, 427-436. Weiss, D.

2008

Vremennaja sootnesennost’ dvojnych glagolov sover9ennogo vida. In: Dinami"eskie modeli: Slovo. Predlo#enie. Tekst. Sbornik statej v "est‘ E.V. Padu"evoj, Moskva: Jazyki slavjanskix kul’tur, 155–177.