Ex-neuter-gender nouns in Canadian Doukhobor Russian

This paper addresses the morpho-syntactic forms of ex-neuter-gender of Canadian Doukhobor Russian nouns. Doukhobor Russian is a near-extinct variety of Russian spoken by a small group of elderly Doukhobors (a religious and ethnic minority of Russian origins) residing mostly in British Columbia, Saskatchewan and Alberta. While Standard Russian has three noun genders (feminine, masculine and neuter), this paper demonstrates that in Doukhobor Russian, a neuter gender erosion has been taking place, whereby ex-neuter-gender nouns have been shifting their declension paradigms mostly to feminine F1 declension forms. Ex-neuter-gender words in Doukhobor Russian also mostly agree with pronouns, attributives, and numerals in feminine gender. They also predominantly agree in feminine gender with preterite verb forms. Two major subclasses of ex-neuter-gender nouns are identified: those in which the stem is stressed and those in which the stress is on the inflectional vowel. A stressed vowel in an inflectional suffix of the nominative and the accusative blocks the paradigm shift of ex-neuter-gender nouns, but does not prevent female gender agreement with attributives and preterite verb forms. The discussion identifies parallels of neuter-gender erosion in Doukhobor Russian with similar processes found in 19th and 20th century Russian dialects.


Introduction
This paper focuses on the class of ex-neuter nouns in Canadian Doukhobor Russian (Doukhoborese), a near-extinct unique language variety only spoken in Canada by about 200 individuals of Doukhobor ancestry (Makarova 2012, pp. 90-91). Evidence of a neuter gender declension paradigm is examined in audio recordings of Doukhobor Russian (DR) speakers made by the author between 2010 and 2018.

Gender in world languages
Grammatical gender is a major classificatory morpho-syntactic feature of nouns embedded in their lemmas (Corbett 1991, p. 6335;Peristeri et al. 2018Peristeri et al. , p. 1009; some authors see it as being synonymous with noun classes (Corbett 2013, http://wals.info/chapter/31). Gender bridges lexical and syntactic information, since it serves not only to assign a particular nominal class and inflectional paradigm to nouns, but also to establish agreement of nouns with other lexical classes of words, such as verbs, adjectives, pronouns, numerals, and others, in a sentence or phrase (Corbett 2013, http://wals.info/chapter/31;Kramer 2015, p. 2;Cook 2018, p. 13;Lemmerth and Hopp 2019, p. 21). Probably because it can be represented and identified through multiple "semantic, phonological, morphological, and syntactic cues of extremely variable reliability" (Meisel 2018, p. 659), gender has been referred to as "one of the more puzzling of the grammatical categories" (Paolieri et al. 2019, p. 112).
The grammatical category of gender has been used as one of several typological features for language classification, whereby world languages are divided into gender and non-gender ones (Tsimpli 2014, p. 298). Furthermore, languages differ by "the number of gender values" (ibid.), the categories of words that bear the grammatical category of gender, and the agreement patterns (Meisel 2018, p. 658). Gender is found in about half of the world's languages (Corbett 1991, p. 6335).
Examples of language families with no gender in their nominal systems include Austronesian (such as Indonesian and Maori), Turkic (Turkish, Turkmen), and Uralic (Finnish, Hungarian), see Corbett (1991, p. 6335). Indo-European languages often have nominal gender systems, but they developed in their own idiosyncratic ways. For example, within the Germanic branch, Dutch nouns have common or neuter genders (Kochari and Flecken 2019, p. 240), German nouns are identified as feminine, neuter and masculine. The Romance languages Italian and Spanish have feminine and masculine gender; the grammar systems of the Slavic languages typically include the three major genders feminine, masculine and neuter. On the higher end of the gender numbers continuum are some languages belonging to the Niger-Congo family, such as Fula / Fulfulde / Fulani, which has about 20 genders, depending on dialect. Nouns in this language are grouped into classes including 'wealth, seeds, slave names, domestic male animals', 'humans', 'elephants', and others (Breedveld 1995, pp. 329-330).
Gender can be marked on nominal inflection forms (e.g., Russian, German, Greek), as well as on forms of other words through agreement, Nouns can typically agree with articles and pronouns (Dutch, French, German), and adjectives (German, Greek and French). 1 Gender can be 'transparent' (predictable from morphophonological clues), such as -a at the end of feminine nouns in Romance languages or 'covert' (not predictable from the word's form), as in German nouns (Bordag et al. 2019, p. 131).
According to a number of studies, gender is acquired in early childhood and stored along with semantic and phonological word properties (Levelt et al. 1999, pp. 4-6;Cubelli et al. 2005, pp. 42-43;Paolieri et al. 2019, pp. 112-113). According to some psycholinguistic theories of gender acquisition, gender is stored "as a node at the lemma level to which individual nouns connect" (Lemmerth and Hopp 2019, p. 22). In addition to memorizing the noun's class and its respective declensional paradigm, gender language speakers have to work through building agreement at the NP level (with adjectives and determiners) and sentence level syntax (agreements with verbs, and cross-phrasally, e.g., anaphoric and cataphoric pronouns use), see Lemmerth and Hopp (2019, p. 22). Monolingual children acquiring languages with more simple and transparent gender systems, like Spanish, Italian and Greek, complete gender acquisition very early, at the age of about two years old. By contrast, the complexity and nontransparency of gender assignment in languages like Norwegian and Dutch may delay the completion of gender system development until the child is over 6 years old (Lemmerth and Hopp 2019, p. 22).
Gender acquisition in the native, second and additional languages (including bi-multilinguals) has proved to be so difficult (e.g., Bordag and Pechmann 2007, pp. 69-70;Cook 2018, p. 15;Paolieri et al. 2019, pp. 112-113) that it has been referred to as "a long-recognized hurdle" (Bordag et al. 2019, p. 130). Many recent studies show evidence of cross-linguistic interactions in speech production and perception by bilinguals and multilinguals resulting from the differences in gender categories of their languages (Morales et al. 2016, pp. 294-295;Lemmerth and Hopp 2019, p. 22;Paolieri et al. 2019, p. 112). Grammatical gender has been reported to play a positive role in language comprehension by activating the lexical prediction of nouns through pre-nominal articles, adjectives and pronouns (Kochari and Flecken 2019, p. 240).

Gender in Standard Russian
As is generally known, Standard Russian (StR) differentiates between the three grammatical genders feminine, masculine and neuter. Gender assignment for inanimate objects in Russian is mostly transparent, i.e., predictable from the word's morpho-phonological form (Corbett 1991, p. 6338;Cook 2018, p. 16). In contrast to languages with articles like German, French or Spanish, gender in Russian is marked on inflectional suffixes of nouns. In addition, Russian requires the noun and an adjective, numeral, or pronoun in a noun phrase to be in agreement (Cook 2018, p. 17). At the sentence level, Russian verbs also have to agree with the subject noun in gender 2 in the past tense singular. The major declensional classes of Russian nouns are directly connected with gender, but not synonymous with it. A simplified representation of major gender / noun class distinctions (disregarding some stem variants) is provided here: Russian feminine nouns ending in the nom. in -a / -ja (e.g., muxa 'fly', tetja 'aunt') belong to the declension class which is denoted here as 'F1'. 3 Russian masculine nouns with zero inflectional suffix in the nom. (e.g., stul 'chair', kon' 'stallion'), denoted in this article as 'M2', 4 and neuter nouns 5 ending in -o / -e (e.g., seno 'hay', pole 'field'), denoted as 'N2', form the second declension class. The third declension class is formed by feminine nouns with palatalized stems, like tetrad' 'notebook'. 6 One more class, denoted as 'N3', is comprised of only a few neuter gender nouns with the final letters -mja (like vremja 'time', imja 'name') 7 , cf. Table 9.
It should be noted that Russian neuter gender nouns are very infrequent in the lexicon, as compared to feminine and masculine forms (Schwartz et al. 2015, p. 729). A comparison with gender frequencies of German nouns is provided in Table 1, since German and Russian are often described as languages with similar gender systems (Schwartz et al. 2015, p. 729;Bordag et al. 2019, p. 131).
N2 nouns declension paradigms differ slightly phonetically depending on whether the final word vowel (inflectional suffix) is stressed (N2s) (like in molok o 'milk') or unstressed (N2u) (like in s alo 'pork fat'). Due to unstressed vowel reduction in Russian, the inflectional suffixes of F1 and N2u are homophonic in the nom. Words with stressed final -e vowel are very rare, come as loan words from other languages and typically belong to indeclinables (e.g., sufl e 'soufflé'). As they are not found in DR, they won't be considered here.
For the purposes of this paper, the distinctions in the inflectional paradigms between F1, M2 and N2 are of the highest importance and are therefore provided below in Table 2 which shows the words mama 'mother', stol 'table', mesto 'place', selo 'village', and pole 'field' in transliteration. The homophonies of the inflectional suffixes in the nom. and acc. cases are indicated in square brackets.
Gender distinctions in Russian are more prominent in the singular than in the plural, a feature commonly found in many languages (Audring 2017, p. 59). Therefore, in this article we only focus on the singular forms of nouns, and not on the plural forms that do not significantly differ by gender either within StR or between StR and DR.

Doukhoborese
Doukhobors are a religious and cultural minority who immigrated to Canada from Russia in 1899 (Tarasoff 1982, pp. 39-46). Widely spoken by thousands of people until the late 20th century (Schaarschmidt 2008, pp. 236-237), Canadian DR is now almost extinct. This unique language variety is only spoken at near-native proficiency levels by about 200 individuals of Doukhobor ancestry (Makarova 2012, p. 91). Only a few studies on Doukhoborese have been carried out and not all aspects of the language are covered. Some that are include language history, ritual style and ritual use (Schaarschmidt 2008(Schaarschmidt , pp. 102-122, 2012Makarova 2012, pp. 90-96), Doukhobor vocabulary (Harshenin 1964(Harshenin , 1967, and its sound system (Makarova et al. 2011, pp. 146-151;Makarova 2013, pp. 53-68). It was noted in earlier research that the use of some forms of nouns in Doukhoborese may differ from StR use (Schaarschmidt 2008, pp. 102-122). In particular, the observation that DR neuter nouns appeared to have differences in their inflection paradigm from StR was put forward in earlier research (Makarova 2012, p. 92). However, no detailed studies of any aspect of Doukhobor grammar (including nominal declension) are available up to date. The author of this paper observed instances of neuter gender erosion in the acc. case and its replacement with the feminine gender F1 inflectional suffix -u and feminine gender agreement (in acc. and nom.) in Doukhobor hymns while attending Doukhobor prayer services between 2010 and 2015. The following DR example was found in a recording of a DR hymn: 8 (1) DR nashu acc.sg.f sontsu acc.s.f ne zakrojut nikakija tuchi StR naše acc.sg.n solnce acc.sg.n ne zakrojut nikakie tuči our sun not cover any clouds 'our sun will not be covered by any clouds' Example (2) illustrates agreement with an adjective in a post-nominal position: (2) DR dela nom.sg.f slavnaja nom.sg.f , druz'ja nom.pl StR delo nom.sg.n slavnoe nom.sg.n , druz'ja nom.pl deed glorious friends 'a glorious deed, my friends' 8 All Doukhobor words and phrases have been transcribed using the ALA-LC romanization system with a few exceptions: Firstly, to reflect idiosyncracies of DR, the w symbol was used for the vocalized v glide, and gh-for the voiced velar / glottal approximant not found in StR. Secondly, some minor distinctions from the ALA-LC transliteration were introduced (ja for the cyrillic я, ju for ю, and ts for ц). The ALA-LC symbols for the soft sign ь ( ) and hard sign ъ ( ) were modified to (') and (") to clearly differentiate them from the stress mark (") and quotation marks. Finally, the system was also slightly adapted to reflect pronunciation rather than spelling in DR (a variety that has no writing system). For the transcription of the StR examples the scholarly system has been applied.
In (2), the inflectional suffix of the noun dela 'deed' (delo in StR orthography) has a reduced [@] vowel, which is homophonous between the F and N genders in StR. However, in DR the adjectival inflectional suffix in the feminine form (-aja) disambiguates the noun as being of feminine gender, and in StR-as neuter.
Conversations with as well as subsequent recordings of DR speakers supported the hypothesis of the neuter gender erosion in DR. The class of nouns equivalent to N2 in StR is referred to here as 'ex-neuter'. Indications of neuter gender erosion in earlier research, as well as evidence of the phenomenon in spoken language, were the motivation behind this study.
The research objective of this paper is to describe the declensional paradigm of ex-neuter nouns in DR (equivalents of N2 class in StR) and to establish patterns of pronominal and adjectival / numeral and verbal agreement with ex-neuter nouns. We tried to find answers to the following questions: • What is the pattern of case declension of ex-neuter nouns in DR?
• Are there any subclasses of ex-neuter nouns that can be identified, and more specifically, • do stressed inflectional suffixes affect the inflectional paradigm and agreement?
• What patterns of gender / case agreements of ex-neuter nouns with other phrase / sentence constituents exist?

Theoretical approaches
This study is empirical, and analyzes noun paradigms from structural perspectives. The analysis is mostly based on a traditional morpheme-based model which views a complex word as a combination of its morphemes (Selkirk 1982, pp. 1-12). However, it is also inspired by construction morphology (Booji 2010, p. 1), a theory that was created with the purpose of providing "a better understanding of the relation between syntax, morphology, and the lexicon" (ibid.). Construction morphology emphasizes the interrelationship between lexicon and grammar, and in this way it is well suited for the analysis of gender in Russian. Consequently, it has already been used to analyze the gender inflectional paradigms in Russian (Booji 2010, pp. 10-11).

Materials and methods
The materials employed in this study come from 32 hours of conversations and monologues (semi-structured interviews on topics related to Doukhobor beliefs, lifestyle and participants' childhood experiences) by 30 highly fluent authentic Doukhobor speakers. The data were recorded in Saskatchewan and Brithish Columbia between 2012 and 2018. 9 Authenticity refers to the participants learning the variety as their mother tongue at home and in the community, and not being significantly exposed to StR or other varieties of Russian (i.e., they did not study Russian abroad, did not spend more than 1-3 weeks in a Russian-speaking country, did not study Russian as a foreign / second language in Canada, and did not work or reside with speakers of StR for longer than a few weeks). The participants were 17 men and 13 women whose average age was 79. The youngest participant was 73, and the oldest 96. The records were conducted in field conditions using a ZoomH2n recorder with the sampling frequency of 32KHz. The records were manually transcribed. Instances of the use of ex-neuter-gender nouns and their agreement with other phrase / sentence constituents were extracted from the transcripts and verified by audio records by the author with the help of two more speakers of StR with graduate degrees in linguistics. A total of 230 instances of ex-neuter-gender with their immediate left and right contexts and pronominal, adjectival and verbal agreement forms were extracted from the records. In addition, sample elicitations of a few ex-neuter declension word-forms were attempted with three participants. The number of elicitations was restricted due to time limitations, the age of the participants, and their low interest in and experience with grammatical questions.

Results
The highest frequencies of ex-neuter nouns in the sample are represented in Table 3.
For a word to be entered into Table 3, it had to have an absolute frequency of at least seven, and be found in the speech of at least two different participants. The words wremja 'time', imja 'name' are ex-neuter of a different subclass, not of N2 (cf. Sect. 1); as can be seen in Table 3, ex-neuter-gender nouns are rare in DR (and they are also rare in StR).
The data obtained allowed us to differentiate between the various subclasses of ex-neutergender, as described below.

DR ex-neuter forms with an unstressed vowel [@ @ @] in the inflectional suffix
The examination of the inflectional paradigm was difficult due to the low frequency of exneuter nouns as well as due to morphosyntactic / semantic asymmetries. Depending on the lexical meaning, a particular word could be found predominantly in one declension case, e.g. the prep. case, if the word denoted a location, but not in others. Many nouns were found in the nom. and acc. cases, but they were extremely rare in the dat. and ins. cases. Therefore, the inflectional paradigms had to be partially reconstructed (following other ex-neuter words). An example of the DR ex-neuter word mesta 'place' with an unstressed vowel [@] in the inflectional suffix is provided in Table 4.
As Table 4 demonstrates, in DR speech production as well as in the three available elicitations, all the instances of the word mesta 'place' have acc. forms that follow the StR feminine (F1) paradigm, and not StR neuter (N2). Dat. case word-forms of mesta 'place' also follow the feminine declension pattern. Ins. case forms were not observed in speech production, but the available two elicitation forms also follow the feminine declension pattern. The prep. case has inflectional forms that are uniform across genders in F1, M2 and N2, so gender distinctions in declension paradigms are neutralized in StR as well as in DR. A 'gray area' in the declension paradigm is the gen., in which DR speakers use forms of both StR neuter / masculine and feminine genders, and these results are similar in speech production and elicitations.
Out of a total of 43 occurrences of the word mesta 'place' in the singular, 22 had agreement, and of these, 20 were in the feminine gender, see (3) and (4), and only two had agreement in the masculine / neuter gender (the latter examples were only found in the speech of one participant), see (5). It should be noted that, except for the nom. and acc. cases, StR M2 and N2 forms are homophonous. Thus, it is impossible to differentiate between them in other cases: (  Gen. m"asla / m"asly 3/1 Dat. x (*m"asle) 0 Acc. m"aslu 14 Ins. m"aslaǐ 1 Prep.
-(*m"asle) 0 (5) Mnogha w adnom prep.sg.n meste prep.sg.n many in one place 'Many in one place.' Another sample paradigm of a DR ex-neuter-gender word masla 'butter' with an unstressed [@] vowel in the inflectional suffix is provided in Table 5. The inflectional paradigm for the word masla 'butter' is very similar to the declension of the word mesta 'place'. The acc. forms uniformly follow the F1 declension paradigm, and the gen. case forms are split between the N2 / M2 and F1 declension classes.
Out of eight instances of agreement observed in the sample for the word masla 'butter', seven were in feminine and only one-in neuter gender: (6) ty veziėtu acc.sg.f maslu acc.sg.f w ghorad you take this butter to town 'You take this butter to town.' (7) i masla nom.sg.n , ano svezhee nom.sg.n shtoby bylo past.3sg.n and butter it fresh so.that was 'and butter, so that it would be fresh'

DR ex-neuter forms with the stressed vowel [o] in the inflectional suffix
Only two nouns of the ex-neuter-gender N2 class subgroup with the stress on the vowel of the inflectional suffix were observed. Their declension forms are represented below in Tables 6  and 7. Out of a total of 34 occurrences of the word sel o 'village' in the singular, there were 19 instances of agreement. In 13 cases, the agreement followed feminine forms, see (8) 'they were driven into our village, into Orlovka' As Tables 6 and 7 demonstrate, these words largely maintain the inflectional forms of the N2 declension, i.e., stress on the inflectional suffix blocks the conversion of these forms into F forms.
Of nine instances of usage of the word malak o 'milk' in the sample, there were only 2 instances of agreement, and both of them followed the feminine pattern (10), (11):

DR ex-neuter equivalents of StR words ending in -ie / -'e
A few words with a relatively high frequency in DR (such as sabranija 'meeting', malenija 'prayer service', vaskresenija 'Sunday', penija 'singing') are the equivalents of StR words ending in the orthographic form with ie / 'e letters; these words are very frequent due to their significance in Doukhobor culture. A declension paradigm for these words is represented in Table 8. As can be seen in Table 8, the nom. and acc. cases assume F1 forms (with the exception of one acc. form by one speaker). In the prep. case, reduced forms of the vowel [@] (found in StR F1, M2) are acoustically and perceptually very similar to [i] (StR N2), so it is difficult to make a clear distinction. Only one form of dat. was observed in DR, a form which does not differ from the StR dat. Although the word malenija 'prayer service' was not observed in DR in the ins. case, the form can be reconstructed based on an ins. case form sabranieǐ 'meeting'.
The word malenija 'prayer service' was used in the sample in the singular 28 times, of which there were seven instances of its use with concord or agreement, six instances following the feminine gender pattern, cf. (12), and one the neuter / masculine one, cf. (13). (13) k kazhdamu dat.sg.n maleniju dat.sg.n tri psalmy spajut for each prayer.service three psalms will.sing 'for each prayer service they will sing three psalms'

DR equivalents of StR ex-neuter forms ending in -e
There were only two low frequency words with the inflectional suffix -e in the sample. One of them was the equivalent of StR more 'sea'. The word was not found in the DR sample in the nom. However, the nom. can be reconstructed as *morja 'sea'. The word has the feminine (F1) inflectional form in the acc. (*morju 'sea'). In the ins. case, the word form is morem 'sea' (M2 / N2). In the prep. case, the word assumes a form *more 'sea' with an inflectional suffix -e which is common between F1, N2 and M2. One more lexeme ending with an unstressed -e in the corpus was otche 'father'. This word was only used in the sample in reference to the 'Lord's Prayer', which starts in Russian with otče naš 'Our Father'. In StR this word is also used mostly in the Orthodox version of the same prayer. Due to its use in the beginning of the most commonly recited prayer, the nom. form probably became fossilized as otche 'father' and did not undergo a transition to the feminine form. However, in the acc., the feminine gender form otchu 'father' (frequency = 9) is found more often than the neuter form otche 'father' (frequency = 5). There were no other case forms of this type found in the sample except for the nom. and acc. There was one case of agreement observed with the word more 'sea' following the F pattern (14). The word otche 'father' was used with the feminine concord three times. It was also found with the pronoun nash 'our' (masculine) in a postposition, but this use is idiomatic, as it relates to the beginning of the Lord's prayer otche nash 'Our Father':

DR equivalent of StR 'mixed' -mja declension
In StR, this declension type is comprised by only ten words of neuter gender. Sometimes they are referred to as -mja nouns, because all of them have a stem ending with a palatalized [m'] and an inflectional suffix [-ja]. These nouns mostly take on F3 declension endings (following the declension paradigm found in feminine nouns ending with palatalized stems), and are therefore sometimes assigned to the F3 declension (Zaliznjak 2002). However, in the ins. case, they follow the M2 declension inflectional suffix. They also have -en-suffix added to the stem in all the oblique cases except for the acc. (that has the same form as the nom.). As Table 9 shows, the word wremja 'time' in DR is undergoing inflectional merging of forms with just a few forms following the F1 feminine gender declension. Some of the DR forms of this word (gen. wremja 'time ', prep. wreme 'time') are also found as colloquialisms in Russian. The word imja 'name' was observed four times in the nom. and twice in the acc. form imju 'name' (following the feminine F1 declension type). The word wremja 'time' may be displaying fewer forms that transition to the F declension due to the existence of idiomatic or fossilized expressions like 'during that time' (vėta wremja), etc.
The word wremja 'time' mostly takes on feminine agreement forms (27 instances, cf. (15)), however, it also occurred with neuter / masculine agreement (13 instances, cf. (16)) and twice with 'mixed form' patterns, see (17) The determiner to in this example is in the neuter form, but the noun takes the feminine (F1) acc. form.

Discussion
The results clearly indicate a shift in gender / declension class of ex-neuter singular nouns in Doukhoborese. This shift mostly affects the F1 (feminine gender, declension 1) type and is particularly apparent in ex-neuter nouns with a stress on the stem in the acc. case. A stress on the inflectional suffix largely blocks the paradigm shift, but does not fully prevent a gender shift in agreement. The shift towards feminine gender questions the generic assumption that the masculine is the 'default gender' in Russian (Steinmetz 2006(Steinmetz p. 1420. Multiple shifts in gender / declension classes have been described in world languages. Germanic and particularly West Germanic (English, German, Dutch) nouns are known to have shifted gender (Steinmetz 2006(Steinmetz , p. 1418. In this process, English lost grammatical gender altogether, merged masculine and feminine into one common gender, and German acquired a three-gender system (Kürschner and Nübling 2011, pp. 357-358, 370, 372).
Some evidence of partial shifts to a bi-gender system has also been described without much detail for Central and Southern Russian dialects (Matthews 1950, pp. 139-140;Pšeničnova 1996, p. 101). For example, Matthews (ibid.) points out "a curious phenomenon" of "the coalescence of feminine and neuter forms where the endings of the latter are not stressed. And even when the stressed neuter ending survives, the accompanying adjective invests a feminine form in the nominative and accusative cases (e.g., bal'šaja aknó 'a big window' [. . . ]), though in the others the neuter paradigm prevails (e.g. pad našim aknóm 'under our window' [. . . ])." 10 It has also been observed in other studies that the loss of the neuter which we refer to as 'neuter gender erosion' in other Russian dialects, may involve the partial or complete loss of the neuter gender, the decrease in number of neuter gender nouns, their shift to feminine or masculine declension patterns. It has also been stated that this shift can be observed not only in inflectional forms, but also in the gender agreement of nouns and 'their' adjectives, pronouns, etc. (Bromlej et al. 2005, p. 122). This article suggests the term 'ex-neuter' that reflects not only agreement but also declension paradigm shift. It is also suggested that subclasses of the ex-neuter gender can be differentiated within it (with the stress on the stem, stress on the inflectional vowel -o, and a separate subgroup of -mja ex-neuter).
Neuter gender erosion appears to be a part of a generic simplification of the nominal gender and declension paradigms in Russian (Bromlej et al. 2005, p. 118). Data available on neuter gender erosion in Russian dialects are fragmented and incomplete due to the loss of dialects and their substitution with StR (Bromlej et al. 2005, p. 6). Therefore, some additional evidence of the same phenomenon coming from DR is of special interest for Russian grammar and dialectology. As noted in Sect. 1, DR comes mostly from Central and Southern Russian dialects, and the erosion of the neuter gender is proof that it inherited some features of these dialects.
The 'minority' status of the neuter gender in Russian and its very low frequency (Steinmetz 2006, p. 1420) may explain why it has been eroding in Doukhoborese as well as in some other Russian dialects. It has been suggested that Russian had a neuter-default gender system in the past, but the language has been shifting towards increased 'marginalization' of neuter nouns (Steinmetz 2006(Steinmetz , p. 1420. The 'minority' ex-neuter forms are overgeneralized and follow F1 type, particularly because their nom. forms are homophonic with F1 forms in the case of an unstressed [@] vowel in the inflectional suffix. Neuter nouns have also been demonstrated to be a 'weak spot' of language acquisition for both monolingual and Russian-English bilingual children; their correct use is acquired much later than that of masculine and / or feminine nouns (Schwartz et al. 2015, p. 728) probably for the same reasons (homophony with the F1 nom. and overgeneralization).
Thus, the shift to the F1 type in the ex-neuter in DR is phonetically and grammatically motivated. In a way, it may be more surprising that StR still maintains this class. 11 It is not fully clear why neuter forms are retained in DR in a few exceptional cases. The latter could probably be explained by the effects of literacy, some small amounts of exposure to Russian language education, some idiomatic expressions, particularly word usage intrinsic to Doukhobor prayer services (such as otche nash 'our Father' in the Lord's prayer). An interesting illustration of the impact of StR was presented in the case of one participant whose speech was not included in the analysis reported in this article due to a very strong influence of StR. He was not only exceptionally literate in Russian, but was also a man of letters, an author, a scholar and could cite by heart long passages from Tolstoy and other Russian authors in StR. He started his interview in grammatically impeccable StR, including StR neuter forms. However, as time went by, and the speaker became more tired and also possibly less conscious of the presence of the interviewer, who was a StR speaker, some ex-neuter forms with F1 shift (and other Doukhobor-specific forms) started to appear in his speech when the neuter gender words belonged to conversational speech, and not to literary or philosophical concepts.

Conclusions
The structural grammatical analysis reveals specific characteristics of ex-neuter nouns and their subclasses along with their declension paradigm. Ex-neuter nouns are shown to follow a mostly female gender declension (e.g., bal'shaja sela 'big village', nom. sg., nashu sontsu 'our sun', acc. sg. plat'ju mnogha bylo 'there were plenty of clothes', gen. / partitive sg.) with some exceptions (e.g., otche nash 'Our Father'). A subclass of DR ex-neuter nouns is identified as the equivalent of neuter gender nouns in StR that end in a stressed vowel (like okn o 'window', sel o 'village', pleč o 'shoulder') and which do not undergo a F1 shift in the declensional paradigm (inflectional forms), but are still undergoing this process partly, as evidenced by the agreement of these words with other grammatical constituents following F patterns. The suggested terms 'neuter gender erosion' and 'ex-neuter-gender' reflect the gradient rather than the categorical nature of the process, a transition rather than completed 'neuter gender loss'. The two -mja declension type neuter words in DR (wremja and imja) are on their way to becoming indeclinables rather than shifting to the F declension pattern.
Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.