Abstract
Over the past decades, theories of cross-language transfer have widely focused on the proficiency required to transfer a single linguistic modality in one language to that same linguistic modality in another language (e.g., first-language reading to second/foreign-language reading). To move beyond cross-language transfer in a single modality, a new theory, the transfer integration hypothesis proposes that reading and writing are related across learners’ first and second languages. To test the transfer integration hypothesis, this systematic qualitative review synthesizes the findings of peer-reviewed studies on the reading-writing connections across languages. Evidence of cross-language reading-writing connections to support the transfer integration hypothesis was found with much variability resulting from literacy and linguistic factors. These factors include multilinguals’ language experience and grade level, reading and writing measures, language exposure, motivation, language proficiency, and language structure similarities and distance.
Keywords
Literacy, the ability to read and write, is one of the most fundamental competencies underlying learning (World Literacy Foundation, 2022). Multilingualism has become increasingly common worldwide, with more than half of the world’s population navigating more than one language (Gration, 2024; May, 2014). For second/foreign language (L2) learners and multilinguals (multilinguals hereafter to refer to speakers/users of two or more languages), dual-language involvement makes literacy development inherently more complex than monolingual reading and writing development (Koda, 2007).
In addition to learning to read and write in each language, there is a cross-linguistic interaction between L1 skills and L2 print input (Koda, 2007, 2008). That is, L1 skills may alter the process of learning to read and write in an L2 and such cross-language transfer can occur from L2 to L1 as well (Cummins, 1979, 1981). While considerable research has focused on cross-language transfer of reading (e.g., Pasquarella et al., 2015; Zhang & Koda, 2021) and writing (e.g., Gu, 2014), cross-modal relations between L1 and L2 skills (e.g., L1 reading and L2 writing) have received little attention.
The well-established reading-writing connection (Fitzgerald & Shanahan, 2000; Shanahan, 2016) provides strong theoretical support for a monolingual cross-modal relation, as skills in one modality (e.g., reading) can enhance development in the other (e.g., writing). Empirically, the benefits of the reading-writing connection are clear for monolingual speakers (e.g., Boyer & Ehri, 2011). Indeed, teaching reading has been found to strengthen students’ writing quality, words written, and spelling (see Graham et al., 2018, for a review). Writing instruction improves students’ reading comprehension, reading fluency, and word reading (see Graham & Hebert, 2011, for a review). Despite the well-established monolingual reading-writing relation, this relation may not be the same for multilinguals, whose reading and writing are increasingly complex because of the cross-linguistic nature of learning two or more languages (Koda, 2007).
Studies focusing on reading-writing connections in L2 also confirmed such relations (e.g., Schoonen, 2019; Zhang et al., 2014), with L2 reading-writing connection demonstrated to be stronger than that in multilinguals’ L1 (Pae, 2018, 2019). However, in a longitudinal study with Dutch students learning English as a foreign language (EFL) in grades 8 to 10, Schoonen (2019) found that in the early years of secondary education, the L2 reading-writing correlation was higher than it was in L1. Intriguingly, as students gained more proficiency in English, the L2 correlation dropped into the same range as the L1. This may instead suggest that as L2 proficiency increases, the L2 reading-writing relation decreases to parallel the association between L1 reading and writing.
In this article, I aim to propose a new theory to address the cross-linguistic connection between reading and writing—the transfer integration hypothesis (TIH). The TIH posits that reading and writing are related across individuals’ L1 and L2 (i.e., L1 reading is related to L2 writing; L1 writing is related to L2 reading). Such cross-linguistic (i.e., L1 and L2) and cross-modal (i.e., reading and writing) relations are thought to be due to common underlying proficiencies that allow individuals to draw upon a set of skills acquired in one language in the course of learning another language, and the shared knowledge and cognitive systems that underlie both reading and writing abilities. My primary goal is to delve deeper and expand our understanding of the cross-linguistic, cross-modal relation from a grain size perspective. To this end, I first propose the TIH to establish a single integrative framework that can explain reading-writing connections across languages, followed by an examination of existing research to test the TIH. Finally, I discuss future directions to test the theory and specific aspects of it. Throughout the manuscript, I use Ln to indicate the nth language spoken by the individual. For example, L1 is the first language and L2 is the second language.
The Transfer Integration Hypothesis (TIH)
Cross-Linguistic Relations (Transfer)
One central premise of the TIH is that shared knowledge and mechanisms underlie the literacy development of one’s L1 and L2, or sometimes L3. Indeed, for multilinguals, it is argued that there is a system of language faculty (i.e., mind-brain complex) that performs as the functional center for operating different languages (Varghese, 2022). As such, one can activate the language faculty of the brain to constantly vitalize the mentalese—the thought of language—with the language that one uses for communication. The idea is that when using different languages (e.g., L1 or L2), learners utilize the same thoughts formed out of the mentalese (L0) to transform the thoughts into the structures of L1 or L2 (Varghese, 2022). This is aligned with the idea that irrespective of the language in use, the thoughts that accompany the process of listening, speaking, reading, and writing come from the same central system; therefore, there is an integrated source of thoughts even for multilinguals (Baker, 2011). This idea is also manifested by the well-recognized common underlying proficiency that allows learners to draw upon a set of language-independent skills and knowledge acquired in one language in the course of learning in another language (Cummins, 1981, 2000). However, this does not mean that acquisition of L1 and L2 proficiency are identical processes. Instead, although some surface features of each language are distinct, L1 and L2 skills are intrinsically connected and are not separate abilities (Goodrich & Lonigan, 2017; Shen & Goodrich, 2024).
One consequence of L1 and L2 skills drawing upon shared knowledge and systems is that L1 and L2 skills are interrelated and may transfer from one language to another language bidirectionally (i.e., cross-language transfer). L1 literacy acquisition can facilitate L2 literacy development, and L2 literacy acquisition can influence L1 literacy development. However, for such a transfer to occur, there is much to consider due to the developmental intertwinement of two or more languages. As depicted in Figure 1, the TIH posits four linguistic factors that may affect cross-language transfer, including language exposure, motivation, language proficiency, and language structure similarities and distance. Although there might exist additional factors, such as instructional practices and familial and societal language ideologies and policies that may affect the learning of additional languages (Shen et al., 2022), I focus on individual-level factors, and those broader factors are beyond the scope of the hypothesis.
Transfer integration hypothesis.
First, the TIH posits that for cross-language transfer of literacy skills to occur, multilinguals should have sufficient exposure to the languages in academic contexts, such as reading and writing. Indeed, language is typically acquired by constant exposure and use of the language in multiple settings (National Academies of Sciences, Engineering, and Medicine, 2017; Sussman et al., 2023). Evidence of bilingual education programs providing adequate exposure for multilinguals’ in both languages supported this key principle of cross-linguistic transfer, such that multilinguals in bilingual education programs were more likely to have higher literacy skills in L1 and/or L2 (Shen et al., 2022; Steele et al., 2017). Further, when considering language exposure, it is important to take into account the academic environment. If multilinguals receive plentiful exposure and opportunities for academic literacy development in one language, that language may be easier to transfer and facilitate the learning of another language (Gebauer et al., 2013; Kim & Piper, 2019). Evidence is also apparent in cases where there is insufficient L1 language exposure; high exposure to a second language in school alone is not enough to promote cross-language transfer (e.g., Barnett et al., 2007).
Second, language learning motivation is another key factor that may influence the extent of cross-language transfer (Cummins, 1979, 1981). Different from motivation in other domains of learning, language learning motivation includes instrumental (a functional goal), integrative (the desire to identify with the culture of a language community; Gardner & Lambert, 1972), and assimilative (the drive to become an indistinguishable member of a language community; Brown, 1994) motives for language learning. Further, according to expectancy-value theories of motivation, intrinsic motivation includes one’s expectations for successful task performance such as perceived competence and difficulty, as well as one’s valuing of tasks, such as enjoyment and perceived importance of tasks (Eccles & Wigfield, 2002). As such, it is possible that multilinguals who are more motivated to learn the languages and integrate/assimilate to L1 and L2 language communities perceive themselves to be successful language learners, and value language learning and reading and writing are more likely to transfer literacy skills across languages, which leads to better literacy skills. Indeed, perceived competence and value/enjoyment for reading have been demonstrated to not only positively predict L1 (Cartwright et al., 2016) and L2 (Takase, 2007) literacy acquisition, but also transfer across languages and thus influence the cross-language transfer of literacy skills (Takase, 2007; Yang, 2009). The TIH posits that motivation to learn to read and write in different languages may serve a moderating role, such that multilinguals with high motivation levels are more likely to benefit from cross-language transfer of literacy skills.
Third, for multilinguals to transfer literacy skills from one language to another, they must acquire and maintain a threshold level of language proficiency (threshold hypothesis; Alderson, 1984; Cummins, 1979; Kecskes & Papp, 2003). When language proficiency is lower than this threshold, it becomes unlikely that L1 literacy skills will be transferred to L2 literacy skills, and vice versa. The TIH further posits that it is not only the oral language proficiency (e.g., vocabulary and grammar) but also the written language proficiency (e.g., reading and writing) that can affect the extent of the cross-language transfer. Although oral language is fundamental to reading and writing (Lesaux et al., 2010), to allow literacy skills to transfer across languages, multilinguals’ written language proficiency plays a more critical role. For instance, for heritage language speakers who are mostly illiterate in their heritage language (Valdés, 2005), cross-language transfer of literacy skills is unlikely to occur. The TIH further posits that such a threshold level of proficiency should be achieved in multilinguals’ L1 and/or L2, depending on their language dominance and language of instruction. By default, multilinguals are more likely to achieve higher proficiency in their dominant societal language/language of instruction; when this is the case, proficiency in the other language is most likely to affect their cross-language transfer. However, there are cases in which the language of instruction does not match language dominance (e.g., French immersion instruction in Canada); then, proficiency in L1 and L2 may both play significant roles in facilitating cross-language transfer (Gebauer et al., 2013).
Finally, language distance, the extent to which linguistic features differ from each other, is proposed to be a determining factor in cross-language transfer. In other words, cross-language transfer hinges upon similarities and differences between L1 and L2 (Chiswick & Miller, 2005; Koda, 2007, 2008). When the L1 and L2 are closely related, shared structural features (i.e., orthography, phonology, or semantics) allow L1 skills to function in L2 literacy acquisition more easily and with less adjustment and L2 exposure. In contrast, when the L1 and L2 are distantly related, L1 skills are not readily transferrable to facilitate L2 learning without significant modification and adjustment, as well as a great amount of L2 exposure (Chung et al., 2019).
Cross-Modal Relations (Integration)
The TIH was built upon well-researched knowledge base on cross-modal relations (i.e., reading-writing connections); that is, even though reading and writing are not identical skills, they are related since shared knowledge and cognitive systems underlie both reading and writing abilities (Abbott et al., 2010; Kim, 2020, 2022; Shanahan, 2016). Different theoretical models addressing monolingual reading-writing relations from a shared knowledge perspective have concluded diverse sources of shared processes and skills. According to Fitzgerald and Shanahan (2000), four basic types of knowledge tend to be shared between reading and writing: pragmatical metaknowledge, domain knowledge, knowledge about text attributes, and procedural knowledge (shared cognition model; Fitzgerald & Shanahan, 2000; Swanson & Berninger, 1996). The interactive dynamic literacy model (Kim, 2020, 2022) focused on the highly similar skills and knowledge underlying reading and writing, including content and discourse knowledge; social-emotional systems; higher-order cognition and regulation; oral language skills at the discourse, sentence, and lexical levels; word reading and spelling; phonological, morphological, and orthographic awareness; and domain-general cognitions.
The perspective of shared knowledge, which focuses on underlying cognitive processing and language subskills, helps to explain the existence of reading-writing relations at different stages of their development. These potential subskills previously reported to underlie reading-writing correlations may be even more critical in L2 reading and writing than in L1, considering that the reading and writing processes in one’s L2 are more dependent on linguistic subskills and proficiency of their L2 (Schoonen, 2019). The TIH was not designed with the intention to elaborate on sources of shared knowledge and skills underlying reading-writing relations, as models of monolingual reading-writing connections have explicitly done so; instead, the focus of the TIH is an articulation of reading and writing skills at different grain sizes that can transfer across languages, such that Lx reading is associated with Ly writing (see Figure 1).
Specifically, the TIH posits that there are both within- and cross-grain size bidirectional associations between reading and writing across languages (see Figure 1). Within-grain size relations include reading-writing connections within the lexical (i.e., word reading and spelling), sentence (i.e., written sentence comprehension and written sentence production), and discourse (i.e., reading comprehension and written composition) levels. These within-grain size relations are posited to differ as a function of grain size; that is, the relation is the strongest at the lexical level, with word reading and spelling drawing on a small set of similar skills (Kim et al., 2023; Shanahan, 2016), and weakest at the discourse level, with reading comprehension and written composition being more likely to entail more diverse processes and skills (Kim, 2022; Kim et al., 2023). Cross-grain size relations include reading-writing associations across lexical, sentence, and discourse levels, as illustrated in Figure 1.
Another key tenet of the TIH is that the cross-modal relations vary as a function of language experience and grade level. Language experience can vary for multilinguals learning languages in different contexts. Multilinguals with early L2 learning experience (e.g., simultaneous bilinguals) are likely to start developing their lexical-level skills in both their L1 and L2 during early grades; therefore, they may demonstrate stronger within-grain size lexical-level relations between reading and writing. This is because in early grades, most learners are still in the process of developing lexical-level skills, whereas at later grades, they would be expected to have developed sufficient word reading and spelling skills and therefore less variations across individuals (Kim et al., 2023). For those who start L2 learning at a later grade (e.g., second/foreign language learners) when they have mastered higher-level (e.g., discourse-level) skills in L1, there might exist stronger cross-grain size cross-modal associations. That is, because they are at the phase of developing lower-level skills in their L2, cross-modal relations are more likely to occur between higher-level L1 skills and lower-level L2 skills.
Finally, the TIH also posits that the magnitudes of the cross-modal relations may vary depending on how reading and writing are measured in L1 and L2, such that measurement differences such as different tasks and genres used to measure reading and writing may also influence the reading-writing relations across languages. In the TIH, reading and writing are general constructs encompassing three levels of grain sizes. However, even within each grain size, reading and writing can be tested differently. For example, word reading can be assessed using real-word decoding, pseudoword decoding, word reading accuracy, or word reading fluency; each may tap into different aspects of the word reading construct. Therefore, the transfer of these skills is likely to show variations (Pasquarella et al., 2015). At larger grain sizes, such as discourse level, reading comprehension can be assessed using multiple tests, including multiple-choice, cloze, open-ended questions, or free recall; different assessment formats are likely to influence one’s reading comprehension performance (Cao & Kim, 2021). Specific to written composition, there is a range of genres and dimensions that are widely used in the assessment of written composition. For instance, narrative and argumentative essays are likely to be evaluated differently since an argumentative essay’s rating is mostly based on the effectiveness of an individual’s argument (Ferretti et al., 2000). Further, multiple dimensions of scoring, such as overall quality, length, correctness, contextual spelling, syntactic complexity, and mechanics (Shen & Coker, 2023) differentially tap language and component skills (Coker et al., 2018); therefore, reading-writing relations tend to differ when different measures were used (Kim et al., 2023).
The Present Review
The TIH provides a promising framework for creating continuity and conversations between researchers and practitioners across programs and curricula to benefit multilinguals. In the current systematic literature review, I test the TIH by synthesizing findings on the reading-writing connections across languages. I sought to address two major research aims: (1) to examine the relations between reading and writing across multilinguals’ first (L1) and second/foreign (L2) languages (and in some cases the third language [L3]) and (2) to investigate factors that tend to affect the cross-language reading-writing relations. According to the TIH, I hypothesized that multilinguals’ reading and writing skills are interrelated across languages; however, the extent of such relations may vary depending on factors such as language exposure, motivation, language proficiency, language structure similarities and distance, grain sizes, language experience and grade level, and differences in measurement.
Method
Search Strategy
The current review adheres to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) procedures and guidelines (Moher et al., 2009). The literature search was conducted using PsycInfo, Educational Resources Information Center (ERIC), Web of Science, and Google Scholar databases with keywords for studies published before January 2023. Searches were conducted using the following search terms as keywords in the article abstract. Because of our focus on reading-writing connection and multilinguals, searches included both constraints on reading and writing/spelling and multilingualism characteristics, such as L2 learners and English as a second language. Therefore, the search queries used were “reading writing” located in article abstracts (AB) AND “L2 OR “second language” OR “foreign language” OR bilingual OR “English Language Learner” OR “English Learner” OR “English as a Second Language” OR “English as a Foreign Language” OR EL OR ESL OR EFL” located in AB. In the second search, “reading writing” was replaced with the term “reading spelling” and search queries were repeated. The terms used in the database searches were broad in order to capture more studies on the relations between reading and writing measured in various types of assessments, as well as the heterogeneous population of multilinguals worldwide. The search was conducted by the author, with consultation from the university librarians. Excel spreadsheets were used to log the results of the search and screening processes and inclusion/exclusion decisions.
Inclusion and Exclusion Criteria
To be eligible for inclusion in this review, studies had to meet the following criteria. First, studies included multilinguals without disabilities reported. Here, multilinguals include second language learners and foreign language learners, as well as multilinguals. Second language learners are those who are learning the dominant language of society at school but coming from a heritage language-speaking family; foreign language learners are those who are learning a foreign language in their own country; and multilinguals refer to those who learn to speak multilingual languages from an early age. Second, studies assessed multilinguals’ reading and writing performance at the word, sentence, or text levels in both languages. Studies may include reading assessment at one grain size (e.g., word reading) in one language and writing test at one grain size (e.g., written composition) in another language; however, reading and writing skills should be included cross-linguistically (e.g., L1 word reading and L2 written composition). Third, studies need to be empirical, using quantitative methods in data analyses, such as correlations, regression, and structural equation modeling analyses. Finally, studies were published in English in a peer-reviewed journal.
Studies were excluded if they (a) included multilinguals with disabilities, (b) mixed group of heritage language speakers and L2 learners, (c) assessed skills in one modality only (e.g., L1 and L2 reading skills), (d) assessed skills in one language only (e.g., L2 reading and L2 writing skills), (e) only included self-reported reading and writing performance, (f) adopted data analyses approaches other than quantitative methods, or (g) were published in non-English languages or in journals that do not require peer review. I excluded multilinguals with disabilities because the disabilities they experience may alter the cross-language relations of reading and writing. For example, multilinguals with dyslexia may have difficulty reading in one or both languages, which may affect the cross-language relation. I further excluded studies with mixed groups of heritage language speakers and L2 learners because L1 and L2 may not be identifiable, making it hard to explore the factors affecting the cross-language transfer.
Search Results
An overview of search, screening, and identification procedures is presented in Figure 2. In total, 3,564 studies were found in the initial search. After excluding 826 duplicates across databases, 2,738 studies’ titles and abstracts were screened for inclusion. At this stage, studies were excluded if they were qualitative (e.g., Fitzgerald & Noblit, 1999), not peer-reviewed (e.g., Ahmadi, 2012), or systematic reviews themselves (e.g., Lim et al., 2022), as well as if they did not include multilinguals (e.g., Achugar et al., 2007) or reading and writing assessments (e.g., O’Brien et al., 2020). As a result, the full texts of 79 studies were retrieved and reviewed for inclusion. Upon full-text review, 43 studies were excluded as they did not meet inclusion criteria. Specifically, one study included multilinguals with disabilities (i.e., dyslexia; van Setten et al., 2017); one study included a mixed group of heritage language speakers and L2 learners, such that either language was the L1 for a subsample of the group and L2 for the rest (He et al., 2022); and two studies included both English native-speaking and heritage language speaking children in French immersion programs and examined the relations between English and French (Chung et al., 2018, 2023), so English can be identified as L1 or L2 or even L3 and French can be identified as L2 or L3. Nine studies did not include reading and writing measures in multiple languages (e.g., Goodrich et al., 2016; Sparks & Patton, 2013; Zaretsky, 2020). Thirty-one studies did not examine the cross-language relations between reading and writing, albeit the availability of cross-linguistic measures (e.g., Soltero-González et al., 2016; Sparrow et al., 2014).
PRISMA chart of the screening process for the current systematic review.
A total of 36 studies in 27 journals met the inclusion criteria in the database. All journals with included study/studies through database search were hand searched and screened. This resulted in an additional 5 studies that met the inclusion criteria. The final sample included 41 studies with 52 reports. These reports included 19 different L1s (English, Chinese, Korean, Hebrew, Hungarian, Russian, Spanish, Cantonese, Arabic, Chichewa, Indian language, Italian, Japanese, Malay, Nepali, Urdu, Persian, Portuguese, and Tamil) and 6 different L2s (English, French, German, Spanish, Hebrew, and Chinese), with 5 additional reports including English as the L3. Participants included those from preschool to adulthood.
Data Coding
The author and four research assistants independently reviewed all included articles and met to discuss discrepancies in coding. Any discrepancies in coding were resolved through discussion. The final coding in the online supplementary materials reflects the consensus. Exact agreement reliabilities for the various codes ranged from 95% to 99%. Coding included the study type (i.e., correlational or interventional), characteristics of subjects (e.g., sample size, grade level), L1 and L2/L3, language exposure, reading and writing measures (e.g., word reading, written composition), relations examined (e.g., L1 spelling-L2 word reading), type of quantitative analyses (e.g., multiple regression, correlation), and findings related to our research aims. Later, I report statistically significant findings from the reviewed studies. If there exists an association, it signals that such an association is statistically significant.
Positionality Statement
This review is shaped by my positionality as a bilingual educator and researcher with language proficiency in Mandarin Chinese, English, and some Spanish. It is also shaped by my classroom teaching experience in a dual language immersion program. During my time as a teacher, I witnessed how children faced difficulties in reading and writing in a language different from their native language, with considerable individual differences at an early age/stage of language learning. I pursued my doctoral degree to further understand the mechanisms underlying these individual differences. The knowledge I gained from my classroom teaching experience and doctoral training shaped my personal understanding of literacy development in two languages, which I view as a vision of literacy practices that prioritizes the inclusion of heritage language development, particularly in the context of learners speaking a heritage language in the United States. Recognizing the complexity of linguistic and cultural identity, I seek ways to promote multilingual children’s literacy development in both languages. In this review, I leveraged reading-writing connections, as a long-established theory and practice in monolingual literature and integrated it with cross-language transfer.
Results
Is There a Relation Between Reading and Writing Across Multilinguals’ L1, L2, and L3?
Cross-Language Reading-Writing Connections at Different Grain Sizes
Lexical Level—Word Reading and Spelling
Table 1 presents the reports and findings of cross-language reading-writing relations at the lexical level, organized by associations between L1 reading and L2 writing or L1 writing and L2 reading. Among 23 reports examining the association between L1 word reading and L2 spelling, 17 reports (73.9%) supported this positive relation, 4 reports (17.4%) failed to support this relation, and 2 reports (8.7%) resulted in mixed findings. Specifically, reports provided evidence of a cross-sectional association between L1 word reading and L2 spelling among children learning English as an L2 and spoke Russian (r = .38) [1], Arabic (r = .52) [4], Italian (r = .47) [11], Portuguese (r = .60) [10], Chinese (rs = .19–.55) [12, 23, 36, 38], Malay (rs = .39–.72) [23], Tamil (rs = .28–.60) [23], and Hebrew (r = .27) [26] as their L1. This cross-linguistic relation between L1 word reading and L2 spelling also existed among English-speaking children learning Hebrew as an L2 (rs = .52–.72) [13]. Findings indicate weak to strong positive correlations cross-sectionally. Further, weak to strong positive longitudinal associations between L1 word reading and L2 spelling were also reported among English L2 learners from Chinese (rs = .19–.48) [23, 38], Malay (rs = .39–.72) [23], and Tamil (rs = .28–.60) [23] L1 backgrounds, as well as English L1 individuals learning Spanish, French, or German as L2s (rs = .48–.76) [30–35]. However, L1 word reading was not correlated with L2 spelling for Russian-Hebrew [3], Chinese-English [21, 40], and Persian-English children [5]. In two other studies, although a weak to moderate positive correlation between L1 word reading and L2 spelling was reported (rs = .27–.36), the association disappeared after controlling for other variables [20, 26].
Cross-Language Reading-Writing Connections at the Lexical Level
Among 24 reports focusing on the cross-language relation between L1 spelling and L2 word reading, 18 of them (75%) provided evidence of a positive association, 2 (8.3%) failed to support such association, and 4 (16.7%) reported mixed findings. Specifically, a weak to strong positive cross-sectional positive association between L1 spelling and L2 word reading was evidenced among English L2 learners who spoke Russian (r = .32) [1], Hebrew (rs = .32–.36) [3, 26], Arabic (r = .56) [4], Korean (rs = .52–.73) [7], Italian (r = .39 ) [11], Chinese (rs = .12–.42) [23, 36, 38], Malay (rs = .42–.72) [23], and Tamil (rs = .19–.38) [23] as their L1. Such positive relation was also evident among Russian-Hebrew (β = .42) [3] and English-Hebrew (rs = .29–.39) [13] bilinguals. In addition, reports also provided evidence of a weak to strong positive longitudinal correlation between L1 spelling and L2 word reading among Korean-English (rs = .52–.73) [7], Chinese-English (rs = .12–.38) [23, 38], Malay-English (rs = .42–.72) [23], Tamil-English (rs = .19–.38) [23] bilinguals, as well as English L1 individuals learning Spanish, French, or German as L2s (rs = .41–.73) [30-35]. However, two studies failed to find a correlation between L1 spelling and L2 word reading after including control variables [12, 40]. Further, four studies found a weak to strong positive correlation (rs = .12–.55), but such associations may be dependent on the control variables included, such as grade level [5, 16, 20, 23].
Taken together, most evidence showed positive concurrent and longitudinal cross-linguistic associations between word reading and spelling across individuals speaking different language pairs; this relation, however, may differ in correspondence with age and the skills controlled for in statistical analysis. Further, due to the considerable variability among studies, the magnitude of lexical-level reading-writing connection also varied from weak to strong.
Discourse Level—Reading Comprehension and Written Composition
Table 2 presents a summary of studies examining cross-language reading-writing connections at the discourse level. Among 15 reports on the association between L1 reading comprehension and L2 written composition, 7 of them (46.7%) provided evidence of a positive cross-language relation, 3 (20%) failed to support this relation, and 5 (33.3%) reported mixed findings. Specifically, L1 reading comprehension was positively correlated with L2 written composition cross-sectionally, suggesting a moderate to strong reading-writing association among Hebrew EFL seventh graders (r = .59) [2]; Hungarian EFL and German as a foreign language (GFL) sixth, eighth (rs = .37–.52) [22], and tenth graders (rs = .32–.46) [9]; Korean middle (r = .38) [25] and high school EFL students (r = .34) [24]; and Hong Kong EFL tenth graders (rs = .45–.49) [41]. However, three other reports provided no evidence of cross-language relations between L1 reading comprehension and L2 written composition [6, 19]. First, L1 reading comprehension was not associated with L2 argumentative essay composition among Korean college EFL students of high L2 proficiency [19]. One study also failed to observe significant cross-language associations between L1 reading and L2 writing among either Chinese or Japanese adult ESL learners [6]. In terms of mixed findings, for Korean-English students in grades 8 and 11, although a moderate positive correlation was found between L1 reading comprehension and L2 written composition (r = .43), such a relation disappeared after controlling for L2 proficiency [18]. Further, studies of a sample of English-speaking students learning Spanish, French, and German (L2) who were followed over 10 years showed that longitudinally, L1 reading comprehension averaged across grades 1–5 was moderately correlated with (r = .45) but did not predict high school L2 written composition after controlling for other L1 skills in elementary school [33], and such relations were dependent on when the early L1 reading comprehension was measured [34]; cross-sectionally, high school L1 reading comprehension was strongly correlated with (r = .54) but also did not predict L2 written composition after controlling for elementary L1 literacy skills, cognitive ability in L1, and L2 aptitude [33].
Cross-Language Reading-Writing Connections at the Discourse Level
Among nine reports examining the cross-linguistic relation between L1 written composition and L2 reading comprehension, two (22.2%) revealed significant positive relations. Such positive associations were found among Russian-Hebrew high schoolers (β = .23) [15] and Hungarian EFL eighth graders (r = .55) [22], suggesting a weak to strong reading-writing association. In contrast, three other reports (33.3%) failed to observe significant relations between L1 written composition and L2 reading comprehension. One study found no cross-language relations between L1 writing and L2 reading among either Chinese or Japanese adult ESL learners [6]. Additionally, L1 argumentative writing and L2 reading comprehension were not correlated among Korean college EFL students [19]. Four other reports (44.4%) provided mixed findings. One showed moderate positive correlations (rs = .34–.44) but the associations disappeared after controlling for L2 proficiency for Korean-English eighth and eleventh graders [18]. Likewise, another study also revealed a strong positive correlation (r = .55), but such a relation disappeared after controlling for L1 reading and L2 proficiency [25]. The study further showed that relations differ with different genres of L1 writing for Korean high school EFL students. That is, L1 narrative writing but not argumentative writing was weakly correlated with L2 reading comprehension (r = .27) [24].
In conclusion, there is some evidence supporting the cross-linguistic associations between reading and writing at the discourse level; however, variations in control variables, grade level, and genre used in writing may come into play to largely affect the existence and magnitude of such associations.
Cross-Language Reading-Writing Connections Across Grain Sizes
Table 3 presents a summary of findings of studies focusing on cross-language reading-writing relations across different grain sizes. The reviewed literature has indeed demonstrated significant reading-writing connections across the lexical, sentence, and discourse levels but the findings are somewhat mixed. One report specifically focused on the association between L1 word reading and L2 sentence writing, which found that L1 word reading accuracy and fluency in grade 1 were longitudinally associated with Hong Kong EFL children’s L2 sentence writing fluency in grade 3 and grade 4 (rs = .35–.53), indicating a moderate to strong association; however, after controlling for other L1 cognitive and linguistic skills, L1 word reading fluency in grade 1 (β = .34), but not accuracy, tended to have a moderate effect on grade 4 sentence writing fluency [29].
Cross-Language Reading-Writing Connections Across Grain Sizes
Among five reports examining the relation between L1 word reading and L2 written composition, three (60%) supported a positive association. One study found weak correlations between grade 1 L1 word reading and L2 written composition at grade 6, measured by writing accuracy (r = .27) and persuasion scores (r = .23) for English-French bilingual children [27]. Two studies of English-speaking high schoolers learning Spanish, French, or German for 2 years, who were followed over 10 years from grade 1 supported moderate associations between early L1 word reading skills averaged across elementary school and later L2 written composition abilities in high school (rs = .39–.40) [32, 33]. However, two other reports (40%) reported mixed findings, depending on multilinguals’ grade level or age. One found that L1 word reading measured at the end of grades 2, 3, and 5 (rs = .45–.48) but not at the beginning and end of grade 1 were moderately correlated with high school L2 written composition [34]. Another also showed a weak longitudinal relation between L1 word reading and L2 written composition, such that L1 word reading tested at ages 5, 7, and 8 (rs = .19–.20) but not at ages 6 and 9, were weakly correlated with L2 written composition at age 9 among Chinese EFL children [39].
Six (66.7%) out of nine studies provided evidence of a positive association between L1 reading comprehension and L2 spelling. Such relation was found among Hebrew EFL learners (r = .47) [2], Chinese EFL children (r = .53) [20], and high school English-speaking students learning Spanish, French, or German as an L2 (rs = .31–.62) [31, 32, 34, 35], suggesting a moderate to strong reading-writing connection. Specifically, L1 reading comprehension tested at the beginning of grade 1, and the end of grades 1, 2, 3, and 5, were moderately to strongly correlated with high school L2 spelling (rs = .31–.53) [30, 31, 34]. However, one study didn’t find an association between L1 reading comprehension at grade 3 and high school L2 spelling [30]. L1 reading comprehension averaged across grades 1–5 was strongly correlated with high school L2 spelling (r = .62) [32, 33, 35]; however, such relation is likely to attenuate after controlling for other L1 skills in elementary school [33]. In addition, high school L1 reading comprehension was correlated strongly and concurrently with L2 spelling (r = .56), with such association disappearing after controlling for elementary L1 literacy skills and cognitive abilities, and L2 aptitude [33]. Further, one study failed to support such cross-linguistic association; specifically, for Russian-Hebrew children in grade 6, L1 reading comprehension was not correlated with L2 spelling [3].
Five (83.3%) out of six studies also provided evidence of a positive relation between L1 spelling and L2 reading comprehension. Specifically, L1 spelling was concurrently and strongly correlated with L2 reading comprehension among Hong Kong EFL children (r = .51) [20]. Additionally, L1 spelling tested in elementary school had consistently moderate to strong associations with L2 reading comprehension for Spanish, French, or German L2 high schoolers (rs = .34–.52) [32–35]. In contrast, one report failed to observe a relation between L1 spelling and L2 reading comprehension among Russian-Hebrew bilingual sixth graders [3].
One study speaks to the relation between L1 written composition and L2 word reading and found that L1 written composition at age nine was weakly to moderately correlated with L2 word reading at all five ages (ages 5–9), with the correlation being stronger in later years (rs = .24–.36) [39].
Evidence supporting a cross-language association was also reported in one study [28] exploring the relation between Chichewa (L1) and English (L2) among Malawi students in grades 2 and 3, which used composite L1 and L2 measures across grain sizes. Specifically, L1 and L2 reading skills in both grades were assessed using word recognition, sentence reading, and word reading accuracy. Writing skills were measured with dictation and writing accuracy, with one exception. Grade 3 Chichewa writing was assessed with story writing, using composing and mechanics scores. Fall L1 reading significantly moderately predicted spring L2 writing among grade 2 students only (β = .31). Fall L1 writing, however, had a small effect on spring L2 reading for grade 3 students only (β = .11). As such, this difference from grade 2 to grade 3 may be developmental or the result of a difference between the writing tasks. It is also surmised that it may take longer for bilingual students to reach an adequate level of L1 writing competence. As children become more proficient in both languages, L1 writing may be a more powerful contributor to L2 reading.
In conclusion, the majority of the evidence supported a cross-linguistic association between reading and writing across grain sizes with a magnitude ranging from weak to strong association; however, a number of studies also reported mixed findings; that is, such cross-language association appears to be dependent on grade level, age, and control variables used.
Reading-Writing Connections Across Three Languages
Thirteen reports extended this line of reading-writing relation inquiry by investigating the degree of transfer of cross-language associations across three languages (Table 4), and five reports (38.5%) provided evidence of a cross-linguistic reading-writing connection across three languages. For instance, a report [3] found that L1 reading comprehension moderately contributed to L3 spelling (β = .36) and L1 spelling also moderately contributed to L3 reading comprehension (β = .32) in a group of native Russian speakers learning English as an L3 and Hebrew as L2. However, they failed to observe a relation between L1/L3 word reading and L3/L1 spelling. Another report [26] observed weak but positive correlations between L1/L3 word reading and L3/L1 spelling for Arabic L1 speakers learning English as an L3 (rs = .17–.25); however, the relation between L1 word reading and L3 spelling disappeared after controlling for other L1 or L1 and L3 skills. Further, a report [40] observed moderate to strong correlations between L2/L3 word reading and L3/L2 spelling among Chinese L2–English L3 children speaking diverse L1s in Hong Kong (rs = .38–.56). Two reports [14, 15] recruited Russian (L1)-speaking immigrant 11th graders studying English (L3) as a subject in Israeli schools, where Hebrew (L2) was the dominant societal language. When examining the contribution of L1 letter writing to L2 and L3 reading comprehension, one [15] found that after controlling for sex, socioeconomic status, and age of arrival in Israel, L1 written composition did not predict L3 reading comprehension. Additionally, L2/L3 written composition explained a significant amount of variance in L3/L2 reading comprehension (βs = .20–.28). Even after they [14] controlled for linguistic and social-psychological variables (such as current language use, home languages, attitudes toward L1, parental engagement, and motivation), the reading-writing relations remained significant (βs = .18–.20). Notably, neither study found a transfer effect of reading to writing from L2 to L3 or from L3 to L2.
Cross-Language Reading-Writing Connections Across Three Languages
These findings suggest that when L3 writing was assessed at word level, L2 word reading and L1 reading comprehension were associated with L3 writing [3, 40]. When writing was measured at discourse level, writing ability in either L2 or L3 was related to success in L3 or L2 reading, respectively [14, 15]. Results demonstrated that L2 and L3 are linguistically interdependent. In addition, while the development of L1 writing was associated with students’ performance in reading L2 texts, it was not related to their L3 reading, indicating a differential degree of relation across the three languages. This may be due to a stronger degree of interdependence between L1 and L2 than between L1 and L3, which may, in turn, be attributed to the different proficiency across L1, L2, and L3.
In conclusion, the review of the studies focusing on reading-writing connections across three languages showed mixed findings, with some studies reporting significant positive associations ranging from weak to strong, some failing to observe such association, and other studies reporting mixed findings depending on control variables and measures used in the studies.
Causal Evidence
To date, only five reports have examined the causal relation between reading and writing across languages, and one (20%) provided causal evidence (Table 5). Specifically, one report [17] tested the causal relation between L1 summary essay writing and L2 reading comprehension. They aimed to identify the unique contributions of summary essay writing and/or mapping in L1 tasks to support L2 science text comprehension. Korean college EFL learners read an L2 science text, received treatment of mapping, writing, both mapping and writing in either L1 or L2, and then completed a reading comprehension posttest. Compared to L2 treatment groups, L1 groups performed consistently better in L2 reading comprehension. Among the three L1 treatments, the combined use of L1 mapping and writing was most effective in improving learners’ L2 reading and then L1 writing (Cohen’s d = 1.37), and the least effective was L1 mapping alone. Findings suggest that L1 essay writing can provide low L2 proficiency students with additional support to improve their L2 reading comprehension.
Causal Evidence of Cross-Language Reading-Writing Connections
Four reports speak to the causal cross-linguistic impact of reading instruction on spelling. One study [37] used a comprehensive, integrated English or Spanish reading intervention curriculum to improve Spanish-English bilingual children’s English and Spanish literacy skills. Results showed that not only L1 reading intervention did not have a significant impact on children’s L2 spelling, but also L2 reading intervention did not significantly improve children’s L1 spelling. It is notable that while comparing the cross-linguistic impact of reading intervention on spelling, word reading at screening was used as a covariate. The authors attributed this finding to the lack of opportunity to read in the other language while receiving reading instruction in one language. A follow-up study further demonstrated that reading intervention in children’s L1 or L2 did not improve their L2 or L1 spelling skills one year later, respectively [8].
In conclusion, most evidence failed to support the causal relations between reading and writing across languages; however, four out of five reports used the same general reading intervention, which may be less effective in promoting cross-linguistic transfer. The one study focusing on L1 essay writing intervention reported a significant causal link between L1 essay writing and L2 reading comprehension, indicating the potential of using essay writing to promote cross-linguistic transfer.
What Are the Factors Affecting the Cross-Language Reading-Writing Relations?
Table 6 presents the factors hypothesized in the TIH, available evidence, and conclusions. Results of all studies and specific findings of the associations between factors and the cross-linguistic relations are presented in the online supplementary material. Below, I summarized the results of the reviewed studies for each of the factors.
Factors Affecting the Cross-Language Reading-Writing Connections
Language Exposure
Information on language exposure of participants in the reviewed studies is provided in the online supplementary materials. There are numerous nuanced differences among the reviewed studies in terms of language exposure, ranging from total L2 immersion, where children were immersed in a total L2 environment at school [27], to only a year of L2 learning experience as a subject [1]. These variations, accompanied by other differences among the reviewed studies (e.g., relations examined, age of the participants, measures used), made the findings more mixed when interpreting the relation between language exposure and the extent of the cross-language transfer of reading and writing. For example, for preschoolers with exposure to both L1 and L2 instruction at school, one study [23] observed consistent cross-linguistic associations between word reading and spelling across preschool years (rs = .12–.72), suggesting weak to strong relations, whereas another report [21] failed to find such cross-linguistic associations, either cross-sectionally or longitudinally. Further, language exposure in EFL and ESL contexts, albeit very different (i.e., EFL learners are more likely to have more L1 exposure than L2 exposure and more established L1 literacy, compared to ESL learners), did not show different relations with the extent of the cross-linguistic transfer. Reports of participants in EFL contexts showed both significant (e.g., [1, 3, 7, 9, 18, 22, 25]; rs = .28–.73) and nonsignificant (e.g., [3, 12, 19, 21, 24, 26]) relations between reading and writing cross-linguistically; the same is also applied to ESL participants showing significant (e.g., [4, 10, 11]; rs = .39–.60) and nonsignificant relations (e.g., [5, 6]). Note that the variation of the magnitudes of reading-writing connection was larger for EFLs (weak to strong associations), compared to ESLs (moderate to strong associations). Moreover, no single study has specifically examined whether language exposure can alter the extent of cross-linguistic transfer between reading and writing. As such, due to the considerable variations and lack of studies testing the direct moderating effect, no conclusions can be made to testify to this hypothesis of the TIH.
Motivation
Two studies have included motivation as a construct. Specifically, one study [32] found weak to moderate correlations (rs = .27–.46) between L2 motivation in high school and both elementary school L1 reading and writing skills and high school L2 reading and writing skills (except L2 spelling). Further, L2 motivation was one of the best predictors to contribute to high school L2 reading comprehension (β = .23) but not L2 written composition, L2 word reading, or L2 spelling; however, only 9% of the variance in L2 reading comprehension was explained by L2 motivation. This suggests that L2 motivation alone is unlikely to be sufficient for mastering an L2 in high school. Notably, the L2 learning motivation construct included in [32] tapped onto instrumental, integrative, and assimilative motives, as well as one’s enjoyment and perceived importance of L2 learning. However, one study [14] found that motivation did not predict L2 or L3 reading or writing skills, after controlling for other demographic, linguistic, and social-psychological variables. The motivation construct in [14] encompassed an individual’s attitudes, intensity, and effort to learn a language; different aspects of the motivation constructs were included as individual predictors to predict L2 and L3 reading comprehension and written composition skills, such as perceived multilingual functioning, attitudes toward L1, integrative orientation, attitudes toward speakers of the language and learning situation, instrumental orientation, motivational intensity, and attitudes toward learning the language. None of these aspects were significant predictors. Further, no evidence of a moderating effect was found since no study has focused on the relations between motivational levels and the extent of the cross-linguistic transfer of literacy skills. Thus, language learning motivation alone may show weak to moderate associations with literacy skills, but such associations may attenuate with additional skills and factors taken into account; no evidence supported this hypothesis in the TIH.
Language Proficiency
Among the reviewed studies, nine reports provided a narrative description of multilinguals’ L2 proficiency level; a positive cross-linguistic reading-writing relation was observed for low ([12, 17, 31]; cf. [8, 37]), intermediate [25], low to high [31], and high [1] L2 proficiency. However, including multilinguals with low-intermediate to high L2 proficiency, one study [6] did not find reading-writing connections across languages.
Studies examining the relation between language proficiency and literacy skills have also used different indicators to measure multilinguals’ proficiency levels. First, vocabulary was used to indicate multilinguals’ language proficiency in 14 studies. Two studies generally demonstrated that both L1 and L2 proficiency were related to L1 and L2 reading and writing skills (rs = .23–.66) [20, 39], suggesting a weak to strong association. A series of studies of foreign language learners in the United States revealed weak to strong associations between L1 proficiency and both L1 and L2 literacy skills (rs = .28–.73) [30–35]. Some other studies showed that L2 proficiency was moderately to strongly associated with L2 reading and writing skills (rs = .46–.80) [5, 19, 40] but not with L1 reading and writing skills [5, 19, 40]; L1 proficiency, similarly, showed a moderate to strong association with L1 reading and writing skills (rs = .30–.70) [5, 21, 40] but not to L2 skills [5, 21, 27, 40]. However, one study [13] did not find an association between L2 proficiency and either L1 or L2 literacy skills. Second, using both vocabulary and grammar to indicate L2 proficiency, one study [24] revealed weak to strong positive associations between L2 proficiency and L1 and L2 reading and writing abilities (rs = .27–.71). Third, studies using listening skills as an indicator of language proficiency showed a weak to strong correlation between L2 listening comprehension skills and both L1 and L2 reading and writing skills (rs = .19–.75) [9, 22]; L1 listening comprehension skills was also moderately to strongly correlated with both L1 and L2 literacy skills (rs = .31–.73) [32, 33, 35]. Fourth, using a more comprehensive measure of linguistic comprehension, one report [27] found that L1 linguistic comprehension in grade 1 predicted L2 persuasive writing in grade 6. Finally, some studies used either listening and speaking or all four skills—reading, writing, speaking, and listening—to indicate multilinguals’ L2 proficiency and revealed moderate to strong relations between L2 proficiency and L1 and L2 reading and writing abilities (rs = .30–.81) [30–34].
One study [18] specifically investigated the moderating effect of L2 proficiency (vocabulary and grammar) and showed that for Korean EFLs in grade 8 and grade 11, L2 proficiency did not moderate the relation between L1 reading comprehension and L2 writing; while L2 proficiency similarly did not moderate the association between L1 writing and L2 reading for grade 11 students, a weak moderating effect (β = −.11) was observed for grade 8 students, with such relation being stronger for students with a lower level of L2 proficiency.
In conclusion, cross-linguistic reading-writing connections were observed among multilinguals with varying levels of L2 proficiency. Most evidence also supported that language proficiency in L1 and L2 appears to be related to multilinguals’ literacy skills. One study directly testing the moderating effect produced mixed findings; such moderating effect of motivation on cross-language reading-writing connections may be dependent on the constructs and relations under examination, as well as multilinguals’ grade level.
Language Structure Similarities and Distance
The reviewed studies included multilinguals speaking diverse language pairs. Generally, multilinguals using languages that share writing systems are more likely to transfer their literacy skills across languages, although the role of language structure similarities and distance were not straightforward. For example, one study [23] observed strongest associations between word reading and spelling across languages for Malay-English bilinguals (both languages are alphabetic; rs = .39–.72), weaker associations for Tamil-English bilinguals (Tamil is Abugida [alphasyllabary]; rs = .19–.60), and the weakest associations for Chinese-English bilinguals (Chinese is logographic; rs = .12–.48). Further, reports that failed to observe cross-linguistic reading-writing connections often included participants speaking distantly related languages, such as Russian-Hebrew (Russian is alphabetic and Hebrew is Abjad; [3]), Persian-English (Persian is semi-syllabary; [5]), Chinese-English [6, 12, 21, 40], Korean-English (Korean is alphabetic or logographic; [19]), and Japanese-English (Japanese is syllabary or logographic; [6]). However, positive cross-linguistic reading-writing associations were also reported in studies of participants speaking these distantly related language pairs, including Russian-Hebrew (βs = .18–.28) [15], Hebrew-English (e.g., [3, 13, 26]; rs = .27–.72), Chinese-English (e.g., [12, 20, 38, 41]; rs = .19–.55), Korean-English (e.g., [7, 18, 24, 25]; rs = .27–.73), and Arabic-English (e.g., [4, 26]; rs = .17–.56), indicating a weak to strong reading-writing connection across languages. As such, there is evidence supporting that language structure similarities and differences may alter the strength of the cross-language relations between reading and writing; however, with considerate variability in other factors of the reviewed studies, the evidence is mixed.
Language Experience and Grade Level
Multilinguals’ grade level/age and language learning experience are presented in the online supplementary material. Most studies focusing on multilinguals with early L2 learning experience investigated the within-grain size relations between reading and writing at the lexical level; majority provided evidence of a positive cross-linguistic relations for preschool (rs = .12–.72) [23], elementary (rs = .17–.72) [4, 5, 10, 11, 13, 20, 26, 36, 38], and middle school (rs = .32–.35) [3] students, suggesting a weak to strong association. However, one study [1] focusing on multilingual college students with later L2 learning experience also supported moderate lexical-level relations (rs = .32–.38). Studies focusing on multilinguals with later L2 learning experience produced some mixed findings. Specifically, within-grain size discourse relations between reading and writing across languages were found in one study [17] but not in another [6]. Across grain sizes, one report [33] revealed that L1 reading comprehension was not associated with L2 spelling or written composition, after controlling for elementary L1 literacy skills, cognitive ability in L1, and L2 aptitude. As such, although multilinguals with early L2 learning experience showed more consistent associations between lexical-level reading and writing skills, the evidence is mixed for multilinguals with later L2 learning experience. This suggests that this hypothesis of the TIH is not fully supported.
Reading and Writing Measures
Different types of reading comprehension tasks and various genres of written composition appear to be associated differently across languages. The reviewed studies showed that reading comprehension measured by multiple-choice tasks was more likely to transfer to written composition skills across languages (rs = .27–.73 [2, 17, 18, 20, 22, 24, 25, 32-34]; cf. [19]), compared to when reading comprehension was measured by cloze [6]. Additionally, one study [41] used short answer tasks of single- and multiple-text passages to measure multilinguals’ reading comprehension and revealed moderate positive associations with written composition across languages (rs = .45–.49). Moreover, in terms of writing in different genres, it appears that essay writing eliciting narrative/expository/descriptive responses was more likely to be weakly to strongly associated with reading comprehension across languages (rs = .27–.59) [2, 9, 14, 15, 18, 22, 24, 25, 32–34], compared to writing requiring argumentative skills ([6, 19, 24]; cf. [27]). Similarly, integrated writing, including writing after passage reading, was likely to show a moderate cross-linguistic relation (rs = .45–.49) [41]. Therefore, the findings of the reviewed studies supported this hypothesis of the TIH, such that measurement differences, such as different tasks and genres used to measure reading comprehension and written composition, can influence the reading-writing relations across languages.
Discussion
The overarching goal was to test the TIH in the context of the current literature by synthesizing findings on the reading-writing connections across languages. This review of the current research can help us determine (1) whether the relation between reading and writing across multilinguals’ languages existed and (2) factors that affected such cross-linguistic reading-writing connections. This is the first step toward a theoretical and pedagogical literacy framework for facilitating reading-writing relations in multilinguals. Establishing the nature of reading-writing connections can provide important and constructive pedagogical perspectives for reading and writing in multilinguals’ both languages. Determining the factors associated with the relation between reading and writing across multilinguals’ L1 and L2 can provide major advances in our understanding of the mechanisms responsible for literacy development and language transfer.
Cross-Language Reading-Writing Connections
Most studies reviewed provided evidence of a cross-linguistic association between reading and writing. Such evidence has supported a cross-linguistic reading-writing connection at both lexical level (i.e., word reading and spelling; e.g., [1, 10, 12, 13, 23, 26, 38]) and discourse level (i.e., reading comprehension and written composition; e.g., [2, 9, 18, 24, 25, 41]). Moreover, studies also found a cross-linguistic association between reading and writing across different grain sizes, such that lexical-level literacy skills were associated with discourse-level abilities (e.g., [20, 27, 33, 34, 39]) and sentence-level literacy skills (e.g., [29]).
It is also notable that more evidence supported a cross-language reading and writing association at the lexical level. That is, word reading and spelling tend to show a more consistent relation across different languages, regardless of the distance between two languages, when compared to discourse-level or cross-level associations. Indeed, of the 52 reports reviewed, about half provided evidence of a lexical-level reading-writing connection across languages. This is consistent with the TIH positing that reading and writing are more likely to be related at the lower levels because they both build on a constrained set of knowledge and skills.
Additionally, an association between reading and writing across multilinguals’ three languages was also found (e.g., [14, 15, 40]), suggesting that reading and writing are more likely to be interdependent between L1 and L2, as well as L2 and L3. One study also provided causal evidence of the significant impact of L1 essay writing on L2 reading comprehension [17]. These findings supported the TIH, such that reading and writing relations existed across multilinguals’ different languages. However, the reviewed studies also provided some mixed (e.g., [16, 18]) and contradicting findings that revealed a lack of connections between reading and writing across languages (e.g., [6, 8, 19, 21, 37]). As such, there are likely different factors at play that may explain such variability of findings.
Factors Related to Cross-Language Reading-Writing Connections
Language Exposure
According to the TIH, multilinguals should have sufficient exposure to the languages in academic contexts to enable the cross-language transfer of literacy skills. The reviewed studies provided mixed evidence in terms of the role of language exposure in facilitating the cross-language transfer between reading and writing. A positive relation was found among children who were immersed in a total L2 environment at school [27], as well as those with only a year of L2 learning experience [1]. Further, both EFL and ESL multilinguals showed significant and nonsignificant associations between reading and writing across languages. No single study has directly tested the moderating effect of language exposure or has included multilinguals with different language exposure. As such, no conclusions can be made in terms of the role of language exposure in the extent of cross-language transfer. Future studies may directly focus on language exposure as a moderator to test whether there exists a moderating effect on the cross-language relations between reading and writing; studies may also include different groups of multilinguals with various language exposure to examine group differences in the extent of such cross-language transfer.
Motivation
Language learning motivation is posited to be a key factor that may influence the extent of cross-language transfer, such that multilinguals with higher levels of language learning motivation are more likely to transfer literacy skills across languages. However, due to the lack of research on the role of motivation, I did not find evidence of a moderating effect of motivation, although one study demonstrated positive associations between L2 learning motivation and L1 and L2 literacy skills [32]. As such, no conclusions can be made to testify to this hypothesis of the TIH. Studies including motivation as a moderator are necessary to examine whether language learning motivation can alter the extent of cross-language transfer between reading and writing. Given the complexity of the language learning motivation construct, different aspects of motivation measured in both L1 and L2 may be included in future studies.
Language Proficiency
The TIH further posits that multilinguals need to achieve a certain level of language proficiency to facilitate cross-language transfer. The reviewed studies demonstrated a positive cross-language, cross-modal relation for multilinguals with various levels of L2 proficiency, although L1 and L2 proficiency are generally associated with multilinguals’ literacy skills. One study observed a moderating effect of L2 proficiency on the relation between L1 written composition and L2 reading comprehension [18]. The association between L1 writing and L2 reading was stronger for grade 8 EFL students with a lower level of L2 proficiency, suggesting that multilinguals with relatively low L2 proficiency were more likely to transfer their L1 writing skills to L2 reading skills. However, this does not indicate that multilinguals with low L2 proficiency in this study did not reach an adequate level of L2 proficiency since they had been exposed to L2 literacy for five years. In addition, language proficiency did not moderate the relation between L1 reading and L2 writing, suggesting that the role of motivation may be dependent on the relations under examination. Further, the moderating effect of L2 proficiency only existed among eighth graders but not among eleventh graders; this indicates that age, grade level, or language exposure differences may also contribute to the existence of the moderating effect of language proficiency.
However, evidence also suggests that L1 writing plays a greater role in L2 literacy compared to that of L3. Thus, there may be a stronger degree of interdependence between L1 and L2 than between L1 and L3. This is likely to be attributed to the variations of proficiency levels across learners’ L1, L2, and L3, as trilingual learners achieved a higher level of proficiency in their L2 than in their L3. Indeed, trilingual learners’ L1 writing was found to be associated with their L2 reading but not L3 reading [14, 15]. Likewise, in one study [28], different patterns of transfer may be accounted for by the different levels of performance in students’ L1 reading and L1 writing. L1 writing only explained significant variance in L2 reading among third graders. It may take longer for bilinguals to reach an adequate level of L1 writing competency to initiate the transfer from L1 writing to L2 reading and writing skills.
As such, although existing evidence suggests that cross-language transfer between reading and writing may require different levels of proficiency to be achieved, the studies reviewed produced mixed findings. The single study focusing on the moderation of language proficiency only focused on L2 proficiency [18]. This calls for future research investigating proficiency in both L1 and L2 and its moderating effects on cross-language transfer. In addition, language proficiency can be assessed in many different ways. The TIH posits that it is not only the oral language proficiency (e.g., vocabulary and grammar) but also the written language proficiency (e.g., reading and writing) that may affect the extent of the cross-language transfer. No direct evidence is available concerning the role of written language proficiency; future studies may examine the different aspects of proficiency in different languages.
Language Structure Similarities and Distance
According to the TIH, the extent to which language structure and features differ from each other can be a key factor affecting cross-linguistic reading-writing relations. Such cross-language transfer is more likely to occur between closely related languages compared to more distantly related languages. Indeed, evidence showed stronger associations between two closely associated languages [23]. Moreover, studies showing no evidence of cross-language associations between reading and writing recruited L2 English learners who mostly spoke L1 languages that are considered linguistically distant from English with distinguished orthographic features (Chiswick & Miller, 2005), such as Chinese, Japanese, and Korean. Cross-language reading-writing transfer, however, also occurred between distantly related languages in the studies reviewed here. The positive cross-linguistic relation between reading and writing was observed across L2 learners speaking diverse language pairs regardless of the linguistic distance between the two languages (e.g., [12, 23, 33, 39]).
Together, these findings suggest that language structure similarities and distance may play a role in the cross-language transfer of reading and writing, but the effect is not straightforward due to the variabilities of the reviewed studies. To further test this hypothesis of the TIH, studies may include participants speaking different language pairs (e.g., [23]) assessed with comprehensive reading and writing measures. Future research may also quantify language distance between languages and use the language distance indicator as a moderator when examining the cross-language reading-writing relations.
Language Experience and Grade Level
The TIH further posits that cross-linguistic reading-writing relations vary as a function of language experience and grade level. That is, multilinguals with early L2 learning experience may demonstrate stronger within-grain size lexical-level associations, whereas those with later L2 learning experience may demonstrate stronger cross-grain size relations. Results generally supported positive lexical-level relations for multilinguals with early L2 learning experience (e.g., [4, 10, 11, 20, 26, 36]); for multilinguals with later L2 learning experience, however, findings provided little evidence to support the TIH regarding the cross-grain size relations [33]. Therefore, no conclusions can be made to testify this hypothesis of the TIH. Studies comparing multilinguals with early and later L2 learning experiences may provide further evidence to testify the hypothesis.
Reading and Writing Measures
Lastly, the TIH postulates that the magnitude of the cross-linguistic reading-writing associations may vary depending on literacy measures. Reading and writing may be associated differently with different types of reading comprehension tasks and various genres of written composition. Findings supported the TIH in that reading comprehension measured by specific tasks (e.g., multiple choice) was more likely to transfer to writing across languages (e.g., [2, 17, 18, 20, 22, 24, 25]); writing in narrative genres and integrative writing were also more likely to transfer to reading across languages (e.g., [2, 17, 18, 20, 22, 24, 25]). As such, evidence supported the TIH regarding the role of reading and writing measures in cross-language transfer. Future studies may further examine whether the magnitudes of reading-writing relations across languages may be altered by reading and writing measures by including multiple reading comprehension (e.g., passage reading, multiple choice, cloze, reading fluency) and written composition (e.g., integrated reading and writing, argumentative writing, narrative writing) measures. Further, written composition can be scored in different ways, such as accuracy, length, quality, argumentative indicators, and so on.
How This Review Informs TIH
Taken together, in the TIH, it is proposed that reading and writing skills are interrelated across languages. Here, the literature on reading-writing connections across languages is directly synthesized. With some variability, evidence does suggest that reading and writing skills are interrelated across multilinguals’ L1 and L2 (and in some cases L3). This supports the core premise of the TIH, which posits that cross-language transfer occurs between reading and writing skills. While the TIH posits that a variety of factors may influence the cross-language relations between reading and writing, the review of existing studies reveals mixed or contradictory evidence regarding the factors that influence the extent of cross-language, cross-modal transfer. Some studies suggest that linguistic and literacy factors (e.g., reading and writing measures) play crucial roles, while others provide limited or inconsistent support. This variability may be attributed to the limited investigation of moderating variables, with only one study addressing the moderating effect of language proficiency [18]. These gaps in the literature highlight the need for further research to explore additional moderating factors that are important in literacy development—such as cognitive skills and experiential influences—which could refine and strengthen the TIH.
Limitations and Future Directions
This timely and necessary review of the literature on cross-language reading and writing relations would have benefited from an increased volume of peer-reviewed published literature. Evidence of a cross-language reading-writing relation beyond word level is scarce. The lack of peer-reviewed published literature limited the scope of this review, as unpublished work such as reports and dissertations was not considered in my search. As such, the estimates of the relation between reading and writing across languages could be an overestimation of these relations.
Future work is also warranted to focus on the mechanisms of the cross-linguistic reading-writing connections. Investigation into the moderating relations of various factors proposed in the TIH is necessary for us to gain a comprehensive understanding of such complex connections. However, extant literature has not extensively and systematically examined these linguistic and literacy factors. The TIH can serve as a starting point for such investigations to facilitate both theoretical and pedagogical progress in language and literacy learning for multilinguals.
Another future direction to test the TIH is to focus more on the associations at the sentence levels since no evidence was found due to the lack of studies. In the current review, no study has examined the within-grain size associations between sentence reading comprehension and sentence writing. Given the importance of sentence-level skills in reading and writing (e.g., MacKay et al., 2021; Wong, 2018), future studies may investigate such sentence-level associations to further test the TIH and gain more practical insights about instruction on sentence-level information.
While not a limitation per se, due to my focus on literacy outcomes, I only included quantitative studies in this review to test the TIH. However, for multilinguals, the process learning to construct their literacy skills in both languages is equally important. Although this process-oriented view of cross-language relation is out of the scope of the TIH, the TIH aligns with the concept that emphasizes the process of language learning and advocates using multilinguals’ full linguistic repertoire to make meaning and negotiate communicative contexts, and thus resources from all languages (L1 and L2 and even L3) can be leveraged to promote their learning in different languages (García & Wei, 2014). Future research may look into the process of reading and writing from a qualitative lens.
Furthermore, the language learning, exposure, and proficiency background of the participants were determined less clearly in most studies reviewed here. As multilinguals are not a homogeneous group when it comes to language experience and proficiency, this is important information to consider when interpreting research findings. As such, less is known about the role of proficiency in the relations between reading and writing across languages. Including a thorough description of multilinguals’ language learning experience when carrying out cross-language studies of reading-writing connections may provide additional insights into the cross-linguistic reading-writing connection.
Lastly, the author conducted the search. Although I sought consultation from university librarians, the reliability of the search procedure was not established in a way similar to how I dealt with the data coding procedure. Future studies including a collaborative team may further refine this process to include multiple researchers in the search process.
Conclusion
The transfer integration hypothesis is needed as a starting point for conversations about how practitioners should approach multilingual language and literacy teaching. Evidence supported that reading and writing are cross-linguistically associated for multilinguals, suggesting that multilinguals develop their literacy skills in either L1 or L2 by making connections between their L1 and L2 literacy skills. As such, the integration of both skills in multilingual classrooms may be essential to advance multilinguals toward proficient use of both skills. Teachers may draw on multilinguals’ L1 and L2 literacy skills to improve their L2 and L1 skills in the other literacy modality, and these cross-linguistic resources are valuable assets in performing literacy tasks in either language. Therefore, educational practices that reinforce the reciprocal relation between reading and writing, such as summary writing after reading source materials and text-structure analysis followed by writing, may be beneficial. Moreover, the significant relation between L1 and L2 reading and writing skills underscores the importance of L1 reading and writing experiences in developing L2 literacy skills. Therefore, multilinguals may benefit from ongoing school support in their L1 and encouragement to bring their experiences in L1 reading and writing to L2 reading and writing processes and vice versa. Findings, although mixed, indicate that there are many factors that can influence the extent of cross-language transfer of literacy skills. Researchers and practitioners should thus consider cross-linguistic constraints in the process of bilingual literacy development and make efforts to facilitate the cross-language transfer of literacy skills across modalities (i.e., reading and writing).
Note. The numbers in square brackets refer to the unique identifier for every paper included in the data set resulting from this literature review. The full list is provided in the online supplementary materials.
Supplemental Material
sj-docx-2-rer-10.3102_00346543251318999 – Supplemental material for Moving Beyond Cross-Language Transfer in a Single Modality: The Transfer Integration Hypothesis
Supplemental material, sj-docx-2-rer-10.3102_00346543251318999 for Moving Beyond Cross-Language Transfer in a Single Modality: The Transfer Integration Hypothesis by Ye Shen in Review of Educational Research
Supplemental Material
sj-xlsx-1-rer-10.3102_00346543251318999 – Supplemental material for Moving Beyond Cross-Language Transfer in a Single Modality: The Transfer Integration Hypothesis
Supplemental material, sj-xlsx-1-rer-10.3102_00346543251318999 for Moving Beyond Cross-Language Transfer in a Single Modality: The Transfer Integration Hypothesis by Ye Shen in Review of Educational Research
Footnotes
Acknowledgements
I would like to express my very great appreciation to Dr. Stephanie N. Del Tufo for her patient guidance, enthusiastic encouragement, and valuable and constructive feedback on this work. I also thank University of Delaware College of Education and Human Development faculty members for their helpful comments and advice on earlier versions of this manuscript. In addition, I appreciate the University of Delaware librarians for their guidance on the literature search process. I would also like to express my gratitude to my research assistants, who helped with data coding: Lori Chen, Kelly Chan, Grace Chen, and Shannon Xu.
Funding
Funding to support this research was provided by the Arnsdorf Summer Fellowship, University of Delaware Doctoral Fellowship, and the University of Delaware Research Foundation.
Author
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
