Does language transfer explain it all? The case of first language change in Russian-English bilinguals

Full Text

1. Introduction

Over the last few decades numerous studies of bilingualism and second language acquisition (SLA) have provided ample evidence of the instrumental and cognitive benefits of achieving fluency in a foreign language. Bilinguals are no longer viewed as imperfect hybrids of two monolinguals; rather, the high value and undeniable advantages of knowing more than one language overwrite the costs of being bilingual, such as unavoidable changes in the first language that often happen as a result of speaking a second language.

The bilingual’s two languages do interact and influence each other. The study of such influences falls in the domain of language transfer or cross-linguistic influence (CLI), which has been a field of extensive research in the past few decades (e.g. Odlin 1989, 1990, Larsen-Freeman & Long 1991, Selinker 1992, Ellis 1994, Kellerman 1995, Gas & Selinker 1994, 2008, Jarvis & Pavlenko 2008). Cross-linguistic influence usually is studied from the perspective of foreign language acquisition and first language attrition. Odlin (1989) defines transfer as “the influence resulting from the similarities and differences between the target language and any other language that has been previously (and perhaps imperfectly) acquired” (27). Furthermore, an earlier study comparing L2 acquisition and L1 forgetting in two groups of participants, Russian English attriters and English speakers learning Russian as L2, suggested that CLI can indeed be the cause of similar syntactic transfers both in L2 acquisition and L1 attrition (Isurin 2005).

The present paper discusses findings from an empirical study that looked into grammatical changes of Russian (L1) under the possible influence of English (L2) in a group of Russian-English bilinguals residing in the U.S. and it mainly focused on two variables – length of immigration and the daily exposure to L1. It aims to contribute to the accumulated knowledge of how first languages changes under the influence of another language in bilingual speakers.

2. Cross-linguistic influence: the state of the art

As mentioned earlier, traditionally CLI is studied from the perspective of second language acquisition and first language attrition. The field of SLA puts great emphasis on studying possible negative influences that the native language can exert on the newly acquired language. Javadi-Safa (2018) presents an exhaustive overview of different hypotheses and models within the CLI framework and the historical overview of research on transfer, predominantly as it applies to a facilitative transfer in second language acquisition (for more information see the original publication). Here I will limit myself to a brief discussion of a few major theories that underlined the investigation of language transfer in the last few decades.

CLI can affect not only the L2 learner’s performance in the target language but also can become a cause of changes (commonly known as L1 attrition/ loss) in the individual’s native language. In the past, studies on transfer from L2 to L1 relied on once well-established, however lately abandoned, theories, such as Contrastive Analysis and Markedness (e.g. Seliger & Vago 1991, Seliger 1996). According to this notion, if two language grammars come into contact and the L2 category is less marked than the corresponding L1 category, then the latter is likely to be replaced by a rule transferred from L2. In other words, transfer will always move in the direction of the less marked category.

The last few decades of the 20^th century also have seen much research conducted within the framework of universal grammar (UG). According to this approach, L1 changes often result from an attempt to simplify the grammatical sentence in order to avoid redundancies in the two languages (Levine 1996, Seliger 1996). According to Seliger (1996: 617), “forgetting within L1 is not random forgetting but guided by a principle of arriving at the most parsimonious grammar that can service both languages.”

The first decade of the 21^st century has witnessed a few attempts to redefine the concept of transfer as it applies to L1 attrition. Cook (2003) was among the first to offer new terminology, such as L1 change (rather than L1 attrition) and L2 effect (rather than transfer). In light of this re-conceptualization of the whole notion of bilingualism, he promoted the idea of L2 users becoming the norm in the modern world and suggested that the bilingual’s linguistic performance – as imperfect as it might be – exhibits distinctive characteristics and has its own rights. Around the same time, Pavlenko (2000, 2003) proposed a classificatory CLI framework describing different instances of transfer, such as lexical borrowings, semantic shift, restructuring transfer, et cetera.

Jarvis and Pavlenko (2008) take a fresh look at the field of transfer and add new perspectives, such as directionality of transfer. CLI is no longer bound by forward (from L1 to L2) or reverse (from L2 to L1) direction but also can be lateral (from L2 to L3) and bidirectional. Bidirectional transfer is defined as the mutual influence of the two languages within the same individual. The authors call for a new approach to the study of transfer and show the need to have a more interdisciplinary approach in order to identify and measure CLI in its context. In addition, they delineate two general approaches to the study of transfer: intrasubjective and intersubjective. The former approach concerns single case studies, either experimental or longitudinal, whereas the latter involves a group of language speakers. The aim of the intersubjective approach is to identify common trends exhibited by a group of bilinguals whose L1 and L2 remain the same.

However, one of the unresolved issues in the field of CLI concerns the very terminology used in the literature. Scholars often use the terms transfer, cross-linguistic influence or linguistic interference interchangeably. Moreover, there is no consensus on defining transfer. A collection of works on code-switching (Isurin, Winford & de Bot 2009) raised the question of separating transfer from code-switching or borrowing. The contribution by Odlin (2009) continued the debate of the right terminology from the linguistic perspective, while Marian (2009) showed a way of looking at code-switching and transfer from a psycholinguistic perspective with the use of rigid experimental instruments. Also, Marian and Kaushanskaya (2007), in their study involving Russian-English bilinguals, demonstrated a clearly identified trend of code-switching on nouns (overt change) and linguistic transfer on verbs (covert change). To illustrate the latter, participants would produce in English (L2) the following utterances that suggest transfer from Russian (L1): walking through a street for ‘crossing the street,’ I was going to the first grade for ‘I was in first grade,’ or we left from Kiev sometime in April for ‘we left Kiev…’

To summarize, a flurry of research on CLI in the last few decades has identified a few contested territories, such as the attempt to find the right theoretical framework and reconcile conflicting terminology used throughout the field. The present article does not aim at solving these fundamental problems. Instead, it takes a functional approach to the linguistic analysis, adopts the intersubjective approach proposed by Jarvis and Pavlenko (2008), and works on the assumption that transfer is a covert change in the individual’s L1 as a result of the L2 influence. In this light, the next section will look at recent findings related to those registered in the present study.

3. Cross-linguistic influence: findings relevant to the present study

The present study concerns changes in L1 under the influence of L2 in two domains, namely, a change in word order and null subject.

3.1. Word order

 One of the major concerns of early SLA studies on language transfer was how word order in L2 might be influenced by the structural differences of word order in L1 (Rutherford 1983, Sharwood Smith 1986, 1990, Zobl 1986a, 1986b, Fathman & LoCoco 1989, Odlin 1990). The results of the studies looking into word order transfer were conflicting. Some indicated that L2 acquisition is affected by SVO (subject-verb-object) in L1 (Zobl 1982), while others disputed this notion (Rutherford 1983). Scholars working within the UG framework maintained that the initial word order acquisition is guided by universal principles rather than cross-linguistic influence (Klein & Perdue 1993, Zanoon 2016). However, Odlin (1990) argued that there is no universal constraint on transfer of the basic word order, a claim that was previously supported by findings in Trevise’s (1986) study. One of the recent ERP (event-related brain potential) studies looked into how L1-Spanish speakers acquiring Basque as L2 process non-canonical SVO and OVS (object-verb-subject) in Basque (canonical word order in Basque is SOV and in Spanish – SVO). The results indicated that bilinguals heavily relied on their L1 while processing non-canonical Basque word orders, which further supported the earlier finding about the role of the word order in L1 that might govern the acquisition of L2. The authors concluded that “it could be that non-native speakers never process the grammatical structure of L2 in a native-like manner, mostly if they already can do it transferring the cues of their L1 grammar” (Erdocia & Laka 2018: 8).

Besides SLA studies, the role of CLI in word order was studied in the bilingual context. A recent study on word order transfer in 117 Dutch-English bilingual children demonstrated a role of CLI in acceptability judgment task (Bosch & Unsworth 2020). While English is an SVO language, Dutch is a V2 (“verb second”) language, meaning that if an element other than the subjects appears in clause-initial position, subject-verb inversion takes place. Although none of their participants produced VS word orders in English, they were more likely than monolinguals to accept V2 word orders as correct in the grammaticality judgment task. In addition, bilinguals were more likely to accept ungrammatical V2 word orders in English if they were exposed to relatively more Dutch than English. Thus the authors concluded that in some cases CLI may be more persistent than previously thought.

However, transfer of word order cannot only affect acquisition of foreign languages but also can lead to changes in the native languages. The research in this area has not been as extensive as in SLA. Merino’s (1983) study showed deterioration of the performance on word order in Spanish (L1) with an increase in performance of the corresponding category in English (L2). Liu, Bates, and Li’s (1992) study found that late Chinese-English bilinguals transferred English-like word-order strategies to the interpretation of sentences in Chinese. Word order information in L1 comprehension was found to be particularly vulnerable when there was a phonological similarity between L1 and L2 verbs used in a grammaticality judgment task (Altenberg 1991). In Shaufelli’s (1996) study on Turkish immigrants in the Netherlands, the analysis of the elicited spontaneous speech showed more variability in word order, whereas the sentence interpretation task showed that the participants relied more on word order. The author suggested that word order changes could result from L2 (Dutch) transfer as well as the language’s internal pressure. The latter concerns internal restructuring of the native language that leads to a more consistent pattern.

In her study, Isurin (2005) looked at two sets of bilinguals whose languages remained the same – Russian and English – but the order of acquisition was the reverse. In other words, Russian-English bilinguals undergoing changes in their L1 and English learners of Russian as L2 participated in the study. Typologically, English and Russian differ in terms of permissible word orders. While the former has a fixed SVO (subject-verb-object) order, the latter allows six mathematically possible combinations, with SVO being basic and pragmatically the most neutral (Comrie 1979). The study analyzed the data from two sources: a longitudinal study on a Russian child adopted by an American family and an experimental study where story elicitation was used as the main psycholinguistic tool. The results of the longitudinal study suggested that the fast process of L1 attrition in a child was accompanied by a restructuring of her word order, with the VS (verb-subject) pattern almost disappearing and SVO remaining a dominant word order by the end of the observation period (a year after Russian input abruptly ended). This finding encouraged the author to look further into this particular phenomenon. The experimental part of the study involved three groups of participants: monolingual Russian speakers, Russian-English bilinguals residing in the U.S., and English-speaking learners of Russian as L2. Picture-based story telling was used as an elicitation technique. Although there was no clear evidence for word order transfer in the group of Russian-English bilinguals, the results of the study indicated that a tendency to use VS order less frequently may be a function of longer residence in the L2 country and higher exposure to L2. Conversely, the use of the inverted (VS) word order in L2 learners of Russian increased with higher fluency in L2. What may be of particular relevance to the present study is that L2 learners of Russian demonstrated a tendency to more frequently use so-called existential sentences where VS use is required. Russian uses the inverted VS order for existential and presentational constructions (see Polinsky 2006, for further references). The author suggested that such a sentence is one of the earliest learned in L2 Russian classes and L2 learners may rely on it heavily during the early stages of L2 acquisition. The reduced use of a variety of word orders in speakers of ‘American Russian’ – first and second generation immigrants – also was reported by Polinsky (2006). Moreover, Harrison (2011) identified instances of word order transfer in
Polish-English bilingual children.

3.2. Null subject or pro-drop

Another area of concern for the present study is so-called null subject, also known as pro-drop, which is a drop of an overt realization of subject from the grammatical surface of the sentence. Although Russian is not a language allowing pro-drop in all instances and shows constraints on where null pronominals can occur (Franks 1995, Lindseth 1998), it nevertheless allows for a null subject, which happens not only in colloquial Russian. For example, the sentence prishli, poeli i legli spat’ (came, ate, and went to sleep) would be correct grammatically and the corresponding referents (pronouns ‘we’ or ‘they’) easily could be recovered based on the inflected verb ending and the referent previously used in the context. English, on the other hand, does not allow for pro-drop, and using null subjects is ungrammatical in most instances other than the omission of a personal pronoun at the beginning of an English sentence in instances of colloquial usage (informal questions and answers), in imperatives, and lately in informal cell phone text messaging.

A few previous studies reported changes in this particular feature as a result of the individual’s exposure to another language. Gürel (2004) studied a group of native speakers of Turkish who had been living in North America (U.S. and Canada) for a prolonged period of time and found clear evidence of changes in pro-drop, as it is used in Turkish. The author discussed her findings within the framework of the Activation Threshold Hypothesis and argued that such change in the use of null subjects happens as a result of a higher level of activation of a competing language system (L2) which does not allow for this particular syntactic property. Harrison (2011) registered decline in pro-drop in the speech of Polish-English bilingual children and attributed it to transfer from English. Isurin’s (2011) study of Russian immigrants in the U.S. provided further evidence of change regarding null subject that happen with more exposure to L2. Her participants were Russian-English bilinguals, monolingual English speakers, and a control group of Russian monolinguals. Bilinguals differed in terms of their daily exposure to English and general level of integration. Significantly less frequent use of null subjects in Russian was attributed to the factor of L2 exposure. The author studied a phenomenon of null pronominal use by bilinguals from a perspective different from that commonly used in the field of CLI. Pro-drop was viewed as one of the characteristic features of languages associated with so-called collectivist cultures (contrasted with individualist cultures) – a line of thinking that emerged over the last few decades within a collectivism and individualism construct in cross-cultural psychology. Despite clear evidence of the diminished use of null subjects among well integrated bilinguals with low level of exposure to Russian the author concluded that changes regarding pro-drop probably should be better studied within the CLI framework.

In one of the most recent longitudinal case studies on pro-drop in a contact situation, a Bulgarian – German bilingual with 17 years of residence in Germany was tested at four points during the five yearlong investigation and showed a significantly higher rate of overt pronominal subjects in L1 (L1 is pro-drop language; L2 – non-pro-drop) than the rates of ten monolingual controls when tested in Germany. However, after three weeks of L1-reexposure in Bulgaria attrition effects disappeared and the overt subject rate fell within the monolinguals’ range. The study adopted a psycholinguistic approach considering both language dominance shift and attrition as modifications of the availability of linguistic structures for ongoing language processing. The results did not support the predictions and the authors concluded “that peculiarities of performance observed in L1 attrition are probably depending much more on language mode and activation states than on restructuring of linguistic representations” (Kӧpke & Genevska-Hanke 2018: 13).

If pro-drop disappears or becomes less frequent as a result of the bilingual’s exposure to another language which does not allow for it, how can we explain findings coming from studies where both languages allow for pro-drop, yet the occurrence of null subjects in L1 changes? Similar changes in the use of null subjects were registered in two studies involving Spanish (L1) in Italy and Italian (L1) in Spain (Sorace 2004, Sorace & Serratrice 2009). Both languages in question allow for the use of null subjects; however, changes occurred in the use of null subjects in respective L1s. Moreover, Russian speakers in the U.S. and Israel showed similar patterns of the disappearance of pro-drop, despite the fact that in Hebrew null subjects in the first and second person are preferable or even required (Dubinina & Polinsky 2013). Dubinina and Polinsky (2013) join Sorace and her colleagues in supporting the Interface Hypothesis that suggests that such an unexpected change actually may result from the fact that a language spoken outside of its native environment may undergo changes that pertain to the interface at which a particular linguistic feature is located. In the case of pro-drop, it is the interface between the syntactic rule (null pronominals can occur only in specific syntactic positions) and the information sentence (the pronoun’s referent and the referent of the nearest topic should be the same). In addition, the authors suggest that less processing effort is required in order to overtly state a lexical pronoun. Since the cognitive load placed on the bilingual speaker operating between the two languages is usually high, we may add that such processing economy indeed may be a mechanism behind those changes that cannot be explained within the CLI framework. The authors agree that we do not have much information as to why such changes occur in the first place, especially in cases where both languages allow for the same feature, such as null subject.

To summarize, prior studies on syntactic changes in L1 under the influence of L2 reported instances of word order and null subject changes. The study presented in this article aims at bringing additional evidence of L1 changes under the influence of L2 in the above two domains and it specifically looks into two factors, length of immigration and the amount of daily exposure to L1, that may account for the registered changes. 

4. Study design and methodology

4.1. Participants

Fifty participants took part in the study: 20 monolingual Russians and 30 Russian-English bilinguals. The pool of Russian monolinguals was recruited in St. Petersburg, Russia. The age of the participants in this group ranged
from 27 to 57. All bilingual participants emigrated from Russia after their L1 was fully acquired and had lived in the U.S. for 10–31 years by the time the study was conducted. The age of this group ranged from 30 to 76. Most of the participants in the study were college educated or had some post-secondary degree. The study controlled for two variables: amount of daily exposure to L1 and length of residence in the U.S.

4.2. Materials and procedure

A semi-structured interview on autobiographical memories was used as an elicitation technique (such technique was previously used to study code-switching and transfer in a group of Russian-English bilinguals by Marian and Kaushanskaya 2007). The number of questions and the order of their presentation remained constant throughout the study. The interview questions concerned memories associated with birthday celebration, New Year/ Christmas celebration, and vacation time. Memories on recent events as well as those from the participant’s childhood were elicited. All participants were interviewed individually by the researcher; the interviews were recorded and later transcribed for analysis. The elicited data were analyzed both quantitatively and qualitatively.

4.3. Predictions

Based on the evidence gathered by prior studies the following predictions were made. Bilinguals’ narratives will show two trends. First, the use of null subjects will become less frequent. Second, the use of the inverted VS order will become less frequent. If observed, both changes can be attributed to CLI from English that generally does not allow for null subject or for the inverted word order. In line with prior findings from studies on L1 attrition, the amount of daily exposure to L1 rather than the length of residence in the L2 country is expected to have an effect on the above changes (see Schmid 2004 for the detailed overview of the above two factors in the literature on L1 attrition).

5. Results

Altogether, 300 narrative excerpts related to same topics were produced by the participants and later analyzed. The total length of narratives produced by monolinguals was 5,270 words (M = 263.5) and the total length of bilinguals’ narratives was 14,179 words (M = 472.6). The difference in the length of the speech sample partially can relate to the elicitation method. All monolingual participants and the majority of bilingual participants (N = 20) were interviewed over the phone. However, much shorter narratives elicited from monolingual speakers can be attributed to a certain sense of alert that Russians residing in Russia may still have when being interviewed over the phone by an American-based researcher.

The analysis of null subjects as well as the analysis of word order was based on the number of clauses produced by each individual participant. In other words, the ratio of VS order and null subjects produced by each participant was calculated based on the number of clauses produced by this participant. In this analysis, a clause was defined as a production unit containing maximally but not minimally one verb. The definition of clause was adopted from Broersma, Isurin, de Bot & Butlena (2009). The total number of clauses for monolinguals was 990 (M = 49.5), and for bilinguals 2,418 (M = 80.6). 

5.1. Word order change

As mentioned earlier in this paper, the word order in Russian is not fixed, meaning that there are six different word order structures possible without any change in the meaning of the sentence; whereas English has a fixed word order, with the VS order violating the grammaticality of a sentence. One of the goals of the study was to look into word order change, especially the use of the inverted verb-subject pattern. The analysis of narratives revealed a remarkable difference in the use of this particular order between the two groups. Below are the excerpts from monolingual and bilingual narratives demonstrating an observed change.

Russian monolinguals:

(1) Все было как обычно, как это положено у русских: сначала баня, потом застолье, потом в 12 часов слушали президента, и вышли к елке на улицу, где уже собралось наше дачное сообщество (VS). Был салют (VS), были танцы (VS), ходили друг к другу, поздравляли.

All was like always, as it is customary for Russians: first, a bathhouse, then a sit-down dinner, then (we) listened to the President at 12 o’clock, and (we) went outside where all our dacha community got together. There was a firework, there were dances, (we) went to see each other, congratulated.

(2) Нам накрывали такой большой стол родители (VS), я помню, что когда вот было время (VS), что было мало что в магазинах (VS), и мне мама купила … принесла очень большой торт, и там вот был ежик такой печеный (VS)… Parents would set a big table for us. I remember, when there was a time that there was little in stores, my mom bought… brought me a very big cake, and there was a pastry hedgehog there…  

Russian bilinguals:

(3)  Последнее рождество … мы замечательным образом проспали. Поскольку моя жена работала, а я был со своим маленьким сыном. Поэтому мы с ним вовремя легли спать.
Last Christmas… We slept through it in a wonderful way. Since my wife worked I was with my little son. So he and I went to sleep on time.

(4)  Я вообще не отмечаю свой день рождения, мы с моей женой идем куда-нибудь в ресторан, и я не отмечаю их вообще. Потому что я перевалил видимо ту черту, когда мои дни рождения приносили мне радость, поскольку мой каждый день рождения делает меня, грубо говоря, старше. И вообще, я думаю, что за последние 20 лет я свой день рождения не отмечал.
I do not celebrate my birthdays at all, my wife and I go to a restaurant, and I do not celebrate them at all. Because I am past that stage where my birthdays were fun to me, since every birthday makes me, so to speak, older. And I think that in the last 20 years I have not celebrated my birthdays.

As Figure 1 demonstrates, bilinguals produced fewer VS orders in their discourse than monolinguals (M = 5.65 vs. M = 8.76). A t-test analysis indicated that this difference was reliable t (1.48) = 2.51, p < .01.

Figure 1. Inverted word order

Length of immigration can be viewed as a relatively arbitrary measure, as often it is the researcher’s choice where to draw the line separating long-timers from recent immigrants. Thus in this study, it was decided to split groups in three different ways to see whether the results will change depending on a particular regrouping. If we find that no matter how we regroup the participants the measure does not show a change, we may suggest that the found effect indeed is reliable. First, the data were split into two groups, participants who spent 6–17 years in the US and those who spent 21–31 years. The mean VS% was higher for those in the 21–31 years condition (6.16) compared to the mean for those with 6–17 years (5.5). A Bonferroni t-test proved unreliable t < 0.4. Then the participants were split into two different groups, those who spent 6–15 and 16–31 years in the US. The mean VS% was nearly identical across the two groups (6–17 year condition, M = 5.67; 16–31 year condition, M= 5.63). A Bonferroni t-test proved unreliable t < 0.21. Finally, the data were split into three groups based on 6–12, 13–17, and 21–31 year condition split. The mean VS% was higher for those in the 21–31 year condition (M = 6.16) than it was for either of the other two groups (13–17 year condition, M = 5.36; 6–12 year condition, M = 5.76). A one-way Analysis of Variance (ANOVA) with year as the single factor and VS% as the dependent measure proved unreliable, suggesting that the means were highly similar across all three year-conditions, F < 1.0 (Table 1). The results suggest that the length of immigration does not play a role in the bilinguals’ use of the inverted word order.

Table 1. Length of immigration versus the VS use

Next, the variable of daily exposure to Russian was considered. First the participants were split into two groups: those with 40% or more exposure to Russian (M = 59) and those with 30% and less (M = 16). The mean VS% for those with 40% or more exposure (M = 4.52) was smaller than the mean for those with 30% or less exposure (M = 6.41). This effect proved unreliable based on a t-test with an applied Bonferroni correction, t (1, 28) = 1.21, p < .12. Then the participants were evenly split into two group conditions (note: the closer in size the two groups are, the more statistical power the analysis contains): those who had less than 20% Russian input and those who had more than 30%. The mean VS% for those with 30% or more exposure to Russian (4.23) was smaller than the mean for those with 20% or less exposure (7.08). This effect proved reliable based on a t-test with an applied Bonferroni correction, t (1, 28) = 1.92, p < .05. Conversely, daily exposure to Russian produced an effect opposite to our predictions; that is, participants with limited exposure to Russian showed significantly more inverted word orders in their discourse than those who had a high exposure to Russian in the US.

5.2. Existential sentences

A closer look at the word order patterns across the two groups showed that monolinguals use existential sentences much more often than bilinguals. The example of existential sentences in Russian will be a sentence like:

(5)  Там было (V) много людей (S)
      There were many people there.

An existential sentence always requires the word order inversion (VS) in Russian. A further analysis of existential sentences revealed that bilinguals’ frequent use of a demonstrative pronoun “это” (this, it) at the beginning of the sentence that otherwise might have been an existential sentence results in the direct word order and makes the sentence less grammatical:

(6)  Там было (V) много людей (S)
       There were many people there.

Compare with:

(*) Это (S) было много людей (V) ¹
         It was many people.

Below are excerpts from the bilinguals’ narratives demonstrating the above trend:

(7)  Но это была в подавляющем большинстве родительская компания, несмотря на то, что они были с детьми, это были – когда я была совсем маленькая – это были не мои подружки со школы, а это был круг друзей родителей со своими детьми. И только уже в старшей школе, я начала приглашать своих подружек. Да и это тоже было всегда много народу.
But it was mostly my parents’ company, despite that they were with their children. These were – when I was really little – these were not my girlfriends from school, but this was a circle of my parents’ friends and their children. Only in high school I started inviting my friends. So this was also a lot of people.

(8)  Это было деревня, (*) это было работал (*). Мы с ребятами ходили в колхоз, зарабатывали деньги. Это вот было школьное время. Иногда, в раннем таком пионерском возрасте, это были пионерские лагеря. В студенческие – это были стройотряды.
This was a village, (*) this was worked (*). My friends and I went to collective farms, made some money. This was a school time. Sometimes, in early pioneer age, this was pioneer camps. In student years – this was student construction brigades.

Existential sentence in Russian involves the verb “быть” (to be) in its conjugated forms “было, был, былa,” for three genders (neuter, masculine, feminine), and plural form “были” in the past tense. Since the speech samples analyzed in this study concerned autobiographical memories, most of the elicited narratives were in the past tense.

On average, bilinguals produced the demonstrative pronoun “это” at the beginning of a clause that had the potential of being an existential sentence twice as often as monolinguals (10.7% vs. 4.7%). Table 2 illustrates a number of instances where the demonstrative pronoun “это“ used at the beginning of a clause as well as four forms of the verb “быть” were produced by each group. 

Table 2. Existential sentences

Bilinguals produced the demonstrative pronoun “это” at the beginning of a clause four times more often than monolinguals and used all four forms of the verb “быть” more often than monolinguals. However, it would be incorrect to claim that the use of the demonstrative pronoun was always associated with the existential sentence. Oftentimes, the pronoun was used in those cases where the impersonal sentence would be more grammatically correct. The following excerpts illustrate the two observed trends, namely, (9) the use of the pronoun in the existential sentence and (10) in the impersonal sentence:

(9)  Каникулы – да я все время занимался спортом с 11 лет, поэтому для меня все каникулы – это были горнолыжные сборы, то в горы, то в Калговово под Ленинградом, т.е. это была работа, это было веселое какое-то настроение, это были завтраки в столовых…
Holidays – well, I always did some sports since I was 11 years old, so all my holidays – this was skiing camps, either in mountains or in Kavgolovo near Leningrad. In other words, this was work, this was high spirits, and these were breakfasts in cafeteria…

(*)   это тоже было всегда много народу.
       There were many people too.

(10) Это было очень замечательно.
It was wonderful.
Это было очень приятно.
It was very nice.
Я только помню, что это всегда было очень весело.
I only remember that always it was much fun.
Это было существенно более событийно.
It was more eventful.
Это было очень скромненькo.
It was very modest.

Although the above examples (9, 10) taken out of context cannot be considered grammatically incorrect and most native speakers of Russian would accept them as grammatical, the absence of a clear referent for the demonstrative pronoun это indicates that bilinguals consistently use it as a strategy. The observed trend of such frequent use of the demonstrative pronoun at the beginning of sentences that otherwise would have been impersonal and, as such, more correct in standard Russian, shows that bilinguals overwhelmingly adopt a new discourse strategy. This strategy can relate to transfer from English where impersonal sentences begin with the expletive ‘it’ (это), whereas, in standard Russian, impersonal sentences in the past tense begin with the neuter form of the verb ‘to be’ (было).

5.3. Null subjects

The analysis of narratives revealed numerous instances of null subjects in monolinguals’ discourse and its less frequent use by bilinguals. The following excerpts show examples:

Russian monolinguals:

(11) Новый Год справляли в кругу семьи, прихо… пришли друзья… ну весело, выпили, закусили, посмотрели телевизор.
[We] celebrated New Year in a family circle, friends came… well, it was fun, [we] had drinks, ate, and watched a TV.

(12) Новый Год? Ну замечательно традиционно, сначала поздравляем родителей... Поздравляем родителей, так сказать, провожаем с ними Новый Год, старый то есть, старый. Потом до боя курантов возвращаемся домой и уже там, скажем, в тесном семейном кругу у елки встречаем Новый Год.
New Year? Wonderful, traditional, first [we] congratulate our parents. [We] congratulate our parents, so to speak, say goodbye to the New Year, that is, old year…. Then before the Kremlin clock strikes [we] return home and there, say, in a close family circle (we) celebrate New Year by the Christmas tree.

(13) Новый Год? Замечательно справляла. Сначала сидела дома с родителями, с ними встретила, потом пошла к друзьям, вот… у них там было весело, танцевали, ели-пили и общались, гуляли.
New Year? [I] celebrated it great. First, [I] was at home with my parents, [I] celebrated with them, then [I] went to my friends’, well it was fun there, (we) danced, drank, ate, talked, went out for a walk.

Russian bilinguals:

(14) Мы справляли и Новый Год и Pождество в этом году в компании русских друзей. Рождество мы справляли в одной компании, а Новый Год в другой.
We celebrated both New Year and Christmas. We celebrated Christmas in one company and New Year in another.

(15) Мы справляли Новый Год, но мы справляли его в России, потому что мы уезжали туда на 3 недели. Мы справляли его в кругу семьи, с моими родственниками, купили елку, приготовили стол, как обычно в России мы справляли Новый Год и раньше.
We celebrated New Year but we celebrated it in Russia, because we left there for three weeks. We celebrated it in a circle of our family, with my relatives, bought a Christmas tree; put together a table, as always in Russia we celebrated New Year before.

(16) В этом году мы были на Гавайях, в отпуске. И мы справляли это естественно на Гавайях. Для начала мы пошли в японский ресторан, наша любимая кухня, а Новый Год встретили у нас в отеле, потому что это было на берегу океана, где были фейерверки.
This year we were in Hawaii, vacation time. Naturally, we celebrated it in Hawaii. First, we went to a Japanese restaurant, our favorite cuisine. And [we] celebrated New Year at our hotel, because it was by the ocean where there were lots of fireworks.

Clearly, the above monolinguals’ excerpts demonstrate how Russians use null subject, especially when the agent is the first person pronoun, whereas bilinguals’ excerpts almost always have an overt pronominal in the subject position. The difference in discourse patterns is particularly striking since both sets of excerpts relate to the same topic and have the same opening line. The results of the quantitative analysis showed the same trend.

The pro-drop mean was higher in the monolingual condition (M = 13.01) compared with the bilingual condition (M = 9.91). Findings from a t-test indicated that this difference was unreliable, t (1.48) = 1.39, p < 1.0. See Figure 2.

Figure 2. Pro‐drop

Next, two independent variables, length of immigration and amount of daily exposure to Russian, were tested. As in the word order analysis, the bilingual data were analyzed against two variables, the length of immigration and the amount of daily exposure to Russian. Steps similar to those in the analysis of VS were made here. First, bilinguals were split into two groups, those who spent 6–17 and 21–31 years in the US. The mean pro-drop score was lower for those in the 21–31 years condition (M = 7.88) compared to the mean for those in the 6–17 years condition (M = 10.52). A Bonferroni t-test proved unreliable t < 0.85. In order to increase the statistical power owing to the increased number of participants in the higher number of years condition, the participants were split into two different groups, those who spent 6–15 and 16–31 years in the US. The mean pro-drop score was slightly higher in the 6–15 years condition (M = 10.48) than in the 16–31 years condition (M=9.04). A Bonferroni t-test proved unreliable t < 0.55. A final split into 6–14 and 15–31 years conditions produced similar results. The mean pro-drop score was larger for those in the 15–31 years condition (M = 10.35) compared to the mean for those in the 6–14 years condition (M = 9.24). A Bonferroni t-test proved unreliable t < 0.4. In other words, the length of immigration does not seem to play a role in the observed syntactic change despite our attempt to see whether a different grouping will make a difference.

However, when the amount of daily exposure to Russian was used as an independent variable, the following results were obtained. Splitting the data into two conditions, those participants who spend 10–30% or less (M = 16) and 40–80% or more of their daily time speaking Russian (M = 59) (note: 92% of participants in that group had more than 50% Russian input), showed that the mean pro-drop score for those with 40% or higher exposure to Russian (M = 14.47) was more than twice the mean for those with 30% or less exposure (M = 6.86). This effect proved reliable based on a t-test with an applied Bonferroni correction, t (1, 28) = 3.13, p < .005. When the data were split slightly differently, that is participants with less than 20% of daily exposure to Russian and those who have more than 30% of exposure, the mean pro-drop score for those with 30% or higher exposure (M=13.25) was larger than that of the mean for those with 20% or less exposure (M=6.56). This effect also proved reliable based on a t-test with an applied Bonferroni correction, t (1,28) = 2.72, p < .01. The results of these tests showed that the less frequent use of overt pronominals in Russian may indeed be a function of the amount of Russian input in the individual’s daily life. In the immigrant setting, the diminished input in L1 directly translates into a higher input of L2, especially in the group of our bilinguals, i.e., those bilinguals who had limited Russian input were professionally employed and highly integrated individuals. 

6. Discussion

The present study looked at first language changes in Russian-English bilinguals and those changes were considered within the CLI framework. A few particular instances of language change were under scrutiny.

The analysis of syntactic changes in the bilinguals’ discourse identified a less frequent use of null pronominals by bilinguals. Although Russian is not a typical pro-drop language and has constraints on when and where the subject can be dropped from the grammatical surface of the sentence, null pronominals are commonly used by native speakers, as was illustrated by the group of monolinguals in the present study. The difference in the use of pro-drop by the two groups of participants encouraged us to look more closely at those factors that may contribute to a change in the native language. Length of residence in the L2 country and daily exposure to L1 are traditionally considered as main factors in L1 attrition research. Predictably and in line with prior studies (Schmid 2004, Kӧpke & Genevska-Hanke 2018), the length of immigration did not produce a significant effect, whereas the amount of daily Russian input showed that the less frequent use of null pronominals indeed might be associated with limited exposure to Russian. The latter translates into a more frequent use of English, which, in turn, suggests that language transfer could account for the registered change in the bilinguals’ discourse. The obtained finding adds to the evidence reported in prior studies where the two languages in contact differed in terms of this particular feature. However, due to the nature of the two languages involved in the present study, it does not provide an answer to why such change can occur in those contact situations where both languages allow for pro-drop. The explanation of that phenomenon by Sorace and colleagues (Sorace 2004, Sorace & Serratrice 2009, as cited in Dubinina & Polinsky 2006) remains the most plausible. Indeed, the cognitive load imposed by processing two languages could result in a bilingual’s stating overtly a lexical pronoun, which would require less processing effort. In the immigrant setting – especially among immigrants who are professionally employed (this was the case of all participants with a low level of Russian input in our study) – limited exposure to Russian can translate directly into a higher input of English. In accordance with this line of thinking, the observed change could occur as a result of a cross-linguistic influence as well as bilinguals’ effort to reduce the cognitive load by using an overt pronoun in those instances where null pronominal would be allowed.

Another observed change in the bilinguals’ discourse concerned word order and, in particular, instances of the inverted verb-subject order. In line with our predictions, the two groups of participants – monolinguals and bilinguals – did differ in their use of this feature and the difference was statistically reliable. As is the case with the less frequent use of null pronominals, the VS use was not affected by the length of immigration. When the factor of Russian input was considered the effect was reliable but going in the direction opposite to our expectations. In other words, bilinguals use the VS order much less than monolinguals and participants with very limited exposure to Russian use it more often than those who have a higher Russian input. How can we explain this dichotomy? Clearly, this finding cannot easily be explained within the CLI framework. Limited exposure to Russian assumes higher exposure to English, a language that has very rigid constraints on the VS order use. Why did transfer from English not happen in this group of speakers? At the same time, high exposure to Russian in the U.S. provides ample chances for participants to hear and use this feature. Why did these speakers use the VS order less frequently? Starting with the second question, we may suggest that the Russian input in the U.S. is different from the Russian input in Russia. Frequent interaction in Russian with other Russian immigrants may not be a sufficient input of standard Russian. In the absence of continuous, firsthand daily contact with the language spoken in the L1 country the language of the immigrant community may change, thereby providing ample linguistic input of what we consider L1 input but which, in reality, is different from the L1 input in the L1 country. Thus, participants with a high level of exposure to Russian may be exposed to the language that had changed in general. Taking this argument a step further we can suggest that either transfer from English affected that input or some internal language changes leading to simplification of certain linguistic features occurred. The L1 input containing covert linguistic changes may be viewed as a source of “transfer,” with the term “transfer” being re-conceptualized to extend it to within language influence. However, this suggestion remains rather speculative due to the insufficient data on the Russian language spoken in Russian immigrant communities in the U.S. In light of this argument, a well preserved use of the VS order by participants with very limited exposure to Russian – whether this exposure means standard Russian or “American Russian” – speaks against cross-linguistic influence from English.

The reduced use of the inverted word order led us to look closely at those particular instances where the observed change occurred. Numerous instances of existential sentences in the monolingual discourse and less frequent use of such sentences in the bilingual discourse revealed an interesting trend. Bilinguals seem to overuse the demonstrative pronoun ‘это’ at the beginning of a sentence that otherwise would have been an existential sentence requiring the inverted word order. The demonstrative takes the subject position and changes the word order into direct, albeit this is less grammatical in standard Russian. A further look into this particular change showed that bilinguals often use this pronoun at the beginning of a sentence that otherwise would have been an impersonal sentence. Based on the anecdotal evidence from teaching Russian to English-speaking students, a similar transfer/strategy has been observed among L2 learners who tend to overuse a Russian demonstrative pronoun which often results in odd sounding sentences, such as Это было важно знать (что-то) instead of a more grammatically correct sentence Было важно знать (что-то) (It was important to know). Moreover, in our study, bilinguals tended to use all four forms of the verb ‘быть’ more often than monolinguals. The above finding can suggest a few things. First, we can see a clear sign of transfer from English where impersonal sentences start with the expletive ‘it’ (это), such as ‘It was interesting.’ However, in the case of existential sentences, such transfer is not clearly identified (cf.: ‘There were many people’). The English adverbial ‘there’ does not translate directly into the Russian demonstrative pronoun это. Another suggestion that we can make here is that bilinguals’ discourse becomes less lexically diverse and this may lead to the overuse of certain basic verbal forms, such as the verb ‘to be,’ or even to the overuse of certain grammatical sentences, such as sentences beginning with the demonstrative это followed by the conjugated forms of the verb быть, whether it changes an existential sentence or an impersonal sentence. In order to substantiate such claim, we should have conducted an analysis involving lexical token ratio, which was outside the scope of the present study. Finally, we can suggest that bilinguals’ strategy of using the demonstrative pronoun at the beginning of a sentence could result from an attempt to reduce processing effort: by introducing a clearly defined subject at the beginning of the sentence the bilingual speaker reduces the processing load involving the choice of an inverted word order.

Going back to the predictions made at the onset of the study, CLI can account for the registered changes in the use of null pronominals but it fails to explain a less frequent use of the VS order. As expected, the amount of the L1 input plays a role in both changes, but the directionality of the effect was not as straightforward as predicted. This suggests that language transfer should be studied within a much broader framework where psycholinguistic techniques and methodologies are combined with a more traditional comparative analysis. Also, such changes should be studied from the perspective of a bi-directional transfer where both L1 attrition and L2 acquisition are scrutinized. 

7. Conclusion

The present study has contributed additional evidence of L2 to L1 transfer to the existing knowledge of covert syntactic changes taking place in the individual’s first language under the influence of the second language. However, the findings of this study provided only partial support for cross-linguistic influence as a sole source of the registered changes. As expected, Russian-English bilinguals have demonstrated a clear trend of using fewer null subjects than their monolingual counterparts and the less frequent use of null subject was related to the amount of Russian input. Predictably, the length of immigration did not play a role in the observed syntactic change. However, the less frequent use of the inverted word order by bilinguals revealed an unexpected finding that cannot be interpreted within the CLI framework. The amount of Russian input produced an effect opposite to our expectations, showing that very limited exposure to Russian actually preserves the use of the inverted word order.

To conclude, the present study conducted within the traditional CLI framework failed to provide sufficient support for transfer as the sole reason for L1 changes in bilinguals living outside the L1 environment. Instead, we suggested that a psycholinguistic perspective on language change and language transfer should be incorporated in the more traditional comparative CLI framework. Finally, the present study showed the undeniable benefit of combining two methods of analysis – quantitative and qualitative.

^{(*) indicates ungrammatical sentences.}

About the authors

Ludmila Isurin

Author for correspondence.
Email: isurin.1@osu.edu
ORCID iD: 0000-0003-2808-3277

Professor in the Department of Slavic and East European Languages and Cultures

References