In view of contemporary, rapidly changing requirements and conditions for successful social participation, the mastery of reading and writing skills is key to educational success and future career prosperity (European Commission/EACEA/Eurydice, 2016). In a migration context, the development of literacy skills may take place in more than one language (e.g., the majority language, foreign language/s learned at school, heritage language/s). Being a part of such complex multilingual repertoires, heritage language (HL) is defined as a language other than the majority language spoken by immigrants and their children (see Montrul, 2012, p. 170; Valdés, 2000). Literacy skills in an HL do not hamper the development of literacy skills in the majority language, according to Usanova and Schnoor (2021). Moreover, students’ biliterate skills as cultural capital have a potential to be converted into economic capital and may positively influence bilingual’s future earnings (Agirdag, 2014a; 2014b; Rumbaut, 2014). Thus, once developed, literacy skills in an HL have the potential to become a valuable resource contributing to future career prosperities of immigrant students. Considering the given institutional settings for majority language learning, students achieve the most prominent literacy skills in the majority language if compared to the other language/s in their repertoires (Usanova & Schnoor, 2021). In addition, literacy skills in an HL show high variability among bilinguals (Polinsky & Kagan, 2007). Due to the restricted opportunities to acquire literacy in an HL in a migration context, there is no single benchmark for the extent of biliterate skills but instead there are continua of competencies that may vary among the languages or specific skills (Dworin, 2006; Hornberger, 2003).
Most studies conducted on literacy skills in bilinguals focus either on writing or on reading, with only a few focusing on both skills (e.g., Brehmer & Mehlhorn, 2018; Brehmer et al., in press; Dworin, 2003; 2006; Klinger et al., 2019b; Sparrow et al., 2014). Studies conducted on writing in more than one language have shown that writing skills in different languages are positively interrelated (Schoonen et al., 2003; 2011; Soltero-González et al., 2012; Sparrow et al., 2014; Uluçam-Wegmann et al., 2019; Usanova, 2019; Usanova & Schnoor, 2021). The studies which considered both skills in bilingual students revealed that the acquisition sequence for both skills may deviate from typical monolingual sequences (Dworin, 2006). Dworin (2006) reports that bilinguals may even develop writing ability in the second language prior to oral abilities by applying their literacy skills in the majority language to literacy learning in the HL. Furthermore, reading and writing skills might follow different developmental paths (Dworin, 2003; 2006). The divergence in the development of both literacy skills in bilinguals is highlighted by the differences in the relation of each skill to the contextual factors as well as by the dynamics and instability of influences within the reading-writing relationship (Klinger et al., 2019a). Furthermore, the relationship between reading and writing skills differs between the majority language (ML) and HL: whereas the directionality of influences in the ML may come from reading to writing (receptive to productive), in HLs, the influence from writing to reading (productive to receptive) has been found (Klinger et al., 2019b). The reasons for these different developmental paths in reading and writing in bilinguals remain largely unexplored.
One of the possible influences on the differential development of both literacy skills in an HL may come from the settings in which the development of biliteracy skills takes place. Whereas the acquisition of literacy in the majority and foreign languages occurs predominantly in institutional settings, students’ literacy in their HL often develops without any institutional support. In Germany, students may attend different forms of HL classes: organized by the federal states (taking place at schools), by consulates, or by private organizers (e.g., private courses in churches, communities, private associations) (Brehmer & Mehlhorn, 2018; Lengyel & Neumann, 2017). The number of schools that provide HL classes is rather low and unequally distributed among the German federal states (Brehmer & Mehlhorn, 2018, p. 261). Overall, HL classes are provided only in a limited number of languages. There is no systematic overview of the languages taught in HL classes in Germany, and the number of languages varies among the German federal states (Lengyel & Neumann, 2017). Furthermore, the content, methods, and qualification of teachers in HL classes vary extremely (Montanari et al., 2018, p. 220). Typically, students start to attend classes while already having some receptive and productive skills. However, students who attend the HL classes show a high variation in their proficiency levels, which is even higher in literacy skills (Montanari et al., 2018). Moreover, as Montanari et al. (2018) show, the continuity of the attendance of HL classes varies among students, and the dropouts may occur at any stage.
The research shows that HL instruction is beneficial for the development of biliteracy, and the quality of instruction is crucial (Sparrow et al., 2014). If attended, the HL classes may positively influence the development of literacy skills in bilinguals. However, the effect may vary for both literacy skills (i.e., reading and writing). The study on the effectiveness of the HL classes conducted by Brehmer and Mehlhorn (2018) reveals that HL classes fostered writing skills (both spelling and the length of produced texts) of German-Russian bilinguals at most if compared to the other proficiencies (e.g., listening, reading comprehension). Still, though the context of migration provides only limited opportunities for institutional support of the literacy skills in the HLs, students may manage to develop reading and writing skills in their HLs (Brehmer et al., in press; Brehmer & Mehlhorn, 2018; Böhmer, 2015).
Apart from contextual influence, the differences in the development of both literacy skills in an HL may be induced by settings where the acquisition of biliteracy involves not only different languages but also different scripts (e.g., Latin/Cyrillic in German-Russian bilinguals in Germany). Script represents the core of literacy which draws on the abilities of decoding and encoding written language (Alves et al., 2020, p. 2; Olson, 1999). Thus, scriptual skills represent a building block that is fundamental to both reading and writing literacy and allow for crossing the bridge between spoken and written modalities. Previous research on biscriptual bilinguals has shown that even without formal instruction in the HL, bilinguals are able to distinguish between different scripts and are aware of their differences (Kenner et al., 2004). If formal instruction is given, children may also experience possible cognitive benefits from the simultaneous learning of both scripts (Kenner et al., 2004).
To specify the biliteracy development in two scripts, the term biscriptuality has been used (Usanova, 2019). According to research on biscriptual writing, students’ biscriptuality represents a continuum of scriptual skills, with two poles representing two different scripts (e.g., Latin and Cyrillic) with the mixing of both scripts placed between both poles. The Continuum of Biscriptuality framework proposes that biscriptual bilinguals, while developing their scriptual skills, apply different strategies in writing (intrascriptual vs. interscriptual graphemes’ mixing, transliteration and transcription). According to this framework, students’ biscriptuality is related to and provides insights into students’ biliteracy. Thus, studies reveal that students’ scriptual skills also reflect their writing skills in the HL, with students who write in Cyrillic script showing the highest skills in writing, followed by those who mix both scripts and those who compose texts by using the Latin script only (Usanova, 2016; 2019). Furthermore, this approach assumes that students’ biscriptuality is dynamic, and the movement alongside the continuum of scriptual skills can take place in either direction (Usanova, 2019).
Thus far, biscriptuality has been primarily investigated with regard to writing. Less is known about biscriptuality as complex literacy profiles involving scriptual skills in reading and writing. Furthermore, whereas cross-sectional studies investigated instances of script choice or scriptual skills at a certain period, the development of scriptual skills from the longitudinal perspective remains to be explored. Previous cross-sectional research on biscriptual writing noted the dynamics across the continuum of biscriptual skills in a migration context with the development of scriptual skills potentially going in either direction (Usanova, 2019). However, a longitudinal study is needed to get insights into the actual process of moving across the Continuum of Biscriptuality as well as to determine the directions of such movement. The current study aims to explore the role of script in reading and writing in Russian as an HL in biscriptual German-Russian bilinguals and to contribute to the still rare research on developmental paths in both literacy skills in an HL of biscriptual bilinguals by answering the two research questions:
The current study draws on the data from the German panel study “Multilingual Development: A Longitudinal Perspective” (MEZ) (Gogolin et al., 2017), which was funded by the German Federal Ministry of Education and Research (BMBF). MEZ investigated individual and contextual conditions that influence multilingual skills in students of higher school-age. MEZ was a longitudinal cohort-sequence study with two starting cohorts (grade 7 and grade 9 students) and with four waves of data collection over three years (2016 to 2018). The panel comprised more than 2,000 students from the German secondary educational system including students with Russian, Turkish, and monolingual German language backgrounds (Usanova & Schnoor, 2021). Only those students who had been in the German educational system from no later than the third grade were invited to participate in the study. The study design involved testing in several languages (English as the first, and French and Russian as the second foreign languages learned at school alongside with German as the majority language, and Russian or Turkish as HLs). To meet these requirements of the design, students’ receptive and productive language skills were tested in these languages. Individual and contextual data were collected as well.
The current study analyzes the receptive and productive language data for the group of German-Russian bilinguals from three waves and from both starting cohorts. The average age of the students was 14.54 years (SD = 1.2). At the first wave, students were in the middle of grade 7 (first cohort grade 7) and the middle of grade 9 (second cohort grade 9). In the second wave, students were at the beginning of grade 8 (first cohort) and the beginning of grade 10 (second cohort). In the third wave, students were at the end of grade 8 (first cohort) and the end of grade 10 (second cohort). Only those students who participated in all three waves were included in the current study. The overall sample of German-Russian bilinguals included in the current study consisted of 131 students. The selected data comprise the written language data in Russian as well as the background data from student and parental questionnaires. Furthermore, data on students’ cognitive ability were included into the analysis.
All students included in the current study carried out reading and writing tasks in HL Russian. The tasks were administered in schools. Students had six minutes for the reading speed and comprehension task and 30 minutes for the writing tasks (IEA Hamburg, 2017).
To test students’ reading skills, a translated version of the German reading speed and comprehension test LGVT 5–12+ (Schneider et al., 2017) was used. This test consists of three parallel versions of fairytale texts distributed among three waves. Students were required to read a fairytale and fill in the gaps by choosing one of the four suggested test items. The score for reading comprehension was calculated based on the number of correctly chosen items. In case of a wrong solution, one point was deducted from the general score.
Furthermore, all students were pretested on their scriptual skills in Russian. They were required to read a short passage of a C-test and asked to fill in the gaps. This C-test, like a classical C-test, is based on the deletion of the second half of every third word. The reconstruction of the systematically deleted part of words in a text as given within a C-test may serve as an indicator for general language proficiency (Eckes & Grotjahn, 2006). Those students who could not read the text in Cyrillic had an option to mark the field, which stated, “I cannot read in Cyrillic script.” Based on this pretest, students were divided into two groups, which got either the Cyrillic version of the reading test (those who managed the pretest) or the transcribed version of the same text (those who answered that they could not read in Cyrillic script). The transcription of the reading test was conducted based on the Duden conventions for the transcription of Russian into German. The choice of this system is based on previous research indicating that Latin graphemes used in writing by German-Russian bilinguals show high correspondence with Duden conventions (Usanova, 2019), as these are written with Latin graphemes and are primarily based on reproduction of the sound (Mulisch, 1996).
The writing prompt used in the study was developed by the FörMig study (Dirim & Döll, 2009). Since MEZ was designed as a panel study, the elaborated writing assignments parallel to the original FörMig prompt were used to measure students’ writing skills over time and to avoid motivational as well as learning effects (Gogolin et al., in press; Klinger et al., 2019a). The writing tasks required students to write down a trial version of an article for a youth journal on the topics “How to make a gingerbread house” (first wave), “How to make string lights” (second wave), and “How to make a boomerang” (third wave). All writing tasks were developed based on the same principles: they involved nine pictures and aimed at eliciting expository text types (Gogolin et al., in press). The writing tasks differed in their difficulty, with the easiest task used in the first wave, followed by the more difficult tasks in the second and the third wave, respectively (Schnoor, in press). We calculated a general score for writing based on five empirical indicators referring to the basic dimensions of language: a rating score for task accomplishment (textual-pragmatic dimension), types of verbs, nouns, adjectives, and conjunctions, and elements of academic language (lexico-syntactic dimension), and the number of words (productivity) (Gogolin et al., in press).
Those students who could not write in Cyrillic script were allowed to compose their texts using the Latin script. Based on the script chosen for writing (see Table 1), students were classified into the following groups: Latin (students who used the Latin script only), Cyrillic (students who used the Cyrillic script only), and Mix (students who mixed both Latin and Cyrillic scripts in writing, starting from at least three instances of script mixing within a text).
Since the current study focuses on students’ scriptual skills in reading and writing in Russian as an HL, only the choice of script in reading and writing was considered for the current analyses. The general scores were not analyzed.
The current study investigated students’ scriptual skills in reading and writing in Russian as an HL in German-Russian bilinguals. To identify the profiles of scriptural skills existing in the investigated population and to trace the patterns of changes in scriptual profiles over time, we conducted the Latent Transition Analysis, or LTA (Collins & Lanza, 2010). LTA is a special version of Latent Class Analysis (LCA) for longitudinal data (Goodman, 1974; Lazarsfeld & Henry, 1968; Vermunt & Magidson, 2009; Wolfe, 1970) and is a person-oriented statistical modeling approach to identifying unobserved subgroups of persons in the investigated population. This grouping is, per definition, unobserved but estimated as a categorical latent variable based on observed categorical indicator variables. Unlike in LCA, where these groups are called classes, these subgroups are called statuses in LTA to emphasize that participants’ affiliation can change over time. For example, participants who only use the Latin script at time t could mix the alphabets at t+1 and use the Cyrillic script only at t+2.
We used two categorical variables as observed indicators for the estimation of the latent variable: “Script used in Russian reading (1 = Latin; 2 = Cyrillic)” and “Script used in Russian writing (1 = Latin; 2 = Mix; 3 = Cyrillic)”. Both variables were measured at each of the three waves of data collection. The array of possible combinations of these two variables’ categories at three timepoints resulted in a contingency table with 216 cells. This contingency table served as the empirical data, which the LTA model was fitted to, in order to identify homogeneous subgroups of participants. Due to the a priori unknown number of subgroups in the population, we identified the best number of statuses to represent the empirical data by estimating a set of LTA models (2 to 5 statuses) and compared them by model fit indices, interpretability, and parsimony (Collins & Lanza, 2010; Nylund et al., 2007).
Since we used only two variables as indicators, we could not start with LCA models prior to LTA to determine the best fitting model at each wave separately, because our indicators did not provide enough empirical information to estimate cross-sectional models with more than one class. However, the contingency table for the longitudinal data contains enough cells (W = 216) for more complex models. Table 2 shows the results for the models with one to five latent statuses.
|NUMBER OF LATENT STATUSES||G2||DF||AIC||BIC|
|4||not well identified|
|5||not well identified|
The likelihood ratio statistic G2 (Agresti, 1990) is an indicator for the absolute model fit. It reflects how well an LTA model fits the observed data (see Collins & Lanza, 2010, pp. 81ff.). Due to the sparseness of the data (N/W < 5), the G2 statistic does not follow a χ2-distribution. Thus, the associated p-values were not reliable and are not reported (Köhler, 1986; Köhler & Larntz, 1980). Instead, we relied on evaluating indices for relative model fit. Relative fit indices allow comparing opposing models by their balance between model fit (i.e., accuracy) and parsimony. The Akaike Information Criterion (AIC, Akaike, 1987) suggested the model with three statuses, whereas the Bayesian Information Criterion (BIC, Schwartz, 1987) suggested the model with two statuses. The models with four or five statuses were not well identified in terms of the uniqueness of the maximum likelihood solution and the degree of the latent status separation. Besides the statistical indices of model evaluation, the interpretability of the results is key for model selection. Based on the theory on biscriptuality, we chose the three-status model because it allowed us to differentiate a Mixed status.
Also, due to the relatively small sample size (n = 131), we could not estimate separate models for our participants’ starting cohorts to check if they were comparable. Instead, we examined whether there was equality of the cohorts’ distribution before (concerning the indicator variables) and after (concerning status patterns) the model estimation. None of these tests showed significant scores under the null hypotheses of equality of the distribution. Thus, we analyzed the total sample of 131 participants in the three-status LTA model.
Based on students’ scriptual skills in reading and writing, students were assigned to groups with different levels of scriptual skills. Thus, students were divided into two groups for reading: those who read in Cyrillic script and those who read in Latin script (the Cyrillic and the Latin groups). Similarly, students were divided into three groups for writing: those who compose texts with Latin letters (Latin group), those who mix the Latin and the Cyrillic scripts in writing (Mix group), and those who apply the Cyrillic script (Cyrillic group).
Table 3 reveals the means and standard deviations for continuous variables and percentages for dichotomous variables for the overall sample (n = 131) as well as for the group’s subsamples divided by cohort: grade 7 (n = 65) and grade 9 (n = 66), by script used in reading: Latin (n = 36) and Cyrillic (n = 95) as well as by script used in writing: Latin (n = 56), Mix (n = 27), and Cyrillic (n = 48). The outcomes for all conducted comparisons are provided in the online appendix.
|MEAN/% (SD)||MEAN/% (SD)||MEAN/% (SD)|
|0.0||GRADE 7 (n = 65)||GRADE 9 (n = 66)||LATIN (n = 36)||CYRILLIC (n = 95)||LATIN (n = 56)||MIX (n = 27)||CYRILLIC (n = 48)|
|Reading Russian t1||0.0||5.7 (10.3)||7.1 (9.7)||1.5 (4.1)||8.3 (10.9)||***||2.1 (3.8)||8.3 (12.4)||10.4 (11.5)||***|
|Writing Russian t1||0.0||26.4 (14.0)||30.7 (13.0)||22.3 (11.6)||31.0 (13.6)||**||23.8 (10.4)||29.2 (13.7)||33.9 (15.0)||**|
|Grade 7 (%, ref. Grade 9)||0.0||–||–||61.1||45.3||51.8||51.9||45.8|
|Female (%, ref. Male)||0.0||70.8||75.8||63.9||76.8||67.9||77.8||77.1|
|Cognitive ability||6.1||16.9 (6.3)||18.3 (4.3)||17.4 (4.8)||17.8 (5.6)||17.4 (4.9)||16.2 (6.0)||18.8 (5.5)|
|School track: Gymnasium (%)||0.0||58.5||53.0||58.3||54.7||50.0||51.9||64.6|
|Migration background||9.9||1.8 (0.4)||1.7 (0.5)||1.8 (0.4)||1.7 (0.5)||1.8 (0.4)||1.6 (0.5)||1.8 (0.4)|
|Language acquisition||38.2||1.2 (0.7)||1.1 (0.5)||1.1 (0.6)||1.2 (0.6)||1.1 (0.5)||1.2 (0.7)||1.2 (0.7)|
|Heritage language classes in school (%)||4.6||23.0||29.7||5.7||34.4||**||10.9||26.9||45.5||**|
|Heritage language classes outside school (%)||4.6||14.8||15.6||5.7||18.9||9.1||11.5||25.0|
|Parents with high education (%)||38.9||78.0||74.4||73.9||77.2||81.1||66.7||76.0|
|Russian language used mostly in the family (%)||20.6||58.7||77.5||*||59.3||72.7||61.4||77.3||75.6|
The students of grade 7 and grade 9 within our bilingual sample tend not to deviate regarding most of the investigated characteristics. Students from grade 9 perform better in reading (M = 7.1) compared to students from grade 7 (M = 5.7), but the difference is not significant (t = 1.83, p = .070, g = 0.32). In writing, students from grade 9 also perform better (M = 30.7) than students from grade 7 (M = 27.4) but, similarly to the results for reading, this difference is not significant (t = 0.85, p = .40, g = 0.16). The absence of significant differences between the grades regarding literacy skills in the HL is not surprising. Unlike literacy skills in the majority language, where we expect better-developed literacy skills in students of higher grades, literacy skills in the HLs are not necessarily better in students of higher grades compared to a younger group of students. Instead, literacy skills in the HLs may show higher variability within the grades. This variability in literacy skills is induced by rather restricted learning opportunities for the literacy skills in the HL given within the migration situation if compared to the literacy skills in the ML, which students develop based on systematical formal learning in school. Thus, depending on their learning experiences, students of both grades may develop rather similar literacy skills in the HL.
The only significant difference between the students from both grades was found regarding Russian language use within the family. Here, 58.7% of students from grade 7 tend to use Russian for the communication within the family, whereas 77.5% of students from grade 9 do so (χ2 = 4.29, p = .04, Phi = 0.04). Overall, however, the given similarities between both cohorts regarding most of the investigated characteristics (Table 3) allow us to consider both cohorts jointly in the analysis of scriptual skills conducted within the current study.
Table 3 shows that scriptual groups deviate regarding their reading and writing skills. Thus, students who read in Cyrillic script achieve significantly better results in reading comprehension (M = 8.3) and writing (M = 31.0) in Russian compared to students reading the Russian task in the Latin script (M = 1.5 in reading comprehension; M = 21.3 in writing): t = 1.47, p < .001, g = 0.72. Students who write in Cyrillic also perform best in reading comprehension (M = 10.4), followed by students who mix both scripts in writing (M = 8.3) and those who write in Latin script only (M = 2.1): F[2, 128] = 11.01, p < .001, Eta = 0.38. Similarly, students who write in Cyrillic perform better in writing (M = 33.9) compared to students who mix both scripts in writing (M = 29.2), and those who write with Latin script only (M = 23.8): F[2, 128] = 8.00, p = .001, Eta = 0.33.
Concerning the additional variables, no significant differences were found between most of the investigated variables and the script used both in reading and writing (Table 3). Noteworthy, the German-Russian bilinguals who read in Cyrillic are more likely to attend HL classes offered within the school: 34.4% of those who read in Cyrillic reported attending HL classes inside school, whereas only 5.7% of the German-Russian bilinguals who read in Latin do so (χ2  = 10.71, p = .001, Phi = 0.29). Similarly, a difference in the attendance of the HL classes exists among the scriptual groups in writing. Thus, the largest share of students attending HL classes in school was found among those who write in Cyrillic (45.5%), followed by the Mix (26.9%), and the Latin (10.9%) groups (χ2 = 15.02, p = .001, Cramer-V = 0.35). Furthermore, a small share of participants within each of the scriptual groups indicated attending HL classes both inside and outside of school: those who read in Latin (2.8%, n = 1), read in Cyrillic (4.2%, n = 4), write in Latin (1.8%, n = 1), mix both scripts in writing (3.7%, n = 1), and write in Cyrillic (6.25%, n = 3).
Our LTA model of scriptual skills in Russian as an HL revealed three statuses of biscriptuality at three waves of data collection: Latin, Mixed, and Cyrillic. These groups evolved over time by participants switching statuses. Bilinguals in the Latin status had a high probability of using the Latin script for reading (64%) and writing (94%) in Russian. The bilinguals in the Mixed status were characterized by a high probability of using the Cyrillic script in reading (98%) and mixing both scripts in writing (97%). Bilinguals in the Cyrillic status were characterized by a high probability of using the Cyrillic script in reading (97%) and writing (87%).
In LTA, three parameters are essential for interpretation: item-response probabilities, latent status prevalences, and transition probabilities (Collins & Lanza 2010, p. 39 ff., p. 187 ff.). Table 4 shows the parameter estimates on a probability scale, ranging from 0 (impossible) to 1 (certain), representing an event’s likelihood to occur. For the readers’ convenience, we report these probabilities in percent. The item-response probabilities represent the possibility associated with the indicators’ categories and conditional membership in a particular latent status. These parameters are the key measures in an LTA because they characterize the found profiles and allow setting the latent statuses’ meaning. To ensure that this meaning stays the same over time, the item-response probabilities are constraint to be equal at each wave (measurement invariance). Overall, the item-response probabilities in our study clearly separated the latent statuses’ meanings, which is a crucial factor of the models’ quality.
|STATUS 1 (“LATIN”)||STATUS 2 (“MIXED”)||STATUS 3 (“CYRILLIC”)|
|1) Item-response probabilities|
|2) Latent Status prevalences|
|Wave 1||.44 (n = 58)||.19 (n = 25)||.37 (n = 48)|
|Wave 2||.37 (n = 48)||.04 (n = 6)||.59 (n = 77)|
|Wave 3||.37 (n = 49)||.42 (n = 55)||.21 (n = 27)|
|3) Transition probabilities|
|From status in wave 1 …||… to status in wave 2|
|From status in wave 2 …||… to status in wave 3|
To trace the change in the bilinguals’ membership in the found statuses over time, latent status prevalences were analyzed. Latent status prevalences are the estimated probabilities of the latent variables’ categories (latent statuses) at each time point. In other words, they represent the estimated group size of the statuses at any point in time. In the first wave, Latin was the most prevalent status, containing 44% of the sample, followed by Cyrillic (37%) and Mixed (19%) statuses. This distribution, however, changed in the second wave, where Cyrillic became the most prevalent status (59%), followed by Latin status (37%). In contrast, Mixed status (4%) was unlikely in the second wave. In the third wave, the picture changed again, and Mixed (42%) became the most prevalent status, followed by Latin (37%) and Cyrillic (21%) statuses. These changes in the latent status prevalences indicate that biscriptuality is not a stable phenomenon and that (some) participants are likely to change their status over time.
In a further step, we investigated which groups were more likely to change their status over time. Transition probabilities presented in Table 4 reveal the participants’ possibilities to transit to a latent status at t+1 conditional on their status membership at t. Thus, in our study, transition probabilities express the likelihood of specific patterns of change in scriptual development in Russian across three waves. There are always t-1 transition probability matrices estimated in an LTA model. Table 4 shows two transition probability matrices estimated in our study. The first matrix contains the conditional transition probabilities from the first to the second wave. Members of Latin had a high probability to stay in their status (85%) and a low probability to transit into either Mixed (5%) or Cyrillic (10%) statuses in the second wave. Besides, Latin is the only status to which there is a 0% probability for the transition from other statuses in the second wave. Members of Mixed, in turn, have a low probability (16%) of remaining in their status and a high probability (84%) of transition to Cyrillic in the second wave. Members of Cyrillic, instead, do not show any change from the first to the second wave. The probability of staying in Cyrillic is estimated as the highest (100%). The second matrix contains the conditional transition probabilities from the second to the third wave. The members of Latin now have the highest probability of staying in their status (100%) in the third wave. The members of Mixed, on the other hand, have a high probability of changing their status (63%) in the third wave. They had a probability of 39% of changing their status to Cyrillic and a probability of 24% of changing their status to Latin. Members of Cyrillic also had a probability of 63% of changing their status to Mixed in the third wave.
The current study analyzed the development of scriptual skills in reading and writing in Russian as an HL in biscriptual German-Russian adolescents. The Continuum of Biscriptuality (Usanova, 2019) was applied as a theoretical framework. According to this theoretical approach, students’ biscriptuality represents a continuum of scriptual skills. This framework proposes that biscriptuality is a dynamic phenomenon and that students change their location alongside the continuum in either direction when developing their scriptual skills. Our study aimed to empirically verify this theoretical approach by answering the two research questions: (1) What profiles of scriptual skills in reading and writing exist in biscriptual German-Russian bilinguals as locations on the Continuum of Biscriptuality? (2) What patterns of change in scriptual profiles may occur over time? We conducted Latent Transition Analysis (LTA) to identify subgroups of scriptual skills in the investigated population of German-Russian biscriptual bilinguals. In LTA, such subgroups are called statuses to emphasize that participants’ affiliation may change over time, i.e., their location on the Continuum of Biscriptuality may change.
We found that the model with three statuses of scriptual skills in Russian as an HL provided the best fit. The Latin status members had relatively high probability (68%) of using the Latin script for reading and high probability of applying the Latin script for writing in Russian (94%). Members of the Mixed status had high probabilities of using the Cyrillic script for reading (98% probability) and mixing both scripts when writing (97%). Members of the Cyrillic status had high probabilities of using the Cyrillic script for reading and writing (97% and 87%, respectively). These findings support the Continuum of Biscriptuality and show that the bilinguals’ scriptual skills are not evenly distributed across the continuum but cluster at specific locations with two different scripts marking the poles of the continuum and the Mixed status placed between the poles.
Furthermore, our study contributes to this theoretical framework by providing empirical findings on the balance of scriptual skills in reading and writing within the found statuses. The findings show a general tendency of scriptual skills in reading and writing to overlap at both poles of the continuum (Latin and Cyrillic statuses), and to diverge at the space between the poles (as given within the Mixed status). The found convergence and divergence of scriptual skills in reading and writing in different scriptual groups reflect the acquisition order of both literacy skills with receptive skills (reading) preceding productive skills (writing) as shown by the previous research (e.g., Ahmed & Wagner, 2020). These findings also support the research on biliteracy showing a diversity of pathways for developing literacy profiles of reading and writing skills (Dworin, 2003; 2006).
In the second research question, we addressed the patterns of change in scriptual profiles in reading and writing in biscriptual adolescents over time. Latent status prevalences indicated that different statuses took the leading position in each of the three waves. Latin (44%) was the most prevalent status in the first wave. In the second wave, Cyrillic (59%) status dominated. In the third wave, the picture changed again, and Mixed (42%) took the leading position. These changes in status prevalences show that scriptual skills are not static and students’ biscriptuality represents a dynamic phenomenon as proposed by the Continuum of Biscriptuality (Usanova, 2019).
Interestingly, the found profiles deviated in their transition probabilities for the change of the status. Thus, the transition matrices suggested different transition patterns for different statuses at different time points. The members of Latin had the highest probability to stay in their status, whereas the Mixed group had the highest probabilities of changing their status across the three waves. In the Cyrillic group, different tendencies could be observed: the stability in status was given in the second wave and the fluctuation in the third wave. These results indicate the movement across the Continuum of Biscriptuality to be prevalent among the Cyrillic and the Mixed groups. The findings also reveal the general movements towards biscriptuality, as shown by the change of the status from Mixed to Cyrillic in the second wave. These results highlight the role of applying different scripts as a bridge to biliteracy in biscriptual bilinguals (Al-Azami et al., 2010; Brehmer & Usanova, 2017; Usanova, 2019). However, remarkable bidirectionality in the changes of statuses in writing in the Cyrillic and Mixed groups was also observed in the current study. The theoretical framework of biscriptuality suggests that the movement alongside the Continuum of Biscriptuality may take place in either direction. The empirical findings of our study confirm this theoretical assumption and reveal that scriptual skills in an HL may be a subject of change if considered from the developmental perspective. The drawback from Cyrillic to Mixed status between the second and the third waves reflects the vulnerability of literacy skills within the migration situation reported by previous research. Overall, students’ scriptual skills in reading turned out to be more stable than students’ scriptual skills in writing, showing fewer dynamics across three waves.
Our study did not aim at clarifying the reasons for the given change in scriptual profiles. However, two aspects should be considered for the clarification of the found dynamics in the change of statuses.
First, the movement on the Continuum of Biscriptuality may be subject to the change in the test situation, and scriptual skills might be vulnerable to the difference in the task difficulty. In our study, the writing prompt used in the third wave was more difficult than the prompts used in the first two waves (Schnoor, in press), which might have influenced the change in the application of script in writing and so that the prevalence of the Mixed status as well as high transition probabilities from the Cyrillic to Mixed in the third wave. Studies have shown scripts and orthographies of both languages being simultaneously activated in different-script bilinguals (Mishra & Singh, 2014; Moon & Jiang, 2012), so that the choice of scriptual resources for writing is always there and may provide the solutions when biscriptual bilinguals experience difficulties with the writing of certain graphemes. Thus, biscriptual bilinguals have the opportunity to use the scriptual resources of both languages and transfer the graphemes from one language to fulfill the task with a higher level of difficulty.
Second, the minor movement from the Latin group towards both other groups in the second wave and the absence of the movement in this group in the third wave indicate that external resources, such as formal language learning, are needed for the development of scriptual skills to take place. Previous research has shown a positive trajectory towards biliteracy in bilingual children participating in a biliteracy program (Sparrow et al., 2014) and those attending HL classes (Brehmer & Mehlhorn, 2018). In line with previous research (Brehmer & Mehlhorn, 2018), the results of the current study revealed a significant association between students’ scriptual skills in reading and writing and the attendance of the HL classes inside of school. This finding highlights the importance of HL classes as being a crucial source for the development of biliteracy within a migration situation. However, the results of our study also show the tendency for HL courses to be not the sole factor determining the acquisition of literacy skills in the HLs in the context of migration since less than 50% of those who read and those who wrote in Cyrillic script attended the HL classes. Therefore, further studies should clarify which resources students use to develop their literacy in HLs and which settings substitute institutional learning within a migration situation while promoting the development of literacy skills in HLs. These resources may be crucial for launching the status change from the Latin via Mixed to the Cyrillic group fostering the positive trajectories of biscriptuality and biliteracy.
This work was supported by the German Federal Ministry of Education and Research under Grant [MARE (01JG2101)].
The authors have no competing interests to declare.
Agirdag, O. (2014a). The long-term effects of bilingualism on children of immigration: student bilingualism and future earnings. International Journal of Bilingual Education and Bilingualism, 17(4), 449–464. DOI: https://doi.org/10.1080/13670050.2013.816264
Agirdag, O. (2014b). The literal cost of language assimilation for the children of immigration: The effects of bilingualism on labor market outcomes. In R. M. Callahan & P. C. Gándara (Eds.), The Bilingual Advantage: Language, Literacy, and the Labor Market. (pp. 160–181). Multilingual Matters. DOI: https://doi.org/10.21832/9781783092437-008
Ahmed, Y., & Wagner, R. K. (2020). A “simple” illustration of a joint model of reading and writing using Meta-Analytic Structural Equation Modeling (MASEM). In R. Alves, T. Limpo & R. Joshi (Eds.), Reading-Writing Connections. Literacy Studies (Perspectives from Cognitive Neurosciences, Linguistics, Psychology and Education, 19, pp. 55–75). Springer. DOI: https://doi.org/10.1007/978-3-030-38811-9_4
Akaike, H. (1987). Factor analysis and AIC. Psychometrika, 52(3), 317–332. DOI: https://doi.org/10.1007/BF02294359
Al-Azami, S., Kenner, C., Ruby, M., & Gregory, E. (2010). Transliteration as a bridge to learning for bilingual children. International Journal of Bilingual Education and Bilingualism, 13(6), 683–700. DOI: https://doi.org/10.1080/13670050903406335
Alves, R. A., Limpo, T., & Joshi, R. M. (2020). Introduction to reading-writing connections: The integration roads ahead. In R. Alves, T. Limpo, & R. Joshi (Eds.), Reading-Writing Connections. Literacy Studies (Perspectives from Cognitive Neurosciences, Linguistics, Psychology and Education, 19, pp. 1–7). Springer. DOI: https://doi.org/10.1007/978-3-030-38811-9_1
Brehmer, B., Krause, M., Savenkova, N., & Usanova, I. (in press). Schreiben und Lesen in der Herkunftssprache Russisch: Ergebnisse der Panelstudie zum Erwerb literaler Fähigkeiten bei russisch-deutschen bilingualen Kindern. In T. Klinger, J. Duarte, I. Gogolin, B. Schnoor, & M. Trebbels (Eds.), Sprachentwicklung im Kontext von Mehrsprachigkeit. Hypothesen, Methoden, Forschungsperspektiven. Springer VS.
Brehmer, B., & Mehlhorn, G. (2018). Unterricht in den Herkunftssprachen Russisch und Polnisch –Einstellungen und Effekte. In G. Mehlhorn & B. Brehmer (Eds.), Potenziale von Herkunftssprachen. Sprachliche und außersprachliche Einflussfaktoren (Forum Sprachlehrforschung, 14, pp. 259–293). Stauffenburg.
Brehmer, B., & Usanova, I. (2017). Biscriptality and heritage language maintenance: Russian in Germany. In H. Peukert & I. Gogolin (Eds.), Dynamics of linguistic diversity (Hamburg Studies on Linguistic Diversity 6, pp. 99–121). John Benjamins. DOI: https://doi.org/10.1075/hsld.6.06bre
Collins, L. M., & Lanza, S. T. (2010). Latent Class and Latent Transition Analysis. With Applications in the Social, Behavioral, and Health Sciences (Wiley Series in Probability and Statistics). Wiley. DOI: https://doi.org/10.1002/9780470567333
Dirim, I., & Döll, M. (2009). ‘Bumerang’ – Erfassung der Sprachkompetenzen im Übergang von der Schule in den Beruf – vergleichende Beobachtungen zum Türkischen und Deutschen am Beispiel einer Schülerin. In D. Lengyel, H. H. Reich, H.-J. Roth, & M. Döll (Eds.), Von der Sprachdiagnose zur Sprachförderung (pp. 139–146). Waxmann.
Dworin, J. E. (2003). Insights into biliteracy development: Toward a bidirectional theory of bilingual pedagogy. Journal of Hispanic Higher Education, 2(2), 171–186. DOI: https://doi.org/10.1177/1538192702250621
Dworin, J. E. (2006). The family stories project: Using funds of knowledge for writing. The Reading Teacher, 59(6), 510–520. DOI: https://doi.org/10.1598/RT.59.6.1
Eckes, T., & Grotjahn, R. (2006). A closer look at the construct validity of C-tests. Language Testing, 23, 290–325. DOI: https://doi.org/10.1191/0265532206lt330oa
European Commission/EACEA/Eurydice. 2016. Structural Indicators on Achievement in Basic Skills in Europe – 2016. Eurydice Report. Luxembourg: Publications Office of the European Union. Retrieved December 2, 2020, from: https://op.europa.eu/de/publication-detail/-/publication/94fc76f3-f997-11e6-8a35-01aa75ed71a1.
Gogolin, I., Klinger, T., Lagemann, M., & Schnoor, B. (2017). Indikation, Konzeption und Untersuchungsdesign des Projekts Mehrsprachigkeitsentwicklung im Zeitverlauf (MEZ) (MEZ Arbeitspapiere No. Nr. 1). Hamburg. Retrieved December 2, 2020, from Universität Hamburg website: http://www.pedocs.de/volltexte/2017/14825/pdf/Gogolin_et_al_2017_Indikation_Konzeption_Untersuchungsdesign.pdf.
Gogolin, I., Klinger, T., Schnoor, B., & Usanova, I. (in press). The competence model of writing skills in MEZ. In H. Brandt, M. Krause, & I. Usanova (Eds.), Language Development in Linguistically Diverse Settings: Cross disciplinary findings from the Project “Multilingual Development: A Longitudinal Perspective” (MEZ). Springer VS.
Goodman, L. A. (1974). Exploratory latent structure analysis using both identifiable and unidentifiable models. Biometrika, 61(2), 215–231. DOI: https://doi.org/10.1093/biomet/61.2.215
Hornberger, N. H. (2003). Continua of Biliteracy: An Ecological Framework for Educational Policy, Research, and Practice in Multilingual Settings. Multilingual Matters. DOI: https://doi.org/10.21832/9781853596568
Kenner, C., Kress, G., Al-Khatib, H., Kam, R., & Tsai, K.-C. (2004). Finding the keys to biliteracy: How young children interpret different writing systems. Language and Education, 18(2), 124–144. DOI: https://doi.org/10.1080/09500780408666871
Klinger, T., Gogolin, I., Schnoor, B., & Usanova, I. (2019b, September). Multilingual development in a longitudinal perspective: Interconnections of reading and writing development in more than one language [Paper presentation]. European Conference on Educational Research 2019, Hamburg, Germany.
Klinger, T., Usanova, I., & Gogolin, I. (2019a). Entwicklung rezeptiver und produktiver schriftsprachlicher Fähigkeiten im Deutschen. Zeitschrift für Erziehungswissenschaft (ZfE), 22(1), 75–103. DOI: https://doi.org/10.1007/s11618-018-00862-0
Lengyel, D., & Neumann, U. (2017). Herkunftssprachlicher Unterricht in Hamburg – Eine Studie zur Bedeutung des herkunftssprachlichen Unterrichts aus Elternsicht (HUBE). Projektbericht. Hamburg: Universität Hamburg.
Mishra, R. K., & Singh, N. (2014). Language non-selective activation of orthography during spoken word processing in Hindi-English sequential bilinguals: An eye tracking visual world study. Reading and Writing, 27(1), 129–151. DOI: https://doi.org/10.1007/s11145-013-9436-5
Montanari, E., Abel, R., Graßer, B., & Tschudinovski, L. (2018). Die Herkunftssprache Russisch im bilingualen Erwerb im Grundschulalter. In G. Mehlhorn & B. Brehmer (Eds.), Potenziale von Herkunftssprachen. Sprachliche und außersprachliche Einflussfaktoren (Forum Sprachlehrforschung, 14, pp. 219–239). Stauffenburg.
Montrul, S. (2012). Bilingualism and the heritage language speaker. In T. Bhatia & W. Ritchie (Eds.), The Handbook of Bilingualism and Multilingualism (2nd ed., pp. 168–189). Wiley-Blackwell. DOI: https://doi.org/10.1002/9781118332382.ch7
Moon, J., & Jiang, N. (2012). Non-selective lexical access in different-script bilinguals. Bilingualism. Language and Cognition, 15(1), 173–180. DOI: https://doi.org/10.1017/S1366728911000022
Nylund, K. L., Asparouhov, T., & Muthén, B. O. (2007). Deciding on the number of classes in Latent Class Analysis and Growth Mixture Modeling: A Monte Carlo simulation study. Structural Equation Modeling, 14(4), 535–569. DOI: https://doi.org/10.1080/10705510701575396
Polinsky, M., & Kagan, O. (2007). Heritage languages: In the “wild” and in the classroom. Language and Linguistics Compass, 1, 368–395. DOI: https://doi.org/10.1111/j.1749-818X.2007.00022.x
Rumbaut, R. G. (2014). English Plus: Exploring the socioeconomic benefits of bilingualism in southern California. In R. M. Callahan & P. C. Gándara (Eds.), The Bilingual Advantage: Language, Literacy, and the Labor Market (pp. 234–257). Multilingual Matters. DOI: https://doi.org/10.21832/9781783092437-009
Schnoor, B. (in press). Vergleichbarkeit der Impulsversionen der “MEZ-Schreibaufgabe Jugendliche”. In H. Brandt, M. Krause, & I. Usanova (Eds.), Language Development in Linguistically Diverse Settings: Cross disciplinary findings from the Project “Multilingual Development: A Longitudinal Perspective (MEZ). Springer VS.
Schoonen, R., Gelderen, A. V., Stoel, R. D., Hulstijn, J., & Glopper, K. D. (2011). Modeling the development of L1 and EFL writing proficiency of secondary school students. Language Learning, 61(1), 31–79. DOI: https://doi.org/10.1111/j.1467-9922.2010.00590.x
Schoonen, R., Van Gelderen, A., De Glopper, K., Hulstijn, J., Simis, A., Snellings, P., & Stevenson, M. (2003). First language and second language writing: The role of linguistic fluency, linguistic knowledge and metacognitive knowledge. Language Learning, 53(1), 165–202. DOI: https://doi.org/10.1111/1467-9922.00213
Soltero-González, L., Escamilla, K., & Hopewell, S. (2012). Changing teachers’ perceptions about the writing abilities of emerging bilingual students: towards a holistic bilingual perspective on writing assessment. International Journal of Bilingual Education and Bilingualism, 15(1), 71–94. DOI: https://doi.org/10.1080/13670050.2011.604712
Sparrow, W., Butvilofsky, S., Escamilla, K., Hopewell, S., & Tolento, T. (2014). Examining the longitudinal biliterate trajectory of emerging bilingual learners in a paired literacy instructional model. Bilingual Research Journal, 37(1), 24–42. DOI: https://doi.org/10.1080/15235882.2014.893271
Ulucam-Wegmann, I., Akın, S., Boubakri, C., & Roll, H. (2019). Schreiben und Mehrschriftlichkeit – eine Analyse schriftsprachlicher Fähigkeiten im Deutschen und Türkischen in der Sekundarstufe I und Ansätze für eine koordinierte Förderung mehrsprachiger Literalität. In H. Roll, M. Bernhardt, C. Enzenbach, H. E. Fischer, E. Gürsoy, H. Krabbe, M. Lang, S. Manzel, & I. Uluçam-Wegmann (Eds.), Schreiben im Fachunterricht der Sekundarstufe I unter Einbeziehung des Türkischen. Empirische Befunde aus den Fächern Geschichte, Physik, Technik, Politik, Deutsch und Türkisch, (Mehrsprachigkeit, 48, pp. 51–129). Waxmann.
Usanova, I. (2016). Transfer in bilingual and (bi)scriptual writing: can German-Russian bilinguals profit from their heritage language? The interaction of different languages and different scripts in German-Russian bilinguals. In P. Rosenberg & C. Schroeder (Eds.), Mehrsprachigkeit als Ressource in der Schriftlichkeit (pp. 159–177). Walter de Gruyter. DOI: https://doi.org/10.1515/9783110401578-010
Usanova, I. (2019). Biscriptuality: Writing Skills Among German-Russian Adolescents. John Benjamins. DOI: https://doi.org/10.1075/hsld.8
Usanova, I., & Schnoor, B. (2021). Exploring multiliteracies in multilingual students: Profiles of multilingual writing skills. Bilingual Research Journal. DOI: https://doi.org/10.1080/15235882.2021.1890649
Valdés, G. (2000). The teaching of heritage languages: an introduction for Slavic-teaching professionals. In O. Kagan & B. Rifkin (Eds.), The Learning and Teaching of Slavic Languages and Cultures (pp. 375–403). Slavica.
Vermunt, J. K., & Magidson, J. (2009). Latent Class Cluster Analysis. In J. A. Hagenaars & A. L. McCutcheon (Eds.), Applied Latent Class Analysis (This digitally printed version 2009, pp. 89–106). Cambridge Univ. Press. DOI: https://doi.org/10.1017/CBO9780511499531.004
Wolfe, J. H. (1970). Pattern clustering by multivariate mixture analysis. Multivariate Behavioral Research, 5, 329–350. DOI: https://doi.org/10.1207/s15327906mbr0503_6