Features

This is a brief state-of-the-art report on the 12 features (F 1-12) which we have chosen to investigate with scrutiny in the project. Some of them can readily be subdivided into sub-features (e.g., F7 ‘Animacy’ and F9 ‘Non-canonical case marking’).

 

2.1.1. Phonology/Phonetics: (F1) restrictiveness of o and e (akan’e and jakan’e). 

In subareas of East Slavic, e.g. the Pskov group of Russian or northern Belarusian, the occurrence of o and e is phonotactically restricted to stressed syllables only. A phonologically similar phenomenon is also found in the High Latvian dialect (= Latgalian; cf. Seržant 2010) and, probably, East Lithuanian varieties as well as, in a broader perspective, Mordvinian (Veenker 1967; Stipa 1973).

2.1.2. Morphonology(F 2-3)

(F2) Consonant alternations

Morphologically conditioned stem final consonant alternations are a typical property of Slavic and, to a lesser extent, Baltic. They pervade both inflectional paradigms and suffixal derivation of the main parts of speech. In the BSCZ, morphonological consonant alternations are a profitable domain to investigate the results of contact for the following reasons:

(a) The Slavic varieties of the BSCZ differ from each other in details. The different shape of variants can be considered to be an indicator of the direction (or dominance) of inner-Slavic contact. E.g., Belarusian dialects demonstrate a change {dj} --> {d3} in the present tense stem (1sg) of the i-conjugation, whereas Russian shows {d‘} --> {3} and Polish {d‘} --> {dz}. In rural mixed Belarusian the Russian-like alternation seems to have become predominant.

(b) Lithuanian sharply differs from surrounding Slavic varieties in the treatment of consonant clusters. It strictly avoids gemination, encountered in East Slavic, shows the elimination of non-last consonants in clusters of similar articulation (e.g., iš-si-sukti [išs’isuk‘t’i --> is’isuk‘t’i] ‘to extricate oneself’; cf. LG 1997: 74), and regularly unifies consonants differing by the feature [± palatalized] according to the palatalized vs. non-palatalized character of the final consonant, which, in turn, depends on the following [± front] vowel (e.g., [k’ir’p’t’i] ‘to cut hairs’ vs. [k’irpo] ‘s/he / they cut.past hairs’). The rule is thus basically phonological. Contrastingly, in Slavic contact varieties such rules are usually conditioned morphologically so that clusters combining palatalized and non-palatalized consonants irrespective of the ±front character of the following vowel are frequent (e.g., Russ. [vzgl’at] ‘glance, look’). It seems plausible to assume that these rules get blurred in Baltic-Slavic contact. Indeed, Turska (1995 [1982]: 63-68) noticed Lithuanian-like assimilations in the Polish speech of her pre-war multilingual informants. Furthermore, the palatalization of Lithuanian consonants does not know restrictions and is, to a much larger extent (compared to neighbouring Slavic languages), ruled phonologically (by the [± front] character of the following vowel). On the other hand, the otherwise consequent opposition of palatalized : non-palatalized consonants in all Slavic varieties of the BSCZ has exceptions (e.g., [š] and [3] are never palatalized). Again, palatalized pronunciation of such consonants has been observed for Belarusian dialects in contact with Lithuanian; cf., for instance, Lekomceva (1972: 128) and Sudnik (1975: 186).

(c) In Lithuanian, the phonologically ruled alternation of palatalized vs. non-palatalized consonants in many cases has resulted from the elimination of [j] after the respective consonants, while *[tj, dj] have changed into [t‡š, d‡3] (cf. Otrębski 1968; Zinkevičius 1988). Both the distribution and the diachronic structural background for these alternations differ from the structural and contact-induced factors leading to the [± palatalized] opposition in North Slavic and Lithuanian conceived of as the center of an area with unitary innovations (cf. Holzer 2001: 35, 43, 47; Stadnik 2002: 173-177). However, this has not prevented the phonotactic rules of assimilation (including palatalization) from gaining different status in the phonology of the involved Baltic and Slavic varieties (see above). Moreover, since the results of these changes often look alike in the contemporary contact varieties of the BSCZ, it becomes easy for bi-/multilingual speakers to identify them so that analogical leveling is likely to occur. This issue has remained almost uninvestigated.

(d) Differences in morphonological consonant alternations exist also between East Lithuanian dialects and Latgalian. The latter appears to be much closer to Belarusian and East Slavic in general (cf. Nau 2011: 17f.). Together with Latgalian vowel alternation types (see below), this suggests that different types of contact with Slavic have led to inner-(East)Baltic differences of morphonological processes which in part have brought different Baltic varieties of the BSCZ closer to their Slavic neighbors and have dissimilated them from their genealogically closest “sisters”.

(F3) Vowel alternations

In Lithuanian, vowel alternations abound both in inflectional and derivational morphology. For the most part, they either represent ablaut phenomena inherited from IE or have become conditioned by stress and apophony (cf. LG 1997: 61-73). In contrast to this, East Slavic and Polish vowel alternations are of a very restricted and predominantly morphonological nature: Ablaut phenomena can be conditioned only by a fronted vowel, e.g. in Polish (e.g., świat ‘world’ -->w świec-ie ‘in the world’). Otherwise, i.e. unless conditioned by stress (e.g., akan’e; see F1), they seem to always co-occur with an alternation (or addition) of suffixes. Most prominently, this combined alternation figures in the derivation of ipf. verb stems from pf. ones, e.g. Pol. zarob-i-ć -->zarab-ia-ć ‘to earn (money)’. Recently, Nau has found a very similar combination of alternations in Latgalian: If the root is enhanced by a suffix that begins with a front vowel, back or central root vowels get fronted and the root-final consonant is palatalized and predictably alternates in line with inner-(East)Baltic sound shifts distinguishing Lithuanian and Latvian (g > dz etc.); e.g. zyrg-s 'horse' > zirdz-eņš ‘little horse’. In the formation of new stems and the building of word-forms we thus observe a strive for harmony as a strong tendency: suffixes with a soft shape combine with soft stems and suffixes with a hard shape combine with hard stems (Nau 2011: 17f.). This tendency is of merely phonological nature while, in the Slavic varieties, which have been in contact with Latgalian, superficially and functionally similar alternations depend on morphological conditions. Such observations allow for the hypothesis that this type of vowel change (not the specific vowels involved) was calqued from a Slavic (Polish?) model, but this process has become more pervasive in the morphological and syllable structure of the Latgalian replica than in the Slavic model(s).

2.1.3. Morphology (F4-F6)[1]

(F4) Gender (different stages in the loss of the neuter)

With regard to partial or total loss of the neuter gender the varieties of the BSCZ can clearly be seen as a transitional region within a larger area stretching from the gender-less Finnic languages in the North(east) including the more typical (“deep”) varieties of the Livonian dialect of Latvian (dialect of Kurland, i.e. outside the BSCZ) to dialects with three target and control genders in the South(west), above all Polish. Latvian and Latgalian have lost the neuter gender altogether. Lithuanian has retained its remnants in predicative adjectives and participles while Old Prussian, an extinct West Baltic language that was spoken at the edge of the BSCZ, not only preserved several neuter nouns, but also even showed (in some texts of the 16th-17th cc.) agreement between a modifier and its NP along neuter gender, though the neuter gender was disappearing as well (Petit 2010). Northern Belarusian dialects show various degrees of this loss; it obviously begins with a loss of neuter-specific endings in the plural (this loss has entered into the standard grammar). In more advanced stages, encountered marginally in regional Polish varieties (‘polszczyzna kresowa’), agreement with adjectives and in pronominal anaphora show that former neuter nouns have been reinterpreted as masculine or feminine. This happened in Latvian, Latgalian, Lithuanian and Old Prussian, too (Petit 2010). At different spots, Belarusian shows these patterns to differing degrees and with varying consistency; their consistency appears to increase toward the north(east). In the BSCZ proper, this loss may be partially explained by a Baltic (Lithuanian) substrate or adstrate (cf. Wiemer 2004: 511-515). Nothing specific is known about neighboring Russian dialects of the Pskov-Novgorod region, but lack of agreement in resultative constructions in the dialects farther to the north (e.g., Trava.fem skošen / skošeno ‘The grass is [= has been] mowed‘) suggests that these dialects have at least partially lost their neuter due to a Finnic substrate (cf. Wiemer/Giger 2005: ch. 3).

(F5)      Derivational affixes

It is certainly derivational morphology that is the most prominent common structural feature of Baltic and Slavic; coincidences concern both the functional range of derivational affixes in all main parts of speech and their etymology (cf., among others, Stang 1966: 18, Martynov 1973). Calquing and even borrowing (= PAT and MAT in the terms of Matras and Sakel; cf., e.g., Sakel 2007) of derivational affixes have often been observed in Lithuanian varieties of the BSCZ, their token-frequency is high (for the hitherto most systematic investigation cf. Wiemer 2009). In turn, borrowing or calquing of Lithuanian affixes in Slavic varieties have hardly ever been observed, and a perusal of transcripts of speech printed in published work hardly yields examples either. Apart from this asymmetry, calquing and even borrowing are often difficult to establish because many derivational affixes in Slavic and Baltic are cognates (the most prominent example is the prefix Balt. da- / Slav. do-, cf. Breidaks 2007 [1972], followed by Balt. pa-, pri- / Slav. po-, pri-). Moreover, whether cognates or not, the analysis of Lithuanian speech in islands in or at the border with Belarus makes one believe that multidialectal Lithuanian speakers use both roots and affixes of either Baltic or Slavic provenance as though they belonged to a set of mutually exchangeable morphemes at one’s free disposal, i.e. to a common lexical diasystem. Polysemy patterns that copy Slavic models arise for both roots and affixes as well as for their combinations (extended stems) (cf. Wiemer 2009); cf. also Wiemer et al. (2004) for two modals.

Polysemy copying was described by Breu (2003) for Molisean Slavic. Heavy polysemy copying was also focused on by Weiss (2009) concerning German and Polish modals, by Giger (forthcoming) in his research on the recipient passive in West Slavic (particularly the Sorbian languages) and by Nau (2012**) regarding the example of convergent lexicalization patterns of modal auxiliaries and particles in the northern part of the BSCZ. But, so far, no-one has dealt more systematically with polysemy copying in the domain of word formation and derivation, which is so proliferous in the Baltic and Slavic languages and obviously highly sensitive to contact-influence both on the level of form (morphemes) and of meaning (function). Gast/van der Auwera (forthcoming) have proposed ‘semantic map assimilation’ as a model for capturing polysemy copying (in affixal derivation, among others) that works mutually between contact varieties in Mesoamerica. As the processes and results of polysemy copying in the BSCZ seem to be asymmetrical (see above), they yield an excellent playground for putting Gast/van der Auwera’s model to the test by applying it to a much richer variation of morphological and semantic influence in a different constellation of contact.

(F6) Verbs marked with a reflexive marker (RM)

The distribution of the RM as a valency-changing and word-formation device is well known from Geniušienė‘s (1983; 1987) seminal work in at least Russian and Lithuanian (to a large extent also in Latvian); on Lithuanian cf. also Holvoet/Semėnienė (2004: 41-54) from a slightly different perspective. Nevertheless, the functional distribution of RM-verbs in Slavic and Baltic dialects has remained uninvestigated. As for contact phenomena, the occurrence of RM-marked transitive verbs used as genuine passives (otherwise non-existent in Lithuanian) has been noticed casually for Lithuanian dialects obviously influenced by Belarusian, as has the use of the RM to mark reflexive-benefactives (productive in Lithuanian but almost inexistent in East Slavic and Polish) in the Belarusian speech of Belarusian-Lithuanian bilinguals in Hervjaty (Lith. Gervėčiai) (Wiemer 2004: 502f. with further references).

From the morphotactic perspective, (East) Slavic influence has been made responsible for the violation of the tmesis-rule (by which the placement of the RM abides in standard Lithuanian) in Lithuanian dialects of the BSCZ (e.g., až-juok-ė-si instead of už-si-juok-ė ‘(s/he, they) laughedʼ; cf. Wiemer 2009: 354 and passim, with further references). The question is whether the violation of this rule could be explained by a conspiracy of contact and internal factors, namely, an expectable tendency along a morphologization cline: tmesis reflects an intermediary stage between cliticization and agglutination, its abandonment results in an agglutinated postfix (as has happened to East Slavic and Latvian RM); thus the development is toward greater morphological tightness, known from grammaticalization. Moreover, in the whole BSCZ tmesis of the RM has been preserved only in standard Lithuanian (and varieties close to it) and in central Latgalian (apart from Curonian) varieties, whereas several other Latgalian varieties show an intermediate stage adding the RM either twice (into the tmesis slot and after the ending) or randomly to either the tmesis slot or to the slot after the ending (Endzelīns 1951: 910).

(F7)      Animacy distinctions in morphosyntax

Slavic languages encode distinctions on the animacy scale for some morphological NP types via differential object marking (DOM) as well as by differential subject marking (DSM). In general, DSM has been related to so-called ‘quirky subjects’. In Slavic however DSM no “quirkiness” of subjects is involved, since the variation results only from a redistribution of endings of the nominative case.

DOM and DSM have been the subject of cross-linguistic reseach in recent years (cf., inter alia, Aissen 2003). In contrast to other languages in which DOM is conditioned by multi-dimensional factors (e.g., definiteness and/or specificity, or discourse prominence; cf. Bossong 1998; Leonetti 2004; von Heusinger/Kaiser 2007), the peculiarity of the Slavic group lies in the fact that animacy has become the exclusive triggering factor in DOM and DSM. Remarkable, furthermore, is the fact that the types of DOM and DSM explained below are not at all encountered in Baltic. This also holds true for Baltic varieties which have otherwise experienced heavy Slavic influence on various levels of their grammar; the only exception might be Latgalian (N. Nau, p.c.).

Concerning the morphological level, there is a considerable degree of inner-Slavic variation  regarding whether animacy distinctions are marked only in the accusative (= prototypical object case, acc=gen, i.e. the genitive ending is used as marker of the accusative) or also in the nominative plural (= prototypical subject case; special case-number endings for virile or animate masculine nouns). This variation becomes particularly obvious in the Slavic varieties of the BSCZ and is reflected in rural Belarusian dialects (cf. Wiemer 2004: 507-511; Erker, forthcoming). Furthermore, the degree to which these distinctions have proceeded in the lexicon along the animacy (or ontological) hierarchy differ, too; it is in particular local Belarusian varieties that we find different manifestations between the systems of standard Russian and standard Polish (Wiemer/Erker, forthcoming: 2.2.2). A thorough check of animacy marking along the ontological hierarchy and with all contact varieties under control is necessary also because this evolution has shown to be non-unidirectional, and its regional manifestation in the BSCZ and adjacent areas has been subject to changing motivations, including interference with the lexically conditioned choice of case (akk vs. gen) required by verbs of different semantic classes and/or by quantification (cf. Kucała 1978; Krys‘ko 1994; 22006). The latter changes have also affected Lithuanian and their semantic motivations are also alive in Finnic,(cf. Holvoet 1991: ch. 6-8 + 13; Koptjevskaja-Tamm/Wälchli 2001: 646-656; Ambrazas 2006: 280f.). Finally, in Slavic we encounter allomorphic variation depending on which forms of which former declensional classes partake in animacy marking, especially in the nominative plural (-i/y vs. -ov’e etc.).

As for the diachrony of animacy-driven DOM (and also DSM), one may state that the acc=gen-assignment began with those object NPs that were high on the animacy scale in Slavic and morphologically did not distinguish between accusative and nominative. However, the distinguishability between two arguments (A and O) of a transitive clause (as formulated in Comrie 1989; Dixon 1994; Woolford 2001; de Hoop/Lamers 2006) does not explain the whole story and animacy-based distinctions must have been superior (cf. Kucała 1978: 162-166, Krys’ko 1994: 156-160 for a detailed argument). There is, furthermore, an implicational hierarchy: object sg.m (historically o-declension) >object pl>subject pl predicting that if one of these positions distinguishes animacy, then the positions left to it are sensitive to animacy as well (but not vice versa). This hierarchy can again be refined in the plural (particularly for acc.pl) by distinguishing the degree to which the distinction proceeds down the animacy scale.

A totally different type of an animacy-driven DOM is found with so-called impersonals (i.e. in passive-like constructions without a structurally assigned nominative) in the whole Eastern part of the CBA (Baltic, Finnic, Old and contemporary North Russian): The accusative object alters with the nominative object if it is expressed by a personal pronoun (cf. Timberlake 1974); see also F 12.

 

2.1.4. Syntax (F 8-12)

(F8) Verb particles

Verb particles are a relatively new feature in both Baltic and Slavic where they “rival” with prefixes as bounders and lexical modifiers, which are an inherited feature. According to Wälchli (2001: 414) verb particles are a good example of bidirectional transfer of language structure between Southern Finnic and Baltic. Finnic and Low German can be seen as models for the calquing of particles as modifiers of verbal action (‘satellites’ in Talmy’s terminology) in Baltic (in particular Latvian) while, reversely, Latvian triggered the evolution of such verbal particles from merely lexical modification toward bounders (in the sense of Bybee/Dahl 1989) in Estonian (Wälchli 2001: 433). The spread of both verbal particles and prefixes now constitutes overlapping clines running into each other (roughly NE—SW); due to both token and type frequency, Estonian and Latvian have to be considered as the epicenter of this complex cline. East Slavic and Polish have been affected by this cline, too, but West Slavic (i.e. also Polish) has additionally been influenced from the West by German so that the Slavic and Baltic region between Germanic and Finnic finds itself “encircled” by languages with a predisposition to verb particles (Wiemer, forthcoming c). In the midst of this SW—NE continuum, we find the BSCZ; transitional between the epicenter and the BSCZ are northern Lithuanian dialects. The exact distribution and frequency of verb particles in Latgalian is unknown (N. Nau, p.c.); the same holds for NW-Russian dialects.

 

(F9) Constructions containing non-canonical realizations of the highest ranked argument (HRA)

A bulk of predicates of the Eastern CBA encode non-prototypical transitivity, which typically surfaces as lack of control over the subject-like participant, not on a constituent of the VP but rather by assigning an oblique case to the HRA. To a certain extent, this subclass of predicates is productive in the area (Seržant, to appear a). This is a feature that is, in fact, atypical for accusative languages (Malchukov 2006). One can preliminarily distinguish between the following types/classes: (i) different kinds of experiencer predicates with DatExp-Nom/Acc/GenStim case frame including verbs of lacking and succeeding; (ii) the ”ad me (est)” possessive construction (cf. Clancy 2010); (iii) a clausal case frame for which inanimate-Actor demotion and non-promotion of the second argument of a causative predicate is characteristic; e.g. Russ.

 

(1)           Lodku        uneslo                                  tečeniem.ins

boat:acc    carry_away:pfv:past.neut.sg    stream.instr.sg

'The boat drifted down the stream’ (lit. ‘(it) brought the boat down with/by the stream’).

 

(iv) genitive of indefinite quantity and the genitive of negation that override canonical subject marking; (v) different kinds of modal constructions (typically deontic) that invert the case frame of a verb and assign dative to the HRA of the verb; (vi) the so-called possessive perfect that assigns dative or dative-like case (or PPs) to the HRA of the lexical verb. Non-canonical lexically conditioned case-marking on the HRA itself does not sufficiently justify the assumption of contact-induced convergence. This changes if we account for the fact that these predicates in the BSCZ not only show a considerable diversity and number, but, at the same time, also semantic (functional) and, partially, morphosyntactic coherence across languages (Seržant, to appear a). In the BSCZ+Finnic (= Eastern Circum Baltic) area, very little attention has been paid to experiencer predicates (see type (i)) that attest a development toward losing subject properties by their nominative stimulus argument. This tendency represents an areal phenomenon. It is slightly less well attested in Lithuanian (Seržant, to appear d) and in Russian (Kozlova 2009; Bonč-Osmolovskaja et al. 2009).

 

(F10)Resultatives, Perfects, Evidentials

The languages of the Eastern CBA show striking similarities among their perfect constructions both from a diachronic and a synchronic perspective. All perfects attested in the area are formed on the basis of copular constructions containing a resultative participle: while active, or agent-oriented, participles (< *-u¾es- for both Baltic and Slavic, -l- for Slavic and -nut-/nyt- for Finnic) yield active perfects, Undergoer-oriented (or ‘inactive’) participles (in Baltic with -t-, in Slavic also -n-suffix, in Finnic with -ttu-suffix) yield passive or impersonal perfects (cf. Wiemer 2006b; Wiemer/Giger 2005). The languages differ as to how many morphosyntactic active properties they assign to the impersonal construction: Thus Polish displays active morphosyntax (Kibort 2004; Wiemer, forthcoming) – and it is not a perfect at all – while Belarusian or Ukrainian show a more archaic state with several passive properties (Lavine 2005); the Lithuanian counterpart is probably an adjustment to Polish (Danylenko 2006). In the literature there is an ongoing debate about the exact paths of the acquisition of the accusative case-marking by the Undergoer argument across the languages and about how it ceased to be syntactically promoted in this construction. Since it is Polish that historically first attests accusative case-marking, it has often been assumed to be the source for accusative case-marking in Ukrainian, Belarusian (Shevelov 1991: 201; Moser 1998: 340) and Lithuanian (Danylenko 2006) whereas natural drift towards acquisition of prototypical object case-marking and/or loss of the ability to be syntactically promoted - i.e. an independent development - is assumed for Ukrainian and Belarusian in Danylenko (2005a: 364f.), Jung (2007: 150; 2009: 214) and for Lithuanian in Fraenkel (1928: 97f.) and Holvoet (2001).

Another property of the perfect constructions in Baltic and Estonian – but not in Slavic (even in intimate contact with Baltic or Estonian) – is their extension to evidential functions. The development most probably started with perfect active-participles (Wälchli 2000). In Lithuanian, the impersonal construction has also been reinterpreted as an evidential, but it is predominantly used with inferential function (Wiemer 2006a; 2006b: 275-277, 299-303; 2007). We thus observe a clear distribution by subareas (and languages), which “harmonizes” with the findings in Stolz (1991) concerning the particularly close assimilation of Latvian and Estonian.

Furthermore, the impersonal construction attracts non-prototypical agents that can be encoded by means of an oblique (cf. the agentivity hierarchy of the cases/PPs used in Danylenko 2005a). The generalization of a dative-like case (or PP) can be found in North Russian, Latvian, Estonian, Karelian and Votian, while the Lithuanian evidential (with ta/ma-participles; cf. Holvoet 2007: 92ff; Wiemer 2004: 310-313; 2006a; 2006b; 2007) employs a genitive of adnominal origin. It has to be stressed that these constructions are not passives (Holvoet 2007: 92ff; Lavine 2010).

 

(F11) The genitive of indefinite quantity (GIQ) and the genitive of negation (GN)

A firmly established feature of the Eastern part of the CBA (including the BSCZ) is the morphological encoding of partial affectedness: structural case assignment is overridden by specific case-marking patterns (cf. Koptjevskaja-Tamm 2001; Koptjevskaja-Tamm/Wälchli 2001: 649-669). Although GIQ and GN show some functional overlap (e.g., both denote decreased referentiality) and might have originated from the same source category, they have been entrenched to different degrees in the languages of the area. One may, therefore, dare to say that they have begun to form intersecting, though different clines (or clusters) within that area.

The GIQ is most pervasive in Finnic and Lithuanian whereas it is less frequent in Standard Russian, even less frequent in Polish and almost extinct in Latvian. The Baltic and Slavic GIQ show traces of an accommodation to the Finnic pattern (cf. Seržant, to appear c). The GN, in turn, is most entrenched in Polish followed by Lithuanian and Latgalian; it is largely optional and triggers different implications in Standard Russian (Timberlake 1986; Babby 2000; Partee&Borschev 2004; Padučeva 1997; 2005, among others) and is almost extinct in Latvian. Both GIQ and GN are less restricted in Northwestern Russian varieties as well as in Belarusian (Karskij 1956: 319, 403; Filin 1972: 514-5).

 

(F12) The nominative object (NO)

In general, the NO is restricted to those contexts lacking an argument and which could be structurally assigned nominative. It is typical for, but not restricted to, verbal non-finite and nominal predication. To some extent in Old Russian and generally in Finnic it also includes the imperative. Timberlake (1974) demonstrated that NO is a feature pertinent to the whole Eastern CBA (i.e. BSCZ, NW-Russian + Finnic). Although a Finnic substrate during the first millenium AD certainly played a role (Mendoza 2008), the beginnings of the nominative marking itself are most probably not contact-induced, as pointed out in Ambrazas (2001). Instead, he assumes the morphological preservation and the syntactic reanalysis from subject to object area recurrent pattern of the NO and shows that a correct relative and absolute chronology of inherited vs. contact-induced properties is crucial. Nowadays, there is a strong tendency to replace the NO with the accusative. This appears to be motivated by the accommodation of the morphosyntactic encoding to the function (cf. Haspelmath’s 2010 Behaviour-before-Coding-Principle). However, there are also competing areal motivations; cf. Danylenko (2005b) who argues for the inheritance of the accusative impersonal in Belarusian and Ukrainian against calquing in Lithuanian.

 

 


[1] In general, and wherever possible, one should pay special attention to variation in the realization of structural (phonological, morphological, syntactic) oppositions. Such cases may often be described as allophonic or allomorphic variation. But this notion can turn out to be inadequate because, for instance, the realization of endings in nominal or verbal inflection of rural mixed Belarusian does not always seem to rest on (near to) complementary distribution. For this reason, we will carefully just speak of ‘variants’. It is essential that the concrete shape of variants can be used as a valuable indicator of contact-induced patterns, in particular within the same family, since we have to assume that speakers often do not just replicate or adapt oppositions from a model but rather borrow items together with their phonetic and morphological shape. This leads to ‘low-level variation’ whose account has the further advantage that token-frequency need not be high (cf. Cornips/Corrigan 2005: 102).