Tver’ Karelian as a new dialect

Abstract

In the article, the development of the Tver’ Karelian dialect is discussed. This new dialect has emerged from a mix of regional dialects of Karelian immigrants from the seventeenth century onwards. Characteristics of a new dialect in Tver’ Karelian are examined on the basis of demographic data and linguistic descriptions. In addition, the unity and internal variation of Tver’ Karelian as well as its relation to other Karelian dialects are taken into consideration. In regard to this kind of comparative study, the article reveals some significant regional shortcomings in the linguistic research material available. A discrepancy could also be found between the areal distribution of certain linguistic features of Tver’ Karelian and the main region of origin indicated by historical documents.

1. Introduction

Tver’ Karelian is a dialect of the Karelian language, spoken in the inner parts of Russia, separate from the original Karelian-speaking region in Karelia. ¹ The area in which it is located lies north of the city of Tver’, which is situated approximately 180 km northwest of Moscow. Tver’ Karelian is an enclave dialect that emerged as a result of migration of Eastern Orthodox Karelians in the seventeenth century. The County of Kexholm (Finn. Käkisalmi) was ceded from Russia to Sweden in 1621, and due to this change of power, a considerable proportion of Eastern Orthodox Karelians in the areas surrounding Lake Ladoga moved to Russia. The displacement of Karelians in the seventeenth century has fundamentally shaped the population and linguistic conditions in Karelia, for example by creating new dialect areas for both Karelian and Finnish. (For the Karelian migration, see Žerbin 1956, 1987; Kuujo 1963; Saloheimo 1973, 2010; Novak 2016:10–12; Koivisto 2018:61–62.)

In the Tver’ region, the population of migrants grew into a vital linguistic community within a few centuries. The aim of this article is to shed light on the historical background of Tver’ Karelian and on the position of Tver’ Karelian among the dialects of the Karelian language. I will discuss the development of Tver’ Karelian into a separate, somewhat different dialect from the rest of Karelian. The examination is based on the fact that Tver’ Karelian dialects have, in practice, developed out of various regional materials representing the individual dialects of people who migrated from different parts of Ladogan Karelia. This type of distinctive, new language variety has been called a new dialect, and I will examine characteristics of a new dialect in Tver’ Karelian mostly on the basis of dialectological sources.

The structure of the article is as follows. Section 2 provides an overview of the Karelian language and language contacts of Karelian. In Section 3, I will present the methods and materials, in particular from the perspective of the theory of new dialects. In Section 4, the settlement history of Tver’ Karelia and the initial situation for the development of the Tver’ Karelian dialect are explored. In Sections 5 and 6, I will examine the development process of a new dialect and its manifestations from the perspective of Tver’ Karelian. Finally, the conclusions are presented in Section 7.

2. On the Karelian language and language contacts

Karelian is a Finnic language spoken in Russia and Finland. Other Finnic (formerly called Baltic-Finnic) languages include Finnish and Estonian as well as Veps, Ludic, Ingrian, Votic, and Livonian. With the exception of Finnish and Estonian, all Finnic languages are now endangered and to some extent already extinct. Karelian is the language most closely related to Finnish, sharing features with the eastern dialects of Finnish in particular. The Finnic languages form a geographical and linguistic continuum, which originally extended along the shores of the Gulf of Finland, proceeding on the northern shore from the western and subsequently eastern dialects of Finnish onwards to Karelian, Ludic, Veps, and Ingrian and – past present-day St Petersburg – along the southern shore of the Gulf of Finland from Ingrian and Votic to Estonian and Livonian. The northeastern branch of the Finnic language family (including Karelian), in particular, has experienced a considerable areal spread during the past millennium, on which language contacts and country history have had a special impact (Grünthal 2020).

There are two main dialects of Karelian: (i) Livvi Karelian (or Olonets Karelian) and (ii) Karelian Proper. Karelian Proper is further divided into the northern dialect, North Karelian (or White Sea Karelian), and the southern dialect, or South Karelian. All these dialects were and are currently spoken in Russia in the Republic of Karelia (Figure 1). In Border Karelia (the hatched area in Figure 1) – the Karelian-speaking region in what was Finnish territory until 1944 – primarily South Karelian and to some extent also Livvi Karelian were spoken. These dialects affected each other in Border Karelia, and in fact they formed a dialect continuum there. There were also intensive contacts between the Karelian and Finnish languages in Border Karelia, as many Karelians moved to Russia in the seventeenth century and a Finnish-speaking population partly replaced them. (For recent overviews on the Karelian language, see Sarhimaa 2022 and Grünthal 2023; for further information on Border Karelian dialects, see Koivisto 2018.)

Figure 1. Karelian dialects and Ludic in the nineteenth century.

The Karelian enclave dialects spoken in Russia outside the original Karelian region are Tver’ Karelian, Tikhvin Karelian, and Valday Karelian (see Figure 2). All three dialects can be traced back to the seventeenth century migration of Karelians, and they are linguistically rather similar. They have traditionally been included in South Karelian (1b in Figure 2), based on their linguistic characteristics and knowledge of their regional origin. Tver’ Karelian is further divided into the southern Tolmachi and the northern Ves’egonsk dialects. In addition, there is a fourth enclave to the south of Tver’ Karelian: the Dërzha subdialect, which differs significantly from Tver’ Karelian and, in effect, all other Karelian dialects. (For Tikhvin Karelian, see Rjagoev 1977; for Valdai Karelian, see Palmeos 1962.)

Figure 2. Karelian language and its enclave dialects.

Russian has been a contact language of Karelian for centuries. Contact took place with the northwest dialects of the northern Russian dialect group, and, in recent centuries, with Standard Russian as well due to facets such as religion and formal education. Karelians’ contact with Slavic people has lasted for more than a thousand years and, for its part, contributed to the gradual differentiation and separation of the Karelian and Finnish languages, which share a common ancestral language. Historically, the Karelian language has been divided between regions belonging to Russia and Sweden (and later Finland), and the state border has largely also corresponded to the border between religions – Eastern Orthodox and Roman Catholic, the latter subsequently Lutheran. Linguistically, the Karelian language has existed and developed within the sphere of influence of Russian, whereas Finnish has been influenced by the Swedish language for centuries.

Tver’ Karelian has been an enclave dialect since the beginning, surrounded by a Russian-speaking area. Tver’ Karelians have apparently not had contact (at least to any significant extent) with Karelians of the other enclaves or with their regions of origin. Overall, it can be noted that a lack of mobility has been characteristic of Tver’ Karelians, and their language has not spread beyond its original boundaries formed in the seventeenth century. (For Tver’ Karelian and its history, see Virtaranta 1961, Virtaranta & Virtaranta 1986, Jeskanen 1998.) The Tver’ Karelian villages remained largely monolingual until the twentieth century, but subsequently Russian spread to the Karelian dialect area as the language of society, administration, and education, and has now almost completely displaced the Karelian language. In modern Tver’ Karelian, the influence of Russian is strong, especially regarding phonetics and phonology (Novak 2019) but also in syntax and vocabulary (Tavi 2022). (For contact with Russian in the Dërzha dialect, see Kehayov et al. 2021.) Tver’ Karelian differs from other Karelian dialects in that it lacks influence from the Finnish language. In the original Karelian region, the dialects of Karelian were exposed to Finnish to varying degrees, particularly from the seventeenth century onwards. The Finnish influence was at its strongest in Border Karelia, where Finnish was spoken alongside the Karelian language until the twentieth century.

It is estimated that there are currently some 20,000 to 30,000 speakers of the Karelian language, a few thousand of whom are in Finland. Since the second half of the twentieth century, the size of the Karelian speech community has been in rapid decline. In the 1930s, Tver’ Karelian had up to 150,000 speakers; at the present time, the numbers have declined to perhaps only a few thousand (for the number of speakers, see Novak 2016:13–15, 203), although the estimates are not precise by any means. The Karelian language spoken in the Republic of Karelia has seen a similar development. The number of Karelian speakers has also decreased in Finland: from over 30,000–40,000 speakers in Border Karelia before the Second World War to a few thousand. After the war, the Finnish Karelian-speaking community was scattered around the country, as Border Karelia was ceded to the Soviet Union and its inhabitants were forced to leave their homes.

Estimating the current number of Karelian speakers both in Russia and in Finland is complicated by factors such as bilingualism, the high variability of proficiency in Karelian, the difficulty of assessing individuals’ language skills, and both the gaps in information and the shortcomings of the data collection concerning Karelian speakers. For all of its speakers, the Karelian language and its use are characterised by bilingualism, where the official language of each country, i.e. Russian or Finnish, is the dominant one. The declining use of the Karelian language over the past century has led to the present situation in which Karelian is classified as ‘definitely endangered’ by UNESCO (Karjalainen et al. 2013:9). This involves a rapid linguistic and ethnic assimilation and eventually a comprehensive language shift.

Since the 1980s, there have been attempts to revitalise diverse dialects of Karelian both in Russia and in Finland. As a part of this effort, standard written languages have been developed for North Karelian and Livvi Karelian in Russia, and a standard is presently being created for South Karelian in Finland. Tver’ Karelian has existed almost exclusively as a spoken language, with the exception of a short-term standardisation project in the 1930s and recent publication activities that sparked in the 2000s. (The linguistic situation of Karelian in Russia is described by Karjalainen et al. 2013 and of Karelian in Finland by Sarhimaa 2016.)

3. Methods and materials

This article combines linguistic facts with the history of Tver’ Karelia and examines their relationship within the theory of new dialects (Trudgill 2004). This examination is based on the key principles of historical linguistics and, on the other hand, research data on new dialects and their typical features and ways of developing. The general background consists of a current understanding of the history of the Karelian language and its relationships with its closely related and neighbouring languages.

In this article, the basis for the study of Tver’ Karelian is historical documentation on the migration of Karelians to Tver’ in the seventeenth century. There is fairly detailed region-specific information on this migration, based on written documents (Žerbin 1956; Saloheimo 1973, 2010). With knowledge of the places of origin and destination and even the number of migrants, these background factors create a rare indication of the starting points of the emerging new dialect. In fact, previous research into Tver’ Karelian has already introduced the methodologically interesting concept of combining data on settlement history with linguistic data, with the aim of examining the dialects’ history of emergence and the related regional features (Sarhimaa & Siilin 1994:269, Joki & Torikka 2001, Kehayov et al. 2021).

Demographic data on the numbers and backgrounds of migrants is considered essential for understanding the development of a new dialect (Dollinger 2008:67). The preconditions for the development of such a dialect are fulfilled in this situation: a multi-dialectal background, the migrants arriving in an area where there were previously no Karelian speakers, and having limited contact with the rest of the Karelian language. This article will focus on investigating the degree of internal variation in modern (i.e. twentieth-century) Tver’ Karelian; a low degree of variation is typical of new dialects in general.

Linguistic materials presented a challenge for the research setting of this article. The tradition of Karelian language studies is not very old. The first studies on Karelian date back to the nineteenth century, and more extensive language materials (texts or vocabulary collections) have only been collected since the end of the nineteenth century. In the twentieth century, text material of all Karelian dialects has been recorded in writing or on tape. There are also extensive lexical sources on Karelian, the most important of which is the Dictionary of Karelian (Finn. Karjalan kielen sanakirja) and its archive, which contains more than 30,000 lexicographical sample cards from Tver’ (Joki & Torikka 2001:464). This accounts for 5% of all lexical material in the archive. ²

The present study’s materials comprise general linguistic descriptions of Karelian as well as dialect geographical studies. In addition, I have collected pieces of information from separate studies on Karelian that, in my view, can shed light on the theme of this article. The main source for the study of the variation and uniformity of Tver’ Karelian is the Dialect Atlas of Karelian (Finn. Karjalan kielen murrekartasto, Bubrih et al. 1997), which comprises 209 dialect maps, mostly covering phonological and morphological features. For Tver’ Karelian, the Atlas includes 35 observation points (villages), and one on the separate Dërzha enclave (marked ‘Zubtsov’ in the maps). ³ In addition to numeral calculations concerning the degree of variation in light of the maps, the method involves overall ‘reading’ of the maps in creating a picture and forming generalisations based on them.

Unfortunately, the Atlas does not contain information on the other enclave dialects (Tikhvin and Valday Karelian) or from Border Karelia, which was Finnish territory in the 1930s when the Atlas materials were collected and therefore not within the reach of Soviet researchers. These regional shortcomings in the material are significant and even fatal in terms of this article’s theme, as Border Karelia is a key region of origin for Tver’ Karelian. Although the dialectal distribution in Karelia has changed in recent centuries, it would be fundamental to compare Tver’ Karelian to the Karelian dialects spoken later in the original area. The two parallel enclave dialects of Karelian would also be important points of comparison for Tver’ Karelian due to the similar background of these enclaves as well as how they emerged.

So far there has been no extensive, consistent research on Tver’ Karelian (or other enclave dialects) or on the subsequent dialects in the region of its origin, the Border Karelian dialects. Despite the shortcomings mentioned above, among the existing linguistic descriptions, the Dialect Atlas of Karelian is the one that offers the best possibilities for a comprehensive examination of the entire Tver’ Karelian dialect area and its comparison with other dialects of Karelian.

The examination of the development of Tver’ Karelian as a new dialect would ultimately require real-time language material from the 1600s and 1700s as well, but no such material exists. At its oldest, Tver’ Karelian is preserved in a translation of the Gospel of Matthew from 1820 (which was reprinted and published in Finland in Ahlqvist 1865; see Haltsonen 1942:176–177). Accordingly, research into the development of Tver’ Karelian must be based on other, more indirect means, which is, in fact, typical of all Finno-Ugric languages that do not have a literary tradition and have been used exclusively as spoken languages. Gathering information for the purposes of historical studies in Karelian is generally challenging due to the limited amount of research focusing on Karelian.

4. Settlement history and the starting points for the development of Tver’ Karelian

The Karelian enclave dialects differ from other Karelian dialects with respect to their background and development. Tver’, Tikhvin, Valday, and Dërzha Karelian emerged due to the mass migration of Karelians to inner Russia in the seventeenth century. A more recent and geographically more delimited enclave dialect of Karelian is Kolvitsa Karelian in the Kola Peninsula, a mixed migrant variety which developed from dialects of individuals speaking North (White Sea) Karelian, in isolation from other forms of Karelian Proper (Kehayov & Kuzmin 2022: 47–51). Parallel cases of enclave dialects in the Finnic languages include the Värmland Finnish dialect area in western Sweden, which emerged through migration from the Finnish region of Savo in the seventeenth century, and Kreevin Votic, a dialect that became extinct in the nineteenth century. This latter dialect was based on the language of the Votes who ended up in Latvia as prisoners of war in the fifteenth century.

4.1 Karelians’ migration

After the County of Kexholm was ceded to Sweden in 1621, a large number of Eastern Orthodox Karelians living in this territory, situated in the area surrounding Lake Ladoga, moved to Russia. The migration of the Karelians to Russia was expedited by unfavourable measures that the new Swedish regime directed towards them, such as enforced conversion to Lutheranism, increased taxes, restrictions on trade, and a generally negative attitude towards people of the Orthodox faith (Virtaranta 1961:36–38). In contrast, beyond the border in Russia, there was living space and farmland, freedom to practise the Orthodox religion, and seven years’ tax exemption for immigrants granted by the Russian administration (Kuujo 1963:54). Karelians living north of Lake Ladoga had been forced to live in a state of constant social unrest and uncertainty after the end of the sixteenth century. In fact, their migration started at that time and continued until the latter half of the seventeenth century. There are documents stating that the last migrants settled in the Tver’ region in 1678 (Karjalainen 1932:537). The greatest migration waves took place after the Treaty of Stolbovo in 1617 and after the war between 1656 and 1658. A large proportion of the first migrants stayed near the border, in Olonets County on the Russian side, but some headed further to the southern shores of Ladoga or towards Moscow, reaching, for example, Tver’ (Kuujo 1963:58).

By the end of the seventeenth century, some 25,000 to 30,000 Karelians, perhaps up to 50,000 of them, had moved to the Tver’ region (Virtaranta 1961:33, 35–36, 1970:461, 463; Saloheimo 2010:11–12). The emergence of the Tver’ Karelian dialect started when the population of speakers of Karelian settled in the new area, cutting off contact with the regions of origin and becoming linguistically intermingled in their new place of residence. On the other hand, the linguistic environment remained at least partly the same; even though each new Karelian village would usually have residents originating from several parishes or villages, neighbours would often make the trek of more than 500 kilometres as a group and end up living close together in Tver’ as well. Still, more than half of the migrant families ended up separated from their former village neighbours (Saloheimo 1973:51–54).

4.2 The linguistic starting point

Based on historical documents on the migration of the Karelians, the linguistic starting point for Tver’ Karelian included the Karelian dialects that were spoken in the Karelian-speaking zone in the territories of Sweden in the seventeenth century, stretching from Border Karelia westwards to the County of North Karelia ⁴ in present-day Finland, and there were also speakers of Karelian living on the northwestern and western coast of Lake Ladoga. Evidence of Karelian being spoken in these regions consists of seventeenth-century documents on Eastern Orthodox Karelians who migrated from the area and, on the other hand, certain Karelian features in the Finnish dialects spoken in these areas in later times.

Delineating a picture of the Karelian language spoken by the seventeenth-century migrants is a challenging task, as the process must rely on modern languages (instead of historical, real-time linguistic material): namely Tver’ Karelian and, respectively, the Karelian dialects spoken in the regions of origin in the twentieth century. There has obviously been further development in both Tver’ and other Karelian dialects, and it is not possible to create a very detailed impression of the original dialects on the basis of their modern forms. Furthermore, the linguistic situation in Border Karelia and its development from the seventeenth century onwards is hard to trace because of demographic and societal factors. After the heavy decline in the population due to the migration, the linguistic continuity was further fragmented in the 1700s, a century characterised by constant unrest and wars in the region.

The migrants’ dialectal variation is a key prerequisite for the assumption of the development of a new dialect. On the basis of historical documents (see Section 3), the roots of Tver’ Karelian can be traced to a variety of South Karelian that was spoken on the Swedish side of the border, i.e. Border Karelia. The dialects of those who inhabited Tver’ in the seventeenth century displayed in all likelihood a considerable amount of variation. There is good reason to assume dialectal variation in the regions of origin, as the area was quite large in geographical terms, albeit sparsely populated in the north. South Karelian is also rich in variation by nature, due to its multiple linguistic sources. However, the linguistic properties of Tver’ Karelian point not only to Border Karelia but also to the South Karelian dialect spoken on the Russian side of the border (see Section 4.3).

Another important trait for the dialectal background of Tver’ Karelian is the fact that two different dialects of Karelian were spoken in the regions of origin: predominantly South Karelian, but also Livvi Karelian in some parts of Border Karelia, particularly in the Salmi parish, as well as on the opposite side of the Russian border. As a dialect, the main features of Tver’ Karelian are linked to South Karelian, but there are also traces of Livvi Karelian, which further enriches the dialectal foundation of Tver’ Karelian. Some originally Livvi Karelian features even cover practically the entire dialect area of Tver’ Karelian (see e.g. Pahomov 2017:231, 235, 243).

Although Livvi Karelian was a regionally narrower dialect than South Karelian, it is possible that in some parts of Tver’ there was a higher percentage of Livvi Karelian speakers (see Sarhimaa & Siilin 1994:269, 272–274, Joki & Torikka 2001:467–469). According to Sarhimaa & Siilin (1994:274), in the 1980s and 1990s there were observable dialectal differences between features of South and Livvi Karelian in the villages of Tver’ that match what we know of the migrants’ dialectal background. This implies that, in Tver’, the linguistic differences between villages may at least partly reflect old dialectal differences between the regions of origin (Sarhimaa & Siilin 1994:269). At the same time, a Livvi Karelian feature in Tver’ Karelian may also originate in South Karelian speakers, as Livvi Karelian had strong influence on the South Karelian dialect long before the seventeenth century, possibly even as a substrate (Itkonen 1971:182, Larjavaara 1986:113–114).

4.3 The South Karelian origins of Tver’ Karelian

South Karelian has been influenced by contacts with its geographically close neighbour languages or dialects, Livvi Karelian and Ludic, which is reflected areally in South Karelian. Miikul Pahomov (2017:224) has used the contact features to divide South Karelian into two areal subdialects: South Karelian on the Ludic side (eastern) and South Karelian on the Livvi Karelian side (western). The parishes included in Ludic-side South Karelian are Suikujärvi, Tunkua, Paatene, Mäntyselkä, and eastern Porajärvi (see Figure 3). ⁵ The South Karelian area borders on the southwest with a Ludic speaking area, and there is demonstrable influence from Ludic (also as a substrate) in the eastern, Ludic-side South Karelian area (see Wiik 1998:19, Pahomov 2017). The parishes included in Livvi-side South Karelian are Repola and western Porajärvi as well as the parishes which were a part of Swedish, later Finnish, Border Karelia: Suojärvi, Suistamo, Korpiselkä, Impilahti, and eastern parts of Ilomantsi (see Figure 3).

Figure 3. Parish names (in Finnish) in the speaking areas of Karelian and Ludic.

Livvi-side South Karelian in Border Karelia represents largely the region where, according to documents, people left for Russia in the seventeenth century. However, it is noteworthy that Pahomov presents a number of linguistic features of Tver’ Karelian (and the other two Karelian enclave dialects) that are typical of Ludic-side South Karelian (see e.g. Pahomov 2017:224, 281–282). This is unexpected in that this South Karelian region remained continuously on the Russian side of the border even after the Treaty of Stolbovo in 1617, and migration from there to Tver’ is not known to have taken place, at least not on a large scale.

There is also information in individual studies on Karelian that refers to more eastern, Russian- and Ludic-side regions of origin for Tver’ Karelian. For example, Virtaranta (1961:42–44, 1970:465) has presented the distribution of the Tver’ Karelian sibilants /s/, /ʃ/, /z/, and /ʒ/ as in line with the representation of the dialect in the Paatene region. Larjavaara (1986:113, 332) has found the entire demonstrative pronoun system of all of enclave Karelian to be similar to that of South Karelian in the ‘Seesjärvi Southern Group’, which for example includes the Mäntyselkä dialect. The parishes of Paatene and Mäntyselkä both represent Ludic-side South Karelian and form a distinctive dialect group that differs in many respects from western South Karelian.

In addition, the variant ɨa of the diphthong ua (e.g. mua ∼ mɨa ‘land’; ɨ is a back variant of i) occurs in a limited area in the Seesjärvi South Karelian region as well as in part of Tver’ Karelian (and also in Veps) (Ojansuu 1907:112, Virtaranta 1970:461, Bubrih et al. 1997:Map 4). There are also features of Ludic and Veps languages widely existing in Tver’ Karelian (Ojansuu 1918:99–100, Itkonen 1971:182). They can be explained on the basis of Ludic-side South Karelian, which has adopted contact features from both Ludic and Veps (Pahomov 2017:265–268).

As shown, there is a significant contradiction between certain linguistic features of Tver’ Karelian (corresponding with those of Ludic-side South Karelian) and the main region of origin indicated by historical documents (i.e. Livvi-side South Karelian on the Swedish side of the border). There is no documentation – on the extent or numbers – on Karelians’ migration from the Russian side of the border. Instead, known historical records (see Žerbin 1956; Saloheimo 1973, 2010) indicate a major migration from areas that fell under Swedish rule (Kexholm County) in the seventeenth century. However, it is known on the basis of some documents and through Tver’ Karelian oral tradition that there was also migration from Olonets County on the Russian side of the border (see Karjalainen 1932:536, Haltsonen 1942:175, Virtaranta 1961:41–47). ⁶

In light of the numerous Ludic-side Karelian features in Tver’ Karelian, it was possible to reconsider and re-evaluate the current overall understanding of the regions of origin of Tver’ Karelian. Either a significant number of migrants originated on the Russian side of the border, or the characteristics of South Karelian, which was spoken in Kexholm County in Sweden in the seventeenth century, could have had more features of Ludic-side South Karelian. Then again, in the latter case, one would expect to continue to be able to see the features in question – at least to some extent – in the later dialects of Border Karelia, documented in the twentieth century. This, however, is not the case. A more plausible explanation would be that domestic migration within Russia was simply not registered officially.

5. Emergence and development of a new dialect

The study of Tver’ Karelian as a new dialect can be supported by evidence of other languages. An illuminating parallel is offered by Colonial Englishes (Trudgill 2004), which have emerged as a result of the migration of English speakers and are spoken around the world, including North America, Australia, New Zealand, and South Africa. These new world settlements were initially established by speakers of various local British Isles dialects, who created a new language community and whose individual languages were shaped by the following generations into a new, more unified local variety of English, i.e. a new dialect. These new dialects of the same origin are very similar in terms of their linguistic features, even in different parts of the world (Trudgill 2004:26).

The starting point for the development of a new dialect is a variety of dialects, linguistic diversity of the migrant group, and the resulting linguistic variation in the new community of speakers. What is also essential in the emergence of a new dialect is that the area where the migrants arrive is not inhabited by speakers of their native language or one of its cognate languages or dialects, and instead they arrive at a tabula rasa, which in a way creates ‘laboratory conditions’ for the emergence of a new variety (Trudgill 2004:26, Kerswill & Trudgill 2005:196). In the Tver’ Karelian area, migrants are known to have arrived at villages left empty by wars and plague. There were also no Karelians living near the Tver’ enclave. The migration to Tver’ had taken people far enough away from the original Karelian-speaking area, to which contacts had practically been severed. The migration was also temporally restricted in such a way that no new Karelian population migrated to the Tver’ region after the seventeenth century.

The new migrant community of speakers created a linguistic mixture where individuals from different places spoke mutually understandable but divergent dialects of the same language. However, in the case of Karelian, the differences in dialects and regional variants were surely not as great as in the varieties of English spoken in the British Isles. The first stage of a new dialect developing in a language community involves an ample variation of linguistic features, caused by the varying linguistic backgrounds of the individual speakers. The development and establishment of a new dialect hinges on the subsequent generation, the migrants’ children (Trudgill 2004:27, 29, 101), who have a wide choice in their language use considering the diversity of the previous generation’s language. Already in the second generation, individual variation decreases, with forms and features undergoing selection and a unification of representations taking place. By the third generation at the latest, i.e. in approximately one hundred years, the new dialect will be established (see Trudgill 2004:23, Kerswill & Trudgill 2005:200–201, Dollinger 2008:139). Trudgill (2004:31–33, 158–160) reports on spoken linguistic material from New Zealand that was recorded of immigrant adults with various dialects and the two generations following them. The report reveals the frankly drastic development of the language between the immigrating generation and the following two, in less than a century.

The principle leading to the equalisation, unification, and reduction of variation (principle of accommodation) in new dialects has, according to Trudgill, been the natural interactive tendency to ‘talk like others’ (Trudgill 2004:27). Language users have enjoyed ample freedom in their choices due to the lack of a peer group that they would have needed to adapt to (Trudgill 2004:103). The emergence of a new dialect is ultimately ‘a product of unconscious choices’ (Hickey 2003:215). The selection and adaptation of features from a high rate of variation to a lower rate is not systematic in any way and instead rather arbitrary (Trudgill 2004:158–159). However, in a situation with much variation, the feature used by the majority usually prevails (Kerswill & Trudgill 2005:148, 198).

With new dialects stabilising by the third generation, Tver’ Karelian would have been established by the beginning of the nineteenth century at the latest. By then it would have ceased to undergo the kind of radical change and unification that is characteristic of the formative phases of a new dialect. There is no linguistic material from the period of the development process, but the closest view of that time is offered by the translation of the Gospel of Matthew into Tver’ Karelian from the year 1820. It has been pointed out by Virtaranta (1973:319) and Gromova (2003:46) that the translation is linguistically very similar to twentieth-century Tver’ Karelian. This assessment would match a scenario in which the expected new-dialect development had taken place and already finished by the beginning of the nineteenth century.

6. On the uniformity of Tver’ and other Karelian dialects

According to the theory of new dialects, a dialect enclave will have less dialectal variation (both individually and regionally) compared to ordinary (baseline) dialects of the same language. As a new dialect it would be initially based on a larger range of dialects and variations, becoming more unified and less varied after this starting point. Contemporary observations of the lack of regional dialectal differences have for example been made of the early stages of American English in the eighteenth century (Trudgill 2004:22).

In this section I will analyse the uniformity and the degree of variation in Tver’ Karelian dialects in comparison to other Karelian dialects. The presumption of a new dialect will gain support if Tver’ Karelian is demonstrably more homogeneous and less varied than its dialects of origin. The analysis is based on the Dialect Atlas of Karelian (Bubrih et al. 1997). Previously, researchers have presented some general estimates of the uniformity of the Tver’ dialect area. Vihtori Alava described in the 1890s that, in Tver’, the Karelian dialect is somewhat uniform without major dialectal differences (see Haltsonen 1942:182). Nearly 100 years later, Pertti Virtaranta stated that even though the area inhabited by Tver’ Karelians is large, the internal dialectal variation is not significant (Virtaranta 1982:53). On the other hand, dialectal differences between villages in Tver’ have been brought up, especially in the vocabulary (Sarhimaa & Siilin 1994:268–269, Joki & Torikka 2001:466–469).

So far, not much research has been done on dialectal differences within the Tver’ Karelian dialect area (with the exception of Wiik 1998, Joki & Torikka 2001, Novak et al. 2019). Kalevi Wiik has studied differences between Karelian dialects on the basis of the Dialect Atlas, concluding with a division of Tver’ Karelian into four dialect groups (Wiik 1998:49), with Dërzha as a fifth group that is separate and clearly different from the rest. Of the four groups, the two to the south represent the Tolmachi dialect and the two to the north represent the Ves’egonsk dialect.

I will compare the degree of variation in Tver’ Karelian with other dialects of Karelian. Each map of the Atlas has been studied in respect to the number of variants of the linguistic feature in question, occurring within the dialect area monitored. The corresponding numbers within each dialect area are presented in Table 1. ⁷ In regard to the number of villages in each of the dialect areas, they can be considered as comparable to each other (see the numbers of villages in note 3).

Table 1. The degree of uniformity in Karelian dialects based on the Dialect Atlas of Karelian (Bubrih et al. 1997)

In the calculations based on the Atlas, no distinction has been made between common and rare representations of a variant on the map (except for Tver’ Karelian in the written report below). In a subtler analysis, however, this aspect should also be considered. A minimum value for a variant to be taken into account in the calculations is two occurrences within the dialect area in question. (A single occurrence could simply be random in nature, and therefore such sole occurrences of a variant were ignored.)

As indicated in Table 1, the most uniform and least varied of Karelian dialects is North Karelian, with 132 maps in the Atlas where the representation of an examined feature is non-varying within the North Karelian dialect area. The uniformity of North Karelian has also been suggested by Novak et al. (2019:411). The second in uniformity is Tver’ Karelian with 116 (see below) and the third Livvi Karelian with 101 maps representing full uniformity within this dialect area. South Karelian, for its part, is the least uniform dialect of Karelian.

The analysis of the degree of variation in Tver’ Karelian was supported by a preliminary examination carried out by Kiira Myller at the University of Eastern Finland in 2015. The results of the analysis show Tver’ Karelian to be less varied than South Karelian. For Tver’ Karelian there are 116 maps in the Atlas where the representation of an examined feature is uniform and non-varying, i.e. there is only one variant in use in the entire Tver’ Karelian area. On the other hand, there are 66 maps with variation between two variants of a feature and 25 maps with variation between more than two variants in Tver’. That is, more than half of the maps indicate full uniformity and the rest some degree of internal variation in Tver’ Karelian. However, a closer inspection revealed that maps with two (or more) variants in Tver’, in fact, may also indicate significant uniformity, due to the fact that very often there is a variant that is widely used, whereas another variant of the same feature may only appear in a very limited area, as in a few villages.

Examining vocabulary (Maps 186–209 in the Atlas) separately gives a slightly different result regarding the uniformity of Tver’ Karelian compared to an examination of phonological features. Of the 24 maps of the Atlas that concern vocabulary, 15 indicate variations of two or more words with the same meaning (i.e. synonymy) in the Tver’ region. A more comprehensive investigation of variation in Tver’ vocabulary should be conducted on the basis of the Dictionary of Karelian (Karjalan kielen sanakirja) and the lexical archive of Karelian dialects.

According to my analysis, the most inconsistent dialect area is South Karelian, displaying a higher degree of variation than any other Karelian dialect; the historical development of South Karelian on the border of main dialects of Karelian provides an explanation for this (Novak et al. 2019:411). In the analysis, I have divided South Karelian into two subgroups, based on former studies of South Karelian and on a preliminary survey of the Atlas, roughly following the division into Livvi-side and Ludic-side South Karelian presented in Section 4.3. In particular, the Paatene and Mäntyselkä parishes stand out from the rest of South Karelian and form a group of their own here. I also present calculations on South Karelian as a whole.

The least uniformity and highest degree of variation in Karelian dialects is represented in South Karelian as a single whole (51 non-varying features); this position remains when divided in two. In both subgroups, (i) Paatene and Mäntyselkä and (ii) the rest of South Karelian, the number of one-variant representations (80 and 72) is lower and the degree of variation higher than in Tver’ Karelian (116). This implies that Tver’ Karelian is more uniform and less varied than the modern representatives of the dialect of its main regions of origin. The difference between them is significant.

The two subdialects of Tver’ Karelian enclave dialect, Tolmachi (Villages 11–35 in the map of Tver’ in the Atlas) and Ves’egonsk dialects (Villages 1–10), are in effect separate, the population being areally discontinuous between them (this point was presented by a reviewer of this article). In addition, a historical map presented by Saloheimo (1973: 479) indicating the number of Karelian families in the entire Tver’ area in 1651 shows a geographical gap in population between the Tolmachi and Ves’egonsk areas. However, there seems to be no such discontinuity linguistically. Wiik regards the Tver’ Karelian dialect area as a unitary whole and points to relatively few differences and fuzzy dialect boundaries within it (Wiik 1998:46, 59), excluding the boundary with Dërzha. This view suggesting unitarity between the two subdialects is supported by an overall examination of the individual maps and the distribution of variants in them, which I carried out from the perspective of the Ves’egonsk subdialect. This examination revealed abundant similarities (shared feature-wise representations of variants) between the two subdialects rather than any numerous or striking differences. There are few maps that show a variant typical solely of the Ves’egonsk area, not existent in the Tolmachi area (approximately ten maps between Maps 76 to 99, all of them describing the variation of sibilants). Instead, the distribution of a variant usually covers both subdialects of Tver’ Karelian to a greater or lesser extent and the variants crisscross within both subdialects, Tolmachi and Ves’egonsk. However, studying the vocabulary of these two subdialects may provide some differing results for this.

Thus, in light of the Dialect Atlas of Karelian, Tver’ Karelian seems to be linguistically rather uniform between its subdialects, with no striking differences or clear-cut dividing lines between variants. There are practically no distinctively Tolmachi-bound or Ves’egonsk-bound variants of linguistic features visible in the maps. Historically, the reasonable uniformity between Tolmachi and Ves’egonsk dialects may be explained either by former mutual contacts within a unitary Tver’ Karelian enclave or, more probably, through separate developmental processes of a new dialect in both subdialects, which has accordingly not produced significant differences between the two areal groups.

The overall view displayed by the Dialect Atlas of Karelian is, of course, only approximate. It should be noted that the maps focus on phonological phenomena while other levels of language (morphology and vocabulary) receive less attention. This is still largely consistent with the theory of new-dialect formation, which concerns phonology almost exclusively (Hickey 2003:214, Dollinger 2008:xvii). However, some phonetic features may be disproportionately weighted in the Atlas, such as the representation of sibilants described in 28 maps (Maps 73 to 100). It has been suggested by Novak et al. (2019:401) that morphological differences between the main dialects of Karelian are considerably smaller than phonetic and phonological differences. Another restrictive factor in analysing variation is that the Atlas shows no variation village-wise, although such variation most probably exists.

The uniformity and decrease in variation that characterise a new dialect are most often manifested as the spread and establishment of one variant in the entire new-dialect area. There are also cases where Tver’ Karelian representations differ from other Karelian dialects, representing either (i) preservation of old features (lost elsewhere in Karelian), (ii) separate development (based on original materials, i.e. interdialect forms), or (iii) influence from Russian on the Karelian language system. Enclave dialect communities may generate special interdialect forms, meaning new variant forms created by contact between different dialects (of the same language). Interdialect forms do not appear in the same form in any of these contact dialects (Trudgill 2004:86–87, Kerswill & Trudgill 2005:199). A good candidate for an interdialect form in Tver’, Tikhvin, and Valday Karelian is the reflexive verb type with the derivative affix čče (e.g. pezie-če-n wash-refl-1sg ‘I wash myself’; see Koivisto 1995:66, 86–90), which does not have a similarly regular and conjugatively productive counterpart elsewhere in Karelian.

7. Conclusion

In this article, I examined the Tver’ Karelian dialect as well as its emergence and development from the point of view of Trudgill’s (2004) new-dialect theory. Based on its historical and linguistic background and means of emergence, Tver’ Karelian can be interpreted as a new dialect. The migration of Karelians to Tver’ took place during a restricted period of time, and there are no records of Karelian populations moving to the region subsequently. The migrants ended up sufficiently far away that contact with their region of origin was practically broken. Since the migrants settled, Tver’ Karelian has been separate from the rest of Karelian, surrounded by a Russian-speaking region.

The emergence of Karelian enclave dialects does not represent the most common way of language development, but there are parallel cases of new-dialect formation globally. Peoples have shifted and migrated throughout history, carrying their languages with them to new locations. Mass migrations have often taken place out of necessity, due to pressures such as power politics, wars, persecution, famine, or changes in administrative borders. Migration resulting in a permanent demographic change always implies linguistic consequences. Major one-time migrations or shifts from one region to another and the linguistic processes connected to them pose interesting challenges to linguistics. In examining and describing them, the traditional methods of such disciplines as historical linguistics, dialect research, or sociolinguistics may not necessarily be sufficient.

In the analysis and description of the development of a new dialect, an essential starting point is settlement history. The migration of Karelians to new areas in the seventeenth century is attested in historical documents, but there is no linguistic material available on the period of migration or the new dialect emerging. Therefore, the presumption that Tver’ Karelian is a new dialect was examined from the perspective of twentieth-century Tver’ Karelian and on the basis of linguistic materials from this period. Initial steps in the examination of the relationship between Tver’ Karelian and its dialects of origin were already taken by Sarhimaa and Siilin (1994) as well as Joki and Torikka (2001).

On the basis of historical records on the migration, I assumed that different types of dialect mixes came about in Tver’ based on the migrants’ areas of origin. A polygenetic, multidialectal background is typical of the emergence of a new dialect. I concluded that Tver’ Karelian is founded on a broader dialectal spectrum than is shown in the later Tver’ Karelian documented in the nineteenth and twentieth centuries. Further development has made it more uniform and reduced its variation. I also draw an overall parallel between Tver’ and other Karelian enclave dialects on the basis of their shared means of emergence and the linguistic features that unite them.

With the materials available on Tver’ Karelian and the methods they allow, finding features characteristic of the development of a new dialect was considerably challenging. In contrast, research into new English dialects, for example, has had access to concrete contemporary linguistic material, usually in the form of written texts (e.g. Dollinger 2008 on Canadian English) but also, exceptionally, audio recordings of spoken material (Hickey 2003:214, Gordon et al. 2004 on New Zealand English). In the case of Tver’ Karelian there is no information available on the linguistic details or dialectal differences in the Tver’ region in the initial situation, whereas we know what kind of English was spoken in England, Scotland, and Ireland in the eighteenth and nineteenth centuries. There was also no written use of Karelian, whereas the new-dialect development of English may have been impacted by the written form as well.

There are various conjectures and uncertainties related to the development of Tver’ Karelian as a new dialect. However, the preconditions and founding elements for the assumed development are valid, supported by linguistic considerations and settlement history. This article concentrated on the uniformity and degree of variation of Tver’ Karelian. The examination was primarily based on a single source of material, the Dialect Atlas of Karelian, which offers a mapping of numerous linguistic features covering most dialects of Karelian (with the areal exceptions mentioned previously in this article). Some implications of mismatch between areal and linguistic aspects were also detected regarding the migrants’ region of origin and the linguistic features of Tver’ Karelian. The contents of this article can be seen as a significant suggestion for further, more detailed study of the subject.