Lost in Between: the Case of Russian Heritage Speakers, Part One
Ludmila Isurin, Ohio State University, and Tanya Ivanova-Sullivan, University of New Mexico
The present paper looks at the growing population of Russian heritage speakers from a linguistic and psycholinguistic perspective. The study attempts to clarify further the notion of heritage language by comparing the linguistic performance of heritage speakers with that of monolinguals and second language learners. The amount of exposure to L1/L2, the age at which immigration to the U.S. occurred, degree of literacy in Russian, and metalinguistic awareness were among the sociolinguistic factors considered in the present study. The qualitative in-group and cross-group analyses revealed syntactic and morphological features that characterize Russian as a heritage language. The performance of heritage speakers on the narrative task differed from that of Russian monolinguals and American learners of Russian.
Over the past 15 years there has been growing interest in the language of heritage speakers (HS). Researchers have increasingly emphasized the special linguistic behavior and needs of HS, which are claimed to be distinct from those of traditional students. However, there is still a lack of empirical research on HS. Studies on Russian heritage speakers range from comparing the language use of the four waves of Russian immigration (Andrews, 1999; Zemskaja, 2001) to giving a descriptive analysis of the problems exhibited by Russian heritage speakers in the American classroom (Andrews, 2000; Bermel & Kagan, 2000). Other American scholars have reached practical solutions for teaching heritage speakers their home language in the U.S. (Kagan, Akishina, & Robin, 2002). In 2004-2005 two major professional venues, AATSEEL in Philadelphia, and AILA in Madison, Wisconsin, hosted panels on Russian in Diaspora. These events can be viewed as an additional indicator of the growing interest in the Russian language and its speakers overseas. However, the insufficient amount of research done in the field, the lack of empirical evidence (Ke, 1998), and the very problem of defining a "heritage speaker" (Kondo-Brown, 2002) have been identified as unresolved issues in previous studies.
The present paper looks at the growing population of Russian heritage speakers from linguistic and psycholinguistic perspectives.1 By adopting such an approach, we further define the notion of heritage language by comparing the linguistic performance of heritage speakers with that of monolingual speakers as well as second language (L2) learners.
2. Experimental Study
The subjects of our study have been described in the literature as speakers of American Russian (cf. Polinsky 2000a), i.e., either immigrants from the former USSR who came to the U.S. before the critical age of 10, or people born in the U.S. to Russian-speaking parents. Research participants meeting these criteria were defined as heritage speakers of Russian (n=7). Two other groups of subjects participated in the study: a) English speakers who are advanced learners of Russian (n=11), and b) a control group of monolingual Russian speakers (n=5). The heritage speakers and L2 learners of Russian were recruited from the pool of students at Ohio State University. The group of L2 learners consisted of English-speaking graduate students who had majored in Russian and whose academic career required a knowledge of the Russian language. It should be noted that an oral proficiency test was not administered to the L2 learners. However, we worked under the assumption that a proficiency level no lower than intermediate high is a prerequisite for admission to the graduate program from which our L2 participants were drawn. The participants in the heritage speaker group were given extra credit for their participation. The control group was recruited from monolingual speakers of Russian residing in St. Petersburg, Russia. The age, social, and educational background of all participants were comparable.
2.2.1. Background Questionnaire
A background questionnaire administered prior to the study was aimed at gathering language-related information. Students were asked to respond to questions on topics including the amount of their exposure to L1/L2, age at immigration to the U.S., language spoken to parents, degree of literacy in Russian, contacts within Russia, motivation to maintain/learn Russian, and metalinguistic awareness.
A children's picture book, A boy, a dog, a frog and a friend by Mayer and Mayer (1978), provided the test stimuli. The book is one in a series of so-called frog stories that has been successfully used in psycholinguistic studies on narrative discourse. The frog methodology was first used in a cross-linguistic developmental study on 5 different languages (Berman & Slobin, 1994) and later adopted by other scholars in studies on heritage learners language (Polinsky, 2008) and first language attrition (Olshtain & Barzilay, 1991; Isurin, 2005).
We used a controlled narration task to elicit speech samples. Subjects were offered the children's picture book and asked to tell a story based on the events chronologically depicted in the book. However, we have somewhat changed the original methodology. The focus of the present research was not on eliciting connected discourse and finding the critical features of a narrative mode of discourse (Berman & Slobin, 1994). Rather, our goal was to detect morphosyntactic difficulties experienced in unrehearsed narration. Thus, contrary to the methodology used in previous studies in which participants were asked to look through the entire book prior to telling the story (Berman & Slobin, 1994; Olshtain & Barzilay, 1991; Polinsky, 2008),2 our participants were specifically instructed not to do so. Instead, they were to tell the story as the events unfolded via each picture. We believe that the use of unrehearsed narration methodology reveals linguistic problems that possibly would not have been detected had the subjects been given a chance to think about the content in advance. Advanced planning could also lead to avoidance strategies (Olshtain & Barzilay, 1991). On the other hand, we are aware that there are certain risks in using such a task; namely, the very nature of the narration of unfolding events depicted in a book could influence certain syntactic patterns in the participants' grammar that otherwise may not have surfaced. We will address these problems in the sections below on the findings of our study.
Subjects were interviewed individually. Prior to the interview, they were asked to fill out the background questionnaire. They were then asked to narrate a story page by page (remember that they were not allowed to see how the events unfold in the book in advance). Their performance was tape recorded and later transcribed and analyzed. No answers were cued or prompted by the researchers. The data were analyzed qualitatively. The focus of the in-group analysis was to find those lexical, syntactic, and morphological features that characterize the linguistic performance of Russian heritage speakers. In a cross-group analysis we then attempted to shed light on the linguistic uniqueness of heritage speakers and their place in the continuum of language speakers.
3. Findings: Heritage Speakers3
Table 1 presents our findings of deviations from standard Russian by HS in both percentages (based on the total number of occurrences of any given category) and raw numbers (the actual number of errors in each category). The order of categories is descending, i.e. from the category with the most errors to the one with the fewest, an arrangement that helps us grasp the bigger picture of the speakers' performance.
Moreover, we would expect to have some specific morphosyntactic categories elicited under such a task and not others.4 We therefore limit our analysis to only a few morphosyntactic categories that occurred most frequently in the speech of our participants and that may indicate general trends in heritage speakers along the continuum from monolingual speakers of Russian to L2 learners. Since these morphosyntactic categories will be compared across the three groups, we will pursue a more detailed analysis of those categories when comparing the HS performance with another experimental group, L2 learners.
The lexical knowledge of the heritage speakers was not an object of study in this work. A complete study of the lexicon would require a different approach (i.e., different task and materials) or larger and more versatile samples. However, we realize that vocabulary is an important component of the overall profile of heritage speakers and that its range could interfere with the morphosyntactic data. Thus, we will briefly discuss the data on vocabulary in a section below.
Table 1. HS Deviations from Standard Russian: Major Findings
|Category||Instances in %||Instances in Numbers||Number of Speakers|
|Word order (VS)||2.1%||8||6|
3.1. Grammatical Findings
The HS narratives reveal alternations of tense and aspect within the same utterance that are uncharacteristic of standard Russian. Note the following examples:
|'Then the boy is getting dressed / got dressed and the dog is coming / came back to the shore.'|
|'The fish took the fishhook and the boy pulled the fish out.'|
|'The boy got upset and apparently was afraid that the dog would drown.'|
|'The turtle remained in the water, it was swimming.'|
We observed very few examples of aspectual forms that would be clearly non-permissible in Standard Russian, such as the use of perfective to express iterative action, e.g.:
|'He was pulling and pulling the fishing rod.'|
Overall, however, the HS narration recorded in this study suggests that these speakers tend to use perfective verbs in the past tense to express accomplished actions and imperfective verbs in the present tense for ongoing actions. Polinsky, in her article on full and incomplete language acquisition, also refers to lexicalized perfectives and imperfectives, which HS choose based on telicity, i.e. the "completedness" of the action (Polinsky, 2000b). In a recent paper on heritage language narratives, Polinsky (2008) expands on the role of telicity in aspect by offering a plausible mechanism for retention of just one member of an aspectual pair. Measuring the frequency of HS use of the perfective versus imperfective in HS narratives, she concludes that "the choice of the single aspectual form, lexicalized in American Russian, may be determined by the more frequent conceptualization of a given event as telic or atelic." We should add that in our study the role of the conceptualization of telicity in the expression of aspect is further complicated by the additional component of tense. We could view the aspect/tense interplay as not only a marker of telicity but also in certain instances a narration strategy where the foreground and background events or actions are marked by both aspect and tense. In other words, perfective/preterit foreground the event while imperfective/present is reserved for background (secondary to the main plot) events. Nevertheless, in order to confirm such a hypothesis, this phenomenon needs to be explored further not only in narratives but in other types of discourse as well.5
In addition to the lexicalization of one member of an aspectual pair, Polinsky's results show combinations of a "light" verb (a semantically less salient auxiliary, such as begin) plus a "content" verb to express perfectivity. Such instances were not found in our HS data but were present in the L2 narratives across the board. Constructions such as načinaet padat' ('begins to fall'), načinaet snjat' odeždu ('begins to get undressed'), eščë raz načinaet plakat' i kričat' ('once more begins to cry and yell'), načinaet ssorit'sja ('begins to argue'), and načala uxodit' ('begins to leave') indicate inchoative (i.e. incipient) actions in some cases, whereas in others they are used to express an already unfolding action.
A comparison with the monolingual group in St. Petersburg shows the Standard Russian mechanisms for expressing inchoative actions: 1) the light verb stat' ('to become/begin') in the past tense, stal pomogat' ('began to help'), stala smotret' ('began to look'), stal nabljudat' ('began to observe'), stal tjanut' ('began to pull') and 2) perfective verbs with the prefixes za- or po- denoting the beginning of an action, such as zakričat' ('to begin to shout'), zagovorit' ('to begin to speak'), pobežat' ('to set off at a run') and poslušat' ('to listen, cock an ear').
Heritage speakers' performance also reveals recurring problems in the use of case. Case errors are present in the speech of both heritage speakers and the L2 learners. However, there are significant differences in the nature of these errors and the possible reasons for their occurrence. In this section, we outline the most notable case errors in the heritage speakers' performance, arranged according to category:
188.8.131.52. Incorrect Case of Personal Pronouns
a) Dative instead of instrumental:
|To him||hard||fish||to take||with||he-PRON:DAT|
|'It is hard (for him) to take the fish with him.'|
b) Accusative instead of instrumental:
|' …and the dog followed him there.'|
|'And after that, the boy is going home with the turtle and the dog is following him.'|
|'Then the boy fell in the river and the dog and the frog ran after him.'|
The L2 narratives revealed a somewhat different pattern of pronominal usage: fewer case errors but also significantly less occurrence of personal pronouns overall. The eleven L2 narratives show 20 instances of personal pronouns in various constructions (3 of them incorrect) versus 50 in the seven HS narratives (5 of them incorrect, but 4 of them were attested in the narrative of a single speaker born in the U.S.). One reason for the much smaller number of pronouns in the L2 narratives could be due to an 'avoidance' strategy (e.g., one L2 speaker did not use personal pronouns at all).7 The use or avoidance of certain vocabulary items and the resulting effects on morphosyntactic analysis are a fundamental problem in such experimental studies. Therefore, before describing other examples of HS case errors, we will briefly discuss the lexical competence of the three groups of participants.
The range of vocabulary used by HS falls somewhere in between that of L2 and monolinguals. HS as a group showed very few lexical gaps,8 mainly in verbs designating specific actions (see the section on lexical extension). In certain cases, HS used circumlocution to fill in those gaps, as shown in the following example:
|'And then the boy wanted (I don't remember how to say that), so he made a hole in the ground.'|
In the above example the verb that the speaker is looking for is vyryt' ('to dig a hole'), which he substitutes with the more general verb sdelat' ('to make').
The narratives of the L2 learners show similar circumlocution but on a larger scale:
|'He was preparing the place to lay down the turtle—some sort of cemetery.'|
|'And he made a grave for the turtle.'|
|'And he is digging.'|
There were also some instances of conceptually inappropriate use of emotion words in the HS narratives. Note the following examples:
|'The boy was offended/hurt and together with his dog ran after his fishing pole.'|
|'But the turtle is no less harsh for being small and bit the dog's paw.'|
By contrast, L2 learners' narratives demonstrated fewer errors in this area, probably due to the smaller number of emotion words used by that group of participants.
Vocabulary data across these two groups show that both used more or less the same core high-frequency vocabulary to express the events in the pictures. Whenever L2 learners encountered difficulties in finding a word, they would resort to circumlocution in Russian or use the English equivalents. It should be noted that English equivalents were used only with nouns, while circumlocution was prevalent when a specific verb had to be retrieved. The HS, however, did not exhibit quite the same patterns. They clearly had fewer lexical gaps and in only one case (the speaker born in the U.S.) did they resort to English words to fill in for the Russian. In addition, they had a more diversified range of vocabulary than the L2 learners, including the use of synonyms. Not only did they describe the events chronologically but they also interpreted them, using a full range of lexical, syntactic, and discourse means such as adjectives, various types of subordinate clauses, particles, interjections, etc. Comparison with the monolinguals' narratives reveals the same overall interpretive strategies, albeit on a larger scale among the latter.
All in all, the HS' range of lexicon was closer to that of the monolingual group than to the L2 speakers'. This fact, together with the longer duration of the HS narratives in comparison to the L2 group, could be viewed as partially responsible for the final results of our study, and particularly for some morphosyntactic patterns in the areas of pronominal case and verbal aspect. We will return to this topic in the section below on cross-group analysis.
184.108.40.206. Case Errors in Nouns and Adjectives
a) Lack of 'direction-location' distinction, expressed by accusative and prepositional cases, respectively:9
|'[The turtle] pulled the boy through the water to the other side.'|
|'…and [the turtle] also pulled them out to the other side.'|
|'…the dog and the frog [went] back to the water.'|
|' The dog went back to the shore.'|
b) Lack of instrumental case in predicative function and after the verb stat' ('to become') (this verb requires instrumental case in Standard Russian):
|'They all became friends.'|
c) Relative clauses with kotoryj ('who', 'that', 'which' in subordinate clauses):
|'And they were pulled ashore by a big turtle, which they had caught instead of a fish.'|
In Standard Russian, the relative pronoun kotoryj refers back to the noun in the matrix clause; it is co-referential with that noun but receives its case according to the structure of the subordinate clause. According to this rule, in the above-listed example from a HS narrative, kotoryj should be in accusative rather than nominative case. We could suggest two possible explanations for this error—the close proximity of the referent of kotoryj in the previous clause, and/or the nominative as a 'frozen' form for kotoryj in the grammar of this particular speaker. However, it is impossible to draw firm conclusions since there were only two instances of relative clauses with kotoryj in this speaker's narrative, one of which involves the correct case form of the relative pronoun:
|'So Sema picked up the dog and carried it to the river together with the turtle, which is still in its paw.'|
Nevertheless, the scarcity of relative clauses with kotoryj in the HS narratives and the occasional case errors with kotoryj where it is used could be important signs of a decline in relativizing structures.10
220.127.116.11. Confusion of Oblique Cases
a) Dative instead of accusative:
|'Then the little boy and the dog went to the sea.'|
Despite the obvious error, we suspect that the confusion here pertains to the choice of preposition rather than to the case ending itself. The Agent characteristic [+ animate] in this sentence, the semantics of motion, and the lexical semantics of the word more are all factors that would contribute to the selection in Standard Russian of the preposition k ('toward') plus dative morju.11
b) Prepositional or dative (the forms are the same for feminine nouns) instead of genitive:
|'The boy is sitting by the river.'|
c) Instrumental instead of prepositional:
|'They are leaving the shore with the little turtle in their hands.'|
The above examples show the presence of oblique case forms in the grammar of heritage speakers but a lack of grammatical competence as to how and when to use them. Polinsky (2000a) argues that examples of this sort are a step towards the complete disappearance of the oblique cases in heritage speaker grammar.13 However, our study presents a more complicated picture. A comparison of our results with those obtained by Polinsky (2008) shows several differences. First, we did not observe a reanalysis of the accusative as the case of indirect objects. We indeed found some instances in which the participants used one oblique case instead of another, but by no means generalizing the accusative as the only case for indirect objects. Second, we never found nominative case occurring after prepositions. The latter is clearly demonstrated in Polinsky's data. Third, the correct forms of cases in the speech of our participants cannot be termed 'chunks', as Polinsky defines forms such as v dome, v lesu, utrom, v vode, ('in the house,' 'in the woods,' 'in the morning,' 'in the water'). In short, 'restructuring of cases,' as Polinsky describes it, did not emerge in our data. Our findings may be better explained as a reanalysis of case functions such as direction, location, means, etc. Additional experimental study with more robust tasks such as translation and grammatical judgment is needed to account fully for the system of cases in HS language.
3.1.3. Word Order
While SV word order is more frequent in Russian overall, VS is particularly common in narratives and story-telling. We therefore viewed VS word order in this study as an important indicator of language proficiency. A discussion of the correlation between word order and age at arrival in the U.S. is offered in the section on the cross- group analysis (see Section 5).
The following examples illustrate VS word order use by the heritage speakers.
|'He pulled very hard, so hard that even he himself fell into the river.'|
|'A little boy is sitting at the river.'|
|'and now Petka is the one drowning.'|
|Looks like||that||takes (bait)-IMPERF:PRES||fish-NOM|
|'It looks like the fish is taking the bait.'|
|'Crossing the shore, she is rushing now to the woods.'|
|'and the little boy pulled the dog aside.'|
3.1.4. Occasionalisms on the Morphosyntactic Level
Idiosyncratic morphosyntactic errors by HS participants were also observed. Despite their infrequent occurrence, they may provide a valuable input for analysis. We call these 'occasionalisms' due to the lack of frequency with which they appeared, but we are aware that in a narrative based on a different task, they might be quite regular.
18.104.22.168. Coordinate and Subordinate Clauses
Interestingly, the HS data show a preference for coordinate rather than subordinate clauses. We did not do a corresponding analysis on the other two groups (L2 learners and monolinguals), hence we limit ourselves to the suggestion that a preference for coordinate clauses was not surprising given that the task involved story telling (i.e., looking at one or two pictures at a time), which presupposes a narrative discourse in which events unfold in a successive manner.
22.214.171.124. Reference-tracking across Clauses14
|'The boy was running back with the dog and he was holding the dog.'|
|'He caught the fish and started pulling.'|
|He-PRON:3SG:MASC||wants||her||to save||of course||and|
|'He wants to save it, of course, but he does not want to get his clothes wet.'|
In the three examples listed above, the coordinate clauses do not require repetition of the referent. In fact, monolingual speakers of Russian and English would almost certainly use a null copy in their respective languages. A plausible explanation for the redundant pronominal referent in the second and third sentences could be that these sentences describe events depicted in two different pictures (the second example additionally involves a page-turn). Thus, the repetition of the referent could be caused by the mere nature of narration, i.e., a description of each picture starts with the subject of the action even when this subject is the same as in the previous picture. On the other hand, this repetition might be a discourse-related device used by HS to secure complete understanding of the message.15
|'The boy was surprised and he became happy, because the turtle lived, and that's all.'|
The above example, composed of two coordinate clauses, would not require referent repetition in Russian but probably would in English. In addition, considering the nature of the described events (i.e., emotional states), it is perhaps not surprising that the speaker chose referent redundancy in order to emphasize the two emotional states of the subject.
|'Suddenly he feels that he caught a big fish and can barely drag it to the shore.'|
Here, the repetition of the referent in the subordinate clause would be obligatory in English but is unnecessary in Russian. Thus, this could be a case of language transfer.
|were getting ready||to bury||turtle||as||became clear||that|
|'They dug out a hole and just as they were getting ready to bury the turtle, it became clear that it was alive.'|
The subordinate clause in the above example involves a reflexive verb, prigotovilis' ('they got ready'), which together with the presence of the conjunction tol'ko ('just') could have triggered the referent repetition. Moreover, the pronoun in the subordinate clause could be an emphatic device.
Notes, Part 1
1. The research reported in this paper was undertaken in the early spring of 2004 and the results were reported at the AATSEEL Annual Convention in Philadelphia in December, 2004. The current version is largely based on the AATSEEL presentation with a few additions to the bibliography and a broader discussion of certain topics. (back)
2. We would like to thank an anonymous reviewer for bringing this article to our attention. (back)
3. Our findings are based on well-known categories of morphosyntax, which have been previously discussed in different venues, cf. Lynch (2003) on Spanish heritage speakers and a review (Polinsky & Pereltsvaig, 2003) of Zemskaja (2000). The reader may also refer to Polinsky (2000a), and Bermel and Kagan (2000). (back)
4. For example, while Polinsky (2000a) observed a large number of resumptive pronouns in her study of American Russian, we do not observe any such pronouns. This is perhaps due to differences in the task and the setting. (back)
5. We should note that we do not observe aspect/tense interplay in the monolingual Russian narratives. The L2 narratives, on the other hand, clearly exhibit non-mastery of the aspectual distinction, with frequent gaps in one or the other member of the aspectual opposition. (back)
6. These three sentences were produced by one speaker. He might not have the pronominal instrumental form in his grammar. (back)
7. We recognize the need to compare the correct use of personal pronoun cases to that of nouns in the same grammatical environment (prepositions or verbs) in order to fully understand the factors behind their use, but such a task goes beyond the present study. (back)
8. We realize that 'lexical gaps' is a technical term referring to a systematic absence, not to performance. However, we use this term in our study to designate problems with lexical retrieval, which may or may not be indicators of competence problems. (back)
9. We could also argue that together with lack of a 'direction-location' distinction, these speakers have one generalized form for both 'direction' and 'location', namely, the prepositional form. (back)
10. The same tendency (a decline in relativizing structures but verbalized in a different way) is observed by Polinsky (2000b). (back)
11. A different case would be the sentence On idët kupat'sja v more ('He's going into the sea to swim'), where the noun more ('sea') is viewed as a 'container' noun (in cognitive terms). Alternatively, the guest editor of this issue has suggested that the case error above could be due to gender confusion, a phenomenon discussed in the companion article by Maria Polinsky. According to this approach, the speaker was treating neuter more as it if were feminine morja and then declining it like a feminine noun. This explanation is attractive and plausible, but we refrain from further comment because of the lack of other (similar) cases of gender transfer in our HS samples and the inaccessibility of Polinsky's paper prior to our final editing. (back)
12. Note here the self-corrected form of the prepositional case. (back)
13. Cf. Polinsky's (2000a, 2000b) discussion of the reduction of cases in American Russian. (back)
Published: Tuesday, January 08, 2008