Laboratory seminars

This paper summarises a joint study by Yuri Koryakov, Andrey Shluinsky, and myself, see (Khanina et al. 2018a, Khanina et al. 2018b). Through a series of linguistic maps based on published ethnographic data and our fieldwork accounts, we reconstruct the territories in which Forest Enets and Tundra Enets (Samoyedic, Uralic; Central Siberia) have been spoken from the 17th century till today. We analyze in details migrations of the two ethnic groups and the changing language contact scenarios. One of the most intriguing findings of this study is an explanation of the Forest Enets - Tundra Enets puzzle.
There is no unanimity whether they are separate languages or dialects of the same language. Ethnographically, the two linguistic communities are clearly distinct, have different self-nominations, and do not consider themselves as belonging to the same ethnic group. As our field experience has shown, the degree of modern mutual comprehension is not a neutral question and depends on the stance that a speaker takes at the moment of conversation, whether stressing the difference between the two ethnic groups or aiming at reaching his/her communicative goal. Whereas the phonologies of Forest Enets and Tundra Enets suggest a split of at least several hundred years ago, and lexicostatistical calculations go even further by dating the split ca. one thousand years ago, the match between the two Enets grammars is so striking that it contradicts this scenario.
So here the linguistic geography steps in, documenting migrations that led to a secondary convergence in the 19th century of the once more distinct Enets lects, which was later, in the beginning of the 20th century, followed by a secondary divergence. We support this historic hypothesis with a catalogue of all features that separate the two Enets varieties and with linguistic maps reconstructing changes in the territories of the two ethnic groups in the last 300 years.

References
Khanina, Olesya, Koryakov, Yuri & Andrey Shluinsky. 2018a. Enets in space and time: a case study in linguistic geography. Finnisch-Ugrische Mitteilungen 42, 109-135.
Khanina, Olesya, Shluinsky, Andrey & Yuri Koryakov. 2018b. Forest Enets and Tundra Enets: how similar/different are they and why? Paper presented at the 7th international conference on samoyedology, October 2018 (Tartu, Estonia).

Western Africa occupies a central place in the research on multilingualism due to the studies on the sociolinguistic situation in Cameroon (di Carlo 2018) and the Casamance area in Senegal (Lüpke and Storch 2013; Lüpke 2016). The study focuses on yet another case of multiligualism in Western Africa by discussing the multilingualism patterns in the area of Fouta-Djallon plateau in Guinea. The situation will be analyzed in the perspective of communities speaking Kakabe, a minor language spoken in about fifty villages. The involved languages are Kakabe, Maninka, Pular, and, to a lesser extent, Sussu, with Pular belonging to the Atlantic family and the three other languages to the Mande family. In my talk, I will analyze the attested multilingualism patterns in different types of language practices. The study is based on a multi-media oral corpus representative of a variety of genres and containing data that I have been collecting in the region since 2009.

The East-Caucasian languages (Nakh-Daghestanian) show a number of puzzling structures that are challenging from the theoretical point of view: non-finite clauses where all the arguments are encoded in the same way as in independent sentences, backward control, long-distance reflexive pronouns and long-distance agreement in complement clauses. This talk is focused on long-distance agreement in East Caucasian languages. First, I discuss the phenomenon in Qunqi Dargwa. The infinitives and converbs are the only complementation strategies that allow long-distance agreement. In Qunqi, there is a fuzzy boundary between infinitives and indirect mood forms. The converbs are used both with control verbs and with emotive and perception complement-taking verbs. The long-distance agreement pattern is only observed with control verbs. I show that these structures show properties of clause union. Then I consider the data of 19 East-Caucasian languages (mostly based on the data from Kibrik 2005 "Materials to the typology of ergativity"), and discuss the long-distance agreement patterns in those languages. In most of these languages this phenomenon is limited to control constructions, while Tsez and Tsakhur deviate from this generalization.

Polynesia covers a vast territory of the planet. It includes a large number of speech communities which descend from a common ancestor; some them are isolated by thousands of miles of open ocean. No wonder, Polynesia has always been the favorite place for both linguists and anthropologists working on phylogenetics. The standard account of the Eastern Polynesian subgrouping is that the language of Easter Island (Rapanui) forms a branch on its own, coordinated with Central Polynesian languages; CE in turns branches into Tahitic and Marquesic (Roger 1985). It has been becoming more and more evident that Tahitic and Marquesic are not valid subgroups (Vladimir Belikov 2009, Mary Walworth 2014). In my talk, I am going to show that Rapanui, Mangarevan, North and South Marquesan constitute a subgroup within Eastern Polynesian languages. Interestingly enough, this proposal implies some phonological and lexical innovations spreading across the Pacific. The main objective of the talk is to discuss the latter and their implications for the theory of language.

The opposition of the geminate and singleton ejective lateral stop /L'/, reconstructed to proto-Andic, has been lost in various Andic languages due to the phonetic evolution of the simpleton into a different soundtype. The speakers of Rikvani Andi, a dialect of Andi (Andic, East Caucasian) spoken by 800 people in the village of Rikvani in Dagestan, developed a glottalized lateral consonant as a reflex. Glottalized sonorants are a typological rara. While they have been sparsely attested in various areas where glottalic initiation also occurs with stops (ejectives), Rikvani Andi is the only variety of a Caucasian language where it has so far been reported.
The present study is an acoustic analysis of the sound as documented in the field data collected by the HSE team in the village. Data from six speakers's word list elicitation as well as from spontaneous texts from two speakers are used. This study is the first that considers glottalized sonorants in connected free speech. We test several generalizations and observations previously made wrt glottalized sonorants (Um 2001, Maddieson and Larson 2002, Maddieson et al. 2009). We confirm presence of strong variation of the realization of the glottliazed /lˀ/ across speakers in terms of presence of creaky voice, focus of glottalization, intensiity and formant structure. We do not confirm Um's suggestion that the peak of glottal construction (pre- vs. post-glottalized realization) is affected by the poisition of the sound in syllable/word. We conclude that the distinctive feature is the presence of creaky voice and intensity, but these features become less pronounced in free speech. Finding the acoustic cue that serves to distinguish glottalized sonorant requires further acoustic research. Variation across speakers and fading away of the cues in connected speech may suggest that a merger of /lˀ/ and /l/ is on its way.
Various visualization techniques highlighting the nature of /lˀ/ will be discussed.

In August of 2019 we visited the village of Botlikh to study the Botlikh language. Our aim was to collect some data for a small investigation on agreement patterns of ordinal numerals. In addition, we translated several texts recorded by Togo Gudava in the 1950s-1960s, met a number of potential language consultants and learned some new things about the sociolinguistic situation in Botlikh. We will talk about the trip and our future plans for working on this language.

Bantu expansion is among the most important and least understood human migrations. Bantu-speaking populations (240 million people, 500 languages, spanning 9 million km2 [1]) are the result of a huge migration originating in a homeland near the border of Nigeria and Cameroon between 4,000BP and 5,000BP [2,3,4,5,6,7,8]. Although the homeland and the time depth are well established, the migration route is still unclear.

Recent phylogenetic studies [1,9,6,7,8] support the late-split [10,11,12,13,14], which claims that East-Bantu and West-Bantu languages' common ancestor crossed the African Rainforest, splitting after this. It is thought that this crossing was made through the Sangha River Interval (SRI), a N-S savanna opening into the rainforest. However, in dated phylogenies [7], dates don't match consistently: They should have crossed this corridor around 4,000BP, while it was completely open only 2,500BP.

We propose two different hypotheses for competing with the traditional SRI late-split. The first, a coastal savanna corridor [15]. The second, an earlier paath through the rainforest. We compare the hypotheses with a Bayesian phylogeographic approach based on linguistic trees. We use lexical and geographical data for 400+ Bantu and Bantoid languages, inferring the linguistic and geographic history in parallel, by implementing the break-away model [16] in BEAST2 [17]. We conclude that the way through the rainforest happened around 4,000BP.

The paper deals with non-standard word order in the variety of Russian spoken by bilinguals from Daghestan. Specifically, we focus on the occurrence of prepositive genitive modifiers in bilinguals’ speech. Whereas in monolinguals’ Russian the neutral and most frequent word order in noun phrases with a genitive modifier is the order N+GEN, in Daghestanian Russian the opposite order GEN+N often occurs. This phenomenon was mentioned as one of the striking morphosyntactic features of Daghestanian Russian, and its frequent occurrence can be partly explained in terms of syntactic calquing from speakers’ L1s, all featuring an unmarked GEN+N order in noun phrases. However, the picture is far less trivial than it could look at first sight. On the one hand, the word-order pattern GEN+N does not seem to affect equally all types of genitive noun phrases in Daghestanian Russian. On the other hand, similar examples of non-standard word order are sometimes found in monolinguals’ speech too. In the course of the paper, we present the results of our corpus-based investigation of genitive noun phrases in Daghestanian Russian as compared to monolinguals’ spoken Russian, including dialectal varieties. Prepositive genitives appear to be favored by several lexico-semantic and processing features of both the head and the genitive dependent. The strongest factor is kinship semantics: noun phrases that express a kinship relation tend to be prepositive. In monolinguals’ spoken Russian, although prepositive genitives are very infrequent, they sometimes show similar lexico-semantic and processing features. Therefore, we are not dealing with a simple calquing process. Rather, L1 influence is manifested in the strengthening of some tendencies existing in monolinguals’ Russian too.

The paper investigates the role that the rate of bilingualism plays in lexical borrowing. Our data comes from Daghestan, an area of high language density. Based on loanword counts, we isolate two zones of lexical influence, the south, heavily influenced by Azerbaijani, and the north, dominated by Avar. This salience of Avar and Azerbaijani as donor languages is likely to reflect the historical role of these languages as lingua francas in their respective geographical zones. The study supports the idea of Brown (1996, 2011) that contact influence from a lingua franca is higher than from a language only used to communicate with its L1 speakers. In line with the widespread argument that the amount of contact-induced change from a language is proportional to intensity of bilingualism (Thomason & Kaufman 1988), Brown stipulates that the importance of lingua francas as lexical donors must be linked to the high rate of bilingualism in these languages. The bilingualism in Azerbaijani and Avar was indeed high, as the evidence from field research on traditional language repertoires of Daghestanian highlanders shows. On the other hand, the knowledge of two other locally important languages, Chechen and Georgian, which was, at some locations, only slightly lower, did not lead to the same level of lexical transfer; in fact, the amount of Georgian and Chechen borrowings seems disproportionately low. High bilingualism rates are thus not sufficient for a language to become a major lexical donor. At the level of methodology, the paper explores the prospects of using short wordlists as ‘contact probes’, tools for measuring lexical contact. We follow the approach by Haspelmath & Tadmor (2009) and Bowern et al. (2011) in applying a fixed list of concept to quantify lexical contact between languages. Based on field elicitations conducted in a number of villages in the Republic of Daghestan, a list of 160 concepts is shown to be efficient enough to differentiate the degrees of lexical impact from the locally important L2’s to minority languages. The method does not only ensure comparability across contact situations but also provides a level of resolution that is sensitive to differences between villages speaking the same language. By fine-tuning the wordlist to a different linguistic setting, the methodology suggested here may be extended to other geographical areas of intense language contact and become a tool for reconstructing multilingual patterns of the past.

In a recent paper (Naccarato, Panova & Stoynova Forth.), we have examined cases of non-standard word order in the variety of Russian spoken by bilinguals from Daghestan. Specifically, we have restricted our analysis to the noun phrase, and have looked at the occurrence of prepositive genitive modifiers in bilinguals’ speech. As we have shown, whereas in Standard Russian the neutral and most frequent word order in noun phrases with a genitive modifier is the order N+GEN (muž sestry), in Daghestanian Russian the opposite order GEN+N (sestry muž) often occurs. This phenomenon has been partly explained in terms of syntactic calquing from speakers’ L1s, all featuring a neutral GEN+N order in noun phrases. However, such inversion in word order does not seem to equally affect all types of genitive noun phrases in Daghestanian Russian, but appears to correlate significantly with noun phrases featuring kinship semantics. Moreover, similar examples of non-standard word order are sometimes found in monolinguals’ speech too, which makes the picture far less trivial than it could look at first sight. In this talk, we present the latest results of our corpus-based investigation of genitive noun phrases in Daghestanian Russian as compared to monolinguals’ spoken varieties of Russian, with the aim of explaining the factors boosting non-standard word-order realizations.

Botlikh (Avar-Andic, East Caucasian) features a two-fold animacy agreement system including, on the one hand, a set of noun class (i.e. gender) markers representative of many EC languages and, on the other hand, an additional set of dedicated animacy markers which are unique to Botlikh. The dedicated animacy markers can appear on various targets (i.e. negative copulas, interrogative particles, question word formants, attributive clitics, present/future participles, ordinal numerals), and agreement is controlled by either the nominal head or the absolutive argument of the verb. By focusing on ordinal numerals, which appear to mark animacy most consistently, we set the following goals: a) to better understand the agreement patterns of these forms; b) to clarify which referents qualify as animates and which do not. For these purposes, we have created the first draft of a survey which we will discuss during the talk.

Most descriptions of Abaza mention two affixes which express the meaning of refactive (‘again’, ‘once more’, etc.): the suffix -χ and the prefix ata-, which almost always appears only in combination with -χ. I argue that the main difference between these two refactive markers is that the marker -χ “sees” the internal structure of an event and can have scope over any part of it (just the resultant state, or just the process, with or without arguments), while the marker ata-+-χ is “blind” to the internal structure of the situation and can only “copy” the whole event with its arguments.

In this talk we introduce our first pilot database for the DagLoans project. The database contains translations of 160 concepts collected in the field in Daghestan (and Northern Azerbaijan). At present, this includes a total of 24.785 entries from 23 different languages. The database can be used to find the translation of a concept in one or more languages. The most important feature is “Set”: all entries are grouped in sets with other similar words, which allows us to plot the spread of lexical items on the map. The database can be used for conducting quantitative research on lexical convergence as well as for creating geographical maps showing the areas and the intensity of foreign influence.

The most common statistical task is hypothesis testing. When a pair of competing models is fully defined, their definition immediately leads to a measure of how strongly each model supports the data. The ratio of their support is often called the likelihood ratio or the Bayes factor. During the talk I will show how to define different models and compare them with Bayes factor.

In Modern Standard Russian, the prefix/preposition pair u-/u is peculiar with respect to other similar pairs, due to the meaning mismatch between the two. While the prefix u- has an ablative meaning, as shown when it is prefixed to motion verbs, the prepositional phrase u+gen occurs in locative constructions, and other related constructions, such as predicative possession that is expressed via the cross-linguistically common Locative Schema. Etymological considerations show that the meaning preserved by the prefix is older. The only type of occurrence which, according to the literature, preserves the ablative meaning for the u+gen construction preposition is found with verbs of requesting, removing, and buying. Notably, however, in other Slavic languages putative ablative contexts are limited to verbs of requesting. Data from MSR, OCS, Polish and Czech lead to the conclusion that the extension of the u+gen construction to verbs of removing in MSR is based on its use for the encoding of predicative possession. Extension to verbs of buying is better explained through the locative meaning of the construction. As a result of different developments, the u+gen construction has become part of the argument structure of a group of verbs including verbs of asking and requesting, verbs of removing and verbs of buying, which are characterized by the common feature of taking human non-recipient third arguments.

Kunbarlang is a critically endangered polysynthetic language spoken in central Arnhem Land, Northern Territory, by approximately 40 people. It belongs to the non-Pama-Nyungan Gunwinyguan family.

This talk reports on the first comprehensive description of Kunbarlang (although it builds on and extends important unpublished work by Carolyn Coleman and Joy Kinslow Harris). Kunbarlang has very rich verbal morphology that includes complex agreement paradigms, composite TMA system that differs from other Gunwinyguan languages, an array of argument derivation tools, and coverb constructions. The nominal domain, on the contrary, has little morphology and relies heavily on syntactic constructions - for instance, case marking of nouns is analytical.

The talk will give a general overview of the grammar, and then focus on a few selected topics across different areas.

It is generally believed that Russian has no sequence of tenses (SoT) in complement clauses, and the choice of absolute tense over relative is considered to be a typical error in the interlanguage of non-standard speakers of Russian as a foreign language whose native language features SoT, e.g. English. However, all uses of absolute tense in Learner Russian cannot qualify as errors, since Standard Russian shows a great deal of variation in tense assignment in complement clauses. One of the factors that is said to govern tense choice is the semantics of the matrix verb (Barentsen, 1996; Гиро-Вебер, 1975, Schlenker, 2003, inter alia). Specifically, speech and mental verbs are said to strictly require the relative tense, whereas sensory, emotion, and existential matrix verbs allow for both absolute and relative tense patterns. Despite the acknowledged variation, the precise distributional patterns of tenses in complement clauses have been understudied. This paper is a systematic corpus-based study of the variation in tense choice across the semantic classes of the matrix verbs in two language varieties: (i) Standard Russian as represented in the Russian National Corpus and (ii) Learner Russian of anglophone speakers as represented in the Russian Learner Corpus. I examine those clausal complexes where the matrix verb in the past tense and the verb of the complement clause denote simultaneous actions.

The analysis identified a likelihood hierarchy of verbal semantic classes ranging from the least likely to tolerate past tense in the complement clauses to the most likely ones: speech<mental<sensory≈emotion

Within the framework of Multidagestan project, vast amount of sociolinguistic data about traditional small-scale multilingualism was collected in Daghestan. The aim of the project is to trace the change of the multilingual patterns in the 20th century. However, 71 percent of the data were collected in an indirect way, asking people about their relatives. We will discuss the statistical methods that we used to check the robustness of the indirectly collected sociolinguistic data.

The particle =OK, originally Turkic, is attested in all the core members of the Volga-Kama Sprachbund: Chuvash =aχ/ =eχ, Tatar and Bashkir =uk/ =ük, Meadow Mari =ak, Hill Mari =ok and Udmurt =ik. Meadow Mari, Hill Mari and Udmurt have arguably borrowed the particle from Turkic (Bulgar).

=OK is used in contexts many of which may be characterized as emphatic identity contexts: the argument marked by =OK is the same as an argument of a different proposition (≈ Russian že: Masha rabotajet v pole, Masha že sidit s det'mi 'Mary works in the field and it is also she who sits with the children').

However, in different languages =OK exhibits different morphosyntactic restrictions on the constituent to which it may attach, i.e. in Tatar it attaches to demonstratives (šul uk keše 'the same person') but not to proper names (*Märijäm ük 'it is also Mary who...'). In Chuvash, Meadow and Hill Mari and Udmurt the particle can attach to proper names. =OK can also attach to the verb, with different interpretations and again with different morphosyntactic restrictions.

There are similar constructions and lexicalizations with =OK in some of the languages (e.g. reduplication construction of the type V-converb=OK V).

I would like to discuss whether -- and how -- we can approach these similar patterns in terms of contact. From the literature we know what were the strongest bonds in the area (Chuvash-Mari, Tatar-Meadow Mari, Tatar-Bashkir). The question is whether greater and weaker similarity of =OK morphosyntactic, construction and lexicalization patterns between languages corresponds to areal affinity and how to demonstrate it.

In this SMALL discussion, we will present the path taken so far for the methods of inter-dialectal comparison, the point we are currently standing (or stuck) at, and will gratefully take advices as to how to proceed.

We will first remind of the starting point of the project, ie a mehod of isolating inter-dialectal divergence that takes into account inter-speaker variation. Than we will briefly overview the steps we did so far (LogLikelihood, Wilcoxon-Mann-Whitney test, Gries' DP). We will then focus on the very last result we got, evaluating the observed DP value against the simluation and permutation test for the distribution of the DP in a random sample - and whether we can use it for our purposes.

In this paper, I consider several Lezgic languages suffixes (possibly, but not definitely related) that cover a rather wide range of contexts. Some contexts of their use may be qualified as denoting unrealized state-of-affairs (such as conditional clauses, polar questions, and, to a certain extent, indefinite pronouns). Some others fall short of this definition, including indirect questions and other subordinate clauses with WH-words. The set of contexts covered by the markers in question is one and the same in at least three Lezgic languages (Lezgian, Aghul, Tabassaran), and also in Azerbaijani, which raises a question of possible contact origin of this pattern. Some other Lezgic languages employ, in these contexts, several different markers. Kina dialect of Rutul presents an especially interesting case, combining in one morpheme (-jden) the meanings which are unlikely to be associated. In this talk, I will present the case of Kina Rutul in details, discuss possible interpretation and origins of the marker -jden, and compare Kina Rutul with other Lezgic languages.

In this talk, we will discuss modifying participial constructions which is a predominant type of relative clauses in East Caucasian languages. One of the key properties of participles in East Caucasian languages is the lack of syntactic orientation. There is little to no syntactic restrictions on what can be relativized: the gap in the relative clause can correspond to a core argument, a peripheral participant or even a participant that is not part of the verb’s argument structure. Languages also share some common patterns of constructionalization of specififc relative construction (such as name-constructions). On the other hand, there is variation across languages, e.g. more or less strongly articulated preference for S relativization; or more or less widespread use of the resumptives; and language-particular features, e.g. a very high ratio of addresse relativization (in name-constructions) in Agul. After a general overview of the problems related to the study of relativization targets, we concentrate on language-particular case studies and discuss the counts of relativization targets in the corpora of two East Caucasian languages (Agul and Archi). As a comparative background, West Circassian corpus data will be presented. In this language relativization is syntactically oriented, the strategy cannot be classified as participial, and special reflexivizers may be interpreted as obligatorified resumptive pronouns. Finally, we discuss the comparison of corpus counts on the relativized syntactic role in the three languages, and the problems connected to such comparison.

The DagLoans project aims at investigating lexical convergence between East Caucasian languages and their neighbours in quantitative terms. We focus on horizontal interaction, looking at borrowings between languages that are in direct contact and dismissing influence of dominant cultures and distant languages, e.g. Arabic, Persian or Russian. The project consists of two parts - one dealing with lexical matter copy, the other with lexical pattern copy. Today we are only discussing the data of the former.

We deal with lexical matter borrowing in an attempt to compare and quantify horizontal borrowing between languages at different locations. Instead of comparing standard languages, we aim at comparing local varieties and, ideally, village lects. Basing on the Leipzig-Jakarta list and issues of "Отраслевая лексика", we attempt to compose a list of lexical items with a high borrowability rate. The list should be concise enough to be elicited from several speakers during a one day visit to a village, but, on the other hand, long enough to discriminate local varieties and, ideally, village lects.

At present, we work on data from Rutul, Tsunta and Botlikh districts. In the talk on September, 19 we plan to discuss the list and its composition and elicitation techniques we use to decide whether it is an adequate tool  for studying lexical contact rate at local level and how it reflects local geography and data on bilingualism. The talk is based on the data from 6 villages (4 languages) of the Rutul region that are located in the same valley: Khlut (Lezgian), Kiche (Rutul), Rutul (Rutul), Kina (Rutul), Helmets (Tsakhur), and Kusur (Avar).

The project "Atlas of Multilingualism of Daghestan" is based on sociolinguistic interviews recorded in Daghestan by the team of this project over the course of seven years. The aim of the project is to determine the level of bilingualism in Daghestanian mountain villages and describe the sociolinguistic patterns of linguistic convergence of local languages. In addition, the project allows to establish the type of linguistic contact characteristic of neighboring villages, which languages not pertaining to a particular area were spoken by inhabitants, how the command of Russian changed, what role the geographic distance between languages played and how the command of certain languages was distributed among inhabitants of a village. This talk will focus on two topics: first of all, we will show the results of a study of how the command of languages was distributed among men and women and how these dynamics have evolved since the beginning of the 20th century until now. Second, we will discuss some problems and shortcomings of the method used and we will suggest some verification methods.

Rich consonantal inventories are a salient feature of the languages of the Caucasus on the whole and of the languages of Daghestan in particular. Their composition is a subject to a certain variation from language to language, but is overall similar. All languages have ejectives, most languages have labialization, and geminates are not uncommon. On the other hand, acoustic properties of the phonologically identical elements may be substantially different. To document these differences, in the last few years we do systematic field recordings of data from different languages. In this talk we will introduce the aims and methods of the project and will present preliminary results of our analysis of acoustic properties of ejectives as compared to corresponding voiceless stops. We will evaluate the impact of such parameters as closure duration and voice onset time. We will use the data from three different languages - Rutul (Kina dialect), Andi (Zilo dialect) and Mehweb. In the long run, we hope that our project will be able to address the following theoretical question: is the observed intragenetic variation more sensitive to areal or to systemic factors?

The talk presents the Spoken Meadow Mari corpus project. Meadow Mari is a Uralic language spoken in the Volga region by some 375 thousand people. The core of the corpus are recordings made in 2000-2001 by a group of researchers from the Lomonosov Moscow State University. In our talk we will discuss the data we have, possible applications of the project and the target audiences of the corpus, as well as its structure. Making the corpus data presentable involves transcribing, glossing and annotating the data as well as aligning audio and text which should facilitate data analysis.

The Universal Dependencies (UD) is a project dealing with consistent cross-linguistic morphological and syntactic mark-up. The UD is currently in version 2 and covers 52 languages with 10 more languages yet to be included.

With its own annotation principles and abstract inventory for parts of speech, morphosyntactic features and dependency relations, UD aims to facilitate multilingual parser development, crosslingual learning, and parsing research from a language typology perspective. While UD covers 11 language families, it does not include any languages of the Caucasus (including the East Caucasian family). In our talk we will describe the way Mehweb Dargwa (East Caucasian) meets the UD scheme.

The talk presents the "Dialectal variation in Even" project. The project works with two dialects of Even: the easternmost one spoken in the village Sebjan-Küöl in Yakutia and the westernmost one spoken in Kamchatka. The first dialect has been in contact with Yakut for a long time, while the second has possibly been in contact with Koryak and Itelmen. The aim of the project is to discover differences between the two dialects and whether they stem from independent innovations or contact. For this, we use corpora of field recordings collected by Brigitta Pakendorf in the course of 2007-2012. In this talk we will describe the data, argue what differences can be found in them and what statistical methods can be used for this, and present some differences in morphology and syntax that have been found so far.

West Circassian and Kabardian languages represent a dialectal continuum spread in Krasnodar district, republics of Adygea, Karachai-Circassia and Kabardino-Balkaria. During the presentation, we are going to talk about phonetic and sociolinguistic features of different Circassian idioms, present a project of an atlas of Circassian isoglosses and show first maps, which the atlas will include. In addition, we will describe the process of creating a phonetic and grammatical questionnaire and the difficulties associated with that.

The study deals with the intonation patterns in three Non-Standard varieties of Russian. In the first part of the talk we discuss the methodological issues, such as analysis of pitch range, data filtering and data representation. In the second part, we consider a number of case studies, namely Daghestanian Russian, Southern Russian and Jewish Russian. The general goal of the project is to compile an annotated corpus of non-standard Russian with pitch annotation and explore the intonation patterns of non-standard varieties of Russian basing on corpus data.

The corpus of regional variants of Russian spoken in Daghestan is based on transcribed sociolinguistic interviews in Russian with speakers of various Daghestanian languages ​​who live in rural areas. Technically, the corpus is built using the platform and annotation principles developed for the dialectal corpus of Ustja River Basin. The aims of the project include both its maintenance and adding new texts, as well as the use of the corpus for systematic study of morphosyntactic characteristics of Daghestanian Russian. In the talk, we plan to discuss the current state of the corpus, the possibilities of corpus-based research, as well as the problems we met and the perspectives of the project.