Cross-linguistic Transfer in Reading in Multilingual Contexts [1 ed.] 9789027266514, 9789027242778

Citation preview

.) s y d k s e t ( e z Zar chwart a n E le M ila S and

B E N J A M I N S C U R R E N T TO P I C S

tic in g guis s-lin Readin x ts n i C ro s r e o nte f C s Tran lingual i Mult



Cross-linguistic Transfer in Reading in Multilingual Contexts

Benjamins Current Topics issn 1874-0081 Special issues of established journals tend to circulate within the orbit of the subscribers of those journals. For the Benjamins Current Topics series a number of special issues of various journals have been selected containing salient topics of research with the aim of finding new audiences for topically interesting material, bringing such material to a wider readership in book format. For an overview of all books published in this series, please see http://benjamins.com/catalog/bct

Volume 89 Cross-linguistic Transfer in Reading in Multilingual Contexts Edited by Elena Zaretsky and Mila Schwartz These materials were previously published in Written Language & Literacy 17:1 (2014).

Cross-linguistic Transfer in Reading in Multilingual Contexts Edited by

Elena Zaretsky Clark University

Mila Schwartz Oranim Academic College of Education, Israel

John Benjamins Publishing Company Amsterdam / Philadelphia

8

TM

The paper used in this publication meets the minimum requirements of the American National Standard for Information Sciences – Permanence of Paper for Printed Library Materials, ansi z39.48-1984.

doi 10.1075/bct.89 Cataloging-in-Publication Data available from Library of Congress: lccn 2016025271 (print) / 2016039492 (e-book) isbn 978 90 272 4277 8 (Hb) isbn 978 90 272 6651 4 (e-book)

© 2016 – John Benjamins B.V. No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publisher. John Benjamins Publishing Company · https://benjamins.com

Table of contents Foreword: Cross-linguistic transfer in reading in multilingual context – resent research trends Elena Zaretsky & Mila Schwartz Introduction: The cross-language transfer journey – a guide to the perplexed Esther Geva The development of orthographic processing skills in children in early French immersion programs Eva Commissaire, Adrian Pasquarella, Xi Chen & S. Hélène Deacon How do previously acquired languages affect acquisition of English as a foreign language: The case of Circassian Janina Kahn-Horwitz, Sara Kuash, Raphiq Ibrahim & Mila Schwartz Bidirectional cross-linguistic relations of first and second language skills in reading comprehension of Spanish-speaking English learners Alexandra Gottardo, Christine Javier, Fataneh Farnia, Lorinda Mak & Esther Geva Concurrent and longitudinal cross-linguistic transfer of phonological awareness and morphological awareness in Chinese-English bilingual children Yang Cathy Luo, Xi Chen & Esther Geva

vii

1

17

43

65

93

The effects of bilingual education on the English language and literacy outcomes of Chinese-speaking children Kathleen Hipfner-Boucher, Katie Lam & Xi Chen

121

The role of L1 and L2 reading on L1 preservation and positive cross-linguistic transfer among sequential bilinguals Elena Zaretsky

145

Contributors

171

Index

173

Foreword Cross-linguistic transfer in reading in multilingual context – resent research trends Elena Zaretsky & Mila Schwartz Clark University, Worcester, USA / Oranim Academic College of Education, Israel

A group researchers from different countries and speaking different languages participated in a symposium “Language and literacy development in bilingual context: evidence for cross-linguistic transfer”, to share their research and thoughts on this timely subject. The symposium was part of the International Association for the Study of Child Language (IASCL) conference, which took place in ­Montreal, in July 2011. Participants raised a variety of theoretically fascinating questions in an attempt to understand the mechanisms behind cross-linguistic transfer in reading which resulted in extended debates during conference. Special Issue of ­Written Language & Literacy (January 2014) devoted to this topic was our attempt to continue that debate and provide novel data in support of the phenomenon of cross-linguistic transfer. With this book we are trying to reach an even wider audience of researchers and educators who work with children from different linguistic backgrounds in multilingual context. Cook (1991) coined the term “multi-competence” which relates to cross-­ linguistic transfer and is defined as “knowledge of two or more languages in one mind” (Cook 2003: 2). Applying this notion to literacy domain in more than one, two or three languages one expects that literacy in multiple languages must interact at some level, rather than form completely isolated systems, since they are assumed to exist “in one mind”. With the increasing number of bilingual and, in recent years, multilingual children, the notion of multi-competence raises questions about the relationship between different aspects of literacy and different contexts. Researchers are actively engaged in examining the nature of cross-linguistic transfer in language and ­literacy skills taking into consideration different factors that influence language and literacy development. For example, how does the degree of typological proximity affect cross-linguistic transfer? What is the actual mechanism of transfer? How does this mechanism work when two or more languages under the scope of investigation have different degrees of linguistic and orthographic p ­ roximity?

doi 10.1075/bct.89.001for © 2016 John Benjamins Publishing Company

 Cross-linguistic transfer in reading in multilingual contexts

How do the common underlying cognitive and linguistic abilities affect the ­cross-linguistic transfer in the literacy domain? What is the role of cross-linguistic transfer in literacy outcomes for bilingual minority language children within the education context of the majority language? This volume makes an attempt to answer those questions by looking at crosslinguistic transfer in school-age children, children who were part of the immigrant communities and were actively engaged in learning the majority language(s) (L2 or L3) while still maintaining their L1, as well as English-speaking children who were acquiring L2 (French) in the immersion context. Focusing on specific skills required for literacy acquisition, e.g. phonological awareness, depth and breadth of vocabulary, good initial oral language skills, and L1 proficiency, allowed a close examination of skills that are truly cross-linguistic and those that are not. As the reader will discover in the collection of scientific articles presented in this volume, phonological awareness may be a cross-linguistic skill (cf. Luo, Chen & Geva, Hipfner-Boucher, Lam & Chen), while the acquisition of orthographic conventions may depend on the shared characteristics between one or more languages (cf. Commissaire, Pasquarella, Chen & Deacon). In a similar vein, KahnHorwitz, Kuash, Ibrahim & Schwartz show that within a quadrilinguistic context, shared characteristics between L1 and L4 and L2 and L4 promote the acquisition of basic literacy skills in L4. The works of Gottardo, Javier, Mak & Farnia and of Zaretsky show the importance of bidirectional influences when children are exposed to multilingual contexts. Gottardo et al. provide evidence of overlapping L1 and L2 vocabulary skills that facilitate cross-linguistic transfer and the importance of considering contextual factors in this process. Zaretsky relates the notion of cross-linguistic transfer to the solid basis in L1 knowledge in oral and written language modalities. Our focus on school-age children is not accidental: Within the classroom ­situation, the phenomenon of cross-linguistic transfer might be one of the expressions of “funds of knowledge” notion (Moll, Amanti, Neff & Gonzalez 1992). In the context of this book, this notion refers to child’s languages and ­literacy background and learning patterns as a potential for enhancing future success in school. It is our hope that this work will result in continuing awareness of the importance of the classroom practices that facilitate cross-linguistic transfer. As our findings indicate, commonalities among languages, e.g. typological proximity, may promote the acquisition of orthographic and linguistic conventions. Mainstream teachers’ sensitivity to, and a degree of background knowledge of, ambient languages spoken in the multi-lingual classrooms will heighten their ability to increase children’s awareness of the specific shared orthographic and linguistic features between their L1 and L2 (L3) and therefore promote academic achievements.

Foreword 

In conclusion, we hope that the contents of this volume will promote the continuation of research in the field of cross-linguistic transfer in reading in ­multilingual contexts and address challenging theoretical and practical questions associated with this complex subject.

References Cook, Vivian (1991). Second language learning and language teaching. London: Arnold. Cook, Vivian (2003). The changing L1 in the L2 user’s mind. In Vivian Cook (ed.), Effects of the second language on the first, 1–18. Bristol: Multilingual Matters. Moll, Louis, Cathy Amanti, Deborah Neff & Norma Gonzalez (1992). Funds of knowledge for teaching: A qualitative approach to developing strategic connections between homes and classrooms. Theory into Practice 31: 132–141.  doi: 10.1080/00405849209543534

Introduction The cross-language transfer journey – a guide to the perplexed Esther Geva University of Toronto, Canada

1.  Introduction Cross-language transfer has been a central and longstanding theme in the study of language and literacy development in L2 learners. Koda’s (2005) definition of cross- language transfer provides a useful starting point for exploring the nuances of the transfer concept. Koda (2005: 9) explains that “the central assumption underlying the cross-linguistic approach is that L1 experience embeds habits of mind, instilling specific processing mechanisms, which frequently kick in during L2 reading”. While in its earlier versions the focus has been primarily on cross-language transfer in spoken language, with the growing interest in understanding reading processes and reading development in various groups of second language (L2) learners,1 one can see now an exponential increase in the number of studies that examine cross-language transfer in relation to literacy aspects such as word reading and spelling, reading comprehension, and writing. Definitions of L1-L2 transfer vary in the extent to which they emphasize innate vs. acquired, cognitive, and developmental aspects and in the extent to which they attend to facilitating or debilitating outcomes of similarities and d ­ ifferences

.  The scope of this Introduction does not allow me to review in detail how the concept of cross language transfer has been studied in various contexts. I will therefore not attempt to offer a systematic analysis of cross-language transfer in distinct groups of learners, with ­different amounts of exposure to the L1 and the L2, varying in age of acquisition (or attrition), type of programs (e.g. ELL, ESL, EFL, bilingual, heritage programs), and so on. I acknowledge the importance of these factors as do the authors of the research papers included in this volume. Ultimately, the question one needs to ask is how has our understanding of crosslanguage transfer advanced and where we need more research. To avoid clumsiness and given the scope of my task, I will use ‘L2’ as a generic term.

doi 10.1075/bct.89.01gev © 2016 John Benjamins Publishing Company



Esther Geva

between elements in the L1 and the L2. Essentially, two cross-language transfer frameworks (with variants) dominate the field – the contrastive (Lado 1957) or typological framework and the linguistic interdependence framework (Cummins 1981). Most but definitely not all studies on transfer can be situated in either one of these frameworks, though some take a complementary perspective that considers each as relevant to some aspects of L2 learning. In what follows I will describe briefly the core features of each of the frameworks and some variations that emerged later on. I will then provide an overview of the studies included in this volume and the issues they raise with regard to cross-language transfer related to literacy development. 2.  Frameworks in studying cross-language transfer 2.1  L1-L2 transfer – a typological/contrastive perspective The concept of transfer is central for learning theories and it is important to think of the typological or contrastive framework as a close ‘relative’ of behaviorism and learning theories. The concept of transfer has been pivotal to education in general and to L2 and literacy learning specifically, because educators and researchers strive to find out the conditions that allow learning that takes place in one context to permeate and enhance learning in other very similar contexts and performances (through near transfer) or to less similar contexts and performances (through far transfer) (Perkins & Salomon 1992). Thorndike (1923) examined the notion that studying Latin would prepare learners to do better in another subject matter. However, Thorndike (1923) failed to demonstrate transfer and concluded that transfer depends on the existence of ‘identical elements’ in two tasks, so that when the tasks do not share identical elements transfer cannot occur. These early notions of transfer of learning gained momentum in the domain of L2 language learning, and in particular with regard to grammar, with Lado’s Contrastive Analysis hypothesis. Lado’s (1957: 2) notion of transfer was that “Individuals tend to transfer the forms and meanings, and the distribution of forms and meanings of their native language and culture to the foreign language and culture – both productively when attempting to speak the language and to act in the culture, and receptively when attempting to grasp and understand the language and the culture as practiced by natives.”

Lado (1957) maintained that L2 learners rely heavily on their native language when they learn the target language and that cross-linguistic differences in L2 acquisition can be determined on the basis of a systematic analysis and c­ omparison of

Introduction

specific features of each of the languages under study. He believed that on the basis of such a comparison, it should be possible to predict what elements of a given L2 will be challenging to learners coming from specific L1 backgrounds and which ones will not (König & Gast 2008). It is important to remember that transfer can be ‘positive’ or ‘negative’. When specific features in two languages are similar to each other, one can expect to ­demonstrate ‘positive transfer’ involving these structures. Examples of positive transfer include shared cognates between Hebrew and Arabic and graphemephoneme recognition in French by children familiar with the English alphabet. However, when structures are very different from each other, studies strive to demonstrate ‘negative transfer’ or ‘interference’ because learners are likely to rely on what is familiar from their L1 and this will yield predictable errors. Examples of negative transfer include underuse (followed sometimes later on by overuse) of definite articles in English by Chinese learners (Robertson 2000) and the difficulties that Hebrew speakers have with the use of modals in English, because modals are not used in Hebrew. A significant contribution of the contrastive framework is that it highlights the importance of systematic comparisons of linguistic structures in L1 and L2 and considers how specific features of the L1 might affect L2 learning. In general, research conducted within the contrastive framework did not support the specific prediction of errors that L2 learners were expected to make due to negative transfer from the L1 to the L2 (Corder 1967). Gradually, researchers came to realize that learning difficulties of L2 learners do not always arise from predicted crosslinguistics differences and that one needs to attend to psychological-cognitive mechanisms that underlie L2 learning and that can explain errors committed by L2 learners as well as to sociocultural and instructional contexts that play a role in L2 learning (Durgunoğlu & Verhoeven 1998; Jean & Geva 2012). In the 1980s, the transfer framework was modified or augmented with progress in cognitive and developmental psychology, raising attention to the role of contextual and developmental factors in explaining facilitation observed in L2 learners and drawing attention to individual differences (Cummins 1981). This framework evolved into subsequent conceptualizing of L2 theories ­juxtaposing language universal with script dependent hypotheses (Bialystok 2007; Geva & Siegel 2000; Perfetti 1997; Share 2008)2 and consideration of L1 and L2 structural features that may help to pinpoint what is actually transferred and what

.  The script dependent hypothesis predicts that literacy acquisition varies across orthographies as a function of different constraints imposed by different mappings between oral languages and writing systems and differences in ‘orthographic depth’ (Geva & Siegel 2000).





Esther Geva

mechanisms are entailed in the process of L2 language and literacy development. A number of papers in this special issue can be traced to the contrastive or typological framework. 2.2  L1-L2 transfer: The linguistic interdependence hypothesis Another prominent theoretical framework that concerns L1-L2 transfer is the interdependence hypothesis formulated by Cummins (1981). Like the contrastive framework, the interdependence framework has been highly influential in the literature on L2 literacy development and some of the papers in this volume are clearly situated in this framework. In fact, already in the 1950s and 1960s, educators were interested in the relationships between L1 and L2 and developed L1-based measures such as the Modern Language Aptitude Test (MLAT; Carroll & Sapon 1959/2000) that includes tests of vocabulary, sound-symbol relationships, grammatical concepts, and verbal memory that would predict L2 learning aptitude (for a review, see Ganschow & Sparks 2001; Spolsky 1989). The interdependence framework has been influenced by cognitive psychology. It postulates that “To the extent that instruction in Lx is effective in promoting proficiency in Lx, transfer of this proficiency to Ly will occur provided there is adequate exposure to Ly (either in school or environment) and adequate motivation to learn Ly.” (Cummins 1981: 29). Stated differently, learning that has taken place in one’s L1 can ‘transfer’ or facilitate L2 achievement, but this transfer is not automatic. It depends on the extent to which the learner has achieved sufficient language proficiency in the L2 and the extent to which instruction in the L1 has been of high quality (Cummins 2012). Indeed, Cummins (2012) argues that there are advantages for maintaining sufficient L1 proficiency and quality of L2 instruction and that stronger progress in acquiring literacy in the L2 can be made (i.e. positive transfer) when students have had opportunities to develop literacy in their L1 (­ Cummins 2000) (for a more comprehensive review and critique, see Genesee, Geva, Dressler & Kamil 2006). Cummins points out that L2 learning tasks vary in terms of their cognitive demands and the extent to which contextual and language proficiency support can help the L2 learner and that learning situations vary in how facilitative existing L1 skills may be for L2 learning. Cummins (2000: 173) hypothesizes that “academic proficiency transfers across languages such that students who have developed literacy in their first language will tend to make stronger progress in acquiring literacy in their second language” because the academic language skills are developmentally linked to common underlying proficiencies across the languages. These common underlying proficiencies involve skills and knowledge that were learned in the L1 context but can be drawn upon when working in an

Introduction

L2. For example, ESL high school students who have developed grade appropriate conceptual knowledge in the L1 may be able to use this knowledge in L2-based cognitively demanding tasks such as comprehending an academic text in the L2 or solving a math word problem, once appropriate L2 language proficiency has been attained (and provided that the student had enough command of the L1 when these skills were taught). Notice that unlike the contrastive/typological framework that focuses on specific elements in the spoken or written language, Cummins’ framework underscores the importance of distinguishing cognitively and conceptually demanding knowledge from knowledge that is less demanding cognitively and the conditions under which the learners can display this knowledge in their L1 and L2. From a research design perspective, studies conducted within the interdependence framework often utilize correlational and regression-based approaches. Such studies typically seek to demonstrate a correlation between parallel constructs or skills in the L1 and L2 and the relevance of known L1 predictors (e.g. comprehension strategies) for L2 reading comprehension and the results are interpreted as evidence of cross-linguistic transfer from the L1 to the L2. However, it is important to be mindful of the fact that correlation-based approaches can establish an association between variables but not the precise causal nature underlying the relationships. Cummins’ basic idea that there are interdependent relationships among L1 and L2 language and literacy skills has been studied especially in the context of L1-L2 transfer of higher-level conceptual and strategic aspects (Genesee et al. 2006). For example, in a study of middle school children enrolled in a transitional bilingual education program, Royer and Carlo (1991) found that reading comprehension in Spanish correlated with reading comprehension in English. The linguistic interdependence hypothesis has left a strong mark on research and policies concerning L2 literacy development and has shaped the conceptualization of L2 learning. Yet, it has been criticized for lack of specificity concerning the definition of the construct of interdependence and in explaining what is actually being transferred (Genesee et al. 2006). On the basis of a systematic review of studies involving ELLs, Genesee et al. (2006) conclude that it is useful for explaining procedural knowledge that underlies language use for academic or cognitively demanding tasks such as the skills involved in defining the meaning of words or in the use of metacognitive, comprehension monitoring skills. But they conclude that the interdependence framework is not sufficient for identifying what specific skills or abilities are transferred. Genesee et al. (2006) point out that not all aspects of L1 development are equally facilitative for L2 development; facilitation from the L1 to the L2 may occur when there is shared contextual support for the transfer. However, tasks vary in their cognitive demands and this means that in some situations,





Esther Geva

performing well on an L1 task will be mirrored in the L2 but some task situations are not conducive for L1 skills to be facilitative for L2 learning (for a review and critique, see Genesee et al. 2006). 2.2.1  A variant: Common underlying cognitive processes Furthering the linguistic interdependence hypothesis, researchers have started to specify underlying cognitive abilities that support acquisition of both L1 and L2 to conceptually understand and empirically investigate cross-language transfer (e.g. Geva & Ryan 1993). In an attempt to explain and predict L1-L2 relationships, Geva and Ryan (1993) argued that correlations between parallel tasks in L1 and L2 may correlate not just because L1 skills ‘transfer’ to the L2 but because of common underlying cognitive processes that underlie performance in the L1 and L2. They proposed that underlying cognitive processing abilities account for individual differences in the rate and success of language and literacy skills development whether in the L1 or L2 so that correlations between L1 and L2 higher level skills (e.g. the activation of prior knowledge in the L1 and L2) may be better understood by considering individual differences in underlying cognitive processes such as working memory, phonological awareness, and rapid automatized naming that are thought to be part of one’s general, cognitive make-up. These underlying processing components are needed to process text in the L1 or L2 and explain observed L1-L2 correlations among higher order processes. To illustrate, rapid automatized naming (RAN) tasks that measure the speed with which students access the names of highly automatized printed symbols such as numbers or letters (Bowers & Wolf 1993) is expected to be related to reading fluency in L1 and L2 because it is an underlying cognitive process that should be relevant across languages or learning groups (e.g. Chiappe, Stringer, Siegel & Stanovich 2002; Everatt, Smythe, Adams & Ocampo 2000; Wade-Woolley & Siegel 1997; Zadeh, Farnia & Geva 2010). Likewise, across numerous studies L1 and L2 word reading skills often correlate (for a review see Lesaux & Geva 2006). Cross-linguistic correlations on word reading can be understood within the underlying cognitive processes framework by considering phonological awareness, short-term memory, speed of lexical access, and orthographic processes. More specifically, phonological awareness (i.e. the awareness that words consist of ‘sounds’ or that tones are important in Chinese and that these ‘sounds’ can be manipulated) is a metalinguistic skill that underlies word reading. This kind of awareness does not have to be learned separately for each language, and ample research has shown that once it has been acquired it can be used regardless of language L2 proficiency (for a review see Lesaux & Geva 2006).

Introduction

2.2.2  A variant: The transfer facilitation model More recently, Koda (2008) has proposed the Transfer Facilitation Model to explain cross-language transfer in the area of reading development. The model underscores the importance of metalinguistic awareness (e.g. phonemic awareness, morphological awareness) that develops in one language in enhancing the development of reading skills in other languages. According to Koda (2008: 78), transfer is “an  ­automatic activation of well-established first-language competencies, triggered by second-language input”. She too views transfer as a developing, constantly changing interplay between well-established L1 competencies such as metalinguistic awareness and ongoing exposure to L2 print. Moreover, she argues that L1 facilitates L2 acquisition because metalinguistic skills developed in L1 are accessible cross-modally, i.e. from L1 spoken language to L2 reading development. For example, phonological awareness assessed in the L1 can predict word reading skills in the L2, not only when the two languages are typologically similar (e.g. Comeau, Cormier, Grandmaison & Lacroix 1999 for English-French; Durgunoğlu, Nagy & Hancin-Bhatt 1993 for English-Spanish), but also when the two languages are typologically dissimilar (e.g. Geva & Wade-Woolley 2004 for English and Hebrew; Wang, Perfetti & Liu 2005 for English and Chinese). According to Koda, what is transferred is not merely a set of rules but form-function relationships that L2 users have acquired in their L1 and can utilize in their L2 (Koda 2008). 2.3  Other frameworks that complement L1-L2 transfer L2 learning is a complex and lengthy process and neither the contrastive nor the underlying competency versions of L1-L2 transfer on their own can capture the myriad of processes that take place and interact over time. Below I address briefly additional frameworks that help to understand processes that influence L2 learning and that guide this synthesis. 2.3.1  Interlanguage and target language influences Another complex and nuanced framework that takes the transfer framework a step forward is that of ‘interlanguage’ (Selinker 1972). Interlanguage refers to the L2 learner’s evolving system of rules regarding the L2. It develops from various processes that take place as individuals learn the L2. These include transfer from the L1 as well as contrastive interference from the L2 and an overgeneralization of newly encountered rules. Interlanguage is a rule-governed linguistic system that forms and guides the L2 learner and it differs from both the L1 and the L2. Interlanguage can be observed when one considers developmental errors and transfer errors (Paradis, Genesee & Crago 2011). Paradis et al. emphasize that certain errors in children’s L2 occur regardless of children’s L1 backgrounds and caution





Esther Geva

that while one may be inclined to interpret errors in the acquisition of L2 language as evidence of L1 transfer errors, many errors that L2 children commit in the process of learning an L2 can be better understood from a complex perspective that integrates processes that reflect transfer, development, and interlanguage. They propose that errors that reflect negative transfer from the L1 may be more noticeable in early stages of L2 acquisition (Zdorenko & Paradis 2008), whereas developmental error patterns may be more typical of the interlanguage of children whose L2 is better developed already. Clearly, when considering transfer in the context of L2 reading development, it is equally important to adopt a broader developmental perspective. For example, children may make errors on reading comprehension tasks not because of negative transfer but because they simply do not know the meaning of words that are crucial for understanding the text. 2.3.2  Challenging elements in the target language Some aspects of L2 language and literacy development can be accounted simply in relation to challenging features of the target language (Bailey, Madden & Krashen 1974; Dulay & Burt 1973) rather than reflecting negative transfer from the L1. Such features may be challenging to both L1 and L2 learners because they are more complex or deviate from a generalizable rule that works most of the time. That is, some errors may be developmental in nature. Such errors may not simply reflect negative transfer from the L1 to the L2. It is therefore important to find out how native speakers gain mastery on certain linguistic or orthographic elements in comparison with what is also more challenging to L2 learners coming from diverse L1 backgrounds. To illustrate, one study in our lab (Wang & Geva 2003) has shown that when they were in grade 1 and were presented with word pairs such as think and sink, Cantonese L1 ESL children thought that they were the same because they were unable to distinguish the /ɵ/ and /s/ phonemes. But by grade 2, with more exposure to English, this auditory discrimination was not a problem anymore. At the same time, both English as L1 and Cantonese L1 ESL learners took more time to learn how to spell correctly the /ɵ/ as in think in comparison with other simpler orthographic components such as the /s/ in sink. 2.4  Summary As this brief overview of trends in cross-linguistic research suggests, the concept of cross-language transfer in literacy development is complex. In this paper, I have examined the transfer concept from a historical angle, showing how current conceptualization of the affinity between L1 and l2 learning can be traced to the contrastive/typological framework and the interdependence framework. However, the concept of transfer has evolved and it is clear that factors other than transfer

Introduction

help to explain the enormous task that faces L2 learners as they gradually learn to communicate and gain literacy skills in an L2 (or L3, L4). Clearly, it is important to consider levels of proficiency on specific aspects of language (e.g. vocabulary, morphological awareness, syntax) in the L1 and L2 and to attend to the specific components of literacy that are being studied, but it is also helpful to consider the role played by cognition, development, as well as social, affective, contextual, and instructional factors. 3.  The current issue Among the papers in this issue, the ones by Commissaire, Pasquarella, Chen, and Deacon and by Kahn-Horwitz, Kuash, Ibrahim and Schwartz have an affinity to the contrastive, typological framework; they explore specific typologically different or similar linguistic or orthographic components in L1 and L2. The focus is on studying in depth the ease or difficulty that L2 learners coming from typologically different languages encounter when they acquire specific linguistic or orthographic elements in the new language. The idea is that L2 (or L3 or L4) learners will commit specific errors when reading or spelling in the target (or newer) language because their previous linguistic and metalinguistic experiences differ considerably from what they encountered in the previously learned languages (i.e. negative transfer). However, when the crucial elements in the target (or newer) language are similar to those of previous languages learned, fewer errors will be made because the learners can capitalize on their awareness of structures encountered previously in their L1, L2, L3 (i.e. positive transfer). In the context of a French immersion program, Commissaire et al. examined how children develop specific orthographic processing skills from grade 1 to grade 2. They investigated the development of an underlying factor structure of orthographic skills across English and French and the ability of children in grades 1 and 2 to extract regularities from exposure to two scripts (English and French). They also investigated whether orthographic processing skills might emerge as a single factor that generalises across two (related) writing system (i.e. English and French). Results indicate that in both grade levels, children are more accurate with language-shared orthographic units than with orthographic units that are not shared across English and French. More generally, the results suggest that what transfers across languages is not merely knowledge of specific individual features shared between languages, but more general orthographic competence that is partly independent of the orthographic features at stake. This conclusion that supports Koda’s (2008) framework is also echoed in two other papers in this special issue (Gottardo, Javier, Mak & Farnia; Luo, Chen & Geva).



 Esther Geva

The article by Kahn-Horwitz et al. explores the transfer issue in a context that is rarely studied – that of multilingual children. The study involved a comparison of shared and unshared linguistic and orthographic features in two groups of 5th grade learners – multilingual Circassian L1 children learning English, their L4 (with Arabic and Hebrew as the L2 and L3) and Hebrew L1 learning English (their L2). Kahn-Horwitz and her colleagues propose a variant of the contrastive framework which they refer to as the linguistic and orthographic proximity hypothesis. This hypothesis is concerned with transfer in relation to elements in the spoken and written language. It stipulates that specific phonemes or orthographic characteristics that exist in a child’s first or additional language(s) and orthography can facilitate or interfere with the acquisition of orthographic conventions in a new language and orthography, but that facilitation will not be seen on novel elements not previously encountered. Accordingly, transfer (positive or negative) depends on the cumulative effect of degree of proximity between linguistic and orthographic characteristics of the target language and the previously learned language(s). As hypothesized, Circassian L1 speaking children outperformed the Hebrew L1 children in decoding and spelling of specific target orthographic elements in English which are similar to conventions to which they were previously exposed. That is, in their performance in English, Circassian L1 multilingual children have an advantage in acquiring specific linguistic and orthographic characteristics over Hebrew L1 speakers. The facilitation in the multilingual group is not correlational in nature but is seen with regard to specific phonological and orthographic characteristics shared between elements of previously acquired languages and orthographies (i.e. Arabic and Circassian) and the new (target) language. Compared with Hebrew L1 children learning English as L2, these multilingual children can draw on a larger and more varied linguistic and cognitive repertoire of resources and experiences accumulated through prior learning experience with other languages when they acquire specific elements in their L4. Four of the six studies in this collection (Gottardo et al.; Luo et al.; HipfnerBoucher et al.; Zaretsky) have a stronger affinity to the underlying proficiency framework. The studies in this grouping examine how language skills such as phonemic awareness, morphological awareness, and vocabulary contribute crosslinguistically to various L1 and L2 literacy outcomes. An innovative transfer-related feature in the Luo et al. study and the G ­ ottardo et al. study is the conceptualization of transfer at the construct level and in terms of cross-linguistic prediction of reading. The study by Gottardo et al. examined bidirectional transfer of L1 and L2 variables that can predict L1 and L2 reading comprehension cross-linguistically in ELLs. The participants came from Spanish speaking homes and either received all their schooling in English in Canada (early sequential bilinguals) or were late sequential bilinguals who had some

Introduction

f­ ormal education in their L1 before the onset of formal schooling in ­English. The study demonstrates transfer first in that within each construct (i.e. vocabulary and reading), parallel L1 and L2 vocabulary skills overlap, as do parallel L1 and L2 word reading skills. The second level of transfer examined in this study as well as in the Luo et al. study involves cross-language prediction of reading comprehension in comparison to within-language prediction. The study demonstrates the complex set of variables that impinge on inter-language transfer, the importance of considering contextual factors such as L1 attrition on the one hand and extent of exposure to the L2 on the other hand, and most importantly the utility of conceptualizing transfer as a dynamic process. In a similar vein, the Luo, Chen and Geva team asked whether parallel phonological awareness and morphological awareness skills measured in Grade 1 in Chinese (the L1) and English (the L2) formed two constructs, a phonological awareness construct and a morphological awareness construct, and whether L1 and L2 phonological and morphological awareness tasks would contribute to the prediction of word reading skills concurrently (in Grade 1) and longitudinally (in grade 2) within and across languages. The authors have shown that parallel L1 and L2 tasks, designed to measure the same linguistic structures in the L1 and the L2 (e.g. compound morphology awareness, phonemic awareness), indeed yield distinct constructs (i.e. a phonological awareness construct and a morphological awareness construct), a finding that provides evidence for transfer in terms of shared variance between parallel L1 and L2 measures. In addition, transfer was demonstrated at the reading level in that phonological awareness or morphological awareness in one language predicted word reading in the other language. In line with the theoretical framework proposed by Durgunoğlu (2002) and Koda (2008) and the conclusions reached by Commissaire et al. (this volume), results suggest that primary level children’s metalinguistic ability to reflect on and manipulate phonological units in English and Chinese forms a single skill construct and it is this construct that is applied cross-linguistically to facilitate word reading in Chinese and English (though not the other way around). Unlike phonological awareness, morphological awareness appears to be a more language specific construct, however, suggesting that cross-linguistic transfer of compound awareness to word reading is constrained by various factors, including the demands of the reading task, language proficiency, and language typology. The finding that unlike phonological awareness, morphological awareness is a more language specific metalinguistic construct which is less likely to predict reading performance in another language is a conclusion that emerged also from the study by Hipfner-Boucher et al. This team compared the English language and literacy outcomes of Mandarin speaking grade 1 children living in Canada who attended one of three different school settings: English-only, paired bilingual



 Esther Geva

(Mandarin-English), and French immersion. The study examined cross-language patterns of transfer of phonological and morphological awareness to word reading and vocabulary in each of the three school settings. Transfer was demonstrated in that French derivational awareness accounted for a significant portion of the variance in English vocabulary, probably because derived words in French and English share a number of morphological features. Cross-language transfer from Chinese to English was evident in that Chinese phonological awareness predicted ­English word reading in both the Bilingual Paired and English-only groups. However, there was no evidence of transfer of Chinese morphological awareness to English vocabulary or word reading. Once again, in support of Koda (2008) and SaieghHaddad & Geva (2008), the results demonstrate that unlike phonemic awareness, transfer of morphological awareness hinges on the extent of shared morphological structures between the L1 and L2. Finally, Zaretsky’s study sits squarely in the interdependence framework. It takes a ‘bird’s eye’ view of transfer in the oral discourse of bilingual, Russian-­ English children, arguing that the common underlying proficiency draws on language and literacy skills in L1 to support L2 achievement and reduce L1 attrition. Hence, the study focused on exploring transfer at the linguistic level, in the oral language skills of elementary children varying in age, learning English (L2), whose L1 is Russian, coming from relatively high SES, living in the US. Their performance was also compared to that of a similar monolingual group. The study examined in depth the use of morphosyntactic structures and lexical constituents in story retellings in L1 and L2 by 13 Russian-English bilingual children whose is L1 is Russian. Children’s narratives revealed the appropriate use of the same lexical and syntactic processes when retelling a story in English or Russian, suggesting a solid command of vocabulary and syntactic structures and lexicon in both. Moreover, L1 reading skills correlated with syntactic structures and L1 and L2 reading correlated with vocabulary knowledge. The results support the notion that overall proficiency in the L1 plays a large role in learners’ overall proficiency in the L2 (e.g. Cummins 1979; Sparks, Patton, Ganschow & Humbach 2009). 4.  Conclusions Reading through the articles in this issue, it is probably not that difficult to trace the affinity between current research and the two ‘old’ frameworks that have propelled research on L2 reading development for so many years. Importantly, the papers in this special issue represent a new generation of conceptualizations of cross-linguistic transfer in literacy. The studies are now more ­sophisticated

Introduction 

­ ethodologically, they encompass a broader spectrum of languages, and offer m another level of complexity when transfer is studied with participants who are developing language and literacy skills in their third or even fourth language (as two papers in this collection do). Much more attention is being paid to methodological considerations, to measurement issues, to contextual factors such as age of acquisition (Flege, Munro & MacKay 1995), intensity of exposure to language and reading in the L1 and L2, home literacy, and to quality of input. Importantly, one notes a shift from reporting on general correlational relationships to a careful examination of transfer of specific constructs under tightly controlled conditions, seeking to explain causality and attending to dynamic developmental changes over time.

References Bailey, Natalie, Carolyn Madden & Stephen D. Krashen (1974). Is there a ‘natural sequence’ in adult second language learning? Language Learning 24: 235–243.

doi: 10.1111/j.1467-1770.1974.tb00505.x

Bialystok, Ellen (2007). Acquisition of literacy in bilingual children: A framework for research. Language Learning 57: 45–77.  doi: 10.1111/j.1467-9922.2007.00412.x Bowers, Patricia G. & Maryanne Wolf (1993). Theoretical links among naming speed, precise timing mechanisms and orthographic skill in dyslexia. Reading and Writing 5: 69–85.

doi: 10.1007/BF01026919

Carroll, John & Stanley Sapon (2000). Modern Language Aptitude Test (MLAT): Manual. ­Rockville, MD: Second Language Testing Inc. (original work published 1959). Chiappe, Penny, Ron Stringer, Linda S. Siegel & Keith E. Stanovich (2002). Why the timing ­deficit hypothesis does not explain reading disability in adults. Reading and Writing 15: 73–107.  doi: 10.1023/A:1013868304361 Comeau, Liane, Pierre Cormier, Éric Grandmaison & Diane Lacroix (1999). A longitudinal study of phonological processing skills in children learning to read in a second language. Journal of Educational Psychology 91: 29–43.  doi: 10.1037/0022-0663.91.1.29 Corder, Stephen P. (1967). The significance of learners’ errors. International Review of Applied Linguistics in Language Teaching 5: 161–170.  doi: 10.1515/iral.1967.5.1-4.161 Cummins, Jim (1979). Linguistic interdependence and the educational development of bilingual children. Review of Educational Research 49: 222–251.  doi: 10.3102/00346543049002222 Cummins, Jim (1981). The role of primary language development in promoting educational success for language minority students. In California State Department of Education (ed.), Schooling and language minority students: A theoretical framework, 3–49. Sacramento, CA: California Department of Education. Cummins, Jim (2000). Language, power and pedagogy: Bilingual children in the crossfire. ­Clevedon, England: Multilingual Matters. Cummins, Jim (2012). The inter’ of cognitive and sociocultural factors in the development of reading comprehension among immigrant students. Reading and Writing 25: 1973–1990.  doi: 10.1007/s11145-010-9290-7

 Esther Geva Dulay, Heidi C. & Marina K. Burt (1973). Should we teach children syntax? Language Learning 23(2): 245–258.  doi: 10.1111/j.1467-1770.1973.tb00659.x Durgunoğlu, Aydin Y. (2002). Cross-linguistic transfer in literacy development and implications for language learners. Annals of Dyslexia 52: 189–204.  doi: 10.1007/s11881-002-0012-y Durgunoğlu, Aydin Y., William E. Nagy & Barbara J. Hancin-Bhatt (1993). Cross-language transfer of phonological awareness. Journal of Educational Psychology 85: 453–465.

doi: 10.1037/0022-0663.85.3.453

Durgunoğlu, Aydin Y. & Ludo Verhoeven (1998). Epilogue: Multilingualism and literacy development across different cultures. In Aydin Y. Durgunoğlu & Ludo Verhoeven (eds.), ­Literacy development in a multilingual context: Cross-cultural perspectives, 289–298. M ­ ahwah, NJ: Lawrence Erlbaum. Everatt John, Ian Smythe, Ewan Adams & Dina Ocampo (2000). Dyslexia screening measures and bilingualism. Dyslexia 6: 42–56.

doi: 10.1002/(SICI)1099-0909(200001/03)6:1%3C42::AID-DYS157%3E3.0.CO;2-0

Flege, James E., Murray J. Munro & Ian R.A. MacKay (1995). Effects of age of second language learning on the production of English consonants. Speech Communication 16: 1–26.

doi: 10.1016/0167-6393(94)00044-B

Ganschow, Leonore & Richard Sparks (2001). Learning difficulties and foreign language learning: A review of research and instruction. Language Teaching 34: 79–98.

doi: 10.1017/S0261444800015895

Genesee, Fred, Esther Geva, Cheryl Dressler & Michael L. Kamil (2006). Synthesis: Cross-­ linguistic relationships in working memory, phonological processes, and oral language. In Diane August & Timothy Shanahan (eds.), Developing literacy in second-language learners: A report of the National Literacy Panel on language-minority children and youth, 153–174. Mahwah, NJ: Lawrence Erlbaum. Geva, Esther & Ellen B. Ryan (1993). Linguistic and cognitive correlates of academic skills in first and second language. Language Learning 43: 5–42.  doi: 10.1111/j.1467-1770.1993.tb00171.x Geva, Esther & Linda S. Siegel (2000). Orthographic and cognitive factors in the concurrent development of basic reading skills in two languages. Reading and Writing 12(1/2): 1–30.  doi: 10.1023/A:1008017710115

Geva, Esther & Lesly Wade-Woolley (2004). Issues in the assessment of reading disability in second language children. In Ian Smythe, John Everatt & Robin Salter (eds.), International book of dyslexia: A cross language comparison and practice guide, 195–206. Chichester UK: John Wiley. Jean, Maureen & Esther Geva (2012). Through the eyes and from the mouths of young heritagelanguage learners: How children feel and think about their two languages. TESL Canada Journal 29(special issue 6): 49–80. Koda, Keiko (2005). Insights into second language reading: A cross-linguistic approach. New York: Cambridge University Press.  doi: 10.1017/CBO9781139524841 Koda, Keiko (2008). Impacts of prior literacy experience on second-language learning to read. In Keiko Koda & Annette M. Zehler (eds.), Learning to read across languages: Cross-­ linguistic relationships in first- and second-language literacy development, 68–96. Mahwah, NJ: Routledge. König, Ekkehard & Volker Gast (2008). Understanding English-German contrasts (2nd edition revised). Berlin: Erich Schmidt. Lado, Robert (1957). Linguistics across cultures: Applied linguistics for language teachers. Ann Arbor: University of Michigan Press.

Introduction  Lesaux, Nonie & Esther Geva (2006). Synthesis: Development of literacy in ­language-minority students. In Diane August & Timothy Shanahan (eds.), Developing literacy in second-­ language learners: A report of the National Literacy Panel on language-minority children and youth, 53–74. Mahwah, NJ: Lawrence Erlbaum. Paradis, Johanne, Fred Genesee & Martha Crago (2011). Dual language development and disorders: A handbook on bilingualism and second language learning (2nd edition). Baltimore: Paul H. Brooks. Perkins, David N. & Gavriel Salomon (1992). Transfer of learning. In Torsten Husen & T. Neville Postlethwaite (eds.), International encyclopedia of education (2nd edition, Vol. 11), 6452– 6457. Oxford, UK: Pergamon Press. Perfetti, Charles A. (1997). The psycholinguistics of spelling and reading. In Charles A. Perfetti, Laurence Rieben & Michel Fayol (eds.), Learning to spell: Research, theory and practice across languages, 21–38. Mahwah, NJ: Lawrence Erlbaum. Robertson, Daniel (2000). Variability in the use of the English article system by Chinese learners of English. Second Language Research 16: 135–172.  doi: 10.1191/026765800672262975 Royer, James M. & Maria S. Carlo (1991). Transfer of comprehension skills from native to ­second language. Journal of Reading 3: 450–455. Saiegh-Haddad, Elinor & Esther Geva (2008). Morphological awareness, phonological awareness, and reading in English-Arabic bilingual children. Reading and Writing 21(5): 481–504.  doi: 10.1007/s11145-007-9074-x Share, David L. (2008). On the Anglocentricities of current reading research and practice: The perils of over-reliance on an ‘outlier’ orthography. Psychological Bulletin 134: 584–616.

doi: 10.1037/0033-2909.134.4.584

Selinker, Larry (1972). Interlanguage. International Review of Applied Linguistics 10(3): 209–231.  doi: 10.1515/iral.1972.10.1-4.209 Sparks, Richrad, Jon Patton, Leonore Ganschow & Nancy Humbach (2009). Long-term crosslinguistic transfer of skills from L1 to L2. Language Learning 59: 203–243.

doi: 10.1111/j.1467-9922.2009.00504.x

Spolsky, Bernard (1989). Conditions for second language learning. Oxford: Oxford University Press. Thorndike, Edward L. (1923). The influence of first year Latin upon the ability to read English. School Sociology 17: 165–168. Wade-Woolley, Lesly & Linda S. Siegel (1997). The spelling performance of ESL and native speakers of English as a function of reading skills. Reading and Writing 9: 387–406.

doi: 10.1023/A:1007994928645

Wang, Min & Esther Geva (2003). Spelling acquisition of novel English phonemes in Chinese children. Reading and Writing 16: 325–348.  doi: 10.1023/A:1023661927929 Wang, Min, Charles A. Perfetti & Ying Liu (2005). Chinese-English biliteracy acquisition: Crosslanguage and writing system transfer. Cognition 97: 67–88.

doi: 10.1016/j.cognition.2004.10.001

Zadeh, Zoreh, Fataneh Farnia & Esther Geva (2010). Toward modeling reading comprehension and reading fluency in English language learners. Reading and Writing 25: 163–187.

doi: 10.1007/s11145-010-9252-0

Zdorenko,Tatiana & Johanne Paradis (2008).The acquisition of articles in child second language English: Fluctuation, transfer, or both? Second Language Research 24(2): 227–250.

doi: 10.1177/0267658307086302

The development of orthographic processing skills in children in early French immersion programs Eva Commissaire1, Adrian Pasquarella2, Xi Chen3 & S. Hélène Deacon4 1Université

3University

de Strasbourg, France / 2University of Delaware, USA / of Toronto, Canada / 4Dalhousie University, Halifax, Canada

Children learning to read in two languages are faced with orthographic features from both languages, either unique to a language or similar across languages. In the present study, we examined how children develop orthographic processing skills over time (from grade 1 to grade 2) with a sample of Canadian children attending a French immersion program and we investigated the underlying factor structure of orthographic skills across English and French. Two orthographic processing tasks were administered in both languages: lexical orthographic processing (e.g. choose the correct spelling from people–peeple) and sub-lexical orthographic processing (e.g. which is the more word-like vaid–vayd?), which included both language-specific and language-shared orthographic regularities. Children’s performances in sub-lexical tasks increased with grade but were comparable across languages. Further, evidence for a one factor model including all measures suggested that there is a common underlying orthographic processing skill that cuts across measurement and language variables. Keywords:  orthographic processing; reading; French immersion; bilinguals; second language learners

1.  Introduction Worldwide estimates of multilingualism seem to suggest that there are more multilinguals than monolinguals in the world (Tucker 1999). Among the variety of contexts that can lead an individual to acquire a second language, the case of early immersion educational programs is an interesting situation for understanding reading acquisition in two or more writing systems. In early immersion programs, a second language is learned in both written and oral forms while children are

doi 10.1075/bct.89.02com © 2016 John Benjamins Publishing Company

 Eva Commissaire, Adrian Pasquarella, Xi Chen & S. Hélène Deacon

implicitly exposed to their dominant language in the society at large. The present study focuses on the case of Canadian children in a French immersion program who receive French instruction at school and are exposed to English in their daily life. The goal of the present study was to examine how orthographic processing skills, or “the ability to form, store and access orthographic representations” ­(Stanovich & West 1989: 414), in English and French change over time. Further, we evaluate whether these skills are developed specifically for a single writing system or in a manner that is shared across writing systems. A longitudinal design in which we followed the children from grade 1 to grade 2 enabled us to evaluate how these skills and their interrelationships might change over time. The context of biliteracy learning raises the issue of the nature of the underlying processes of reading acquisition. Children in French immersion programs are exposed to multiple orthographic regularities; some of these are common to the two writing systems, while others are specific to one system. Note that here and throughout this manuscript, we use the term writing system to refer to the way in which a specific language is represented on the page.1 For instance, word endings such as 〈ame〉 are shared across French and English in that they occur in both writing systems. In contrast, word endings such as 〈ough〉 occur in English and not French, and 〈ille〉 occurs at the ends of words in French and not English. Accordingly, these orthographic patterns may be considered to be language-specific since they occur in each of these writing systems only. Children’s sensitivity to languageshared and language-specific orthographic regularities may help them to establish orthographic representations. These might, in turn, support the development of fast, precise, and automatic mechanisms of visual word recognition (Ehri 1995; Perfetti 1992). 1.1  Orthographic processing skills Orthographic processing skills, as referring to the ability to remember word spellings and regularities in letter sequences that is involved in the acquisition of ­successful reading2 (Cunningham & Stanovich 1990; Juel, Griffith & Gough 1986),

.  Note that in our previous work (Deacon, Commissaire, Chen & Pasquarella 2013), we used the term ‘writing system’ to refer to the way in which a group of languages that share the same representational system is represented on the page (e.g. English and French are within the same writing system as they are both represented with Roman characters). Here, following a distinction more common in the study of written language, we use the term ‘writing system’ to refer to the specific manner in which a single language is represented on the page. .  Note that the term ‘orthographic processing’ (also called ‘orthographic coding’) may also refer to the on-line process involved during reading isolated words which involves the fast and



The development of orthographic processing skills in children 

has been more and more considered as a multifaceted component that involves orthographic knowledge at various grain sizes (Castles & Nation 2006). Following this view, prior psycholinguistic research has considered two dimensions of orthographic processing skills: lexical and sub-lexical (Deacon, Benere & Castles 2012). Lexical orthographic skills (i.e. at the word level) are commonly assessed via the orthographic choice task which taps into word-level orthographic representations. In lexical tasks, participants are asked to choose the correct spelling between a word and its pseudo-homophone (e.g. rain – rane). The use of homophonic pseudo-word foils is to control for involvement of phonological processing skills so that responses specifically tap what the participant knows about word spellings. Sub-lexical orthographic skills (i.e. at the sub-word level), which have received growing attention, may be assessed by a word-likeliness judgment task with pseudo-word items. Participants are asked to judge which of two pseudowords is more word-like in a given writing system (e.g. daik – dayk where only 〈aik〉 may occur as a word ending in English). Ideally, the two pseudo-words used in the sub-lexical task are homophonic so that the task measures sensitivity to orthographic patterns rather than phonological plausibility or ease of decoding of the pseudo-words. Using sub-lexical orthographic tasks, researchers have shown that both E ­ nglish and French monolingual children are sensitive to the position and identity of doublet consonants as early as in grade 1 (e.g. Cassar & Treiman 1997; Pacton, Perruchet, Fayol & Cleeremans 2001). For instance, Cassar and Treiman (1997) showed that first graders considered nonwords with allowable double consonants (e.g. yill) more word-like than those with non-allowable doublets (e.g. yihh). Sensitivity to the letter-context of a given orthographic pattern has also been reported in both English and French (Pacton, Fayol & Perruchet 2005; Hayes, Treiman & Kessler 2006; Treiman & Kessler 2006; see Deacon, Conrad & Pacton 2008 for a review). Orthographic processing skills, as a whole construct, have been reported to predict significant variance in reading outcomes in monolinguals, after controlling for cognitive abilities, phonological awareness, or print exposure (Barker, Torgesen & Wagner 1992; Cunningham, Perry & Stanovich 2001). Nevertheless, it is not clear whether the lexical and sub-lexical dimensions are connected to reading in a similar way. Some researchers argue that lexical orthographic skills may be too closely related to reading outcomes and, in particular, to isolated word

automatic mapping of the written input to representations of letters, bigrams, and words. The extent to which this on-line reading process is related to the metalinguistic skills or knowledge that are assessed with off-line tasks remains unfortunately little addressed in the field (see Burt 2006).

 Eva Commissaire, Adrian Pasquarella, Xi Chen & S. Hélène Deacon

r­ eading, leading to a circular relationship between the two skills (Burt 2006; Castles & Nation 2006; see also Deacon, Benere & Castles 2012). By avoiding this circular relationship, sub-lexical orthographic skills have been argued to be a more valid measure of orthographic processing (Castles & Nation 2006). There is increasing evidence that orthographic processing at the sub-lexical level is associated with reading development. And yet the issue of whether lexical and sub-lexical subcomponents of orthographic skills tap the same underlying construct is not clear (see Commissaire, Duncan & Casalis 2011). Cunningham, Perry and Stanovich (2001) examined several measures of orthographic processing of lexical and sub-lexical orthographic processes in children from grade 1 to 4. With a principal component analysis, they showed that all tasks were highly loaded onto one principal component which explained 60% of the variance. This finding suggests that orthographic processing skills could be considered as a unified theoretical construct, at least in monolingual children. This finding inspired us to conduct similar analyses in the present study with the performance of children learning to read two languages. 1.2  Orthographic processing skills across languages Deacon, Wade Woolley and Kirby (2009) recently uncovered cross-language transfer of orthographic processing skills to reading in bilingual children learning to read in English and French, two languages represented with the same Roman alphabet. Lexical orthographic processing skills in one language predicted significant variance in reading outcomes in the other language, after multiple controls, in a bidirectional manner (see also Deacon, Commissaire, Chen & Pasquarella 2013; Deacon, Chen, Luo & Ramirez 2011). These findings differ greatly from those in prior research; several prior studies of bilinguals learning to read two languages in different scripts, such as Chinese and English, have found no correlation between L1 and L2 orthographic processing skills or transfer to reading across languages (Abu Rabia 2001; Arab-Moghaddam & Sénéchal 2001; Gottardo, Yan, Siegel & Wade-Woolley 2001; Wang, Park & Lee 2006). Taken as a whole, the evidence to date suggests that orthographic skills operate at a writing-system general level, specifically in that they transfer to reading when the languages are within the same script. As assumed by the linguistic interdependence hypothesis (Cummins 1979), this implies that orthographic processing skills might be considered as general metalinguistic processes that are available when learning to read an L2 of a same alphabet (e.g. Roman). One step forward has been taken by more recent research which has turned to investigating whether sub-lexical dimensions of orthographic processing skills, that is the ability to extract regularities from a written language, are also language-­ general for children learning to read languages within the same script. In a study



The development of orthographic processing skills in children 

with Canadian children attending a French immersion program in grade  1, Deacon, Commissaire et al. (2013) manipulated the ­language-specificity of the orthographic features examined in a sub-lexical orthographic task (i.e. say which pseudo-word is more word-like). In the language-shared condition, the target orthographic patterns were legal in both French and English (e.g. plame – plahme where the target pattern 〈ame〉 occurs in both languages); in the language-­specific condition, the target orthographic patterns were legal only in the language of the task (e.g. rouve – rouvve in the French task where the target rime 〈ouve〉 occurs in French and not English; yead – yeadh in the English task). Children were more accurate with language-shared than language-specific items suggesting that there is benefit of greater exposure. Notably, scores were similar in the French and English versions of the task, suggesting that both explicit and implicit exposure might support the development of orthographic sensitivity. With regard to crosslanguage transfer, language-shared English sub-lexical orthographic processing skills predicted significant variance in French reading outcomes. This finding adds to the growing body of evidence that orthographic processing skills, both at lexical and sub-lexical levels, are related to reading outcomes in children learning to read in languages represented with the same alphabet (Deacon et al. 2009; Deacon, Commissaire et al. 2013). And yet, cross-language transfer was not clear-cut when considering the language-specific component of orthographic processing skills, possibly due to the multiple controls that were included in the regression analysis which reduced statistical power. The findings of transfer of orthographic processing to reading across languages reflect that children who reach high orthographic performances in one language make use of this ability when learning to read in another language. This is underpinned by the idea that the orthographic processing skill itself should be related across languages, at least for children learning two languages that share the same Roman script. Given this, it is perhaps surprising that correlations between performance on orthographic processing tasks across languages are less than reliable. Significant correlations were found across languages for both lexical and sub-lexical (both language-shared and language-specific measures) components of orthographic processing skills in French immersion children attending grade 1 level (Deacon, Commissaire et al. 2013) and for lexical measures in grade 2 French immersion children (Deacon et al. 2009). In contrast, these relationships were not significant in other samples of bilinguals or second language learners (i.e. older children; see Commissaire, Duncan & Casalis 2011; Deacon, Chen et al. 2011). For example, in a study of French-speaking adolescents in grades 6 and 8 who learned English in secondary school, Commissaire and colleagues revealed that only the performance on the lexical measures of orthographic processing was related across languages, after controls for French reading and English vocabulary. In contrast,

 Eva Commissaire, Adrian Pasquarella, Xi Chen & S. Hélène Deacon

performance on the two sub-lexical measures did not correlate. This mixed picture of results to date suggests that it is worthwhile to further evaluate whether orthographic processing skill is related across languages for bilingual children. 1.3  The present study The present study addresses the question of how orthographic processing skills in English and French develop and relate to each other in Canadian children attending a French immersion program. We followed the children longitudinally from grade 1 to 2. In this learning context, the children were learning to read two languages that use the same alphabet, a situation in which cross-language transfer of orthographic processing skills to reading has been observed in previous studies (e.g. Deacon et al. 2009; Deacon, Commissaire et al. 2013; Deacon, Chen et al. 2011). Lexical orthographic skills were assessed by an orthographic choice task (e.g. rain – rane). A word-likeliness judgment task (e.g. vaid – vayd) was used to evaluate sub-lexical orthographic skills. Language-specificity of the orthographic patterns was manipulated in this task so that language-specific (e.g. 〈ough〉 and 〈ille〉 in English and French, respectively) versus language-shared (e.g. 〈ame〉) orthographic features could be contrasted. Our first objective was to evaluate the development of sub-lexical orthographic processing skills, for both language-shared and language-specific features, in the two writing systems to which the children were exposed. Our prior findings with this sample in grade 1 showed that sub-lexical orthographic skills have already developed in grade 1, as indicated by the above chance performance of our sample (Deacon, Commissaire et al. 2013). In our prior study, the children were more accurate with the language-shared than the language-specific components at grade 1 (Deacon, Commissaire et al. 2013). In the present study, we tested whether this advantage for processing of the language-shared over language-specific items would still be observed at grade 2. This was expected to be the case given that exposure to language-shared orthographic patterns would still remain higher than for language-specific patterns. Furthermore, our prior research demonstrated comparable levels of sub-lexical skill in the two languages at grade 1 (Deacon, Commissaire et al. 2013). The present study evaluates the rate of development of sub-lexical orthographic processing skills in the two languages from grade 1 to grade 2. If sensitivity to orthographic regularities arises from explicit teaching of a written language, then we would expect French immersion children to perform better on the French as compared to the English sub-lexical orthographic processing task. On the contrary, if sensitivity to sub-lexical orthographic patterns emerges from implicit exposure to a written language (see Deacon, Conrad & ­Pacton 2008), then one could predict similarity in the development of French and



The development of orthographic processing skills in children 

English sub-lexical orthographic processing given the children’s exposure to both written languages. To summarize, the present follow-up study enabled us to test the developmental course of sub-lexical orthographic processing skills. Our second main objective was to evaluate whether orthographic processing skills develop in a general manner, both across lexical and sub-lexical tasks and across two languages represented within the same script. Prior research from ­Cunningham and colleagues (2001) suggests a single factor for lexical and sub-­lexical measures of orthographic processing skills in monolingual English-­ speaking children and the present study enabled us to test again the degree of unity of these skills. There are no comparable investigations with bilingual children, with substantial variability in the correlations reported to date (e.g. ­Commissaire et al. 2011; Deacon, Chen et al. 2011; Deacon, Commissaire et al. 2013). The present study also offered the opportunity to test whether orthographic processing skills emerge in a manner that generalizes across a given script with a similar approach; using structural equation modeling (SEM), a powerful statistical approach to determine latent variables, we evaluated whether children’s skills in French and English formed a single factor or two separate factors. There could be two plausible patterns for the factor analysis. We could uncover a single underlying construct for the whole set of measures of orthographic processing skills, both lexical and sub-lexical and both English and French. Such a finding would suggest that the ability to extract orthographic regularities is common to both languages. Another possibility is that only the lexical and language-shared sub-lexical components form a single factor across languages, as suggested by the results of both Deacon, Commissaire et al. (2013) and Commissaire et al. (2011). Further, given that the children were followed longitudinally from grade 1 to grade 2, the developmental course of such underlying factor structure could also be examined. 2.  Method 2.1  Participants Seventy-three children (36 males) enrolled in a Canadian early French immersion program located in Southern Ontario participated in this study in the spring semester of grade 1 (as reported in Deacon, Commissaire et al. 2013) and again in the spring semester of grade 2. Children’s mean age was 6.28 (SD = 0.39) in grade 1. Seven children moved to different schools between testing points and did not participate in the grade 2 testing. In early immersion, children receive all school instruction in French at the beginning of senior kindergarten or grade 1 and they begin to receive part of their instruction in English only from grade 3. Accordingly, the participants’ first formal experience in learning to read was in French. As part

 Eva Commissaire, Adrian Pasquarella, Xi Chen & S. Hélène Deacon

of their regular curriculum, the children received direct instruction in phonological awareness as well as in the use of other reading cues. They received no formal instruction in reading or oral language in English in grade 1 or 2. The majority of the sample (40 of the 73 children) spoke only English at home. Twenty-two children spoke English more than 50% of the time but did occasionally use another language at home. Ten children spoke English less than 50% of the time at home and only one child spoke English at home less than 25% of the time. Of the children who spoke a language other than English at home, the most common home languages were Hebrew, Russian, and Mandarin. Informal conversation (such as on the playground) between the children tended to occur in English, given that English was the first language of most children enrolled in this program. Exposure to English written language began early for all participants. Over 90% of parents reported starting to read with their children in English before 2 years of age. The rest of the parents reported starting to read in English with their c­ hildren between 3 to 5 years of age. All parents read with their children at least once or twice a week, with the majority doing so almost every day. Additionally, 35% of the sample reported that their children read independently every day, where an additionally 40% reported independent reading at least one or twice a week. Finally, the remaining 25% reported reading independently at least once or twice a month. All children whose parents responded (98% of the sample) had at least ten ­English books in the home, with the majority having more than 100 children’s books. Further, teachers reported informally that the majority of children were enrolled in English kindergartens prior to beginning French immersion. Thus, there was substantial informal learning of English reading both before and after children entered the French immersion program. 2.2  Measures 2.2.1  Lexical orthographic processing The lexical orthographic task assessed word-specific orthographic skills. Participants were asked to choose between alternative spellings for a target word (based on Olson, Forsberg, Wise & Rack 1994; e.g. dream – dreem for English, jaune – jeaune for French). The alternative spellings were constructed so that they shared phonological overlap with the target word and respected the graphotactic constraints of the language. In grade 1, the English and French tasks were composed of 29 and 22 items, respectively.3 Cronbach’s alpha reliability was .69 for the French task and .88 for .  Examination of the English and French lexical orthographic performances led to ­removing some of the items when too difficult (see Deacon, Commissaire et al. 2013).



The development of orthographic processing skills in children 

the English task. In grade 2, more difficult items were added to both the French and English versions so that ceiling effects were avoided. In grade 2, there were a total of 41 and 34 items for English and French tasks, respectively. All of the items are presented in Appendix A (*for the added items). The items in the ­English and French tasks were matched on length (mean length: 5.64 and 5.73 letters for English and French, respectively), number of syllables (mean number: 1.52 and 1.59 syllables), and size of orthographic neighborhood (mean size: 1.5 and 1.68, respectively). Importantly, according to child-based databases (English: ­Children’s Printed Word Database; Masterson, Stuart, Dixon & Lovejoy 2003; French: ­Manulex; ­Peereman, Lété & Sprenger-Charolles 2007), the words were also matched on printed frequency in the two languages (mean frequency: 131 and 160 occurrences per million for English and French, respectively), t (73) = 1.15, n.s. Conbach’s alpha reliability was .79 for French and .82 for English. 2.2.2  Sub-lexical orthographic processing Participants were asked to choose the more likely spelling for pseudo-words in either English or French (based on Cassar & Treiman 1997; Pacton et al. 2001). In order to evaluate specifically orthographic and not phonological processing, all pseudo-words were homophonic within a pair according to the decoding regularities of the target language of the task (e.g. doard – dowrd for English and doeur – doeure for French). The alternative pseudo-word contained an illegal orthographic pattern according to both English and French graphotactic rules, either due to the illegality of a letter string (e.g. froul – fhroul where 〈fhr-〉 does not occur in either languages) or to the illegality of its position in the word (e.g. cafle – ckafle where 〈ck-〉 does not occur at the beginning of a word). Two sets of items were created for both the English and French tasks by manipulating the language-specificity of the target orthographic pattern: (1) language-specific and (2) language-shared. In the language-specific condition, the target orthographic patterns were legal in the specific target language of the task only. For instance, pairs of items in the English task such as gook – goock contained the target orthographic pattern 〈-ook〉 which is legal in English only. In the French language-specific condition, we used items such as dreille – dreylle where 〈eille〉 is a common body pattern in French. In the language-shared condition, the target orthographic patterns were legal according to both French and English. For instance, the rime 〈ame〉 occurs in both languages and was contrasted to the orthographic pattern 〈ahme〉 which is illegal in both languages. Note that the same orthographic patterns (e.g. 〈ame〉 vs. 〈ahme〉) were used in the English and French tasks although they were combined with different initial consonants to reduce repetitiveness across tasks and to make the items more appropriate for each writing system. All of the items are presented in Appendix B.

 Eva Commissaire, Adrian Pasquarella, Xi Chen & S. Hélène Deacon

Several criteria were followed when creating these pseudo-words. First, as noted earlier, all pseudo-words were homophonic within a pair and therefore, only orthographic processing could help discriminate between the two items. Second, none of the rime patterns corresponded to morphological units (e.g. quickly), avoiding multiple influences on participants’ responses. Third, the items were matched on grapheme to phoneme consistency.4 No difference was found between the language-specific and language-shared items, F < 1, n.s., and there was no interaction with the language of the task, F(1, 52) = 1.693, n.s. There was, however, a trend for higher grapheme to phoneme consistency in the French than in the English items, F(1, 52) = 2.68, p = .11, an observation which is consistent with the overall greater grapheme to phoneme consistency of the French than the English script (Ziegler, Jacobs & Stone 1996; Ziegler, Stone & Jacobs 1997). 2.2.3  English word reading The Letter-Word Identification subtest of the Woodcock-Johnson III battery (Woodcock, McGrew & Mather 2001) was administered to assess English word reading. In total, there are 76 items in the task. The first 16 require children to identify letters and the remaining 60 items corresponding to reading isolated words aloud that increase in difficulty. The task is discontinued when the participants misread six words consecutively. 2.2.4  English vocabulary knowledge The Peabody Picture Vocabulary Test – Fourth Edition (Form A; Dunn & Dunn 2007) was administered to measure English receptive vocabulary. In this task, four pictures are presented and the participant chooses the picture that best corresponds to a target word that is presented orally. Two practice items are used to ensure the instructions are understood and the task is stopped when students incorrectly answer at least eight items in a set of twelve. 2.3  Procedure Measures of lexical and sub-lexical orthographic processing were administered in English and French in both grades 1 and 2. Trained research assistants highly fl ­ uent in both English and French administered the measures to groups of a­ pproximately .  Phoneme to grapheme consistency of the items was also checked (Ziegler et al. 1996, 1997). Yet, given the items were presented visually, this variable was of less relevance than that of grapheme to phoneme consistency. Items in the French and English task were matched on this criterion, with no interaction between the language of the task and language-specificity, all Fs < 1, n.s. Given strong constraints on item selection, we could not avoid having the items in the language-shared items more consistent in terms of phonology to orthography conversion than those in the language-specific condition, F(1, 52) = 4.024, p < .05.



The development of orthographic processing skills in children 

20 children. English measures were administered before French measures and lexical measures were administered before sub-lexical measures; this was held consistently across grades 1 and 2. 3.  Results 3.1  Development of lexical and sub-lexical orthographic processing Table 1 presents descriptive statistics for the lexical and sub-lexical measures of orthographic processing as well as the English word reading and vocabulary ­measures. Cronbach’s alpha as well as means and standard deviations are displayed Table 1.  Descriptive statistics Mean Cronbach’s alpha

# of items

 Raw

 (SD)

% Correct

(SD)

Grade 1 English Lexical OP

.88

 28

19.13

(5.54)

.66

(.19)

English Sub-lexical OP Shared

.71

 14

10.38

(2.81)

.74

(.20)

English Sub-lexical OP Unique

.65

 14

9.64

(2.65)

.69

(.19)

French Lexical OP

.69

 24

16.61

(3.65)

.69

(.16)

French Sub-lexical OP Shared

.80

 14

9.85

(3.39)

.70

(.24)

French Sub-lexical OP Specific

.67

 14

9.36

(2.89)

.67

(.21)

English Word Reading

.95

 76

English Word Reading SS English Vocabulary

.96

208

English Vocabulary SS

23.6

(10.27)

119.71

(17.23)

115.82

(22.08)

104.24

(15.04)

Grade 2 English Lexical OP

.82

42

30.92

(7.19)

.74

(.17)

English Sub-lexical OP Shared

.79

14

11.49

(2.87)

.82

(.21)

English Sub-lexical OP Specific

.69

14

10.44

(2.68)

.75

(.19)

French Lexical OP

.79

34

26.72

(4.75)

.79

(.11)

French Sub-lexical OP Shared

.83

14

11.55

(2.99)

.82

(.21)

French Sub-lexical OP Specific

.67

14

10.59

(2.64)

.76

(.19)

Note: OP = Orthographic Processing; SS = Standard Score

 Eva Commissaire, Adrian Pasquarella, Xi Chen & S. Hélène Deacon

for all measures. Means of percentage correct for the orthographic processing measures were used in the following analyses to aid interpretation of comparisons between measures and conditions. Standard scores are also presented for the English word reading and vocabulary measures; these measures have a standardized mean of 100 and a standard deviation of 15. On average, children scored one standard deviation above the mean on the word reading measures, suggesting an above average performance for the group. On the receptive vocabulary measure, the average standard score was at the average, suggesting appropriate grade level performance for this group of students. 3.1.1  Lexical orthographic processing A 2 (time: Grade 1, 2) × 2 (language: English, French) repeated measures Analysis of Variance (ANOVA) was used to examine changes in lexical orthographic processing for English and French across grades 1 and 2. There were significant main effects of time, F(1, 70) = 86.69, p < .001, η = .553, and language, F(1, 70) = 7.631, p < .001, η = .098, but no significant interaction. Across grades 1 and 2, children’s scores in lexical orthographic processing significantly increased. Moreover, children scored significantly better in French lexical orthographic processing than English lexical orthographic processing. 3.1.2  Sub-lexical orthographic processing A 2 (time: Grade 1, 2) × 2 (language: English, French) × 2 (condition: shared, specific) repeated measures ANOVA was used to examine changes in sub-lexical orthographic processing for English and French across grades 1 and 2. The only significant main effects were condition, F(1, 70) = 42.78, p < .001, η = .379, and time F(1, 70) = 20.26, p < .001, η = .224. There was no main effect of language or any significant interaction. From grade 1 to grade 2, children significantly improved on the sub-lexical orthographic processing measures in English and French. Children consistently scored higher on shared versus unique conditions of orthographic features in English and French across grades 1 and 2. 3.2  C  omparison of factor models of French and English orthographic processing We used structural equation modeling (SEM) to compare factor models to examine the extent to which orthographic processing in French and English formed multiple factors or a single factor in grades 1 and 2. Using maximum likelihood estimation, four different models were tested in grades 1 and 2. The models tested were: (1) a three factor model with a French orthographic specific factor, English orthographic specific factor, and a shared orthographic factor; (2) a two factor



The development of orthographic processing skills in children 

model with French and English orthographic processing as unique factors; (3) a two factor model with lexical and sub-lexical factors across French and English, and (4) a one factor, cross-language orthographic processing factor. Factor models were constructed and tested separately for grades 1 and 2. Theoretical models are shown in Figure 1.5 Table 2 presents fit indices that were used to examine model fit of the four different models. The left side of Table 2 presents grade 1 fit indices and the right side presents grade 2 fit indices. A χ2 to df ratio < 2 and a Comparative Fit Three Factor Model 1

Two Factor Model 2

French Lexical OP

French Lexical OP French Orthographic Processing

French Sublexical OP Specific

French Sublexical OP Shared English Sublexical OP Shared French OP Lexical

French Sublexical OP Specific

English Orthographic Processing

English OP Lexical English Sublexical OP Shared

One Factor Model 4

English OP Lexical

French Lexical OP Lexical Orthographic Processing

French Sublexical OP Specific

English Sublexical OP Shared

French Sublexical OP Shared

English Sublexical OP Specific

English Sublexical OP Shared

French Sublexical OP Specific

English Orthographic Processing

English Sublexical OP Specific

Two Factor Model 3

French Sublexical OP Shared

French Orthographic Processing

Shared Orthographic Processing

English Sublexical OP Specific

French Lexical OP

French Sublexical OP Shared

Sublexical Orthographic Processing

Orthographic Processing

English Sublexical OP Specific English OP Lexical

Figure 1.  Theoretical factor models of orthographic processing

.  The four models were tested for the influences of English word reading and vocabulary knowledge on the measures of lexical and sub-lexical orthographic processing. Given that these variables did not impact on the fit statistics, factor loadings, or parameter estimates, these were not included in the figures or the models presented below to simplify the results.

 Eva Commissaire, Adrian Pasquarella, Xi Chen & S. Hélène Deacon

Table 2.  Model fit indices for grades 1 and 2 Model

Grade 1 χ2 df

p

Grade 2

χ2/df AIC RMSEA CFI χ2

df

p

χ2/df AIC RMSEA CFI

1. Three Factor Model 4.91 4 .297 1.23 50.91

.05

1

6.25 4 .181 1.56 52.25

.07

.99

2. Two Factor Model

6.16 6 .406 1.03 48.16

.01

1

7.6

6 .269 1.27 49.60

.05

.99

3. Two Factor Model

6.22 7 .515 0.89 46.22

.01

1

7.67 7 .363 1.09 47.67

.03

1

4. One Factor Model

6.22 7 .515 0.89 46.22

.01

1

7.64 7 .365 1.09 47.64

.04

1

Note: AIC = Akaike Information Criterion, RMSEA = Root Mean Square Error of Approximation, CFI = comparative Fit Index

Index (CFI)  > .95 suggest good fit. The Root Mean Square Error of Approximation1 (RMSEA) values ≤ .05 suggest good fit and values ≥ .10 suggest poor fit. The Akaike Information Criteria (AIC) was also examined as an additional index where lower numbers are preferred over higher numbers. As the AIC values decrease, the proposed model is a better representation of the data (Arbuckle 2009; Browne & Cudeck 1993; Kenny, Kashy & Cook 2006). All four models had good fit statistics in grades 1 and 2. However, the three factor model (Model 1) and the two factor model with lexical and sub-lexical factors across French and English (Model  3) had correlations among latent factors larger than 1, in both grades 1 and 2. Therefore, these models displayed Heywood Cases making the factor solutions inadmissible. Heywood cases occur when estimates are produced that are outside acceptable (and possible) boundaries, for example, standardized coefficients larger than 1. If a solution cannot be reached, then a model is deemed inadmissible. Often, a solution to Heywood cases is to impose constraints on coefficients so that they are equal; however, these constraints are generally placed on factor loadings (Bollen & Davis 2009; Kenny, Kashy & Cook 2006). Given that the standardized coefficients among latent factors were larger than 1, constraining parameters among the different latent factors is not a logical solution, thus making Models 1 and 3 inadmissible. Further model comparisons were then made between Models 2 and 4 to determine the most parsimonious model which fit the data best. While Models 2 and 4 displayed good fit statistics, there was an advantage for Model 4 as the χ2/df ratio and the AIC were lower than Model 2. The improved fit statistics for Model 4, in comparison to Model 2, suggests that Model 4 is the preferred model as it explains the data best. Model 4 was consistently a better fitting model in both grades 1 and 2. Additionally, the latent factors in Model 2 where strongly correlated at. 79 in grade 1 and .75 in grade 2, suggesting that even when a 2 factor solution was



The development of orthographic processing skills in children 

imposed on the data, there is a high degree of alignment between the latent factors of French and English orthographic processing in both grades 1 and 2. Thus, Model 4 (the one factor model) emerged as the preferred model in both grades 1 and 2 as it was the most parsimonious model that best explained the data. The SEM factor analyses described a one factor solution as providing the best fit statistics in comparison to all other models, suggesting that lexical and sub-lexical orthographic processing in French and English form a common latent factor in the early elementary grades. Figure 2 presents the factor loadings of Model 4 for orthographic processing in French and English in grades 1 and 2. Factor loadings ranged from .56 to .87 and variance explained in the observed variables ranged from .31 to .74 in grade  1. Factor loadings ranged from .55 to .94 and variance explained in the observed variables ranged from .33 to. 89 in grade 2. Figure 2 also displays the significant correlations between error variances. Correlations among error variances refer to method or measurement effects not captured by the common factor (Eid 2000; Eid, Lischetzke & Nussbeck 2006). Error variances for lexical measures between French and English were significantly correlated in grade  1 (β = .52, p < .001) and grade 2 (β = .43, p < .001). Furthermore, the French sub-lexical language-specific and shared conditions correlated in grade 1 (β = .53, p < .001) and grade 2 (β = .54, p < .001). English sub-lexical error variances were not significantly correlated with each other, possibly because of the high amount of variance captured by the latent orthographic processing factor. No other significant correlations were found between error variances between measures.6 4.  Discussion This longitudinal study investigated orthographic processing skills in Canadian children attending a French immersion program from grade 1 to grade 2. We evaluated both lexical orthographic skills (i.e. at the word-level, rain – rane) and sub-lexical orthographic skills (i.e. at the sub-word level, vaid – vayd). We had two main objectives. First, we examined the development of sub-lexical orthographic processing skills in English and French. Second, we examined the extent to which commonalities among the whole set of orthographic measures could be observed.

.  The pattern of error variance correlations was identical across all versions of the factor models tested in Table 2.

 Eva Commissaire, Adrian Pasquarella, Xi Chen & S. Hélène Deacon Grade 1

Grade 2 .47 e1

French Lexical OP

e2

French Sublexical OP Shared

.35

.52 .53 e3

.31 French Sublexical OP Specific .47

e6

e4

e5

English Lexical OP .75 English Sublexical OP Shared .66 English Sublexical OP Specific

.42 e1

French Lexical OP

e2

French Sublexical OP Shared

.68

.69 .60

.43 .54

.56

.68

e3 Orthographic Processing

.30 e4

.87 .81

.50 French Sublexical OP Specific

e5

e6

English Lexical OP .89 English Sublexical OP Shared

.65 .82 .71

.55 .94 .85

.73 English Sublexical OP Specific

Figure 2. One factor model of lexical and sub-lexical orthographic processing in French and English, showing significant paths only

Orthographic Processing



The development of orthographic processing skills in children 

4.1  Development of orthographic processing skills To address our first objective on the development of sub-lexical orthographic skills, above-chance performance shows early development of sub-lexical orthographic processing skill after only a few months of schooling. That said, children continue to learn about orthographic skills through to grade 2 and likely for several years to come (see also Commissaire et al. 2011; Pacton et al. 2001). Further, children were more accurate with the language-shared items over the language-specific items in the sub-lexical task, across both grade levels. As also demonstrated in our prior study at the grade 1 level (Deacon, Commissaire et al. 2013), children in grade 2 scored higher on items with shared letter patterns across English and French as compared with items with specific letter patterns to an individual writing system. This finding could easily be explained in terms of greater input for the language-shared patterns.7 More research is needed to further examine how the degree of (dis)similarities among writing systems could also help to highlight the specificities of each language and thus contribute to the development of orthographic skills. We also compared the rate of development of sub-lexical orthographic processing skills in the children’s two languages. Children reached similar levels of performance across grades 1 and 2 in English and French on the sub-lexical orthographic task. Though a null result should always be interpreted with caution, this result could suggest comparable development of these skills in English and French. In our view, this suggests that explicit exposure to the written language through formal teaching of the language is not the only vehicle through which learning of sub-lexical orthographic regularities occurs; orthographic learning at the sublexical level might also occur through implicit exposure. This interpretation is in line with statistical approaches of orthographic learning which suggest that implicit

.  We thank an anonymous reviewer for pointing out another explanation to this advantage for the language-shared patterns over the language-specific patterns. It could be that language-shared patterns were more common than language-specific ones within each of the writing systems under study. Using the Lexique and the CPWD databases in French and English, we checked our stimuli by examining the properties of the monosyllabic words that contained the target pattern in both English and French. We found that although more words contained the language-shared as compared to the language-specific patterns (p < .05 in both languages), language-shared and language-specific target patterns were matched within a language ­according to the summed frequency of all words that contained the target patterns (p > .10, n.s) and according to the frequency of the higher frequent word that contained that specific target pattern (that could have been reminded to perform the task, p > .10, n.s.). This good matching would tend to support our interpretation of this advantage as reflecting a greater input for the language-shared patterns in bilinguals.

 Eva Commissaire, Adrian Pasquarella, Xi Chen & S. Hélène Deacon

e­ xposure to the written input leads to progressive extraction of the regularities of the orthographic system (Deacon, Conrad & Pacton 2008; Pacton et al. 2001). As we discuss language similarities in sub-lexical orthographic processing skill, we also need to address our finding that the children scored higher on the French as compared to the English lexical orthographic task. Though the items in this task were matched on several dimensions including printed frequency, this matching was based on English and French as L1 databases that possibly do not reflect printed frequencies in immersion programs. So, this advantage that we found for the French as compared to the English task might be explained by other factors such as differences in the familiarity or regularity of the grapheme to phoneme mappings of the items, entailing different levels of difficulty for the two tasks. Further, the advantage in the French lexical task could also be due to the explicit nature of French learning at school. By emphasizing decoding skills, the formal teaching of the French written language would possibly enhance orthographic skills at the word-level (Share 1995), more than when implicitly exposed to a written language as is the case for English. So, it is possible that explicit learning, and associated larger decoding experiences, is important when considering the development of word-level orthographic skills while not for sub-lexical aspects of orthographic skills. Future studies should explore the underpinning processes of orthographic learning in immersion program contexts and compare the processes (e.g. successful decoding, prior letter knowledge) between the two writing systems to which bilingual children are exposed. 4.2  Relationship of orthographic processing across languages Our next goal in the present study was to evaluate the relationships between the different components of orthographic processing across the two languages. A common factor analysis (Widaman 1993, 2007) was used in order to assess the latent factors underlying orthographic processing skills in English and French, at both lexical and sub-lexical levels. We conducted two analyses, one at each grade, to capture potential developmental trends. The best fitting model at both grades was a one factor model for which the pattern of relationships among the measures were similar for grade 1 and grade 2. This pattern suggests that orthographic skills in English and French, the two writing systems that children are exposed to, are at least partly underpinned by similar processes, a result which indirectly supports the finding of cross-language transfer from orthographic processing to reading for children learning to read languages within the same script (Deacon et al. 2009; Deacon, Commissaire et al. 2013). Indeed, the fact that orthographic processing skills in one language may significantly contribute to reading outcomes in another language might be conceived as in part reflecting the commonalities between



The development of orthographic processing skills in children 

orthographic skills across languages. As for the precise level at which cross-­ language transfer may occur, i.e. knowledge and/ or skill levels (Koda 2000), our data suggest that what transfers across languages is not only knowledge of specific individual features shared between writing systems but also at a more g­ eneral skill level, partly independent of the orthographic features at stake – though within the same script. Interestingly, measures of orthographic processing skills at both the lexical and sub-lexical level significantly loaded onto this single latent factor. Further, this was observed for language-shared items as well as language-specific ones in the sub-lexical task. These results seem to suggest that comparable processes underlie the different sub-components of orthographic processing even when the orthographic regularities under investigation are specific to a single writing system. Retrieving word-level orthographic information and recognizing wordlikeliness of sub-lexical orthographic patterns might require, to some extent, the same underlying processing skills. This observation, congruent with findings from Cunningham and colleagues (2001) with monolingual children, presents a unified concept of orthographic skills. However, in a study that examined older English (L2) learners attending grades 6 and 8, Commissaire et al. (2011) found correlations across languages for the lexical but not sub-lexical orthographic measures. Such diverging results need to be addressed by future inquiry into orthographic patterns across different second language learning contexts. The correlations of error variances between English and French lexical measure and the French shared and specific sub-lexical conditions point to measurement effects between lexical and sub-lexical items. Children might be using slightly different strategies when completing the lexical and sub-lexical measures. Vocabulary knowledge and word recognition can aid decisions on the lexical tasks, as many of these words were known by children, while sub-lexical tasks may rely on finer-grain knowledge of orthographic patterns and decoding in French. Although lexical and sub-lexical measures of orthographic skills load onto a single latent factor in the present study, future studies might investigate possible method and task level differences. 4.3  Limitations and implications Limitations in our study must be considered in relation to the measures that were used and the sample under investigation. Our orthographic measures have some strengths and some weaknesses, too. Our finding of improvement on the sublexical tasks between grades 1 and 2, along with the satisfactory psychometric quality of the measures, provides a validation of the sub-lexical tasks used in the present study. These appear to capture orthographic skills and their development

 Eva Commissaire, Adrian Pasquarella, Xi Chen & S. Hélène Deacon

over time, without reaching ceiling levels, at least through to grade 2. Further, the sub-lexical tasks were well controlled on a number of dimensions. That said, some variables could not be controlled, especially due to stimuli constraints and the type of exposure the children received from each language. In the lexical o ­ rthographic tasks, it is possible that there were subtle differences in the familiarity and frequency of the items between English and French, leading to better performance in the French task than the English task. With respect to the sample, the multicultural nature of the Canadian society led to the inclusion of children for whom English was not their first language in the French immersion program. Yet, both reading and vocabulary scores of these children were comparable with English first language children and multiple statistical controls were embedded into the SEM models which ensured that our findings were not biased by this particular aspect of our data. Despite the limitations, our study has important theoretical implications. Findings of the present study confirm our prior study (Deacon, Commissaire et al. 2013) by showing that learners of two languages rapidly develop multiple orthographic skills and this happens at a younger age than predicted by theories of reading acquisition (Ehri 1995). Not only do children quickly learn to decode new words but they also acquire orthographic processing skills rapidly, at both the lexical and the sublexical dimensions (see Martinet, Valdois & Fayol 2004 and Pacton et al. 2001 for additional evidence in monolinguals). This observation is in line with monolingual research by Share (1995) suggesting that relatively little exposure to a visual input is necessary for developing an orthographic representation. The extent to which these orthographic skills may be dependent upon decoding skills remains unanswered. Future research should address this issue by examining the relationships between decoding and/ or phonological skills with both lexical and sub-­lexical dimensions of orthographic skills, in both languages of immersion children. Further, our findings of a one factor model of orthographic processing skill seem to be in favour of a language- general view of orthographic skills, specifically one that emerges across two languages that share the same alphabet. As previously described for other metalinguistic skills such as phonological awareness (see Koda 2007 for a review), orthographic skills might be considered as a common skill underlying both dominant and nondominant literacy experiences, at least in the context of learning two languages represented by the same alphabet. The finding that even language-specific conditions of the sub-lexical orthographic measure loaded onto one single factor seems to suggest that common skills are required to extract orthographic regularities, whether they are shared between the two writing systems or are specific to one of the writing systems. This unified factor structure also increases our understanding of the nature of cross-language transfer of orthographic processing skills to reading (Deacon et al. 2009; Deacon, Chen et al.



The development of orthographic processing skills in children 

2011; Deacon, Commissaire et al. 2013). As previously discussed, the connection between orthographic processing in one language and reading in the other language may be partially based on the association between orthographic processing across the two languages. In terms of educational implications, our finding of early development of orthographic processing skills suggests that it might be useful for teachers to help the children extract the orthographic regularities of each of writing systems in which they are learning to read. Emphasizing these orthographic features might help children extract the similarities and dissimilarities between the two languages, thereby improving their awareness of the formal aspects of the languages. The findings of commonalities between orthographic skills in English and French also suggest that it might be worthwhile to make explicit connections between formal teaching of these two languages, once English is introduced in the curriculum. These suggestions are speculative and require confirmation with intervention studies.

References Abu-Rabia, Salim (2001). Testing the interdependence hypothesis among native adult-bilingual Russian-English students. Journal of Psycholinguistic Research 30: 435–453.

doi: 10.1023/A:1010425825251

Arab-Moghaddam, Narges & Monique Sénéchal (2001). Orthographic and phonological processing skills in reading and spelling in Persian/English bilinguals. International Journal of Behavioral Development 25: 140–147.  doi: 10.1080/01650250042000320 Arbuckle, James L. (2009). Amos 17.0 user’s guide. Chicago: SPSS. Barker, Theodore A., Joseph K. Torgesen & Richard K. Wagner (1992). The role of orthographic processing skills on five different reading tasks. Reading Research Quarterly 27: 334–345.  doi: 10.2307/747673 Bollen, Kenneth A. & Walter R. Davis (2009). Two rules of identification for structural equation models. Structural Equation Modelling 16: 523–536.  doi: 10.1080/10705510903008261 Browne, Michael W. & Robert Cudeck (1993). Alternative ways of assessing model fit. In K ­ enneth A. Bollen & Jeff S. Long (eds.), Testing structural equation models, 136–162. B ­ everly Hills, CA: Sage. Burt, Jennifer S. (2006). What is orthographic processing skill and how does it relate to word identification in reading? Journal of Research in Reading 29: 1–18.

doi: 10.1111/j.1467-9817.2006.00315.x

Cassar, Marie & Rebecca Treiman (1997). The beginnings of orthographic knowledge: Children’s knowledge of double letters in words. Journal of Educational Psychology 89: 631–644.  doi: 10.1037/0022-0663.89.4.631 Castles, Anne & Kate Nation (2006). How does orthographic learning happen? In S. Andrews (ed.), From ink marks to ideas: Challenges and controversies about word recognition and reading, 151–179. London, UK: Psychology Press.

 Eva Commissaire, Adrian Pasquarella, Xi Chen & S. Hélène Deacon Commissaire, Eva, Lynne G. Duncan & Séverine Casalis (2011). Cross-language transfer of orthographic processing skills: A study of French children who learn English at school. Journal of Research in Reading 34: 59–76.  doi: 10.1111/j.1467-9817.2010.01473.x Cummins, James (1979). Linguistic interdependence and the educational development of bilingual children. Review of Educational Research 49: 222–251.  doi: 10.3102/00346543049002222 Cunningham, Anne E., Kathryn E. Perry & Keith E. Stanovich (2001). Converging evidence for the concept of orthographic processing. Reading and Writing 14: 549–568.

doi: 10.1023/A:1011100226798

Cunningham, Anne E. & Keith E. Stanovich (1990). Assessing print exposure and orthographic processing skill in children: A quick measure of reading experience. Journal of Educational Psychology 82: 733–740.  doi: 10.1037/0022-0663.82.4.733 Deacon, S. Helene, Jenna Benere & Anne Castles (2012). Chicken or egg? Untangling the ­relationship between orthographic processing skill and reading accuracy. Cognition 122: 110–117.  doi: 10.1016/j.cognition.2011.09.003 Deacon, S. Helene, Xi Chen, Yang Luo & Gloria Ramirez (2011). Beyond language borders: Orthographic processing and word reading in Spanish-English bilinguals. Journal of Research in Reading 3: 1–17.  doi: 10.1111/j.1467-9817.2010.01487.x Deacon, S. Helene, Eva Commissaire, Xi Chen & Adrian Pasquarella (2013). Learning about print: The development of orthographic processing and its relationship to word reading in first grade children in French immersion. Reading and Writing 26: 1087–1109.

doi: 10.1007/s11145-012-9407-2

Deacon, S. Helene, Nicole Conrad & Sebastien Pacton (2008). A statistical learning perspective on children’s learning about graphotactic and morphological regularities in spelling. Canadian Psychology 49: 118–124.  doi: 10.1037/0708-5591.49.2.118 Deacon, S. Helene, Lesly Wade-Woolley & John R. Kirby (2009). Flexibility in young secondlanguage learners: Examining the language specificity of orthographic processing. Journal of Research in Reading 32: 215–229.  doi: 10.1111/j.1467-9817.2009.01392.x Dunn, Lloyd M. & Leota M. Dunn (2007). Peabody picture vocabulary test (4th edition). Circle Pines, MN: American Guidance Service. Ehri, Linnea C. (1995). Phases of development in learning to read words by sight. Journal of Research in Reading 18: 116–125.  doi: 10.1111/j.1467-9817.1995.tb00077.x Eid, Michael (2000). A multitrait-multimethod model with minimal assumptions. Psychometrika 65: 241–261.  doi: 10.1007/BF02294377 Eid, Michael, Tanka Lischetzke & Fridtjof W. Nussbeck (2006). Structural equation models for multitrait-multimethod data. In Michael Eid & Ed Diener (eds.), Handbook of multimethod measurement in psychology, 283–299. Washington, DC: American Psychological Association.  doi: 10.1037/11383-020 Gottardo, Alexandra, Bernice Yan, Linda S. Siegel & Lesly Wade-Woolley (2001). Factors related to English reading performance in children with Chinese as a first language: More evidence of cross-language transfer of phonological processing. Journal of Educational Psychology 93: 530–542.  doi: 10.1037/0022-0663.93.3.530 Hayes, Heather, Rebecca Treiman & Brett Kessler (2006). Children use vowels to help them spell consonants. Journal of Experimental Child Psychology 94: 27–42.

doi: 10.1016/j.jecp.2005.11.001

Juel, Connie, Priscilla L. Griffith & Philip B. Gough (1986). Acquisition of literacy: A longitudinal study of children in first and second grade. Journal of Educational Psychology 78: 243–255.  doi: 10.1037/0022-0663.78.4.243



The development of orthographic processing skills in children 

Kenny, David A., Deborah A. Kashy & William L. Cook (2006). Dyadic data analysis. New York: Guilford Press. Koda, Keiko (2000). Cross-linguistic variation in L2 morphological awareness. Applied Psycholinguistics 21(3): 297–320.  doi: 10.1017/S0142716400003015 Koda, Keiko (2007). Reading and language learning: Crosslinguistic constraints on second ­language development. Language Learning 57: 1–44.  doi: 10.1111/0023-8333.101997010-i1 Martinet, Catherine, Sylviane Valdois & Michel Fayol (2004). Lexical knowledge develops from the beginning of literacy acquisition. Cognition 91: 11–22.

doi: 10.1016/j.cognition.2003.09.002

Masterson, Jackie, Morag Stuart, Maureen Dixon & Sophie Lovejoy (2003). The Children’s Printed Word Database. Retrieved from www.essex.ac.uk/psychology/cpwd Olson, Richard K., Helen Forsberg, Barbara Wise & John P. Rack (1994). Measurement of word recognition, orthographic and phonological skills. In G. Reid Lyon (ed.), Frames of reference for the assessment of learning disabilities: New views on measurement issues, ­243–277. Baltimore, MD: Paul H. Brookes Publishing Co. Pacton, Sebastien, Michel Fayol & Pierre Perruchet (2005). Children’s implicit learning of ­graphotactic and morphological regularities. Child Development 76: 324–339.

doi: 10.1111/j.1467-8624.2005.00848_a.x

Pacton, Sebastien, Pierre Perruchet, Michel Fayol & Axel Cleeremans (2001). Implicit learning out of the lab: The case of orthographic regularities. Journal of Experimental Psychology: General 130: 401–426.  doi: 10.1037/0096-3445.130.3.401 Peereman, Ronald, Bernard Lété & Liliane Sprenger-Charolles (2007). Manulex-Infra: Distributional characteristics of grapheme-phoneme mappings, infra-lexical and lexical units in child-directed written material. Behavior Research Methods 39: 579–589.

doi: 10.3758/BF03193029

Perfetti, Charles A. (1992). The representation problem in reading acquisition. In Philip B. Gough, Linnea C. Ehri & Rebecca Treiman (eds.), Reading acquisition, 145–174. Hillsdale, NJ: Erlbaum. Share, David L. (1995). Phonological recoding and self-teaching: Sine qua non of reading acquisition. Cognition 55: 151–218.  doi: 10.1016/0010-0277(94)00645-2 Stanovich, Keith E. & Richard F. West (1989). Exposure to print and orthographic processing. Reading Research Quarterly 24: 402–433.  doi: 10.2307/747605 Treiman, Rebecca & Brett Kessler (2006). Spelling as statistical learning: Using consonantal ­context to spell vowels. Journal of Educational Psychology 98: 642–652.

doi: 10.1037/0022-0663.98.3.642

Tucker, G. Richard (1999). A global perspective on bilingualism and bilingual education. ERIC Digest (ERIC Document Reproduction Service No. ED435168). Wang, Min, Yoonjung Park & Kyoung R. Lee (2006). Korean-English biliteracy acquisition: Cross-language phonological and orthographic transfer. Journal of Educational Psychology 98: 148–158.  doi: 10.1037/0022-0663.98.1.148 Widaman, Keith F. (1993). Common factor analysis versus principal component analysis: ­Differential bias in representing model parameters? Multivariate Behavioral Research 28: 263–311.  doi: 10.1207/s15327906mbr2803_1 Widaman, Keith F. (2007). Common factors versus components: Principals and principles, errors and misconceptions. In Robert Cudeck & Robert C. MacCallum (eds.), Factor analysis at 100: Historical developments and future directions, 177–203. Mahwah, NJ: Erlbaum. Woodcock, Richard W., Kevin S. McGrew & Nancy Mather (2001). Woodcock-Johnson III. Itasca, IL: Riverside Publishing.

 Eva Commissaire, Adrian Pasquarella, Xi Chen & S. Hélène Deacon Ziegler, Johannes C., Arthur M. Jacobs & Gregory O. Stone (1996). Statistical analysis of the bidirectional inconsistency of spelling and sound in French. Behavior Research Methods, Instruments & Computers 28: 504–515.  doi: 10.3758/BF03200539 Ziegler, Johannes C., Gregory O. Stone & Arthur M. Jacobs (1997). What’s the pronunciation for OUGH and the spelling for /u/? A database for computing feedforward and feedback inconsistency in English. Behavior Research Methods, Instruments & Computers 29: 600–618.  doi: 10.3758/BF03210615

Appendix A Items in the lexical orthographic processing tasks. Note that *represents those items that were added to the task in grade 2

English Task nostrils-nostrels, need-nead, backward-backword, between-betwean, travel-travle, growngrone, learn-lurn, easy-eazy, hurt-hert, stream-streem, rain-rane, applause-aplause, answeranser, every-evry, turtle-tertle, ghost-goast, explain-explane, true-trew, dream-dreem, roar-rore, believe-beleave, skate-skait, smoke-smoak, mystery-mysterey, thumb-thum, waitwate, pavement-­ pavemant, assure-ashure, engine-enjine, salmon-sammon*, wreath-reath*, wheat-wheet*, scare-scair*, wise-wize*, trousers-trowsers*, basement-baisement*, heavy-hevvy*, take-taik*, several-sevral*, sudden- suddin*, study-studdy*

French Task grimper-grinper, château-châtau, beaucoup-beaucout, jumeau-jummeau, oeuf-euf, corbeau-­ corbau, chaise-cheise, vent-vemp, miel-miele, poulet-poulet, neige-naige, verre-vaire, terreteire, jambon-jembon, patin-patain, chanson-chenson, écran-écren, bavard-bavar, jaune-jeaune, drapeau-drapau, tricot-triquot, bouge-bouje, crapaud-crapau*, achète-achaite*, trompe-tronpe*, durant-duran*, vêtement-vaitement*, sonnerie-sonnerit*, jamais-jamet*, voleur-voloeur*, songe-sonje*, concours-comcours*, vendre-vandre*, lenteur-lanteur*

Appendix B Items in the sub-lexical orthographic processing tasks

English task Language-specific items: yead-yeadh, sween-sweinn, cluff-cluph, nowl-nowll, doard-dowrd, florn-flornne, crell-crelh, dreel-dreell, vish-visch, gook-goock, cruck-cruq, glough-glouph, fikeficke, skoad-skowde Language-shared items: ploin-ployn, cufle-ckufle, stearl-sttearl, frode-fhrode, slage-slaje, knase-knazze, flean-phlean, chowl-chhowl, spime-spiime, stame-stahme, plour-plouur, scoalsckoal, clange-clangge, boogh-bhoogh



The development of orthographic processing skills in children 

French task Language-specific items: rouve-rouvve, jonde-jomde, clabe-clabhe, vratte-vrahte, doeur-doeure, vutte-vuutte, dreille-dreylle, splomme-splaumme, proce-prosce, sulle-suule, fompe-fonpe, blappe-blaape, fouche- foushe, clige-clije Language-shared items: doin-doyn, cafle-ckafle, stime-sttime, froul-fhroul, glage-glaje, klase-klazze, flaipe-phlaipe, chabe-chhabe, vime-viime, plame-plahme, vrour-vrouur, scoilsckoil, drange-drangge, borl-bhorl

How do previously acquired languages affect acquisition of English as a foreign language The case of Circassian Janina Kahn-Horwitz1, 2, Sara Kuash1, Raphiq Ibrahim2 & Mila Schwartz1 1Oranim

Academic College of Education, Israel / 2Edmond J. Safra Brain Research Center for the Study of Learning Disabilities, University of Haifa, Israel

The present study aims to examine the linguistic and orthographic proximity hypothesis in new script acquisition by comparing the performance of Circassian L1 speaking children who are emerging quadri-literates with Hebrew L1 speaking children who are emerging biliterates. Tests in decoding and spelling various English target conventions were conducted. Thirty 10 year old Circassian L1 speaking children were compared to 46 Hebrew L1 speaking children. Results show that the group of Circassian L1 speaking children outperformed the group of Hebrew L1 speaking children and showed a significant advantage in decoding and spelling target orthographic conventions. There were no significant differences between the two groups on decoding and spelling the silent 〈e〉, which provided a challenge for both groups. The results provide support for the linguistic and orthographic proximity hypothesis whereby phonemes and orthographic characteristics that exist in a child’s first or additional language system and writing system facilitate acquisition of orthographic conventions in a new language and writing system. Keywords:  linguistic and orthographic proximity; decoding; spelling; Circassian; Hebrew; EFL

1.  Introduction Different indigenous language communities living side by side in the same geographical area create situations whereby people learn several languages. Israel provides a historical example of language and cultural coexistence. The two official languages of Israel are Hebrew and Arabic, with numerous minority languages spoken by indigenous and immigrant communities. The Circassian community functions within a typically natural multilingual context. Their members speak doi 10.1075/bct.89.03kah © 2016 John Benjamins Publishing Company

 Janina Kahn-Horwitz, Sara Kuash, Raphiq Ibrahim & Mila Schwartz

Circassian as their mother tongue and they acquire both Arabic and Hebrew from the first grade. Arabic is the language of their religion which is Islam and Hebrew is the first language of the country in which they live. In addition, they learn English as a foreign language (Olshtain & Nissim-Amitai 2004: 30–36). Thus, Circassian children are exposed to four languages, Circassian, Arabic, Hebrew, and English, in their oral and written forms (see Table 1). Table 1.  Order of language and literacy acquisition amongst Circassian children Acquisition sequence

Languages and literacies

From birth

Oral Circassian

From first grade

Oral and written Hebrew

From third grade

Oral and written English

From fifth grade

Written Circassian

Modern Standard Arabic (written Arabic)

This research aims to investigate how Circassian L1 speaking children cope with English literacy acquisition where English is a fourth sequential language and compare this acquisition with Hebrew L1 speaking children who acquire E ­ nglish as a second sequential language. In other words, to what extent does oral C ­ ircassian together with Hebrew and Arabic literacy affect English literacy development? This study has significance for policy makers, practitioners, and parents worldwide who reside in multilingual settings and question the timing and cognitive and linguistic benefits of multi-literate experiences. The question of cross language transfer can be examined from a language and literacy acquisition perspective. This paper addresses the question of positive and negative transfer in the context of acquiring the English writing system amongst multilingual and multi-literate children. Transfer will be presented as a process which can either facilitate or hamper the acquisition of an additional language writing system. Cook (n.d.) describes multi-competence as being a state whereby language learners utilize knowledge of their first language and interlanguage in the acquisition process of a second language. In the case of the current study, multicompetence is understood as knowledge of first and additional languages (and writing systems) along with interlanguage in the mind of the language learner in the process of acquiring a further language and writing system. Cook’s notion of multi-competence complements Cummins’ linguistic interdependence hypothesis (Cummins 1978: 131–149). According to Cummins, academic language proficiency transfers across languages such that students who have developed L1 will tend to make greater progress in acquiring L2. Multicompetence, according to Cook, could also explain the impact of acquired L2 on L1. The aforementioned theories provide a background that attempts to explain



How do previously acquired languages affect acquisition of English as a foreign language 

cross-language transfer between basic literacy skills. Two hypotheses that explain this ­transfer in the acquisition of a new writing system are the script dependence hypothesis (Geva & Siegel 2000: 1–30) and the linguistic and orthographic proximity hypothesis (Kahn-Horwitz, Schwartz & Share 2011: 136–156). The script dependence hypothesis relates to the orthography of a given language and claims that specific characteristics of L1 orthographic structure affect the acquisition of the writing system of a second language (L2) – decoding and spelling – due to cross-linguistic transfer (Geva & Siegel 2000: 1–30). Evidence supporting the script dependence hypothesis comes from numerous studies examining cross-linguistic transfer of lower level decoding abilities between English and other languages (e.g. Hebrew-English: Geva & Wade-Woolley 1998: 85–110; Chinese-English: Gottardo, Yan, Siegel & Wade-Woolley 2001: 530–542). The linguistic and orthographic proximity hypothesis examines the degree of proximity between linguistic as well as orthographic characteristics of L1 and L2. Research by Kahn-Horwitz, Schwartz and Share (2011: 136–156) found that Russian-Hebrew speaking bilingual biliterates outperformed Russian-Hebrew speaking bilingual mono-literates and Hebrew speaking mono-lingual mono-­literates for English reading and spelling of short vowels. Russian and English have more in common, both orthographically and linguistically, in contrast to Hebrew and Russian or Hebrew and English. Transfer is more frequent amongst similar languages, which are close linguistically and orthographically (Cenoz & ­Genesee 1998: 16–32; Cenoz & Hoffmann 2003: 1–8). The advantage experienced by the Russian-Hebrew speaking bilingual biliterates may be explained within this paradigm. Russian and English use a more similar vowel system in contrast to the differences in use of vowels amongst Russian-Hebrew and English-Hebrew. It appears that the specific advantage of Russian-Hebrew speaking bilingual biliterates on short vowel decoding and spelling in English may be a result of the similar way the Russian and English scripts graphically represent vowels. In this case, the more typologically similar language might facilitate the L2 acquisition and vice versa. According to the functional assumption, what is transferred is not a set of rules but form-function relationships that L2 users have acquired together with their mapping skills (Koda 2008: 68–93). Hence, linguistic knowledge from a bilingual biliterate or multilingual multi-literate perspective is continually in the process of developing and transfer becomes a dynamic process. Transfer in the context of the acquisition of an additional language and writing system necessarily involves drawing on previously acquired resources (Koda 2008: 68–93). It can be said that the acquisition of a third language and writing system is a repeated process to the extent that acquiring a new writing system each time has characteristics in common with the previous process of acquiring a writing system. Similarities between languages and writing systems should allow third language learners to

 Janina Kahn-Horwitz, Sara Kuash, Raphiq Ibrahim & Mila Schwartz

usefully exploit the resources accumulated through prior learning experience, thereby accelerating first, second, third, and maybe even fourth language acquisition. In the current study, the specific characteristics of acquiring the fourth writing system may be influenced by the fact that the children involved have already acquired spoken L1, spoken and written L2, and written L3. They have gained conscious linguistic knowledge and language learning experience on which they can potentially rely when learning a further language. Thus, they can draw from a larger and more varied linguistic cognitive repertoire than monolingual speakers. As a consequence, L3 perception and production differ from L2 perception and production in the complexity of potential sources for cross linguistic awareness (Gut 2009: 19–38). As a result, Circassian children who are exposed to three languages before acquiring English may show advantages and subsequently better results than children exposed to fewer languages and writing systems.

2.  Typological properties of the four different languages In the following section, we will address typological distance and proximity with regard to acquisition of basic reading and spelling acquisition in English as a foreign language, specifically focusing on linguistic and orthographic categories that are similar and different to the Circassian, Hebrew, and Arabic oral and written languages. 2.1  English English has 26 letters which map onto more than 40 phonemes (Venezky 1999). The English orthography has a complex vowel system with short and long vowel phonemes and di-, tri-and quadrigraphs as well as diphthongs. There are 15 phonemes corresponding to the five vowel graphemes (Frost 2005: 272–295). Each of the single vowel graphemes in English maps onto multiple sounds (e.g. 〈a〉 as in apple, apron or above; 〈o〉 as in opera, oven or open; 〈u〉 as in under, difficult or union). In contrast to the aforementioned grapheme-phoneme correspondences, when considering phoneme-grapheme correspondence, a good example of ­English’s relatively complex mapping would be /i:/, which can be represented by at least six different spellings, e.g. 〈i〉 as in hi, or 〈igh〉 as in light, 〈y〉 as in why, 〈ie〉 as in pie, and the silent 〈e〉 as in the word mine or rhyme. In the case of digraphs, in order to determine the value of the phoneme one needs to consider more than one letter. For example, the phoneme /ð/ is represented by two letters, 〈th〉. The above features make English a deep orthography. As a result, English decoding acquisition is considered to be rather difficult, even for monolingual



How do previously acquired languages affect acquisition of English as a foreign language 

speakers (Seymour, Aro & Erskine 2003: 143–174). English is one of the least transparent of alphabetic writing systems, exceeded perhaps by Hebrew and Arabic when they are written without vowel markings (Frost 2005: 272–295). 2.2  Circassian The spoken Circassian language (called Адыгэбзэ, Adyghabza, by the Circassian people) belongs to the family of Northwest Caucasian languages (Mamser Batsag 2009: 363–402). It is spoken in the homeland in Adygeya (a constituted republic of the Russian Federation) and in many other countries around the world including Israel. In Israel, there are two Circassian villages, Rihania and Kfar Kama (Bram 1999: 205–222). The Circassian language consists of 56 consonants and 7 vowels (〈a〉 /æ/ as in /ænʌ/ ‘table’, 〈э〉 /ʌ/ as in /dʌ/ ‘nut’, 〈e〉 /ɛ/ as in /sɛtʌ/ ‘my father’, 〈o〉 /o/ as in /mor/ ‘that’, 〈y〉 /ʊ /as in /kʊ/ ‘wagon’, 〈и〉 /ɪ/ as in /qɪn/ ‘difficult’ and 〈ы〉 /ɨ/ as in /zɨ/ ‘one’) and it is written using Cyrillic letters (Tcharkuacho 2000). Like English, the Circassian language is read from left to right. Some of the consonants can be contrasted in that they have either plain or glottal stops. In addition, similar to the English language some of the phonemes in the Circassian language are represented by two graphemes, for example the phoneme /dʒ/ 〈дж〉. Some phonemes are also common to Circassian and English, e.g. /t∫/, /w/ and /dʒ/ whereas the phonemes /θ/ and /ð/ represented by the consonant digraph 〈th〉 in English do not exist in Circassian (Tcharkuacho 2000). 2.3  Arabic Arabic is a diglossic language, meaning that there are two varieties of the language: Spoken Arabic, which is used in every day communications and which has no written form; and Modern Standard Arabic, which is used in writing and speaking in formal settings. Modern Standard Arabic is more or less unified, while Spoken Arabic is represented by different dialects in each area (Allaith & Joshi 2011: 1089–1110). Arabic is a Semitic language that is read from right to left. It is comprised of 28 letters that primarily represent consonants but also include three letters that correspond to long vowel phonemes: aleph 〈‫ 〉  ا‬/aː/, yeh 〈‫ 〉ي‬/iː/, and wow 〈‫ 〉و‬/uː/. In addition, there are three short vowel forms that are written as diacritics representing short vowels: fatha- a short diagonal stroke above the letter (e.g. 〈‫ ب‬ َ 〉 /ba/), damma- the diacritic mark above the letter (e.g. 〈‫ ب‬ ُ 〉 /bu/), and kasra- a short diagonal stroke below the letter (e.g. 〈‫ب‬ ِ  〉 /bi/). Arabic is written in cursive form, and letters within words are usually combined. Twenty two out of the 28 consonants have different forms for word ­initial,

 Janina Kahn-Horwitz, Sara Kuash, Raphiq Ibrahim & Mila Schwartz

word medial, and word final positions. Children initially learn to read Arabic through use of a fully vowelized orthography in which all the consonants and vowels are represented in the script. In this ‘shallow’ orthography, the short vowels are included in the form of diacritic markings above and below the consonants. Thus, children initially acquire a phonologically transparent writing system in which each letter corresponds to one phoneme. These texts are easy to decode phonologically since the letters and diacritics have mostly a highly consistent and reliable grapheme-to-phoneme correspondence. The standard form of Arabic used for texts in the higher grades is represented without vowel diacritics, thereby making it a deep orthography (Fender 2008: 19–42). The deep or unvowelized orthography is more complex to decode because decoding cannot take place by direct translation of spelling to pronunciation. In order to decode unvowelized Arabic, there is a need to use morphological, lexical, and contextual information. There are similarities and differences in the development of basic literacy skills in Arabic and English. For example, beginning readers of Arabic and of English must acquire the ability to read alphabetic orthographies. This requires learning the specific grapheme-to-phoneme correspondences, learning to blend letters and phonemes to decode or sound out words, and learning spelling patterns of the writing system. Therefore, initial literacy development in Arabic similar to E ­ nglish requires acquisition of phonological decoding skills in order to read words. In addition to phonological skills and similar to early English literacy acquisition, children acquiring Arabic literacy must remember orthographic patterns of specific words that do not correspond to more direct grapheme-phoneme correspondences (e.g. free versus people) in order to become competent at word recognition (Koda 2008: 68–93). Many Arabic consonantal phonemes are similar to English phonemes. However, there are several Arabic phonemes which do not exist in English; /ʔ/, /ħ/, /x/, /sʕ/, /dʕ/, /tʕ/, /ðʕ/, /ʕ/, /ɣ/, and /q/; and a few English consonantal phonemes which do not exist in Arabic; /g/, /p/, /v/, and /ʧ/. In addition, the digraph consonants 〈sh〉 /ʃ/ and 〈th〉 /ð/, /θ/ exist as phonemes in Arabic but are represented by a grapheme consisting of one consonant. There are no clusters containing more than two consonants in Arabic. In addition, the silent 〈e〉 convention does not exist in Arabic (see Appendix 1). 2.4  Hebrew The Hebrew script, like the Arabic script, is also consonantal. Similar to Arabic, Hebrew is considered a Semitic language that is read from right to left. It has 22 letters, five of which have a different final form. Hebrew vowels can be represented



How do previously acquired languages affect acquisition of English as a foreign language 

by diacritical marks in the form of dots and dashes appearing below, above to the left, or by means of four consonantal letters (‫‘ – )א ה ו י‬mothers of reading’ – which can signify vowels as well as consonants. Similar to the Arabic orthography and unlike the English orthography, the voweled Hebrew script, which is taught from first grade, can be described as shallow in that there is a direct one to one grapheme-phoneme correspondence. An additional similarity between the two Semitic scripts is that they are both considered to be abjads (Daniels & Bright 1996), that is, writing systems whereby the letters denote consonants and the vowels are inferred from lexical knowledge. In contrast to the alphabetic scripts, e.g. English, Russian, Circassian, the consonantal scripts like Hebrew and Arabic in their unpointed (unvoweled) form are considered to be examples of deep orthographies. Homographs are frequently a result of partial vowelization of words. The assumption is that reading abjads such as Arabic and Hebrew requires different strategies from reading alphabetic scripts, where both consonants and vowels are denoted by the letters. In contrast to English and Arabic, the acquisition of grapheme-phoneme correspondence and decoding of voweled Hebrew demands that the learner acquires relatively few orthographic conventions. For spelling, however, Hebrew becomes more challenging as there are numerous phonemes which are represented by two graphemes, e.g. the phoneme /t/ is represented by two different graphemes, 〈‫〉ת‬, 〈‫〉ט‬. Children initially acquire voweled Hebrew and as they become skilled readers they are gradually exposed to unvowelized texts. At this stage, the orthographic script can be considered deep as the reader depends on both contextual and linguistic knowledge for accurate decoding (Frost 2005: 272–295). Differences between the Hebrew and English orthographies include more than one letter representing a phoneme in English. For example, the silent 〈e〉, the digraph consonants 〈th〉 /ð/, /θ/, 〈sh〉 /ʃ/, and 〈ch〉 /ʧ/ and the vowel digraphs 〈ee〉 /iː/ and 〈oo〉 /uː/ (see Appendix 1). 3.  Research Hypothesis With the aforementioned background in mind, we expected there to be differences in acquisition level of English as a result of exposure to multiple languages and writing systems amongst the Circassian L1 versus the Hebrew L1 speaking children. Based on the script dependent hypothesis and linguistic and orthographic proximity hypothesis, we hypothesized that specific English orthographic conventions and linguistic characteristics would be challenging both for Hebrew L1 speaking and Circassian L1 speaking children due to their Semitic linguistic and orthographic background which is different to the English language and

 Janina Kahn-Horwitz, Sara Kuash, Raphiq Ibrahim & Mila Schwartz

orthography. On the other hand, we predicted that the Circassian L1 speaking children would perform better than their Hebrew L1 speaking peers on specific English orthographic conventions due to phonemic proximity with previously acquired languages (Circassian, Hebrew, and Arabic) and writing systems despite the fact that the similar phonemes were not represented as digraphs as in the case of ­English, e.g. the phonemes /ð/ and /θ/ (Appendix 1). To conclude, in this study we expected that Hebrew speaking L1 children who were learning English as a foreign language would experience challenges acquiring the following phonemes and orthographic conventions: the silent 〈e〉, the digraph consonants 〈th〉 /ð/, /θ/, and 〈ch〉 /ʧ/, vowel digraphs 〈ee〉 /iː/ and 〈oo〉 /uː/, the short vowels including the short 〈a〉 /æ/. Research questions: 1. Do Circassian L1 speaking children show advantages on specific English orthographic conventions over Hebrew L1 speaking children due to their wider linguistic and orthographic repertoire? 2. Do Circassian L1 speaking children show similar difficulties to Hebrew L1 speaking children with specific English orthographic conventions which do not exist in Circassian, Hebrew, or Arabic orthographies? 4.  Method 4.1  Participants Participants included 76 Circassian and Hebrew L1 speaking children at the end of fifth grade. Language and literacy characteristics of Circassian L1 speaking children Thirty L1 speaking Circassian children, who comprise the entire population of 5th grade Circassian children in Israel, came from two schools in two villages in the north of Israel. All Circassian children were characterized by a similar middle socioeconomic status (SE) index. This index is calculated by the Israeli Ministry of Education based on parents’ reports on their income, occupation and ranking of family residential area. The government uses this index for funding of schools. In Israeli schools, the SE index is measured on a 10-point scale ranging from 1 to 10, with higher scores indicating lower SE status. In the present study, the SE index for both schools is 6.02 and 6.65, respectively. The consent form filled in by parents included a question about mother education, dominant language and other languages spoken and read at home, and whether children received any private tutoring in English outside of school hours.



How do previously acquired languages affect acquisition of English as a foreign language 

All Circassian children acquire four oral and written languages (see Table 1). From birth, these children are exposed to Circassian (L1). They acquire Modern Standard Arabic and Hebrew in parallel as two second languages (L2) from the first grade. English is the third oral and written language which is acquired from the third grade. Similar to Hebrew L1 speaking children, Circassian children study English for at least two hours a week in third grade and then at least four hours a week from fourth grade. In the third grade, children acquire basic vocabulary, for example, colors, numbers, and animals, and they also start to learn the letters (names and phonemes). In addition, they start reading simple words. The written form of the Circassian language is taught mainly from fifth grade. The language and literacy base-line questionnaire answered by the parents reported that the Circassian children grow up in homes where the main spoken language is Circassian. They use Hebrew from time to time when they speak to each other and few wrote that they use Arabic from time to time. None of them wrote that they use English at home. All parents reported that they possess Hebrew books, most of them possess Arabic books, more than half of them have books in English, and half of them have books in Circassian. It was interesting to note that according to the parents, all children prefer reading books in Hebrew. From these responses it appears that the dominant language at home is Circassian whereas the dominant written language is Hebrew. The Circassian L1 speaking children learn English at school and according to parents’ reports, they are exposed to the English language through a variety of means; mainly through TV, internet, and music and for some also by reading books. None of the Circassian L1 speaking children received any supplementary private tutoring in English outside the school framework. Performance on English literacy tasks by the Circassian participants in the current study were compared to the Hebrew L1 participants in a previous study comparing Russian-Hebrew speaking children to Hebrew L1 speaking children (Kahn-Horwitz et al. 2011: 136–156). Language and literacy characteristics of Hebrew L1 speaking children There were a total of 46 Hebrew L1 speaking children at the end of grade 5 that were compared to the Circassian participants in this research. The children were selected from five schools located in the northern region of Israel. Similar to the Circassian children in this study, the schools were characterized by a similar middle socio-economic index with scores of 4.70, 4.90, 5.05, 5.05, and 5.1, respectively. Participant selection was conducted by approaching parents during parent-teacher meetings. Parents who consented to participate in the research answered questions

 Janina Kahn-Horwitz, Sara Kuash, Raphiq Ibrahim & Mila Schwartz

about years of mother education, the age when the child began to receive English instruction, and whether the child received any private tutoring in English. Similar to their Circassian L1 speaking peers, Hebrew L1 speaking children were selected who began English instruction in Grade 3, did not receive any supplementary private tutoring in English outside of school, and were not diagnosed with learning disabilities. In the context of the two language groups compared in this study, it is important to highlight that English as a foreign language (EFL) studies fall under centralized Israeli Ministry of Education supervision and all children regardless of L1 background follow a similar program in their EFL studies. 4.2  Measures The research design included two sets of measures. The first set of measures were base-line measures comparing general non-verbal ability, Hebrew and English phoneme awareness as a prerequisite for basic literacy skills acquisition, as well as Hebrew word decoding and spelling. Both groups of children were compared on the aforementioned measures. In addition, the Circassian L1 speaking children were tested on basic Arabic phoneme awareness, decoding and spelling measures in order to check their previous language and literacy experience in Arabic. The second set of measures included English orthographic and linguistic target items that were tested via a pseudo-word decoding and spelling measure. 4.2.1  Base-line measures General ability (Raven, Raven & Court 1976: sets A, B, C, D and E), a nonverbal ability test, participants matched one of six graphic patterns to a visual array. Arabic literacy measures Word recognition accuracy (Asadi, Shany, Ben-Simon & Ibrahim 2011). This test included a list of 36 voweled words varying in degree of difficulty. The participants read word after word aloud. Accuracy scores were computed by allocating one point for each correctly pronounced word. The maximum score on this task is 36. Internal consistency (coefficient alpha) is .91. Word spelling (Asadi et.al. 2011). This list of 28 words was read aloud by the tester to each participant. The participants had to spell the words. Accuracy scores were computed by allocating one point for each correctly spelled word. The maximum score on this task is 28. Internal consistency (coefficient alpha) is .90. Phonemic awareness (Asadi et al. 2011). This measure consists of 20 words ranging from one to three syllables. For test administration, the tester pronounced each word aloud and then deleted a phoneme at the beginning, middle, or end of the word. The participant then had to pronounce the remaining (meaningless)



How do previously acquired languages affect acquisition of English as a foreign language 

sequence. For example,‘Say ruz (‘rice’). Now, say ruz without r.’ The correct answer would be uz. Two examples were presented to the participants at the beginning of task administration. The total number of correct responses was calculated. The maximum possible score is 20 and the internal consistency (coefficient alpha) is .92. 4.2.2  Hebrew literacy measures Word recognition accuracy (Shany, Lachman, Shalem, Bahat & Seiger 2003). This test includes a list of 38 voweled words varying in degree of difficulty, ranging from short, monomorphemic words (e.g. ‫ גד‬dag, ‘fish’) to longer, multi-morphemic ones (e.g. ‫מתרווחים‬, mitravxim, ‘they get more comfortable’). The participants read word after word aloud. Accuracy scores were computed by allocating one point for each correctly pronounced word. The maximum score on this task is 38. Internal consistency (coefficient alpha) is .90. Word spelling (Deutsch 1994). This list of 20 words was read aloud by the tester to each participant. The participants had to spell the words. Accuracy scores were computed by allocating one point for each correctly spelled word. The maximum score on this task is 20. Internal consistency (coefficient alpha) is .90. Phonemic awareness (Shany et al. 2003). This measure consists of 20 words ranging from one to two syllables. The words represent the following consonant/ vowel structures: CVC, CCVC, CCVCV, CVCVC, and CVCCV. For test administration, the tester pronounced each word aloud. The participant repeated the word after hearing it. Then the tester deleted a phoneme at the beginning, middle, or end of the word and the participant had to pronounce the remaining (meaningless) sequence. For example, ‘Say gal (‘wave’). Now, say gal without g’. The correct answer would be al. Two examples were presented to the participants at the beginning of task administration. The total number of correct responses was calculated. The maximum possible score is 20 and the internal consistency (coefficient alpha) is .85. 4.2.3  English literacy measure English phoneme deletion (adapted from Rosner 1975). The adapted version of this test included 20 one syllable words that required the deletion of initial, middle, or final phoneme. Each word was presented orally. The participant repeated the word after hearing it. Then the tester deleted a target phoneme and the participant had to pronounce the remaining sequence, for example, ‘Say man. Now, say man without m’. The correct answer would be an. Two examples were presented to the participants at the beginning of task administration. The total number of correct responses was calculated. The maximum possible score is 20 and the internal consistency (coefficient alpha) is .74.

 Janina Kahn-Horwitz, Sara Kuash, Raphiq Ibrahim & Mila Schwartz

4.2.4  English experimental measures Pseudo-word spelling. This test included a list of 29 pseudo-words. These pseudowords contained the following target vowels and consonants that were later analyzed; short vowel graphemes (〈a〉, 〈e〉, 〈i〉, 〈o〉, 〈u〉), consonant digraphs (〈ch〉, 〈th〉), vowel digraphs (〈oo〉, 〈ee〉) and the silent 〈e〉 (see Appendix 2). Each target pseudo-word was read twice by the tester. After hearing the target pseudo-word, participants repeated the stimulus and then wrote it. Scores were computed in two ways: first by allocating one point for each correctly spelled item. Maximum score for this task is 29, internal consistency (coefficient alpha) was .72. Then scores were computed for each variable separately by allocating one point to each correctly spelled variable. The same items of this task were reordered into a pseudo-word reading task in order to examine the relationship between spelling and decoding. Pseudo-word reading. The same above items were presented in a different order for reading. Scores were computed in two ways; first by allocating one point for each correctly read item. Maximum score for this task is 29, internal consistency (coefficient alpha) was .80. Then scores were computed for each variable separately by allocating one point to each correctly read variable. 4.3  Procedure Each Circassian L1 speaking child was tested during three sessions, one session devoted to each language, Arabic, Hebrew, and English. The second author (an ­English teacher and graduate student) together with a Circassian L1 speaking research assistant conducted the research. Each Hebrew L1 speaking child was tested during two sessions, one session devoted to Hebrew and one to English. For all children, testing took place at school in a quiet room. Instructions for both groups were given in their respective mother tongues in order to avoid misunderstandings. Each session lasted between 30–40 minutes and both languages and tests were rotated. The Ravens test of nonverbal intelligence was administrated to all children in small groups of 2–3 children at the end of one of the shorter sessions. 5.  Results The following section will include three subsections. First, bio-social background data, second base-line measure results and third, English orthographic and linguistic target items that were tested using pseudo-word decoding and spelling measures.



How do previously acquired languages affect acquisition of English as a foreign language 

5.1  Bio-social background data Table 2 presents the background data for bio-social and non-verbal ability for the two groups. There were no significant differences between the groups for any of the aforementioned variables. Table 2.  Means and standard deviations for bio-social background data for both groups Variables

χ2

t

Hebrew L1 speaking children (n = 46)

Circassian L1 speaking children (n = 30)

Age (years. months)

10.7 (0.51)

10.09 (0.18)

 .32

–

Gender (boys: girls)

24:22

13:17

–

.57

Nonverbal intelligence

68.3 (10.94)

67.6 (13.28)

–.23

–

Mother education

14.3 (4.54)

13.4 (1.85)

–.81

–

5.2  Base-line measures Phoneme awareness was tested in Hebrew, Arabic, and English for all children. Decoding and spelling were tested in Hebrew and Arabic for the Circassian L1 speaking children and in Hebrew for the Hebrew L1 speaking children (see Table 3). Table 3.  Comparison of Circassian and Hebrew L1 speaking children on English ­phoneme awareness and Arabic and Hebrew phonemic awareness, reading, and spelling presented as percentages Circassian L1 speaking children (n = 30)

Hebrew L1 speaking children (n = 46)

t

English phoneme awareness

97.8 (4.09)

84.7 (15.23)

4.6***

Hebrew word spelling

74.5 (18.26)

90.7 (9.17)

–5.11***

Hebrew word reading

84 (9.07)

93 (5.12)

–5.54***

Hebrew phoneme awareness

92.8 (10.64)

84 (14.93)

2.8**

Arabic word reading

42.3 (26.83)

Arabic word spelling

20.6 (20.86)

Arabic phoneme awareness

62.2 (23.84)

*p < .05, **p < .01, ***p < .001

Results show that the Circassian L1 speaking children outperformed the Hebrew L1 speaking children on the English phoneme awareness task where they reached ceiling effects. Concerning Hebrew, as could be expected, the Hebrew L1 speaking children performed significantly better than their Circassian L1 speaking peers on two

 Janina Kahn-Horwitz, Sara Kuash, Raphiq Ibrahim & Mila Schwartz

(Hebrew word reading and spelling) out of three measures. The relatively lower scores achieved by the Circassian L1 speaking children are understandable considering the fact that Hebrew is acquired as one of their second languages and literacies. Taking this constraint into account, one can still be impressed by the relatively good performance on Hebrew tasks by the Circassian L1 speaking children. The Circassian L1 speaking children scored lowest on the Hebrew spelling measure and this result could be attributed to the complexity of the Hebrew orthography for spelling, e.g. more than one grapheme representing numerous phonemes. At the same time, it is interesting to note that the Circassian L1 speaking children outperformed the Hebrew L1 speaking children on the Hebrew phoneme awareness task. This task reflects superior meta-linguistic awareness of the Circassian L1 speaking children which was expressed both in the results of the English and the Hebrew phoneme tasks. This meta-linguistic advantage amongst children with more than one language has been found in previous research (Bialystok 2001; Schwartz, Geva, Share & Leikin 2007: 25–52). Regarding Arabic, the Circassian L1 speaking children received low scores on all Arabic tasks, which could reflect the fact that Modern Standard Arabic is a second language acquired in parallel to Hebrew after L1, Circassian. In addition, objective challenges in acquiring Modern Standard Arabic, as opposed to the Hebrew orthography, include features such as the many letters that vary in shape according to word position (Ibrahim, Eviatar & Aharon Peretz 2007: 297–317). A final possible explanation for the Circassian L1 speaking children’s low scores on the Arabic measures could be due to limited usage of Modern Standard Arabic outside the school framework. As mentioned above, Modern Standard Arabic is the language used for religious purposes by the Circassian population in Israel. An interesting result is that despite the high performance of the Circassian L1speaking children on the English and Hebrew phoneme awareness tasks, these same children scored rather poorly on the Arabic phoneme awareness task. This result will be related to in the discussion. 5.3  English experimental measures For pseudo-word decoding, the Circassian L1 speaking children outperformed the Hebrew L1 speaking children and showed a significant advantage in almost all of the tests. The Hebrew L1 speaking children outperformed the Circassian L1 speaking children in the decoding of pseudo-words containing vowel digraphs (see Table 4). Regarding spelling of pseudo-words, the Circassian L1 speaking children outperformed the Hebrew L1 speaking children on all measures except for the short vowels and the silent 〈e〉 convention where there were no significant differences between the two groups (see Table 5).



How do previously acquired languages affect acquisition of English as a foreign language 

Table 4.  Comparison of Circassian and Hebrew L1 speaking children on English target pseudo-word decoding represented as percentages Variables (range)

Circassian L1 speaking children (n = 30)

Hebrew L1 speaking children (n = 46)

t

Eng. pseudo-word reading total score (0–29)

68.5 (12.40)

47.9 (22.62)

 4.56***

Eng. pseudo-word reading grapheme short 〈a〉 (0–5)

95.3 (12.52)

53.4 (27.98)

 7.69***

Eng. pseudo-word reading short vowel graphemes (0–14)

78.3 (14.46)

63.8 (18.89)

 3.58**

Eng. pseudo-word reading grapheme 〈ch〉 (0–5)

90.6 (26.64)

73.4 (35.79)

 2.25*

Eng. pseudo-word reading grapheme 〈th〉 (0–5)

87.3 (23.18)

34.4 (37.15)

 6.97***

Eng. pseudo-word reading vowel digraph (0–5)

52 (28.57)

68.6 (30.30)

–2.40*

Eng. pseudo-word reading silent 〈e〉 (0–5)

36 (21.27)

38.6 (33.83)

–.39

*p < .05, **p < .01, ***p < .001

Table 5.  Comparison of Circassian and Hebrew L1 speaking children on English target spelling represented as percentages Variables (range)

Circassian L1 speaking children (n = 30)

Hebrew L1 speaking children (n = 46)

Eng. pseudo-word spelling total score (0–29)

45 (13.08)

30.7 (19.79)

3.49**

Eng. pseudo-word spelling grapheme short 〈a〉 (0–5)

90 (20.17)

61.3 (31.73)

4.39***

Eng. pseudo-word spelling short vowels (0–14)

40.9 (17.48)

48.8 (24.31)

Eng. pseudo-word spelling grapheme 〈ch〉 (0–5)

90.6 (22.11)

66.6 (35.41)

3.30**

Eng. pseudo-word spelling grapheme 〈th〉 (0–5)

81.3 (29.68)

26.2 (35.37)

7.04***

Eng. pseudo-word spelling vowel digraph (0–5)

50.6 (25.58)

34.6 (25)

2.69**

Eng. pseudo-word spelling silent 〈e〉 (0–5)

30 (25.05)

22.2 (24.57)

1.33

**p < .05, **p < .01, ***p < .001

t

–1.54

 Janina Kahn-Horwitz, Sara Kuash, Raphiq Ibrahim & Mila Schwartz

In addition, it is interesting that the Circassian L1 speaking children outperformed the Hebrew L1 speaking children on spelling vowel digraphs, whereas for reading vowel digraphs, the Hebrew L1 speaking children outperformed the Circassian L1 speaking children. 6.  Discussion This study was inspired by an attempt to understand a case of a multilingual and multi-literate community in Israel which may further our understanding of the role of numerous languages and writing systems in acquiring a new language and writing system. The results of this study support the script dependent hypothesis and the linguistic and orthographic proximity hypothesis. Despite similarities between the groups regarding program of EFL instruction, hours of instruction, mother education, and non-verbal intelligence, results reveal that the group of Circassian L1 speaking children outperformed the group of Hebrew L1 speaking children and showed a significant advantage in decoding and spelling almost all of the orthographic conventions measured. Children in both groups were less successful at pseudo-word spelling compared with reading. This is reasonable since spelling is considered to be more difficult than reading. Spelling requires the retrieval rather than the recognition of graphemes (Perfetti, Rieben & Fayol 1997). Despite the difficulty of English spelling, the gap between Circassian L1 speaking children and Hebrew L1 speaking children remained to the advantage of the ­Circassian L1 speaking children. In line with the orthographic and linguistic proximity hypothesis, the Circassian L1 speaking children outperformed the Hebrew L1 speaking children on decoding and on spelling pseudo-words containing the short vowel /æ/ and on decoding the short vowels; /e/, /I/, /o/, /ʌ/. This may be explained by the fact that the Circassian language contains the short vowel /æ/ and the short vowel /ʌ/. Subsequently, they are exposed to the phonemes from birth. In addition, the Arabic language contains long vowels presented in writing as letters and short vowels presented in writing as diacritics (Fender 2008: 19–42). As previously mentioned, the Circassian children study the Arabic language from first grade. The short vowel /æ/ in English is considered to be a long vowel /a/ in Arabic. The Arabic language is taught in school to young Circassian L1 speaking children where the difference between the two vowels, /æ/ and /ʌ/, is stressed in writing and in pronunciation. The advantage demonstrated by the Circassian L1 speaking children in differentiating between long and short vowels may be a benefit from the facilitating effect of knowledge that was gained from the Arabic language whereby the difference between short and long vowel sounds was acquired. The disadvantage ­demonstrated



How do previously acquired languages affect acquisition of English as a foreign language 

by Hebrew L1 speaking children may be a result of only being exposed to the differences between long and short vowels when they start learning English. Despite the fact that the spelling of the short vowel /æ/ was at ceiling level among the Circassian L1 speaking children, the other short vowels clearly were challenging for these children in the same way as for their Hebrew L1 speaking peers. This is reasonable, as previously discussed; spelling is more difficult than reading (Perfetti, Rieben & Fayol 1997). A closer qualitative analysis of the spelling errors made by the Circassian children showed that they mainly confused the graphemes representing the vowels /ɪ/ and /e/ and that they misspelled the grapheme representing the vowel /ʌ/ by writing instead 〈a〉 /æ/. This may be a result of the complexity and richness of the vowel system of the English language and its many variations. Regarding consonant digraphs, the Circassian L1 speaking children outperformed the Hebrew L1 speaking children on reading and spelling pseudo-words containing the phoneme sequence /tʃ/, and phonemes /ð/, and /θ/. This result may be explained by the fact that the Circassian L1 speaking children were previously exposed to the phonemes representing the consonant digraphs 〈ch〉, and 〈th〉 in contrast to the Hebrew L1 speaking children. In addition, as was predicted, there were no significant differences between the two groups on reading and spelling the silent 〈e〉. It was a challenge for both groups and both groups obtained low scores. This result may be explained by the fact that when reading voweled Hebrew or Arabic, the Circassian L1 and Hebrew L1 speaking children are accustomed to acknowledging each vowel whether for pronunciation or spelling including vowels at the end of a word. Most of the children’s mistakes in reading the silent 〈e〉 were that they could not ignore the silent 〈e〉 and read it as a long vowel, for example, reading the word brile as /brili/. This might be because of the influence of previous studied languages, Arabic and Hebrew, where children mostly pronounce the phoneme that represents each letter when they decode. This is supported by results reported by Sun-Alperin and Wang (2011: 591–614) who argue that Spanish speakers, with no English oral vocabulary knowledge, would use the same letter-by-letter decoding strategy as could be used in Spanish. Therefore, their English word reading often would not result in real words because of the inconsistent mapping between graphemes and phonemes in English. Because the mapping from graphemes to phonemes in Spanish is relatively direct, knowing the sounds that map onto the letters is sufficient for proper pronunciation of words. However, as this is not the case with English’s more opaque orthography, knowledge of grapheme-phoneme often does not result in the accurate pronunciation of English words, as it happened with our Circassian and Hebrew L1 speaking children. It should be also noted that spelling the silent 〈e〉 is difficult not only for foreign language learners. The silent 〈e〉 is a

 Janina Kahn-Horwitz, Sara Kuash, Raphiq Ibrahim & Mila Schwartz

characteristic of English graphemic complexity and has been found to be challenging for English L1 speakers as well (Davis & Bryant 2006: 849–856). Interestingly, we found that the Hebrew L1 speaking children outperformed the Circassian L1 speaking children on decoding vowel digraphs (〈oo〉 and 〈ee〉); the Circassian L1 speaking children obtained relatively low scores for these graphemes. Examining the mistakes of the Circassian children reveals that they often read the vowel digraphs as single vowels; they read 〈oo〉 as the phoneme /o/ and 〈ee〉 as the phoneme /e/. This surprising finding occurred despite the fact that the orthographic representations of vowel digraphs are novel for both groups. Further research could possibly shed light on this finding. As opposed to the reading of vowel digraphs, Circassian L1 speaking children outperformed the Hebrew L1 speaking children on spelling them. A close look reveals that the Circassian children maintained their scores in both reading and spelling (around 50% for both) whereas the group of Hebrew speaking children scored significantly lower in spelling. This superiority could be explained by the fact that the Circassian L1 speaking children are more experienced in distinguishing between long and short vowels through their experience with the Arabic language. The pronunciation and spelling of short and long vowels is stressed in teaching Arabic. The Circassian L1 speaking children maintained their scores in spelling vowel digraphs as opposed to reading even though spelling is much more difficult. A possible explanation for this could be that in Arabic, Circassian L1 speaking children also have to write another consonant in order to signify the long vowel. As a result, they are used to signifying the long vowel by adding an additional grapheme. 7.  Conclusions This study showed clear evidence for an advantage regarding specific linguistic and orthographic characteristics amongst Circassian L1 speaking children over Hebrew L1 speaking children. The results of this study support the script dependent hypothesis (Geva & Siegel 2000: 1–30) and more broadly the linguistic and orthographic proximity hypothesis (Kahn-Horwitz, et al. 2011: 136–156). The former hypothesis claims that specific characteristics of first language orthographic ­structure affect second language literary acquisition due to cross-linguistic transfer. The latter hypothesis refers to both linguistic and orthographic characteristics in L1 impacting EFL literacy. Indeed, the results of this study show that the C ­ ircassian L1 speaking children outperformed the Hebrew L1 speaking children in decoding and/ or spelling specific target English graphemes that exist as phonemes in spoken Circassian and as phonemes or sometimes as graphemes in literate Arabic. Finally, it seems that previous literacy background per se is not a sufficient condition for cross-linguistic transfer facilitating literacy acquisition in a new orthography. This



How do previously acquired languages affect acquisition of English as a foreign language 

study illustrated that similar characteristics existing between previously acquired languages and orthographies and the current language and writing system being acquired is what facilitates this new acquisition highlighting multi-competence (Cook n.d.) from a phonological and orthographic perspective. With regard to implications for this field, the results of this study suggest that teachers in general, specifically English foreign language teachers, should be aware of and sensitive to the background knowledge of languages that children have. English teachers should stress similarities and differences between the target scripts and the phonemes and in this way use children’s knowledge that was acquired from previous learned languages and writing systems to enhance the learning process of the new language. The identification of orthographic patterns that are difficult for EFL children who come from different language backgrounds and which we have evidence for being challenging for L1 English children such as the silent 〈e〉 could dictate specific reading and spelling instructional emphases. In sum, teachers should assist multilingual and multi-literate children in exploring the literacy of the languages that they know as supporter languages due to typological proximity with English as a lingua franca. Teachers should bear in mind the strengths and the weaknesses of their multilingual and multi-literate children and teach them English accordingly. Regarding future research, it would be interesting to take the same groups of children when they get to high school and have them conduct the same tests and see if the same advantages remain or do they disappear or perhaps new advantages will be observed.

Acknowledgements This research was supported by a grant from Oranim Academic College of Education Graduate Department to the first and fourth authors. The data was collected by the second author in the framework of an M.Ed. extended final paper completed in the Languages Teaching Program, Oranim Academic College of Education.

References Allaith, Zainab A. & R. Malatesha Joshi (2011). Spelling performance of English consonants among students whose first language is Arabic. Reading and Writing 24: 1089–1110.

doi: 10.1007/s11145-010-9294-3

Asadi, Ibrahim, Michal Shany, Anat Ben-Simon & Raphiq Ibrahim (2011). Individual diagnostic tests in the assessment of learning disabilities in Arabic: Tests and Manual. Haifa: Haifa ­University Publications. Bialystok, Ellen (2001). Bilingualism in development. Cambridge: Cambridge University Press.  doi: 10.1017/CBO9780511605963

 Janina Kahn-Horwitz, Sara Kuash, Raphiq Ibrahim & Mila Schwartz Bram, Chen (1999). Circassian re-immigration to the Caucasus. In Shalva Weil (ed.), Routes and roots: Emigration in a global perspective, 205–222. Jerusalem: Magnes. Cenoz, Jasone & Fred Genesee (1998). Psycholinguistic perspectives on multilingualism and multilingual education. In Jasone Cenoz & Fred Genesee (eds.), Beyond bilingualism: ­Multilingualism and multilingual education, 16–32. Clevedon: Multilingual Matters. Cenoz, Jasone & Charlotte Hoffmann (2003). Acquiring a third language: What role does bilingualism play? International Journal of Bilingualism 7: 1–8.  doi: 10.1177/13670069030070010101 Cook, Vivian (n.d.). Multi-competence. Retrieved from http://homepage.ntlworld.com/vivian.c/ Writings/Papers/MCentry.htm Cummins, Jim (1978). Bilingualism and the development of metalinguistic awareness. Journal of Cross-Cultural Psychology 9: 131–149.  doi: 10.1177/002202217892001 Daniels, Peter T. & William Bright (eds.) (1996). The world’s writing systems. New York: Oxford University Press. Davis, Claire & Peter Bryant (2006). Causal connections in the acquisition of an orthographic rule: A test of Uta Frith’s developmental hypothesis. Journal of Child Psychology and ­Psychiatry 47(8): 849–856.  doi: 10.1111/j.1469-7610.2006.01597.x Deutsch, Avital (1994). Operating of attention systems in the field of the syntax in children with severe reading disabilities. Unpublished doctoral thesis [In Hebrew]. Jerusalem: The Hebrew University. Fender, Michael (2008). Spelling knowledge and reading development: Insights from Arab ESL learners. International Online Journal of Issues in Foreign Language Reading and Literacy 20: 19–42. Frost, Ram (2005). Orthographic systems and skilled word recognition processes in reading. In Margaret J. Snowling & Charles Hulme (eds.), The science of reading: A handbook, 272–295. Oxford: Blackwell.  doi: 10.1002/9780470757642.ch15 Geva, Esther & Linda S. Siegel (2000). Orthographic and cognitive factors in the concurrent development of basic reading skills in two languages. Journal of Reading and Writing 12: 1–30.  doi: 10.1023/A:1008017710115 Geva, Esther & Lesly A.Wade-Woolley (1998). Component processes in becoming ­English-Hebrew biliterate. In Aydin Y. Durgunoğlu & Ludo T. Verhoeven (eds.), Literacy development in a multilingual context: Cross-cultural perspectives, 85–110. Mahwah, NJ: Erlbaum. Gottardo, Alexandra, Bernice Yan, Linda S. Siegel & Lesly A. Wade-Woolley (2001). Factors related to English reading performance in children with Chinese as a first language. Journal of Educational Psychology 9(3): 530–542.  doi: 10.1037/0022-0663.93.3.530 Gut, Ulrike (2009). Cross-linguistic influence in L3 phonological acquisition. International Journal of Multilingualism 7: 19–38.  doi: 10.1080/14790710902972248 Ibrahim, Raphiq, Zohar Eviatar & Judith Aharon-Peretz (2007). Metalinguistic awareness and reading performance: A cross language comparison. The Journal of Psycholinguistic Research 36(4): 297–317.  doi: 10.1007/s10936-006-9046-3 Kahn-Horwitz, Janina, Mila Schwartz & David L. Share (2011). Acquiring the complex English orthography: A Tri-Literacy Advantage? Journal of Research in Reading 34: 136–156.

doi: 10.1111/j.1467-9817.2010.01485.x

Koda, Keiko (2008). Impacts of prior literacy experience on second-language learning to read. In Keiko Koda & Annette M. Zehler (eds.), Learning to read across languages: Cross-­ linguistic relationships in first-and second-language literacy development, 68–93. New York: Routledge.



How do previously acquired languages affect acquisition of English as a foreign language 

Mamser Batsag, Muhamad C. (2009). “‫( الموسوعة التاريخية لألمة الشركسية “ األديغة‬The historical series of the Circassian nation ‘The Adegh’)1: 363–402. Amman, Jordan. Olshtain, Elite & Frieda Nissim-Amitai (2004). Being trilingual or multilingual: Is there a price to pay? In Charlotte Hoffmann & Jehannes Ytsma (eds.), Trilingualism in family, school and community, 30–36. Great Britain: Cromwell Press. Perfetti, Charles A., Laurence Rieben & Michel Fayol (1997). Learning to spell: Research, theory, and practice across languages. Mahwah, NJ: Erlbaum. Raven, John C., Jean Raven & John H. Court (1976). Raven’s progressive matrices. Oxford: Oxford Psychologists Press. Rosner, Jerome (1975). Helping children overcome learning difficulties: A step-by-step guide for parents and teachers. New York: Walker and Company. Schwartz, Mila, Esther Geva, Mark Leikin & David L. Share (2007). Learning to read in English as L3: The cross-linguistic transfer of phonological processing skills. Written Language and Literacy 10(1): 25–52.  doi: 10.1075/wll.10.1.03sch Seymour, Phillip H.K., Mikko Aro & Jane M. Erskine (2003). Foundation literacy acquisition in European orthographies. British Journal of Psychology 94: 143–174.

doi: 10.1348/000712603321661859

Shany, Michal, Dorit Lachman, Tzila Shalem, Amira Bahat & Tali Seiger (2003). Reading test: Word recognition fluency (words-per-minute) and accuracy [in Hebrew]. Israel: Nitzan. Sun-Alperin, M. Kendra & Min Wang (2011). Cross-language transfer of phonological and orthographic processing skills from Spanish L1 to English L2. Journal of Reading and Writing 24: 591–614.  doi: 10.1007/s11145-009-9221-7 Venezky, Richard L. (1999). The American way of spelling: The structure and origins of American English orthography. New York: Guilford Press. Тхьаркъуахъо, Ю.А. (Yu Tcharkuacho) (2000). АдыгэБуквэмэЯшъэфхэр (The secrets of the Circassian alphabet) Maykop: MGTI (МГТИ).

Appendix 1 Comparison between the four languages regarding similarities and differences between target ­English linguistic and orthographic characteristics English

Circassian

Hebrew

Arabic

silent 〈e〉

phoneme x digraph x

phoneme x digraph x

phoneme x digraph x

consonant digraph 〈th〉

phoneme x digraph x

phoneme x digraph x

phoneme  digraph x

consonant digraph 〈sh〉

phoneme  digraph x

phoneme  digraph x

phoneme  digraph x

consonant digraph 〈ch〉

phoneme  digraph x

phoneme x digraph x

phoneme x digraph x

short vowel grapheme 〈a〉

phoneme  grapheme x

phoneme x grapheme x

phoneme  grapheme 

differentiation between long and short vowels



x







 Janina Kahn-Horwitz, Sara Kuash, Raphiq Ibrahim & Mila Schwartz

Appendix 2 Examples of English orthographic conventions embedded within pseudo-word decoding and spelling. Category

Example

1. the short vowel grapheme 〈a〉

chand

2. short vowels graphemes; 〈i〉, 〈o〉, 〈e〉, 〈u〉

fint, shog, jelp, snup

3. consonant digraphs; 〈ch〉 and 〈sh〉

fosh, chand

4. consonant digraph 〈th〉

theg

5. vowel digraphs; 〈oo〉 and 〈ee〉

stoon, theen

6. silent 〈e〉

jime

Bidirectional cross-linguistic relations of first and second language skills in reading comprehension of Spanish-speaking English learners Alexandra Gottardo1, Christine Javier1, Fataneh Farnia2, Lorinda Mak3 & Esther Geva2 1Wilfrid

Laurier University, Waterloo, Canada / of Toronto, Toronto, Canada / 3York University, Toronto, Canada

2University

This study examines the bidirectional, cross-linguistic associations between language and word-level reading skills and reading comprehension for 51 students in grades 4 to 6 who speak Spanish as first language (L1) and English as second language (L2). Within-language predictors of reading comprehension were consistent with the simple view of reading. We found unidirectional cross-linguistic associations between Spanish word reading and English reading comprehension. However, the results do not support a cross-linguistic association between English word reading and Spanish reading comprehension. Specifically, results indicate that although L1 and L2 language and reading constructs correlate, L2 skills do not strongly contribute to L1 reading comprehension. Findings are discussed in terms of possible factors that might influence potential cross-language relations among Spanish and English measures. Keywords:  reading comprehension; bilingualism; cross-linguistic relations

1.  Introduction The role of first language (L1) skills in second language (L2) achievement is often studied by researchers in order to assist learners in acquiring their L2. Higher levels of achievement in terms of speaking and reading in the L2 are considered crucial for the academic success of immigrant children as well as their future economic success. Moreover, recent studies have demonstrated the cognitive benefits of being bilingual (Bialystok 1999). These benefits are specific to active bilingualism in which the speaker maintains skills in the L1 while performing at high levels

doi 10.1075/bct.89.04got © 2016 John Benjamins Publishing Company

 Alexandra Gottardo, Christine Javier, Fataneh Farnia, Lorinda Mak & Esther Geva

in the L2 (Luk, De Sa & Bialystok 2011). Therefore, in addition to examining the influence of the L1 on the L2, it is important to examine the influence of the L2 on the L1. The present study examines cross-linguistic associations among L1 and L2 skills in relation to L1 and L2 reading comprehension. Specifically, L1 and L2 skills related to reading comprehension are examined within and across languages for Spanish-English speakers. 1.1  Context Greater globalization and migration have resulted in an increasing number of bilingual or multilingual individuals. In countries with large numbers of immigrants, the L1 might be a way of interacting with a relatively large segment of the population (e.g. Spanish in the United States). For example, media, customer service, and government services are available in Spanish in the USA (e.g. banking, legal consultation). In Canada, immigrants come from more than 100 countries. Therefore, provision of bilingual education is not possible. However, in contrast to the United States, Canada has an official multiculturalism policy that “… ensures that all citizens can keep their identities, can take pride in their ancestry and have a sense of belonging” (Citizenship and Immigration Canada 2013). Therefore, in urban centres, publically funded school districts offer heritage language programs after school or on the weekends. Additionally, Canadian immigration policy promotes the immigration of people with relatively high levels of education in their country of origin (Citizenship and Immigration Canada 2013). These immigrant parents are more likely to actively encourage educational enrichment activities for their children, one of these activities being supplemental lessons in the L1. This idea of bilingualism as enrichment is supported by research that showed that bilingualism is linked to a number of cognitive benefits. Individuals who maintain their L1 and simultaneously use their L2 in everyday life are reported to have greater cognitive flexibility and decreased effects of aging (Luk et al. 2011). These findings have attracted educators and researchers interested in the effects of maintaining L1 skills for children of immigrants who learn their L2 and additional languages at school. 1.2  Bilingualism Simply put, the term ‘bilingual’ can be widely defined as knowing two languages (Brutt-Griffler & Varghese 2004). This broad definition has led to several operational definitions of bilingualism (Bialystok 2001; Bloomfield 1933; Grosjean 1982, 1998). An additional challenge to defining the concept of bilingualism is the variability in levels of L1 and L2 skills in bilingual individuals (Bialystok 2001; Cook 2003; Gottardo & Grant 2008; Paradis, Genesee & Crago 2011). ­Understanding



Bidirectional cross-linguistic relations of first and second language skills 

bilingualism is important given that being bilingual is more common than being monolingual, when worldwide trends are examined (Grosjean & Li 2012). ­Historically, researchers suggested that young children who learned more than one language at a time would be at risk for poor school achievement (cf. Snow, Burns & Griffin 1998). This misconception is maintained by some laypeople. Although early research showed lower scores for bilingual children compared to their monolingual peers on both languages, these studies were flawed in that the bilingual children were often from first- or second-generation immigrant families from lower socioeconomic backgrounds with very low English proficiency (Hakuta 1988). Furthermore, all measures were administered in English rather than in their language of greatest proficiency and their performance was compared to norming samples largely composed of middle-class, monolingual ­English-speaking children (Hakuta 1988). These inappropriate comparisons make the performance of bilinguals seem ‘sub-standard’. Mounting evidence suggests a reciprocal relationship between first and second language skills (Cook 2003). However, even within groups of bilinguals, L1 and L2 experiences can differ. Therefore, to better understand the language skills of bilinguals, comparisons should be made between L2 groups or within groups of L2 learners across the languages that they speak. Additionally, contextual components related to bilingualism should be considered. The present study examines cross-linguistic relations in Spanish-English speakers in Canada. The Canadian experience offers an opportunity to examine language variables distinct from socio-economic status variables. At present, Canadian immigration policy selects immigrants based on a point system that rewards: high levels of education, work experience, arranged employment in Canada, and personal adaptability as well as encouraging investors and entrepreneurs (Citizenship and Immigration Canada 2013). As a consequence of this unique immigration policy, recent immigrants to Canada tend to be middle class professionals in their country of origin. Therefore, studies examining the English skills of immigrants in Canada allow researchers to examine the effects of being bilingual separately from the variable of socio-economic status. The participants in this study were from middle class families and showed relatively high levels of achievement in English, often scoring in the average range. Therefore, they were not considered to be ­at-risk for academic difficulties. 1.3  Cross-linguistic relations The most commonly discussed cross-linguistic relations are the influences of L1 on L2. This practical line of research has been conducted to determine whether immigrant children will succeed in L2 instruction and to identify learners who

 Alexandra Gottardo, Christine Javier, Fataneh Farnia, Lorinda Mak & Esther Geva

might require additional assistance to succeed. Additionally, educational research considers L1 loss to be inevitable. Therefore, preserving L1 skills that are not part of the dominant culture is not considered the domain of the educational system (Laufer 2003). Cummins (1984, 1991) proposed the Linguistic Interdependence Hypothesis, which stipulates that there is a common underlying proficiency across languages. That is, the proficiency developed in the L1 is a foundation for L2 proficiency. This underlying language proficiency, in turn, facilitates the transfer of literacy skills across languages. Linguistic theory has also examined relations between the L1 and L2 in an attempt to build theoretical models of bilingualism. These linguistic models suggest that because both languages are in one ‘mind’, they must interact. However, the degree and direction of overlap has been the subject of debate in theories of second language acquisition. For example, Cook (2003) has suggested that L1 and L2 relations are bidirectional and has provided evidence of L2 influences on the L1 in highly skilled users of each language. Studies have examined L2 effects on the L1 in the domain of oral language skills. These studies attempted to classify the domains in which the transfer occurs (Jarvis 2003; Porte 2003). For example, analyses of L1 oral language errors showed that L2 semantic and morpho-syntactic structures intruded when the adults spoke their L1 (Pavlenko 2003). The small number of studies that have examined effects of L2 reading on L1 reading have focused on word reading (Jared & Kroll 2001; Jiang 2004). In terms of literacy research, evidence examining relations between specific skills suggests that L2 literacy builds on existing levels of language abilities and literacy skills in the L1 (Durgunoğlu 2002; Geva & Wang 2001). Some skills such as lower level phonological skills and higher level comprehension skills are more likely to transfer (Durgunoğlu 2002; Gottardo, Yan, Siegel & Wade-Woolley 2001; Lindsey, Manis & Bailey 2003; Verhoeven 2007). Skills that deal with linguistic structures such as syntax, morphology, and vocabulary show differential levels of transfer based on whether the languages share features in common in terms of grammatical structures or lexical roots (Geva & Siegel 2000; Gottardo 2002; Pasquarella, Chen, Lam, Luo & Ramirez 2011; Ramirez, Chen, Geva & Kiefer 2010; Verhoeven 2007). These findings suggest a common process underlying literacy that is influenced by linguistic similarities. The directionality of relations among variables has not been elucidated. Alternately, the Competition Model posited by MacWhinney and colleagues examined oral L2 acquisition at the lexical level and suggested that L2 acquisition is parasitic on L1 knowledge (Hernandez, Li & MacWhinney 2005; Li, ­Farkas & MacWhinney 2004). Early in development, L1 lexical knowledge is highly interconnected within a language and is also connected to the concepts it represents. Early in L2 acquisition, L2 lexical items are linked to L1 lexical items. With



Bidirectional cross-linguistic relations of first and second language skills 

e­ xperience, L2 lexical items become more strongly interconnected and are solely activated when the L2 is used. This model implies that there are no strong bidirectional links between L1 and L2 in which L2 influences L1 processing. In terms of reading skills, the Competition Model would predict that L1 skills influence L2 skills, but L2 skills do not influence L1 skills. 1.4  Why reading comprehension? Language proficiencies are implicated in more cognitively demanding tasks such as reading comprehension. Evidence examining relations between specific skills suggests that L2 reading skills build on existing levels of language abilities and literacy skills in the L1 (Durgunoğlu 2002; Geva & Wang 2001). Despite the existence of theoretical models that show an ‘overlap’ between L1 and L2 skills in the lexicon, few studies have examined the relations between L2 oral language and literacy skills and L1 reading comprehension. In terms of reading comprehension, the majority of students in the late elementary grades have progressed to the stage of ‘reading to learn’ (Chall 1996). Therefore, the assumption is that skills related to L1 and L2 reading comprehension are linked (Durgunoğlu, Nagy & Hancin-Bhatt 1993; Gottardo, Yan, S­ iegel & Wade-Woolley 2001; Verhoeven 2007), less research has been conducted on cross-linguistic predictors of reading comprehension. See Lesaux, Siegel, Koda and ­Shanahan (2006) for a discussion on this topic. Walter (2007) questions the relevance of examining ‘transfer’ between L1 to L2 in terms of global reading comprehension skills. She suggests that the focus of research should be on underlying cross-linguistic cognitive processes in the aid of L2 reading comprehension. The current study examines bidirectional, cross-linguistic relations between reading comprehension and the key variables of decoding and vocabulary. One of the most commonly tested models of reading comprehension is the simple view of reading (Gough & Tunmer 1986). Using the simple view of reading to examine development of reading comprehension over time in monolingual speakers has shown that language comprehension plays an increasingly important role in reading comprehension with increasing experience (Catts, Hogan & Adlof 2005; Proctor, August, Carlo & Snow 2006; Proctor, Carlo, August & Snow 2005). In an attempt to update the simple view of reading, current research has focused on ‘unpacking’ language comprehension (Farnia & Geva 2013; Gottardo, Javier & Farnia 2013). Despite critiques regarding the simplicity of the model (Kendeou, Savage & van den Broek 2009; Kirby & Savage 2008; Ouellette & Beers 2010; Savage 2001), the simple view of reading includes key factors related to reading comprehension, specifically decoding and language comprehension. Additionally, the variable labeled

 Alexandra Gottardo, Christine Javier, Fataneh Farnia, Lorinda Mak & Esther Geva

‘language comprehension’ is deceptively simple, in that comprehension requires linguistic skills such as knowledge of semantics, syntax, and morphology as well as verbal memory ability and background knowledge. Other researchers have created more complex models of reading comprehension by including other cognitive variables such as memory or higher level comprehension skills (Cain, Oakhill & Bryant 2004) or by examining the role of text-based variables such as narrative structure on reading comprehension (Jenkins, Fuchs, van den Broek, Espin & Deno 2003; van den Broek, Lynch, Naslund, Ievers-Landis & Verduin 2003). Existing L2 research focuses on the effects of L1 on L2 reading comprehension (Gottardo & Mueller 2009; van Gelderen, Schoonen, Stoel, de Glopper & Hulstijn 2007) or on the development of reading comprehension skills in tandem for each language in young English learners (ELs) (Nakamoto, Lindsey & Manis 2007). In these cases, versions of the simple view of reading are modified slightly. For example, Proctor et al. (2005) found that L2 vocabulary knowledge contributed unique variance to reading comprehension beyond the core variables in the simple view of reading. In an augmented simple view of reading, Farnia and Geva (2013) showed that early and late vocabulary, syntax, and listening comprehension contributed uniquely to reading comprehension. They argued that the contribution of distinct early and late cognitive variables reflected developmental changes in reading and language comprehension in both English as L2 learners and native speakers of English. Nakamoto et al. (2007) found a significant contribution of word reading fluency to the simple view of reading. Based on the available sample size, the current study examined components of comprehension considered to be related to oral language proficiency, specifically L1 and L2 vocabulary and L2 grammatical knowledge as well as L1 and L2 decoding. These variables were examined to determine whether specific L1 and L2 skills overlap and if skills in one language are more predictive of reading comprehension than skills in the other language (Gottardo & Mueller 2009). 1.5  The role of L1 and L2 writing systems The study of cross-linguistic relations in reading introduces another important variable, the nature of the writing system, specifically the way sound-symbol and symbol-sound relations are represented. The properties of the writing system are believed to influence the strategies used to read the words (Ziegler & ­Goswami 2005). Some languages, such as Spanish, provide very regular and consistent letter-sound patterns. Although English has an alphabetic orthography, it has irregular letter-sound relations and a tendency to preserve morphological relations at the cost of letter-sound regularities (e.g. sign, signal) (Plaut, McClelland, ­Seidenberg & Patterson 1996). Additionally, the predictable letter-sound ­patterns



Bidirectional cross-linguistic relations of first and second language skills 

that exist in the English writing system often are accompanied by exceptions (Venezky 2004). However, researchers argue that despite its irregularities, readers eventually acquire high levels of word reading in English using either pattern recognition (Venezky 2004) or networks instantiating statistical probabilities (Plaut et al. 1996). The majority of literacy research that examines cross-linguistic effects on reading is conducted at the level of word reading skills. In these cases, L1 literacy in a writing system with regular letter to sound correspondences facilitates English literacy (DaFontoura & Siegel 1995; Durgunoğlu, Nagy & Hancin-Bhatt 1993). However, little research has been conducted examining effects of L2 on L1 reading specifically in relation to reading comprehension. The present study focused on reading comprehension in Spanish-English bilinguals. All of the students in this study were attending school in an English immersion setting. The present study examined within-language and cross-language relations between L1 and L2 word reading and oral language proficiency, and L1 and L2 reading comprehension using the simple view of reading as a framework. Unlike most research, the present study included relations between L1 and L2 skills and L1 reading comprehension. Finally, in addition to testing a variant of the simple view of reading, unique and overlapping L1 and L2 variance within the constructs of word reading and vocabulary are examined in relation to reading comprehension using commonality analyses. This method is used to determine cross-linguistic relations within constructs.

2.  Method 2.1  Participants Fifty-one students were recruited from a large metropolitan area in Canada. Parti­ cipants were recruited through community centres, churches, and local associations that served Latino families. The participants had a mean age of 11 years (SD = 1 year 1 month). There were 22 males (mean age of 11 years 1 month) and 29 females (mean age of 10 years 11 months) in the sample. The participants had lived in Canada for a mean of 6 years 3 months, but showed high variability in terms of time living in Canada (SD = 3 years 3 months). Therefore, all analyses controlled for time in Canada. The Hollingshead’s (1975) Four Factor Index of Social Status was used to determine socioeconomic status (SES), as it takes into account various factors when assessing one’s SES (Adams & Weakliem 2011). For the purposes of the current study, the seven-point Educational Scale and the nine-point Occupational

 Alexandra Gottardo, Christine Javier, Fataneh Farnia, Lorinda Mak & Esther Geva

Scale of the index were used in order to classify the levels of education and v­ arious occupations listed by the participants’ parents. The educational level of each parent was assigned a numerical code between 1 and 7 with higher numbers indicating a higher level of education. Occupations were assigned a numerical code between 1 and 9 with higher numbers indicating higher status occupations, based on the classifications outlined in the index. Both occupations in their native country and in Canada were considered, since recent immigrants may have difficulty obtaining employment in their fields and may need to upgrade their qualifications upon arriving in Canada (e.g. in the fields of medicine, law, or teaching). The scores on the Hollingshead’s Index range from 8 to 66 with higher scores assumed to correspond to higher status in society (Adams & Weakliem 2011). In the present study, the mean estimation of current socio-economic status was 46.00 (SD  =  12.9). Using this index, status estimations fell between 40 and 54, which signifies that the parents were engaged in “medium business, minor professional and technical” work (Adams & Weakliem 2011: 46). This category suggests that the families in the present study are part of the upper-middle class, the second highest of five social strata. 2.2  Measures Nonverbal reasoning. The Matrix Analogies Test- Expanded Form (MAT-EF: Naglieri 1985) was used as a measure of nonverbal cognitive ability. This test is considered to be culture free and was used to determine whether the participants’ nonverbal reasoning skills were within the average range. Nonverbal reasoning is considered the best estimate of intelligence in L2 learners, who might struggle with higher level linguistic tasks in their L2. Participants were shown patterns and asked to identify which picture best fits with or completes each pattern. A raw score out of 16 was obtained for each of the four subtests; Pattern Completion, Reasoning by Analogy, Serial Reasoning, and Spatial Visualization as well as a total raw score out of 64. The MAT-EF demonstrates a median internal consistency of .83 and a median test-retest reliability of .78 (Naglieri 1985). L1 and L2 Reading. L1 and L2 word reading as well as reading comprehension were assessed using standardized measures in English and standardized measures in Spanish. Word reading. The Word Identification subtest from the Woodcock Reading Mastery Test-Revised (WRMT-R NU; Woodcock 1998) was administered. Participants were asked to read the words out loud from a list. The words increased in difficulty and testing stopped when the participant made six consecutive errors at the end of an easel page. Raw scores and standard scores were calculated as per the manual. This subtest has an internal consistency of .92 (Woodcock 1991).



Bidirectional cross-linguistic relations of first and second language skills 

In Spanish, the Identificacíon de letras y palabras subtest of the Batería III Woodcock-Muñoz (Woodcock & Muñoz-Sandoval 1995) was used as a measure of word reading in the L1. In this task, participants were asked to identify a series of letters and then words of increasing difficulty. Raw scores were calculated and then were converted into standard scores. This test has a mean of 100 and a standard deviation of 15. The split-half reliability using a Spearman-Brown correction is .91 (Woodcock & Muñoz-Sandoval 1995). Reading comprehension. The Neale Analysis of Reading Ability (NARA; Neale 1999) was used as a measure of reading comprehension in English. Participants were asked to read passages out loud and then answer comprehension questions for each. There were a total of six passages, however the total number of passages administered to each participant varied depending on the participant’s reading ability and age for a maximum of 44 questions. The passages increased in difficulty and testing was discontinued when the participant exceeded the maximum number of allowable errors for a given passage. Three scores were obtained for each passage: a comprehension score, an accuracy score, and a reading speed score. However, only the total reading comprehension score was used in the analyses. The NARA has a reliability coefficient of .84 for ages 10 to 12 (Neale 1999) and is known to be a valid measure of reading comprehension (Nation & Snowling 1997; Savage 2001). Reading comprehension was measured in Spanish using the Comprensión de textos, text comprehension subtest of the Batería III Woodcock-Muñoz (­Woodcock & Muñoz-Sandoval 1995). For this test, the participants read the passages silently and completed a sentence using a cloze format. For the easiest passages, the participant was required to select the picture that matched the statement being read. The passages increased in the length and complexity in terms of the target sentences as well as the sophistication of the topics being covered. Testing was discontinued when the participant has made six consecutive errors. Reported reliabilities for English-speaking norms are .88 for age nine (internal consistency reliability coefficients; Woodcock 1991). L1 and L2 vocabulary. Measures of receptive vocabulary were administered in L1 and L2. Previous research with English L2 learners suggests that, “expressive abilities tend to decline abruptly, but receptive […] abilities in L1 are more robust” (Oller, Jarmulowicz, Pearson & Cobo-Lewis 2011: 100). The Peabody Picture Vocabulary Test-Third Edition (PPVT-III; Dunn & Dunn 1997) was administered as a measure of English receptive vocabulary. The participant chose the picture that best described the orally presented word. Raw scores were used in the analyses. The PPVT-III has a split-half reliability that ranges from .86 to .97, with test-retest reliability ranging from .91 to .94 (Dunn  & Dunn 1997).

 Alexandra Gottardo, Christine Javier, Fataneh Farnia, Lorinda Mak & Esther Geva

The Test de Vocabulario en Imagenes Peabody (TVIP; Dunn, Lugo, Padilla & Dunn 1986) was administered as a receptive measure of Spanish vocabulary. The TVIP is the Spanish equivalent of the Peabody Picture Vocabulary Test-Revised (PPVT-R; Dunn & Dunn 1981). Raw scores were used for the analyses. This measure is norm-referenced and has a split-half reliability of .93 (Dunn et al. 1986). Syntactic knowledge. The sentence assembly subtest of the Clinical Evaluation of Language Fundamentals – 4 (CELF-4: Semel, Wiig & Secord 2003) was administered as a measure of syntactic knowledge. The participants were required to reorder stimuli, which were grouped as individual words and word pairs, to formulate two sentences. The sentences must be grammatically correct and semantically meaningful. The randomized stimuli were presented orally and in written form to the participants. This measure has a test-retest reliability of .82. Demographic information. As mentioned above, a questionnaire was completed by the parents asking about parental education and parental occupation. Parents were asked about the language used by the child at home, the child’s age of arrival in Canada, and the child’s educational experience in the L1. Questionnaires were provided in Spanish and in English and parents were allowed to select the language with which they were most comfortable. 2.3  Procedure Participants were tested in English and in Spanish on separate days. The order of testing was counterbalance by language, with some students being tested first in English and others being tested first in Spanish. English testing was conducted by a native English speaker while Spanish testing was conducted by two research assistants, one a native Spanish speaker and one a highly proficient Spanish speaker who was a Spanish instructor for university level courses. 3.  Results Initially, group means were calculated. Then a series of correlational analyses was conducted to determine general relationships across measures and to determine possible candidates for the regression analyses. Subsequently, four sets of regression analyses were conducted to determine within and cross-language relations between the cognitive-linguistic variables and reading comprehension. Finally, commonality analyses were conducted to determine the unique and shared variance of the constructs of word reading and vocabulary knowledge across the L1 and L2 in explaining variance in L1 and L2 reading comprehension.



Bidirectional cross-linguistic relations of first and second language skills 

3.1  Performance on L1 and L2 measures Because many of the measures were standardized, the mean raw scores and mean standard scores are presented. Standard scores can provide a general picture of proficiency in a given language for these bilinguals. The scores should be interpreted with caution as the children may not have had the same experiences as monolinguals in each language. However, if the participants score within the average range on any measure, it confirms that these children are not at-risk for school difficulty in that domain. As seen in Table 1, the standard scores for word reading, vocabulary, and reading comprehension in English were within the average range. The lowest mean score on the English measures was on the CELF-4 measure of syntactic knowledge, mean SS = 89.70, and was still broadly within the average range. For the Spanish measures, mean standard scores were more than one standard deviation above the mean for word reading, mean SS = 121.20. The scores were much lower for reading comprehension, with the mean score being one standard deviation below the mean with a mean SS = 85.37. Table 1.  Mean scores and group comparisons for variables  

Total (N = 51) Mean

SD

132.04

14.23

6.14

3.17

MAT-raw MAT-SS

40.67 97.37

10.38 12.97

NARA comprehension raw

26.65

8.00

Word ID-raw Word ID-SS

72.78 100.40

9.53 10.36

PPVT-raw PPVT-SS

140.06 104.55

22.84 17.00

Sentences CELF4 raw Sentences CELF4 SS

7.69 89.70

4.65 16.73

WLPB-R comprehension raw

18.31

6.06

47.24 121.20

8.18 24.45

88.04

18.40

Age (months) Time in Canada (years)

Spanish Word reading raw Spanish Word reading-SS TVIP-raw

3.2  Relations within and between languages: Partial correlations Partial correlations were calculated for the variables in the study, controlling for age and length of time in Canada (see Table 2). Reading comprehension in English was moderately to highly correlated with all of the key variables in English and

 Alexandra Gottardo, Christine Javier, Fataneh Farnia, Lorinda Mak & Esther Geva

Spanish, specifically word reading, vocabulary, and sentence assembly, rs = .46 to .72, ps < .01. Spanish reading comprehension was correlated with English word reading, Spanish word reading, and vocabulary, rs = .35 to .62, ps < .05. Moderate correlations were found within constructs for word reading and oral language skills, rs = .41 & .52, ps < .01, when examined across languages. Table 2.  Partial correlations among all English and Spanish measures with the effect of age and length of time in Canada partialled out (n = 49) Tasks

1

2

1. MAT

1

.29*

2.  English RC 3.  English Word ID 4.  English PPVT 5.  English SA 6.  Spanish RC 7.  Spanish Word ID

3

4

5

–.03

.28

.06

.72***

.63***

.28

.40**

.30*

1

1

.57*** 1

6

7

   8

.24

.26

.49**

.46**

.35*

.41**

.36*

.54***

.24

.34*

.52***

1

.20

.32*

.22**

.54***

.62***

1

.61***

–.08

1

8.  Spanish TVIP

1

Note: MAT = Matrix Analogies Test; RC = Reading Comprehension (NARA comprehension raw scores in English, WLPB comprehension raw scores in Spanish); Word ID = Word Identification; SA = CELF – Sentence Assembly *p < .05; **p < .01; ***p < .001

3.3  W  ithin-language relations, L2 to L2 & L1 to L1: Variables related to the simple view of reading Two sets of hierarchical regression analyses were conducted to determine the withinlanguage relations between word reading and oral language proficiency (OLP) in Spanish or in English with respect to Spanish or English reading comprehension skills. English reading comprehension was measured by the NARA, while Spanish reading comprehension was measured by the passage comprehension subtest of the WLPB-R. For each of the variables, age and the number of months in Canada were entered as the first step as control variables. For the analyses using English predictors, scores on the PPVT-III measure of vocabulary knowledge and the sentence assembly subtest of the CELF-4 were entered in one step as measures of oral language proficiency. Word reading scores as measured by the Word Identification subtest of the WRMT-R were also entered as a separate step. The order of the final two steps was varied to determine variables uniquely related to reading comprehension. Word reading in English was related to reading comprehension in English when all of the variables were entered in the equation, β = .312, t = 3.45, p < .01 (see Table 3). Additionally, the components of English OLP, specifically sentence



Bidirectional cross-linguistic relations of first and second language skills 

assembly from the CELF-4 and PPVT-III, were unique predictors of English reading comprehension when entered as the final step, β = .271, t = 2.75, p < .01, and β = .442, t = 4.08, p < .001, respectively. Table 3.  Within language relations L2 to L2: English variables related to English reading comprehension Variable

ΔR2

Beta

Final β

1. Age

.108

.331*

.082

.043

–.064

   Time in Canada (mos) 2.  Word Identification

.301

3. OLP

.309

.585***

.312**

   PPVT-III

.442***

   CELF-4 SA

.271**

2. OLP

.540

   PPVT-III

.549***

.442***

   CELF-4 SA

.306**

.271**

3.  Word Id.

.070

.312**

For the analyses, using Spanish predictors of Spanish reading comprehension, scores on the TVIP measure of vocabulary knowledge were entered as the measure of oral language proficiency. Word reading scores from the identification of letters and sounds subtest of the WLPB-R were entered as a separate step. The order of the final two steps was varied to determine variables uniquely related to reading comprehension. Spanish word reading was related to Spanish reading comprehension when it was entered as the second but not as the final step in the equation (see Table 4). Spanish vocabulary as measured by the TVIP was a unique predictor of Spanish reading comprehension, β = .431, t = 3.16, p < .01. 3.4  C  ross-language relations, L1 to L2 & L2 to L1: Variables related to the simple view of reading Two sets of hierarchical regression analyses were conducted to determine the cross-linguistic relations between word reading and OLP specifically from the L1, Spanish, to the L2, English reading comprehension, as well as from the L2, ­English, to the L1, Spanish reading comprehension. Measures of English and Spanish reading comprehension are the same as described above. For each of the variables, age and the number of months in Canada were entered as control variables. For the analyses using Spanish predictors, Spanish letter-word

 Alexandra Gottardo, Christine Javier, Fataneh Farnia, Lorinda Mak & Esther Geva

Table 4.  Within language relations L1 to L1: Spanish variables related to Spanish reading comprehension Variable

ΔR2

1. Age

.223

   Time in Canada (mos) 2.  Word reading

.227

3. OLP

.102

Beta

Final β

.122

–.073

–.443**

–.271*

.537***

.262

   TVIP 2. OLP

.431** .295

   TVIP 3.  Word reading

.582***

.431**

.034

.262

i­dentification, WLPB-R, and Spanish vocabulary, TVIP, were measured. The order of the final two steps was varied to determine variables uniquely related to reading comprehension (see Table 5). Analyses examining the role of Spanish predictors showed that Spanish word reading was uniquely related to English reading comprehension, β = .352, t  =  2.04, p < .05. Although Spanish vocabulary was related to English reading comprehension as the second step, it was not uniquely related when entered as the final step. Table 5.  Cross language relations L1 to L2: Spanish variables related to English reading comprehension Variable

ΔR2

Beta

Final β

1. Age

.108

.331*

.137

.043

.218

.538***

.352*

   Time in Canada (mos) 2.  Word reading

.222

3. OLP

.040

   TVIP 2. OLP

.281 .205

   TVIP 3.  Word reading

.494*** .057

.095 .047*

For the analyses using English predictors, age and the number of months in ­Canada were entered as control variables. Scores on the PPVT-III and the sentence assembly subtest of the CELF-4 were entered in one step as measures of oral language proficiency. Word reading as measured by the Word Identification subtest of



Bidirectional cross-linguistic relations of first and second language skills 

the WRMT-R was also entered as a separate step. The order of the final two steps was varied to determine variables uniquely related to reading comprehension (see Table 6). However, only length of time in Canada was a unique predictor of Spanish reading comprehension, β = –.457, t = –3.54, p < .001. The negative values indicate that a shorter time in Canada, specifically being a new immigrant, was related to better Spanish reading. None of the English linguistic variables emerged as a significant predictor of Spanish reading comprehension, although English word identification showed a trend using non-traditional levels of significance, β = .280, t = 1.94, p = .059. Table 6.  Cross language relations L2 to L1: English variables related to Spanish reading comprehension Variable

ΔR2

1. Age

.223

   Time in Canada (mos) 2.  Word Identification

.097

3. OLP

.011

Beta

Final β

.122

.013

–.443**

–.473**

.329*

.280

   PPVT-III

.081

   CELF-4 SA

.056

2. OLP

.050

   PPVT-III

.174

.081

   CELF-4 SA

.088

.056

3.  Word Id.

.059

.280

3.5  C  ross-language relations within constructs: Word reading and vocabulary Commonality analyses were used to examine the common variance for key constructs across the two languages in relation to reading comprehension. Commonality analyses examined the contributions of predictor variables by partitioning the total variance of the criterion variable, L1 or L2 reading comprehension, into the mutually exclusive components of the unique contribution of each individual predictor variable and the common effects or contributions of combinations of the predictor variables (Mood 1971; Nimon & Reio 2011; Warne 2011). To limit the number of variables in the commonality analyses, only the two control variables, age and months in Canada, and within-construct measures of the word reading and vocabulary across L1 and L2, were used in the analyses. Separate analyses were conducted to determine relations within the target construct across languages.

 Alexandra Gottardo, Christine Javier, Fataneh Farnia, Lorinda Mak & Esther Geva

L1 and L2 word reading and reading comprehension. Commonality analyses were carried out to examine the unique and common contributions of Spanish and English word reading to Spanish and English reading comprehension. The commonality coefficients of all possible combinations of the predictor variables, age, months in Canada, English word reading, and Spanish word reading, and their percentages of the total variance in English reading comprehension are displayed in Table 7. The majority of the regression effect for English reading comprehension was explained by variance that was unique to English word reading, 31.1%, although Spanish word reading explained 14.7% of unique variance. The ­common Table 7.  Commonality coefficients for English reading comprehension with age, years of Canadian residency, English word reading, and Spanish word reading (N = 51) Variance component

U1

Predictor variable Age (% Total)

CAN residency (% Total)

ENG WR (% Total)

.005 (.99)

U2

.016 (3.42)

U3

.149 (31.07)

U4

.070 (14.61)

C12

–.001 (–.154)

C13

.022 (4.63)

C14

.009 (1.87)

–.001 (–.154) .022 (4.63) .009 (1.87)

C23

.029 (5.98)

C24

–.016 (–3.33)

C34

.029 (5.98) –.016 (–3.33) .153 (31.84)

C123

.001 (.22)

.001 (.22)

C124

.000 (.07)

.000 (.07)

C134

.072 (15.02) .072 (15.02)

C234 C1234

SPA WR (% Total)

–.002 (–.51)

.153 (31.84)

.001 (.22) .000 (.07) .072 (15.02)

.072 (15.02)

–.027 (–5.69)

–.027 (–5.69)

–.027 (–5.69)

–.002 (–.51)

–.002 (–.51)

–.002 (–.51)

Unique

.005

.016

.149

.070

Common

.101

–.016

.247

.395

Total

.106

.000

.395

.258

Note: criterion variable = NARA comprehension raw score; C12 = common component of predictor variable 1 and 2, etc. Full model R2 = .479



Bidirectional cross-linguistic relations of first and second language skills 

effects between English word reading and Spanish word reading also yielded an important contribution to English reading comprehension as they accounted for 31.8% of the total variance. The common effects of English and Spanish word reading and age also accounted for 15.0% of the total variance. As a whole, the model that included all four predictors accounted for 47.9% of the variance in English reading comprehension. The results of the commonality analysis examining contributions of the control variables and English and Spanish word reading to Spanish reading comprehension are included in Table 8. The majority of the unique variance was explained by Spanish word reading, 31.2%. The common effects of English and Spanish word reading, and Spanish word reading and months in Canada accounted for 17.4% Table 8.  Commonality coefficients for Spanish reading comprehension with age, years of Canadian residency, English word reading, and Spanish word reading (N = 51) Variance component

U1

Predictor variable Age (% Total)

CAN residency (% Total)

ENG WR (% Total)

SPA WR (% Total)

.002 (.49)

U2

.067 (14.27)

U3

.016 (3.52)

U4

.146 (31.23)

C12

–.001 (–.20)

C13

–.002 (–.40)

C14

–.002 (–.44)

–.001 (–.20) –.002 (–.40) –.002 (–.44)

C23

–.011 (–2.26)

C24

.139 (29.87)

C34

–.011 (–2.26) .139 (29.87) .081 (17.36)

C123

.001 (.205)

.001 (.205)

C124

.006 (1.28)

.006 (1.28)

C134

.016 (3.49)

.081 (17.36)

.001 (.205) .006 (1.28) .016 (3.49)

.016 (3.49)

–.002 (–.41)

–.002 (–.41)

–.002 (–.41)

.009 (1.99)

.009 (1.99)

.009 (1.99)

.009 (1.99)

Unique

.002

.067

.016

.146

Common

.028

.142

.093

.248

Total

.030

.209

.110

.364

C234 C1234

Note: criterion variable = WLPB comprehension raw score; WR = word reading; C12 = common component of predictor variable 1 and 2, etc. Full model R2 = .467

 Alexandra Gottardo, Christine Javier, Fataneh Farnia, Lorinda Mak & Esther Geva

and 29.9%, respectively. Together, the four predictors accounted for 46.7% of the unique variance in Spanish reading comprehension. L1 and L2 vocabulary and reading comprehension. Commonality analyses were carried out to examine the unique and common contributions of L1 and L2 vocabulary to English and Spanish reading comprehension. The commonality coefficients of all possible combinations of the predictor variables, age, months in Canada, English vocabulary, and Spanish vocabulary, and their percentages of the total variance in English reading comprehension are reported in Table 9. The majority of the regression effect was explained by variance that was unique to English vocabulary, 47.9%. The common effects between English vocabulary and Spanish vocabulary yielded an important contribution to English reading Table 9.  Commonality coefficients for English reading comprehension with age, years of Canadian residency, English vocabulary, and Spanish vocabulary (N = 51) Variance component

U1

Predictor variable Age (% Total)

CAN residency (% Total)

ENG Vocab (% Total)

.009 (1.49)

U2

.007 (1.45)

U3

.288 (47.88)

U4

.015 (2.45)

C12

.002 (.26)

C13

.016 (2.74)

C14

.007 (1.12)

.002 (.26) .016 (2.74) .007 (1.12)

C23

.011 (1.78)

C24

.011 (1.87)

C34

.011 (1.78) .11 (1.87) .191 (31.74)

C123

–.004 (–.59)

–.004 (–.59)

C124

.006 (1.02)

.006 (1.02)

C134

.076 (12.58)

C234 C1234

SPA Vocab (% Total)

–.006 (–.996)

.191 (31.74)

–.004 (–.59) .006 (1.02) .076 (12.58)

.076 (12.58)

–.027 (–4.49)

–.027 (–4.49)

–.027 (–4.49)

–.006 (–.996)

–.006 (–.996)

–.006 (–.996)

Unique

.009

.007

.288

.015

Common

.097

–.007

.257

.257

Total

.106

.000

.545

.272

Note: criterion variable = NARA comprehension raw score; C12 = common component of predictor variable 1 and 2, etc. Full model R2 = .601



Bidirectional cross-linguistic relations of first and second language skills 

c­omprehension as it accounted for 31.7% of the total variance. The common effects of English and Spanish vocabulary and age also accounted for 12.6% of the total variance. Together, the four predictors accounted for 60.1% of the variance in English reading comprehension. The commonality coefficients for the predictor variables, age, months in ­Canada, English vocabulary and Spanish vocabulary, and their percentages of the total variance in Spanish reading comprehension were also calculated (see Table 10). The majority of the regression effect was explained by variance that was unique to Spanish vocabulary, 48.9%. The common effects between length of time in Canada and Spanish vocabulary yielded an important contribution to ­Spanish reading comprehension as it accounted for 30.5% of the total variance. The Table 10.  Commonality coefficients for Spanish reading comprehension with age, years of Canadian residency, English vocabulary, and Spanish vocabulary (N = 51) Variance component

U1

Predictor variable Age (% Total)

CAN residency (% Total)

ENG Vocab (% Total)

SPA Vocab (% Total)

.001 (.27)

U2

.070 (13.40)

U3

.007 (1.34)

U4

.257 (48.91)

C12

–.001 (–.22)

C13

.001 (.12)

C14

.003 (.62)

–.001 (.22) .001 (.12) .003 (.62)

C23

.036 (6.80)

C24

.160 (30.52)

C34

.036 (6.80) .160 (30.52) .038 (7.19)

C123

–.000 (–.01)

–.000 (–.01)

C124

.020 (3.74)

.020 (3.74)

C134

.009 (1.78)

.038 (7.19)

–.000 (–.01) .020 (3.74) .009 (1.78)

.009 (1.78)

–.073 (–13.87)

–.073 (–13.87)

–.073 (–13.87)

–.003 (–.60)

–.003 (–.60)

–.003 (–.60)

–.003 (–.60)

Unique

.001

.070

.007

.267

Common

.029

.138

.007

.154

Total

.030

.209

.015

.411

C234 C1234

Note: criterion variable = WLPB comprehension raw score; C12 = common component of predictor variable 1 and 2, etc. Full model R2 = .525

 Alexandra Gottardo, Christine Javier, Fataneh Farnia, Lorinda Mak & Esther Geva

c­ ommon effects between English vocabulary and Spanish vocabulary accounted for 7.2% of the total variance, while the common effects of English and Spanish vocabulary and length of time in Canada accounted for –13.9% of the total variance. The negative number is likely the result of statistical suppression where the common variance between Spanish vocabulary and English vocabulary is higher than the common variance between Spanish vocabulary and English reading comprehension and is lower than the common variance between English vocabulary and reading comprehension. Together, the four predictors accounted for 52.5% of the variance in English reading comprehension. 4.  Discussion The present study examined cross-linguistic relations between oral language proficiency and decoding variables and reading comprehension in Spanish-English speakers. It is important that the study results be interpreted in the context of this particular sample. Specifically, these students were achieving well on the English measures. The students had scores in the average range for most of the English measures, with the lowest scores on English grammatical knowledge. These scores are particularly impressive given that many of the students had only been in Canada for approximately 3½ years. The demographic information provided shows that the participants were from an upper middle class background, which likely contributes to their relatively high scores in the areas of English vocabulary, word reading, and reading comprehension. In terms of the variables related to reading comprehension across and between languages, the majority of the analyses support current theories of reading comprehension based primarily on research with monolingual English speakers. Specifically, in relation to the simple view of reading, the key components related to reading comprehension are word reading and variables related to language comprehension, in this case vocabulary and grammatical knowledge. 4.1  Within-language relations and English reading comprehension The key variables of decoding, vocabulary, and grammatical knowledge are related to English reading comprehension in the current study. These findings are similar to those reported for monolingual English speakers in grades 2 to 7 (Catts et  al. 2005), where oral language skills gradually explained more unique variance in reading comprehension of the older participants than did word level reading. Although decoding explained unique variance in reading comprehension in the earlier grades for monolinguals, it only explained shared variance with oral language in the later grades in Catts and colleagues’ study. Presumably, the



Bidirectional cross-linguistic relations of first and second language skills 

­ articipants of the current study are in a transitional stage and both decoding and p oral language proficiency are related to reading comprehension. Even though the participants span the middle and upper age ranges for the Catts et al.’s study, the predictive pattern that includes both oral language proficiency and decoding in the present study might be related to the students’ English L2 status. Previous research with younger Spanish-speaking ELs in second grade showed that both L2 word reading and L2 oral language proficiency were related to L2 reading comprehension (Gottardo & Mueller 2009). This relationship was considered to be the result of a bottleneck at the levels of decoding and oral language proficiency even for relatively ‘easy’ texts. The results of the present study do not clarify if differences in the variables related to reading comprehension in ELs in the later elementary grades reflect a slight delay in the development of relations among processes. Alternately, the different patterns could reflect a difference in processing with more resources and therefore greater processing capacity being required for L2 reading comprehension throughout the development of reading comprehension skills (Verhoeven & van Leeuwe 2008). 4.2  Within-language relations and Spanish reading comprehension The results are somewhat different for the within-language analyses examining the Spanish variables related to Spanish reading comprehension. In this case, only Spanish vocabulary is related to Spanish reading. This finding is probably due to two factors. One factor is the higher variability on Spanish oral language proficiency measures. This variability may be the result of the scores of the students with lower Spanish proficiency, who had been in Canada for longer periods of time. The second factor is related to the lack of contribution of Spanish decoding to Spanish reading comprehension. This factor may be due to the ceiling scores on Spanish decoding resulting from the transparency and regularity of the language. Consequently, even students with relatively poorer Spanish ability, as measured by Spanish vocabulary and reading comprehension, had high Spanish decoding skills. Therefore, the bottleneck in relation to Spanish reading comprehension was in the area of vocabulary, not decoding. This pattern is similar to findings with monolingual English speakers who have mastered English decoding and are limited by listening comprehension skills (Braze, Tabor, Shankweiler & Mencl 2007; Catts et  al. 2005). However, in the case of Spanish, decoding can be mastered quickly and easily even when other linguistic skills are weaker. 4.3  Cross-language relations: Variables related to the simple view of reading Cross-language relations between the variables and reading comprehension showed a different pattern of results. Only one L1 variable was related to L2 ­reading

 Alexandra Gottardo, Christine Javier, Fataneh Farnia, Lorinda Mak & Esther Geva

c­ omprehension, specifically decoding. None of the L2 variables were related to L1 reading comprehension. Examining unique and overlapping relations across languages but within constructs yielded some interesting findings. Specifically, when using L1 and L2 vocabulary to predict L2 reading comprehension, the majority of variance was explained by English vocabulary and the overlapping variance between English and Spanish vocabulary. Similar results were found when L1 and L2 word reading were examined in relation to L2 reading comprehension. In contrast, the majority of variance in L1 reading comprehension was explained by Spanish vocabulary and the overlapping variance between L1 vocabulary and length of time in Canada. Greater time in Canada was related to lower Spanish vocabulary skills, suggesting language loss (e.g. Grosjean 1998; Guardado 2002). When using word reading as predictors of Spanish reading comprehension, ­Spanish word reading was a unique predictor, as was its shared variance with English word reading and length of time in Canada. The findings of these different analyses are consistent with and link to two bodies of research. The findings of L1 to L2 cross-language relations for lower level processes, specifically alphabetic decoding skills, are consistent with the work of other reading researchers (Durgunoğlu 2002; Verhoeven 2007). Similarly, the lack of transfer at the level of linguistic structures such as lexical knowledge is similar to previous research with children who are ELs (Durgunoğlu 2002; Kroll & de Groot 1997). Additionally, the lack of L2 to L1 relations is supported by ­MacWhinney’s Competition Model (2004) that states that L2 skills are parasitic on L1 skills. No bidirectional transfer was found in the current study when only L2 variables were examined as predictors of L1 reading comprehension, particularly when oral language proficiency was examined as a predictor. Although L2 to L1 cross-linguistic relations were not found for the ‘gross’ measures of vocabulary and grammatical knowledge, it is possible that cross-linguistic effects exist at finer grained levels. For example in terms of lexical knowledge, L1 to L2 effects have been found for cognates or related words (Carlo, August, McLaughlin, Snow, Dressler, Lippman, Lively & White 2004; Jiang 2004). Similar results might be found for the use of cognates to learn new, higher level L1 vocabulary, particularly for Latin based languages or vocabulary items. For example, learning avion in French could facilitate learning aviation in English for EnglishFrench speakers. Other work examining L1 errors based on the L2 has been conducted using fine-grained error analyses with oral language samples and adult participants. Any or all of these factors might contribute to the findings of L2 to L1 effects. Additionally, some research has been conducted on L2 intrusion errors when reading specific, orthographically similar L1 words (Jared & Kroll 2001). The role of language dominance was considered to contribute to these findings. Finally, Walter (2007) suggests that underlying processes are stronger candidates



Bidirectional cross-linguistic relations of first and second language skills 

for cross-linguistic ‘transfer’ than are global comprehension skills. The results of the current study showed shared variability when between-language relations for specific constructs were examined, particularly for lower level constructs. The present results are consistent with Walter’s hypothesis (2007). 5.  Conclusions In summary, within-language predictors of reading comprehension are consistent with theories of reading comprehension developed for monolingual speakers. Additionally, L1 variables that tap lower level processes are related to L2 reading comprehension. However, the results are not as clear when examining L2 variables related to L1 reading comprehension. Specifically, although L1 and L2 skills overlap within each construct, L2 skills are not strongly related to L1 reading comprehension. However, the use of ‘gross’ processing measures might not be sensitive enough to reveal subtle differences based on similarities between the L1 and L2.

Acknowledgements This research was supported by Grant 410-2010-1695 from the Social Sciences and Humanities Research Council of Canada to Alexandra Gottardo. We would like to thank members of the Halton Peel Hispanic Association for their assistance in recruitment. We would like to thank Aline Ferreira, Ana Pinzon, Amani Saleh, Mallory Durran and Pia Morar for their help in data collection and Amna Mirza for her help in manuscript preparation.

References Adams, Julia & David L. Weakliem (2011). August B. Hollingshead’s ‘Four factor index of social status’: From unpublished paper to citation classic. Yale Journal of Sociology 8: 11–51. Bialystok, Ellen (1999). Cognitive complexity and attentional control in the bilingual mind. Child Development 70(3): 636–644.  doi: 10.1111/1467-8624.00046 Bialystok, Ellen (2001). Bilingualism in development: Language, literacy, and cognition. New York: Cambridge University Press.  doi: 10.1017/CBO9780511605963 Bloomfield, Leonard (1933). Language. New York: Holt, Rinehart & Winston. Braze, David, Whitney Tabor, Donald P. Shankweiler & W. Einar Mencl (2007). Speaking up for vocabulary: Reading skill differences in young adults. Journal of Learning Disabilities 40: 226–243.  doi: 10.1177/00222194070400030401 Brutt-Griffler, Janina & Manka Varghese (2004). Bilingualism and language pedagogy. Clevedon, England: Multilingual Matters.

 Alexandra Gottardo, Christine Javier, Fataneh Farnia, Lorinda Mak & Esther Geva Cain, Kate, Jane Oakhill & Peter Bryant (2004). Children’s reading comprehension ability: concurrent prediction by working memory, verbal ability, and component skills. Journal of Educational Psychology 96(1): 31–42.  doi: 10.1037/0022-0663.96.1.31 Carlo, Maria S., Diane August, Barry McLaughlin, Catherine E. Snow, Cheryl Dressler, David N. Lippman, Teresa J. Lively & Claire E. White (2004). Closing the gap: Addressing the vocabulary needs of English language learners in bilingual classrooms. Reading Research Quarterly 39: 188–215.  doi: 10.1598/RRQ.39.2.3 Catts, Hugh W., Tiffany P. Hogan & Suzanne M. Adlof (2005). Developmental changes in reading and reading disabilities. In Hugh W. Catts & Alan G. Kamhi (eds.), The connections between language and reading disabilities, 25–40. Mahwah, NJ: Lawrence Erlbaum. Chall, Jeanne S. (1996). Stages of reading development (2nd edition). Fort Worth, TX: Harcourt-Brace. Cook, Vivian J. (2003). The changing L1 in the L2 user’s mind. In Vivian J. Cook (ed.), Effects of the second language on the first, 1–18. Clevedon, England: Multilingual Matters. Cummins, Jim (1984). Bilingual education and special education: Issues in assessment and pedagogy. San Diego, CA: College Hill Press. Cummins, Jim (1991). Interdependence of first- and second-language proficiency in bilingual children. In Ellen Bialystok (ed.), Language processing in bilingual children, 70–89. ­Cambridge, England: Cambridge University Press.  doi: 10.1017/CBO9780511620652.006 Da Fontoura, Helena A. & Linda S. Siegel (1995). Reading, syntactic, and working memory skills of bilingual Portuguese-English Canadian children. Reading and Writing 7(1): 139–153.  doi: 10.1007/BF01026951 Dunn, Lloyd M. & Leota M. Dunn (1981). Peabody picture vocabulary test (revised edition). Circle Pines, MN: American Guidance Service. Dunn, Lloyd M. & Leota M. Dunn (1997). Peabody picture vocabulary test (3rd edition). Circle Pines, MN: American Guidance Service. Dunn, Lloyd M., Delia E. Lugo, Eligio R. Padilla & Leota M. Dunn (1986). Test de Vocabulario en Imagenes Peabody. Circle Pines, MN: American Guidance Service. Durgunoğlu, Aydin Y. (2002). Cross-linguistic transfer in literacy development and implications for language learners. Annals of Dyslexia 52: 189–204.  doi: 10.1007/s11881-002-0012-y Durgunoğlu, Aydin Y., William E. Nagy & Barbara J. Hancin-Bhatt (1993). Cross-linguistic transfer of phonological awareness. Journal of Educational Psychology 85(3): 452–465.

doi: 10.1037/0022-0663.85.3.453

Farnia, Fataneh & Esther Geva (2013). Growth and predictors of change in English language learners’ reading comprehension. Journal of Research in Reading 36: 389–421.

doi: 10.1111/jrir.12003

Geva, Esther & Linda S. Siegel (2000). Orthographic and cognitive factors in the concurrent development of basic reading skills in two languages. Reading and Writing 12: 1–30.

doi: 10.1023/A:1008017710115

Geva, Esther & Min Wang (2001). The development of basic reading skills in children: A crosslanguage perspective. Annual Review of Applied Linguistics 21: 182–204.

doi: 10.1017/S0267190501000113

Gottardo, Alexandra (2002). The relationship between language and reading skills in SpanishEnglish speakers. Topics in Language Disorders 22: 46–70.

doi: 10.1097/00011363-200211000-00008

Gottardo, Alexandra & Amy Grant (2008). Defining bilingualism. Retrieved from www.literacyencyclopedia.ca



Bidirectional cross-linguistic relations of first and second language skills 

Gottardo, Alexandra, Christine Javier & Fataneh Farnia (2013). Reading comprehension in Spanish-English learners: Unpacking listening comprehension. Talk presented at the Pacific Coast Research Conference. San Diego, CA, February 2013. Gottardo, Alexandra & Julie Mueller (2009). Are first- and second-language factors related in predicting second-language reading comprehension? A study of Spanish-speaking ­children acquiring English as a second language from first to second grade. Journal of Educational Psychology 101: 330–344.  doi: 10.1037/a0014320 Gottardo, Alexandra, Bernice Yan, Linda S. Siegel & Lesly Wade-Woolley (2001). Factors related to English reading performance in children with Chinese as a first language: More evidence of cross- language transfer of phonological processing. Journal of Educational Psychology 93: 530–542.  doi: 10.1037/0022-0663.93.3.530 Gough, Philip B. & William E. Tunmer (1986). Decoding, reading, and reading disability. RASE: Remedial and Special Education 7(1): 6–10.  doi: 10.1177/074193258600700104 Grosjean, François (1982). Life with two languages. Cambridge, MA: Harvard University Press. Grosjean, François (1998). Studying bilinguals: Methodological and conceptual issues. Bilingualism: Language & Cognition 1: 131–149.  doi: 10.1017/S136672899800025X Grosjean, François & Ping Li (2012). The psycholinguistics of bilingualism. London, UK: Wiley-Blackwell. Guardado, Martin (2002). Loss and maintenance of first language skills: Case studies of H ­ ispanic families in Vancouver. Canadian Modern Language Review 58: 341–364.

doi: 10.3138/cmlr.58.3.341

Hakuta, Kenji (1988). A worldly look at language acquisition. PsycCRITIQUES 33(7): 576–578.  doi: 10.1037/030457 Hernandez, Arturo, Ping Li & Brian MacWhinney (2005). The emergence of competing modules in bilingualism. Trends in Cognitive Sciences 9(5): 220–225.  doi: 10.1016/j.tics.2005.03.003 Hollingshead, August B. (1975). Four factor index of social status. New Haven, CT: Yale University Press. Jared, Debra & Judith F. Kroll (2001). Do bilinguals activate phonological representations in one or both of their languages when naming words? Journal of Memory and Language 44: 2–31.  doi: 10.1006/jmla.2000.2747 Jarvis, Scott (2003). Probing the effects of the L2 on the L1: A case study. In Vivian J. Cook (ed.), Effects of the second language on the first, 81–102. Clevedon, England: Multilingual Matters. Jenkins, Joseph R., Lynn S. Fuchs, Paul van den Broek, Christine Espin & Stanley L. Deno (2003). Sources of individual differences in reading comprehension and reading fluency. Journal of Educational Psychology 95(4): 719–729.  doi: 10.1037/0022-0663.95.4.719 Jiang, Nan (2004). Semantic transfer and its implications for vocabulary teaching in a second language. The Modern Language Journal 88: 416–432.  doi: 10.1111/j.0026-7902.2004.00238.x Kendeou, Panayiota, Robert Savage & Paul van den Broek (2009). Revisiting the simple view of reading. British Journal of Educational Psychology 79: 353–370.

doi: 10.1348/978185408X369020

Kirby, John R. & Robert S. Savage (2008). Can the simple view deal with the complexities of reading? Literacy 42: 75–82.  doi: 10.1111/j.1741-4369.2008.00487.x Kroll, Judith F. & Annette M. B. de Groot (1997). Lexical and conceptual memory in the bilingual: Mapping form to meaning in two languages. In Annette M. B. de Groot & Judith F. Kroll (eds.), Tutorials in bilingualism: Psycholinguistic perspectives, 169–199. Mahwah, NJ: Lawrence Erlbaum.

 Alexandra Gottardo, Christine Javier, Fataneh Farnia, Lorinda Mak & Esther Geva Laufer, Batia (2003). The influence of L2 on L1 collocational knowledge and on L1 lexical diversity in free written expression. In Vivian J. Cook (ed.), Effects of the second language on the first, 18–31. Clevedon, England: Multilingual Matters. Lesaux, Nonie K., Keiko Koda, Linda S. Siegel & Timothy Shanahan (2006). Development of literacy. In Diane August & Timothy Shanahan (eds.), Developing literacy in second-­language learners: Report of the national literacy panel on language-minority children and youth, 75–122. Mahwah, NJ: Lawrence Erlbaum. Li, Ping, Igor Farkas & Brian MacWhinney (2004). Early lexical development in a self-­organizing neural network. Neural Network 17(8–9): 1345–1362.  doi: 10.1016/j.neunet.2004.07.004 Lindsey, Kim A., Franklin R. Manis & Caroline E. Bailey (2003). Prediction of first-grade reading in Spanish-speaking English-language learners. Journal of Educational Psychology 95(3): 482–494.  doi: 10.1037/0022-0663.95.3.482 Luk, Gigi, Eric De Sa & Ellen Bialystok (2011). Is there a relation between onset age of bilingualism and enhancement of cognitive control? Bilingualism: Language and Cognition 14(4): 588–595.  doi: 10.1017/S1366728911000010 MacWhinney, Brian (2004). A unified model of language acquisition. In Judith F. Kroll & Annette M. B. de Groot (eds.), Handbook of bilingualism: Psycholinguistic approaches, 49–67. Oxford University Press. Mood, Alexander M. (1971). Partitioning variance in multiple regression analyses as a tool for developing learning models. American Educational Research Journal 8: 191–202.

doi: 10.3102/00028312008002191

Naglieri, Jack A. (1985). The matrix analogies test. New York, NY: The Psychological Corporation. Nakamoto, Jonathan, Kim A. Lindsey & Franklin R. Manis (2007). A longitudinal analysis of English language learners’ word decoding and reading comprehension. Reading and ­Writing 20(7): 691–719.  doi: 10.1007/s11145-006-9045-7 Nation, Kate & Margaret J. Snowling (1997). Assessing reading difficulties: The validity and utility of current measures of reading skill. British Journal of Educational Psychology 67: 359–370.  doi: 10.1111/j.2044-8279.1997.tb01250.x Neale, Marie D. (1999). Neale analysis of reading ability (3rd edition). Melbourne, Victoria, ­Australia: ACER Press. Nimon, Kim & Thomas G. Reio (2011). Regression commonality analysis: A technique for quantitative theory building. Human Resource Development Review 10(3): 329–340.

doi: 10.1177/1534484311411077

Oller, D. Kimbrough, Linda Jarmulowicz, Barbara Z. Pearson & Alan B. Cobo-Lewis (2011). Rapid spoken language shift in early second-language learning: The role of peers and effects on the first language. In Aydin Yucesan Durgunoğlu & Claude Goldenberg (eds.), Language and literacy development in bilingual setting, 94–120. New York: ­Guilford Press. Ouellette, Gene & Ashley Beers (2010). A not-so-simple view of reading: How oral vocabulary and visual-word recognition complicate the story. Reading and Writing 23(2): 189–208.  doi: 10.1007/s11145-008-9159-1 Paradis, Johanne, Fred Genesee & Martha B. Crago (2011). Dual language development and disorders: A handbook on bilingualism and second language learning (2nd edition). Baltimore, MD: Brookes. Pasquarella, Adrian, Xi Chen, Katie Lam, Yang C. Luo & Gloria Ramirez (2011). Cross-language transfer of morphological awareness in Chinese-English bilinguals. Journal of Research in Reading 34(1): 23–42.  doi: 10.1111/j.1467-9817.2010.01484.x



Bidirectional cross-linguistic relations of first and second language skills 

Pavlenko, Aneta (2003). I feel clumsy speaking Russian: L2 influence on L1 narratives of Russian L2 users of English. In Vivian J. Cook (ed.), Effects of the second language on the first, 32–61. Clevedon, England: Multilingual Matters. Plaut, David C., James L. McClelland, Mark S. Seidenberg & Karalyn Patterson (1996). Understanding normal and impaired word reading: Computational principles in quasi-regular domains. Psychological Review 103: 56–115.  doi: 10.1037/0033-295X.103.1.56 Porte, Graeme (2003). English from a distance: Code-mixing and blending in the L1 output of long-term resident overseas EFL teachers. In Vivian J. Cook (ed.), Effects of the second language on the first, 103–119. Clevedon, England: Multilingual Matters. Proctor, C. Patrick, Diane August, Maria S. Carlo & Catherine Snow (2006). The intriguing role of Spanish language vocabulary knowledge in predicting English reading comprehension. Journal of Educational Psychology 98(1): 159–169.  doi: 10.1037/0022-0663.98.1.159 Proctor, C. Patrick, Maria S. Carlo, Diane August & Catherine Snow (2005). Native Spanishspeaking children reading in English: Toward a model of comprehension. Journal of Educational Psychology 97(2): 246–256.  doi: 10.1037/0022-0663.97.2.246 Ramirez, Gloria, Xi Chen, Esther Geva & Heidi Kiefer (2010). Morphological awareness in Spanish-speaking English language learners: Within and cross-language effects on word reading. Reading and Writing 23(3–4): 337–358.  doi: 10.1007/s11145-009-9203-9 Savage, Robert S. (2001). The ‘simple view’ of reading: Some evidence and possible implications. Educational Psychology in Practice 17(1): 17–33.  doi: 10.1080/02667360120039951 Semel, Eleanor, Elisabeth H. Wiig & Wayne A. Secord (2003). Clinical evaluation of language fundamentals. San Antonio, TX: The Psychological Corporation. Snow, Catherine E., M. Susan Burns & Peg Griffin (eds.) (1998). Preventing reading difficulties in young children: A report of the National Research Council. Washington, DC: Academy Press. van den Broek, Paul, Julie S. Lynch, Jan Naslund, Carolyn E. Invers-Landis & Kees Verduin (2003). The development of comprehension of main ideas in narratives: Evidence from the selection of titles. Journal of Educational Psychology 95(4): 707.  doi: 10.1037/0022-0663.95.4.707 van Gelderen, Amos, Rob Schoonen, Reinoud Stoel, Kees de Glopper & Jan Hulstijn (2007). Development of adolescent reading comprehension in language 1 and language 2: A longitudinal analysis of constituent components. Journal of Educational Psychology 99(3): 477–491.  doi: 10.1037/0022-0663.99.3.477 Venezky, Richard L. (2004). In search of the perfect orthography. Written Language and Literacy 7: 139–163. Verhoeven, Ludo (2007). Early bilingualism, language transfer, and phonological awareness. Applied Psycholinguistics 28(3): 425–439.  doi: 10.1017/S0142716407070233 Verhoeven, Ludo & Jan Van Leeuwe (2008). Prediction of the development of reading comprehension: A longitudinal study. Applied Cognitive Psychology 22: 407–423.

doi: 10.1002/acp.1414

Walter, Catherine (2007). First- to second-language reading comprehension: Not transfer, but access. International Journal of Applied Linguistics 17(1): 14–37.

doi: 10.1111/j.1473-4192.2007.00131.x

Warne, Russell T. (2011). Beyond multiple regression: Using commonality analysis to better understand R2 Results. Gifted Child Quarterly 55(4): 313–318.  doi: 10.1177/0016986211422217 Woodcock, Richard W. (1998). Woodcock reading mastery test (revised new norms). Itasca, IL: Riverside Publishing.

.

 Alexandra Gottardo, Christine Javier, Fataneh Farnia, Lorinda Mak & Esther Geva Woodcock, Richard W. (1991). Woodcock language proficiency battery (revised, English and Spanish forms). Chicago: Riverside Publishing. Woodcock, Richard W. & Ana F. Muñoz-Sandoval (1995). Woodcock language proficiency ­battery (revised, Spanish form). Itasca, IL: Riverside Publishing. Ziegler, Johannes C. & Usha Goswami (2005). Reading acquisition, developmental dyslexia, and skilled reading across languages: A psycholinguistic grain size theory. Psychological ­Bulletin 131(1): 3–29.  doi: 10.1037/0033-2909.131.1.3

Concurrent and longitudinal cross-linguistic transfer of phonological awareness and morphological awareness in Chinese-English bilingual children Yang Cathy Luo, Xi Chen & Esther Geva University of Toronto, Canada

The present study was designed to examine concurrent and longitudinal cross-linguistic transfer of phonological awareness and morphological awareness at two levels, the construct level and the reading level. We investigated whether phonological awareness and morphological awareness measured in one language are related to the same constructs measured in another language in Chinese-English bilinguals. Moreover, we assessed the cross-linguistic effects of the two constructs on reading concurrently and one year later in Grade 1. Participants of the study included 91 kindergarten and Grade 1 Chinese-English bilingual children. The children were tested twice, approximately one year apart, on a battery of cognitive and literacy measures in both languages. The data were analyzed with comprehensive path models that included phonological awareness, morphological awareness, and word reading in both languages. Our results demonstrate cross-linguistic transfer of phonological awareness and morphological awareness at the construct level and cross-linguistic transfer of phonological awareness to reading concurrently. Keywords:  transfer; phonological awareness; morphological awareness; word reading; Chinese-English bilingual children

1.  Introduction Both phonological awareness and morphological awareness are important for children’s reading development (e.g. Adams 1990; Bradley & Bryant 1983; ­Carlisle & Nomanbhoy 1993; Deacon & Kirby 2004; Ho & Bryant 1997; Mahony, Singson & Mann 2000; McBride-Chang et al. 2005; Singson, Mahony & Mann 2000). ­Phonological awareness refers to the ability to reflect on and manipulate the phonological structure of a word (e.g. the word book consists of three phonemes,

doi 10.1075/bct.89.05luo © 2016 John Benjamins Publishing Company

 Yang Cathy Luo, Xi Chen & Esther Geva

/b/ /u/ and /k/), while morphological awareness is the ability to reflect on and manipulate the morphemic structure of a word (e.g. the word friendly consists of two morphemes, the stem friend and the suffix ly). Among monolingual children, phonological awareness and morphological awareness have been found to predict concurrent and subsequent reading skills beyond the effects of general abilities such as age, nonverbal intelligence, vocabulary, memory, and/or socio-economic status (e.g. Bryant, MacLean, Bradley & Crossland 1990; Carlisle & Nomanbhoy 1993; Carlisle 1995; Deacon & Kirby 2004; Mahony et al. 2000; ­Wagner, ­Torgesen & Rashotte 1994). Their contributions to reading skills are also independent from each other (e.g. Carlisle & Nomanbhoy 1993; Deacon 2012). For bilingual children, an important question is whether phonological awareness or morphological awareness developed in one of their languages is associated with the same construct and reading success in the other language. Previous research has provided preliminary evidence for cross-linguistic transfer of phonological and morphological awareness. However, the majority of previous studies was limited in scope (investigating either phonological awareness or morphological awareness, but not both) and adopted a concurrent design (e.g. Gottardo, Yan, Siegel & Wade-Woolley 2001; Keung & Ho 2009; Wang, Perfetti & Liu 2005; Wang, Yang & Cheng 2009). Therefore, it is not clear whether metalinguistic skills in one language longitudinally predict literacy development in the other. In the present study, we examined the cross-linguistic transfer of phonological awareness and morphological awareness in Chinese-English bilingual children with a longitudinal design. We focused on transfer at two different levels, the construct level and the reading level. Specifically, we asked whether phonological awareness and morphological awareness measured at Time 1 were correlated with the same constructs across Chinese and English, and whether each construct would contribute to word reading concurrently and word reading measured one year later across languages. 1.1  Cross-linguistic transfer of phonological awareness Phonological awareness is an essential skill for early literacy development (Adams 1990; Bradley & Bryant 1983; Bryant, MacLean, Bradley & Crossland 1990; ­Stanovich 1986, 1992; Wagner & Torgesen 1987). Research has shown that for bilingual children, phonological awareness in one language contributes to literacy development in the other language (e.g. Cisero & Royer 1995; Comeau, Cormier, Grandmaison & Lacroix 1999; Durgunoğlu, Nagy & Hancin-Bhatt 1993; Gottardo et al. 2001; Lafrance & Gottardo 2005). Among different language pairs, it is particularly interesting to study transfer of phonological awareness between Chinese and English, two languages represented by typologically different writing systems.



Cross-linguistic transfer of phonological and morphological awareness 

Despite the typological differences, phonological awareness is important for reading both Chinese and English (e.g. Pasquarella, Chen, Lam, Luo & Ramirez 2011; Ho & Bryant 1997; Hu & Catts 1998; McBride-Chang & Ho 2000; Siok & Fletcher 2001). Moreover, there is overlap in the phonological structure between the two languages. In Chinese as well as in English, a syllable can be divided into an onset and a rhyme, and the phoneme is the smallest unit of sound. These commonalities likely form the foundation for cross-language transfer of phonological awareness across Chinese and English. Several studies have examined the cross-linguistic effects of phonological awareness on literacy outcomes in Chinese-English bilinguals (Chow et al. 2005; Gottardo et al. 2001; Wang et al. 2005, 2009). For example, Gottardo et al. (2001) showed that among Cantonese-speaking Canadian children, Chinese rhyme detection uniquely predicted English real word and pseudoword reading after English phoneme deletion was taken into consideration. In addition, there was cross-­linguistic association at the construct level. Chinese rhyme detection was significant correlated with English rhyme detection and English phoneme deletion. Wang et al. (2009) found that both Chinese onset and tone awareness uniquely predicted English real word reading among Mandarin-speaking American children in Grade 1 after controlling for English phonemic awareness and English compound awareness. However, neither aspect of Chinese phonological awareness was correlated with English phoneme deletion in this study. In another study, Keung and Ho (2009) demonstrated that English rhyme detection was related to Chinese word reading in Cantonese-speaking Hong Kong second graders after controlling for Chinese phonological processing skills. The researchers also reported significant correlations among Chinese rhyme awareness, English rhyme awareness, and English onset awareness. Taken together, these studies showed that phonological awareness had a cross-linguistic effect on reading in Chinese-­English bilinguals, when the measures were administered at the same time point. There is also evidence for transfer at the construct level. With the exception of Wang et al. (2009), the phonological awareness tasks were correlated across ­Chinese and English. A limitation of the studies described above is that they all adopted a concurrent design. As a result, it is not clear whether phonological awareness measured in one language has a sustained effect on phonological awareness and reading in the other language. A large number of studies have confirmed the effectiveness of phonological awareness as a longitudinal predictor of reading in monolingual children (e.g. Bradley & Bryant 1983; Bryant, MacLean, Bradley & Crossland 1990). By contrast, longitudinal studies are rare in cross-linguistic research. To our knowledge, only a couple of studies have tested transfer of phonological awareness in Chinese-English bilinguals with a longitudinal design (Chow et al. 2005; Pan, McBride-Chang, Shu, Zhang & Li 2011). Following a group of young

 Yang Cathy Luo, Xi Chen & Esther Geva

children in Beijing, Pan et al. (2011) found that Chinese syllable awareness at age 5 was related to English word reading at ages 8 and 10, beyond the effects of age, nonverbal reasoning, Chinese character reading, and Chinese cognitive skills. Chow et al. (2005) reported that for Hong Kong kindergartners, Chinese syllable deletion was a significant predictor of English word reading measured 9 months later, after controlling for English word reading at Time 1, age, visual skills, and Chinese vocabulary. These two studies provide preliminary evidence that the cross-­linguistic effect of phonological awareness on word-level reading remains significant over time. However, English phonological awareness was not used as a control variable in either study. The present study adopted a more stringent test that included this control. To summarize, although the previous research has provided strong evidence for transfer of phonological awareness, especially concurrently, there are some unresolved issues. First, with the exception of Wang et al. (2009), none of the previous studies has controlled for morphological awareness when examining transfer of phonological awareness. Given that phonological and morphological awareness have been conceptualized as interrelated component skills underlying literacy development (e.g. Carlisle & Nomanbhoy 1993; Deacon 2012) and have been shown to be correlated in empirical data (e.g. McBride-Chang et  al. 2003), it is important to compare their relative contributions to word reading. Such comparison has been carried out within the same language (e.g. Tong & ­McBride-Chang 2010) and our study extended it to cross-linguistic research. Furthermore, only a couple of studies have investigated cross-linguistic transfer of phonological awareness longitudinally, but both were missing important controls (e.g. within-language phonological awareness). To address these issues, the present study included Chinese and English phonological and morphological awareness in the same model and investigated the transfer of both concurrently and longitudinally. 1.2  Cross-linguistic transfer of morphological awareness There are three types of morphology across most languages: inflection, derivation, and compounding. Inflectional morphemes are word endings denoting meanings without altering the meaning or the part of speech of the root word, such as verb tense (e.g. play → played), aspect (e.g. do → doing), and person (e.g. I finish → she finishes); nouns for number (e.g. one pencil → two pencils); pronouns for possession (e.g. her → hers); and adjectives for comparison (e.g. fast → faster). Derivation involves forming new words that have a different meaning or word class from the base words by applying prefixes and suffixes (e.g. un-, sub-, -ness, -ly). Compounding involves the combination of two or more words in forming new words



Cross-linguistic transfer of phonological and morphological awareness 

(e.g. ice + cream = ice cream) (Katamba 1993; Kuo & Anderson 2006; McBrideChang 2004). Languages vary in terms of morphological structure. While languages written with an alphabet typically contain a large number of inflected and derived words, about 70–80% of Chinese words are compound words (Kuo  & Anderson 2006). Accordingly, there is substantial evidence that compound awareness is related to both character and word reading in Chinese (e.g. McBrideChang, Shu, Zhou, Wat & Wagner 2003; Shu, McBride-Chang, Wu & Liu 2006). A small number of studies have investigated transfer of compound awareness among Chinese-English bilinguals (Pasquarella et al. 2011; Wang, Cheng & Chen 2006; Wang et al. 2009). These studies provide mixed results regarding the cross-linguistic effect of compound awareness on word reading. Wang et al. (2006, 2009) reported in two studies that English compound awareness contributed to Chinese character reading among Mandarin-speaking children in the US. In Wang et al. (2009), the contribution survived controls for Chinese phonological awareness, Chinese compound awareness, and English phonological awareness. In Wang et al. (2006), however, the contribution was only significant with ­Chinese phonological awareness as a control; it was no longer significant after English phonological awareness was entered into the regression model. Pasquarella et al. (2011) conducted a comprehensive test of cross-language effects of morphological awareness among Mandarin-speaking Canadian children in Grades 1 to 4. English compound awareness was found to predict Chinese vocabulary and Chinese reading comprehension, though not Chinese character reading. Cheung et al. (2010) showed a somewhat different pattern of transfer among a sample of Hong Kong that included kindergartners, second graders, and fourth graders. In this study, Chinese compound awareness explained unique variance in English word reading beyond the effects of English phonological and morphological awareness. Despite some inconsistency in regard to the direction of transfer, extant research suggests for Chinese-English bilinguals, compound awareness developed in one of their languages facilitates reading development in the other language. However, all of the previous studies only focused on concurrent relationships. As a result, it is unclear whether the cross-linguistic effect of compound awareness will remain significant over time. Interestingly, several studies have compared the cross-linguistic effect of compound awareness and that of derivational awareness on reading in ChineseEnglish bilinguals (Wang et al. 2006, 2009; Pasquarella et al. 2011; Zhang 2012). All the studies demonstrated transfer of compound awareness, whereas none of them provided evidence for transfer of derivational awareness. As Pasquarella et al. (2011) point out, transfer of compound awareness may be attributed to the common features shared by compounds, especially noun-noun compounds, in the two languages, despite the fact that compounding is a more prominent form

 Yang Cathy Luo, Xi Chen & Esther Geva

of word formation in Chinese than in English. A large percentage of noun-noun compounds in both languages are right-headed (Benczes 2006; Packard 2000), which means that the right morpheme specifies the category and the left morpheme modifies the meaning and identifies the subcategory, e.g. teacup in ­English and 茶杯 in Chinese. Most compounds in both languages are semantically transparent in that the meaning of a compound is predictable from the meaning of the constituent morphemes. On the other hand, derivational morphemes are different in many ways across Chinese and English. For example, there are far fewer derivational morphemes in Chinese, they are less productive, and typically do not label part of speech (Packard 2000). Thus, it seems that the aspect of morphological awareness (compound vs. derivational) that transfers is influenced by the morphological structure of the languages involved. Transfer occurs when the target morphological structure is similar across bilingual children’s two languages. Building on previous research, we chose to examine transfer of compound awareness in Chinese-English bilinguals in the present study. Previous research has produced mixed results with respect to transfer of compound awareness at the construct level in Chinese-English bilinguals. ­Neither study conducted by Wang and colleagues (2006, 2009) observed a significant correlation between compound awareness across Chinese and English. By contrast, Pasquarella et al. (2011) and Cheung et al. (2010) reported a significant cross-linguistic association. Perhaps the strongest evidence of cross-linguistic transfer of compound awareness at the construct level comes from Zhang et al. (2010), who carried out an intervention study among fifth graders in China who learned ­English as a foreign language. In the study, children received a 45-minute training session on either Chinese or English compound awareness. The Chinese intervention group outperformed an untrained control group on an English compound awareness task, suggesting transfer of compound awareness from Chinese to ­English. Transfer from English to Chinese was also detected, but only for those who had a high level of English proficiency. The results of Zhang et al. (2010) point to a causal link between Chinese and English compound awareness. However, because of the mixed results from the naturalistic studies, it remains unclear whether a cross-linguistic connection between compound awareness exists in Chinese-English bilinguals who do not receive explicit interventions. The present study explored this issue. 2.  The present study The present study focused on cross-linguistic transfer of phonological and morphological awareness in Chinese-English bilinguals. Our participants were ­Chinese-English bilinguals in Kindergarten and Grade 1 at the beginning of the study. We tested their phonological awareness, compound awareness, and word



Cross-linguistic transfer of phonological and morphological awareness 

reading in Chinese and English at the beginning of the study and then word reading in the two languages again one year later. The aim was to examine transfer of phonological awareness and compound awareness concurrently and longitudinally. Compared to previous research, our study has several unique features. First, we investigated transfer at two different levels, the construct level and the reading level. Transfer at the construct level occurs through the association between parallel skills in Chinese and English; transfer at the reading level occurs from phonological awareness or morphological awareness in one language to word reading in another language. While previous transfer research has mostly focused on transfer at the reading level, it is important to study transfer at the construct level because it forms the foundation for transfer at the reading level. Second, our study provided a stringent test of transfer by including phonological and morphological awareness in Chinese and English in the same model. These two aspects of metalinguistic awareness were often examined in separate studies or statistical analyses (Cheung et al. 2010; Chow et al. 2005; Gottardo et al. 2001; Keung & Ho 2009; Pan et al. 2011; Wang et al. 2006, 2009). In particular, studies that focused on transfer of phonological awareness typically did not account for the effect of morphological awareness (Chow et al. 2005; Gottardo et al. 2001; Keung & Ho 2009; Pan et al. 2011; Wang et al. 2006). Phonological and morphological awareness tasks tend to be associated, as morphological awareness is a complex construct that involves processing on several different dimensions, including the phonological dimension (Carlisle 2003). Thus, it is important to ask whether each skill makes an independent cross-linguistic contribution to literacy development. Finally, our study adopted a longitudinal design. There have been only two longitudinal transfer studies involving Chinese-English bilinguals and both were missing important control variables (autoregressor of reading and/ or within-language phonological awareness). The present study examined whether there are longitudinal crosslinguistic effects of phonological and compound awareness on reading, after ­controlling for the autoregressor of reading at Time 1 and within-language metalinguistic awareness. 3.  Method 3.1  Participants Participants were 91 typically developing Chinese-English bilingual children, recruited from entry-level Chinese heritage language classes in a large city in ­Canada. The sample consisted of 56 children in kindergarten (Mage = 5.23 years, SD = 0.53 years; 25 girls) and 35 children in Grade 1 (Mage = 6.58 years, SD = 0.33 years; 12 girls) at the beginning of the study (Time 1). The children were tested again one year later (Time 2). Sixty-six children remained in the study at Time 2

 Yang Cathy Luo, Xi Chen & Esther Geva

(42 in Grade 1 and 24 in Grade 2). The rest of the participants moved out of the area with their families. Statistical comparison was conducted between children who remained and those who left on their performance on all given measures at Time 1 and yielded no difference between the two groups (ps > .98). All the participants were of Chinese descent. Their parents were first-­ generation immigrants to Canada. Sixty-two percent of the children were born in Canada and the other 38% immigrated to Canada at an average age of 3 years and 5 months from a Chinese-speaking country or region. According to the parents’ responses to our questionnaire, all the participants attended regular English public schools. They also attended 2.5 hours of heritage language classes every week where they received instruction on both oral and written Chinese. Mandarin ­Chinese was the primary language used between children and parents. Over 80% of the children and 85% of the parents reported speaking Mandarin at home most of the time. On average, the parents had completed a university education. 3.2  Measures The participants were assessed twice, approximately one year apart. At Time 1, the children received a battery of measures assessing their nonverbal reasoning, phonological awareness, morphological awareness, verbal ability, and word-level reading in both Chinese and English. Their parents were asked to fill out a questionnaire regarding family background, home literacy activities, and language use. The children were tested again on word-level reading and verbal ability in the two languages at Time 2. The internal consistency reliabilities (Cronbach’s alphas) for all the measures at both time points are reported in Table 1. Family questionnaire. Parents were asked to complete a questionnaire at Time 1. Data collected with the questionnaire provided descriptive information about our sample’s immigration experience, home language use, and literacy activities (see the ‘Participants’ section). Maternal education was used as a proxy for socio-economic status. This variable was measured with a 6-point scale, where 1 = ­primary school, 2 = junior high school, 3 = high school, 4 = college, 5 = university degree, and 6 = graduate degree. Nonverbal reasoning. Nonverbal reasoning was measured with the Raven Standard Progressive Matrices (SPM; Raven 1958) at Time 1. The SPM comprises 5 sets (A to E) of 12 items each, with items within a set arranged in order of increasing difficulty. Provided that our participants were still at the lowest age range for which the test was intended, only the first three sets (A, B, and C) were used in the present study. For each item, children were shown a picture with a missing section and asked to choose out of the six patterns listed below the one that best matched the missing section.



Cross-linguistic transfer of phonological and morphological awareness 

Verbal ability. A shortened version of the Peabody Picture Vocabulary Test-III A (PPVT-III A; Dunn & Dunn 1997) was used at Time 1 to assess children’s general verbal ability in English. Every third item of the original PPVT-III was selected for the shortened version (e.g. 1st, 4th, 7th…item; a total of 60 items; similar to Wang et al. 2006) to maintain the progression of difficulty in the test and at the same time to reduce testing time within this large test battery. In each item, the experimenter presented four pictures to the child, and then stated a word orally twice. The child heard the word and chose out of the four pictures the one that best described the word. Three practice items were given before the 60 testing items were administered to ensure that the child understood the procedure of the task. 3.3  English measures English word reading. English word reading was assessed at both times with the Letter-Word Identification Subtest from the Woodcock-Johnson III Tests of Achievement (Woodcock, McGrew & Mather 2001). Children were asked to identify 14 letters and to read 62 words of increasing difficulty. The test was discontinued if the child made six consecutive errors. English phonological awareness. Phonological awareness in English was assessed at Time 1 using the Elision subtest of the Comprehensive Test of Phonological Processing (CTOPP; Wagner, Torgesen & Rashotte 1999). This test contained 6 practice and 20 test items. Children were asked to remove a syllable at the initial or final position of a word or a phoneme at the initial, middle, or final position of a syllable to produce new words (e.g. “Say doughnut; now say doughnut without saying dough.” Or “say cat; now say cat without the /k/.”). Testing was terminated when the child made three consecutive errors. English compound awareness. Adapted from McBride-Chang et al. (2005), this test evaluated children’s ability to form new compound words. It was administered at Time 1 only. The test included 2 practice items and 15 test items. In each item, a scenario was orally presented in a two- to four-sentence story. Children were then asked to come up with a novel compound word to describe the object or concept presented by the scenario. For example, “Early in the morning, we can see the sun rising. This is called a sunrise. At night, we might also see the moon rising. What could we call this? (moonrise)”. To minimize the effect of vocabulary, all the answers were novel compounds that did not exist in English. 3.4  Chinese measures Chinese character reading. This test was administered at both times. It consisted of 125 unrelated characters selected from the 12 volumes of the Elementary School Textbooks (Elementary Education Teaching and Research Center, Beijing

 Yang Cathy Luo, Xi Chen & Esther Geva

­ ducation and Science Institute 1996) employed in the Chinese language curriE cula from Grade 1 to Grade 6 in Mainland China. It started with highly frequent characters (e.g. 人 ‘­person’, 小 ‘small’) and moved to less frequent ones (e.g. 蹑 ‘slink’, 黝 ‘­swarthy’). The test was discontinued when the child misread 10 characters consecutively. Chinese phonological awareness. Chinese phonological awareness was assessed at Time 1 with an onset and syllable deletion task. Children were asked to delete either a syllable from the initial, middle, or end position from a multiplesyllable word or the onset from a syllable. Both real and pseudo items were used. For example, children were asked to delete /dao1/ from /tu3dao1mei3/, and to delete /s/ from /se4/. The test contained 4 practice items and 24 test items with 12 syllable deletion items and 12 onset deletion items. Chinese compound awareness. Adapted from McBride-Chang et al. (2005), this test was parallel to the English compound awareness test described above. It also included 2 practice items and 15 test items. For example, “斑马是身上有 斑纹的一种马, 那么身上有斑纹的牛我们叫什么?” ‘Striped horse (Zebra) is a kind of horse with stripes on the body. What should we call a cow with stripes on the body?’ (斑牛 ‘striped cow’). 3.5  Procedure Testing took place in a quiet room in a participant’s regular school or heritage language class. At both Time 1 and Time 2, all measures were administered individually in two sessions on separate days. The English measures were given in one session, while the Chinese measures were given in the other. The order of testing was counterbalanced within each session. The experimenters were ChineseEnglish bilinguals who only administered the tests in their native language. These graduate and undergraduate students in psychology had received extensive training on the testing battery. Instructions were first given in the language of a test. If a child did not understand the original instructions, the experimenter then used the child’s stronger language to explain the procedure.

4.  Results The original dataset contained missing values due to absence at each testing session or attrition from Time 1 to Time 2. Approximately 15% of the values were missing. Little’s MCAR test showed that missingness was random, χ2(100) = 92.82, p = .682. The effect of missingness of each variable on other variables was ­subsequently



Cross-linguistic transfer of phonological and morphological awareness 

examined with t tests. None of the t tests were significant, ts ≤ 2.20, ps ≥ .062, and hence further confirmed the missing-completely-at-random assumption. We imputed data using the SPSS multiple imputation function and synthesized the results based on the suggestion made by Rubin (1987) and Schafer (1997). The imputed data and the original data yielded the same results and therefore, we only reported the results based on imputed data in this section. Table 1 reports means, standard derivations, and ranges of maternal education, nonverbal reasoning, phonological awareness, and compound awareness measured at Time 1, and word-level reading measured at both Time 1 and Time 2 in English and Chinese for children in kindergarten and Grade 1 separately, as well as for the combined group. Raw scores are presented for all measures; standard scores are also presented for the English word reading measure. The skewness and kurtosis of the raw scores were both within a reasonable range, -1.158 ≤ skewness ≤ 1.852 and -2.477 ≤ kurtosis ≤ 2.5. Thus, the assumption of normal distribution was satisfied (Tabachnick & Fidell 2006). We performed a Box’s M test of homogeneity of variance-covariance matrices available in Discriminant Analysis to compare the correlational patterns among the English and Chinese measures between children in kindergarten and Grade 1. The results showed no significant difference in variance-­ covariance patterns between the two grades on the Chinese and English measures, Box’s M = 69.38, F(66, 16889) = 1.16, p = .17, suggesting that the correlational patterns were rather similar. Therefore, the two grades were combined for further analysis. Table 2 presents the cross-linguistic correlations among the English and Chinese measures for the combined group. As shown in Table 2, phonological awareness measured in the two languages was moderately correlated, r = .67, p < .001; so was compound awareness in the two languages, r = .67, p < .001. Word-level reading in English and Chinese demonstrated small but significant correlations concurrently and longitudinally (rs ranging between .24 and .31, ps < .05). Time 1 English phonological awareness was correlated with Time 1 Chinese reading, r = .23, p < .05. Moreover, Time 1 Chinese phonological awareness was correlated with Time 1 and Time 2 English reading, r = .67 and .50 respectively, ps < .05. Time 1 Chinese compound awareness was related to Time 1 and Time 2 English word reading, r = .42, and .30 respectively, ps < .05. We performed path analyses using AMOS 18 to further examine transfer between parallel constructs across the two languages and the direct and indirect cross-linguistic effects of phonological awareness and morphological awareness on reading. Two baseline models were proposed, one (Figure 1) examining the predictions from English metalinguistic skills to Chinese reading and the other

 Yang Cathy Luo, Xi Chen & Esther Geva Table 1. Means and standard deviations of the measures by grade Measures (Raw scores)

Min– Max

Sample specific reliability Cronbach’s α

Kindergarten cohort (n = 56) M

SD

Grade 1 cohort (n = 35)

ANCOVAa (Kindergarten vs. Grade 1)

M

SD

F

Combined (n = 91) M

SD

Maternal Education

1–6

N/A

4.89

0.97

4.89

0.87

N/A

4.89

0.92

Nonverbal Reasoning

0–36

.83

14.84

3.74

21.83

6.10

N/A

17.53

5.86

Verbal Ability

0–60

.54

32.46

4.49

33.83

5.20

0.31

32.99

4.77

T1 Chinese Phonological Awareness

0–24

.95

10.11

7.53

17.89

5.79

5.02*

13.10

7.86

T1 Chinese Compound Awareness

0–15

.86

4.43

3.67

7.87

3.40

2.87

5.76

3.92

T1 Character Reading

0–125

.97

16.70

17.06

29.50

19.62

3.12

21.62

19.04

T2 Character Reading

0–125

.97

25.76

21.17

31.95

29.07

0.15

28.14

21.61

T1 English Phonological Awareness

0–20

.93

4.09

3.78

9.63

5.55

6.22

5.31

T1 English Compound Awareness

0–15

.86

4.43

3.67

7.87

3.40

0.78

5.76

3.92

T1 English Word Reading

0–76

.97

22.31

10.24

36.78

11.91

11.09**

27.87

12.95

T2 English Word Reading

0–76

.96

3.60

13.56***

34.93

12.46

43.53

14.72

38.24

13.32

T1 English Word Reading (Std Scores)

119.02

23.22

113.03

32.46

116.71

27.13

T2 English Word Reading (Std Scores)

124.78

17.32

109.80

27.79

119.11

22.88

Note: T = Time. ***p < .001; **p < .01; *p < .05. a. ANCOVA controlled for nonverbal reasoning



Cross-linguistic transfer of phonological and morphological awareness 

Table 2.  Zero order correlation matrices among the English measures for the combined group Measures

Age

ME

NR

VA

T1 EPA

T1 ECA

T1 EWR

T2 EWR

Age

–

-.07

.63***

.20

.54***

.49**

.58***

.35**

ME

-.07

–

.13

.20

.24*

.21*

.25*

.25*

NR

.63***

.13

–

.14

.44**

.55***

.46**

.31*

VA

.20

.20

.14

–

.29**

.31**

.30**

.29*

T1 CPA

.63***

.17

.55***

.28*

.67***

.56**

.67***

.50***

T1 CCA

.50***

.20

.53***

.06

.44**

.62***

.42**

.30*

T1 CCR

.37**

.06

.30*

.07

.23*

.28*

.31*

.24

T2 CCR

.12

.18

.18

.00

.16

.18

.25*

.14

Note: ***p < .001; **p < .01; *p < .05. T1 = Time 1; T2 = Time 2. ME = Maternal Education; NR = Nonverbal Reasoning; VA = Verbal Ability. EPA = English Phonological Awareness; ECA = English Compound Awareness; EWR = English Word Reading; CPA = Chinese Phonological Awareness; CCA = Chinese Compound Awareness; CCR = Chinese Character Reading. N = 91

examining the predictions from Chinese metalinguistic skills to English reading (Figure 2). Notably, the baseline models only captured the intercorrelations among phonological awareness and morphological awareness skills and within-language relationships from the metalinguistic skills to reading outcomes. On top of the baseline models, direct cross-linguistic relationships from Time 1 phonological awareness and morphological awareness in one language to Time 1 and Time 2 reading in the other language were tested through model justification by comparing models with additional path(s) with the baseline models. Nonverbal reasoning and vocabulary were entered in both baseline models to control for general nonverbal and verbal abilities. These two control variables were selected through a two-step procedure. We first performed a factor analysis on the five possible control variables available in our data (nonverbal reasoning, vocabulary, age, grade, maternal education). These measures loaded on two factors, with nonverbal reasoning, age, and grade on one and vocabulary and maternal education on the other. In the second step, stepwise regressions revealed that nonverbal reasoning and vocabulary were the best predictors of the Chinese and English measures in the models from the two factors respectively. Therefore, these two control variables were entered in our models. To simplify presentations, however, they are not depicted in the figures. Age, grade, and maternal education did not explain additional variances in Chinese or English reading in either model for our sample and therefore were not entered.

 Yang Cathy Luo, Xi Chen & Esther Geva

T1 CPA

T1 CCA T1 EPA

T1 CCR

T1 ECA

T2 CCR

Figure 1.  The English-to-Chinese baseline model Dashed lines depict hypothesized covariance while solid lines illustrate hypothesized paths. EPA = English Phonological Awareness; ECA = English Compound Awareness; CPA = Chinese Phonological Awareness; CCA = Chinese Compound Awareness; CCR = Chinese Character Reading. Verbal and nonverbal abilities were also controlled in the model but eliminated in this simplified presentation.

T1 EPA

T1 ECA T1 CPA

T1 CCA

T1 EWR

T2 EWR

Figure 2.  The Chinese-to-English baseline model Dashed lines depict hypothesized covariance while solid lines illustrate hypothesized paths. CPA = Chinese Phonological Awareness; CCA = Chinese Compound Awareness; EPA = English Phonological Awareness; ECA = English Compound Awareness; EWR = English Word Reading. Verbal and nonverbal abilities were also controlled in the model but eliminated in this simplified presentation.

4.1  The English-to-Chinese model We first examined the concurrent and longitudinal cross-linguistic predictions from English phonological awareness and compound awareness to Chinese reading (Figure 1). In this model, English phonological and compound awareness at Time 1 were hypothesized to contribute to the parallel Chinese measures at Time 1, which in turn predicted Time 1 Chinese character reading. The Time 1 literacy measures were hypothesized to predict the same measures at Time 2. Since phonological awareness and morphological awareness are mutually dependent constructs (e.g. Deacon 2012), phonological awareness and morphological awareness in ­English were modeled to co-vary with each other, and so were the two measures in ­Chinese.



Cross-linguistic transfer of phonological and morphological awareness 

The hypothesized baseline model demonstrated an acceptable fit to the data, χ2 (7) = 9.94, p = .192, CFI = .98, GFI = .97, RMSEA = .07. However, the converged model contained some non-significant paths. These paths were removed to create the most parsimonious model. The most parsimonious model also demonstrated an acceptable model fit, χ2 (16) = 24.45, p = .08, CFI = .98, GFI = .96, RMSEA = .06. It fit the data as well as the baseline model, ∆χ2 (9) = 14.52, p = .105. Based on the most parsimonious model, two alternative models were then hypothesized to examine the concurrent direct prediction respectively from Time  1 English phonological awareness to Time 1 Chinese reading and from Time 1 English compound awareness to Time 1 Chinese reading. The results of the two models were then compared to the most parsimonious model and showed none of these added paths were significant, βs ≤ .10, ps ≥ .250, or significantly improved the model fit, ∆χ2s (1) ≤ 1.25, ps ≥ .264. Similarly, two alternative models were hypothesized to examine the longitudinal direct prediction respectively from Time 1 English phonological awareness to Time 2 Chinese reading and from Time 1 English compound awareness to Time 2 Chinese reading. When compared to the most parsimonious model, the two models did not fit the data significantly better, ∆χ2s (1) ≤ 2.46, ps ≥ .117. The added longitudinal paths were not significant either, βs ≤ .09, ps ≥ .111. Given that none of the above mentioned alternative models significantly improved data fit, our final model, therefore, was the same as the converged most parsimonious model, depicted in Figure 3 with standardized coefficients. Table 3 provides the standardized parameter estimates for the final model with direct, indirect, and total effects. As shown in the figure and table, English compound awareness indirectly affected Chinese reading measured concurrently, β = .17, T1 CPA .27*

.52*** T1 CCA T1 EPA .56*** T1 ECA

.46***

.38** T1 CCR .29*

T2 CCR

Figure 3.  The English-to-Chinese final model (with standardized coefficients) T1 = Time 1. T2 = Time 2. EPA = English Phonological Awareness; ECA = English Compound Awareness; CPA = Chinese Phonological Awareness; CCA = Chinese Compound Awareness; CCR = Chinese Character Reading. Verbal and nonverbal abilities were also controlled in the model but eliminated in this simplified presentation. ***p < .001; **p < .01; *p < .05. N = 91. χ2 (16) = 24.45, p = .08, CFI = .98, GFI = .96, RMSEA = .06.

 Yang Cathy Luo, Xi Chen & Esther Geva

Table 3.  Standardized direct, indirect, and total effects for the English-to-Chinese model Concurrent Direct effects β

Longitudinal

Indirect effects

Total effects

β

β

R2

Direct effects β

R2

Indirect effects

Total effects

β

β

T1 Chinese Character Reading

T2 Chinese Character Reading

NR

–

–

.11

.11

–

–

.03

.03

VA

–

–

–

–

–

–

–

–

T1 CCR

–

–

–

–

.29*

.08

–

.29*

T1 CPA

–

–

–

–

–

–

–

–

T1 CCA

.38**

.14

–

.38**

–

–

.11

.11

T1 EPA

–

–

–

–

–

–

–

–

T1 ECA

–

–

.17*

.17*

–

–

.05

.05

T1 Chinese Phonological Awareness Nonverbal

.30*

.09

–

.30*

VA

–

–

–

–

T1 EPA

.52***

.27

–

.52***

T1 ECA

–

–

–

–

T1 Chinese Compound Awareness Nonverbal

.28*

.08

–

.28*

VA

–

–

–

–

T1 EPA

–

–

–

–

T1 ECA

.46***

.21

–

.46**

Note: ***p < .001; **p < .01; *p < .05. NR = Nonverbal Reasoning; VA = Verbal Ability; EPA = English Phonological Awareness; ECA = English Compound Awareness; CPA = Chinese Phonological Awareness; CCA = Chinese Compound Awareness. N = 91

p = .010. Such an indirect effect operated through Chinese compound awareness, as English compound awareness first significantly predicted Chinese compound awareness, β = .46, p < .001, which in turn significantly contributed to Chinese reading, β = .38, p < .01. English phonological awareness had neither direct nor indirect cross-linguistic effects on Chinese reading. 4.2  The Chinese-to-English model Next, we examined the cross-linguistic predictions from Chinese metalinguistic measures to English literacy outcomes (Figure 2). Time 1 Chinese phonological and compound awareness were hypothesized to contribute to the parallel Eng-



Cross-linguistic transfer of phonological and morphological awareness 

lish measures at Time 1, which in turn predicted English word reading at Time 1. Time 1 English reading was expected to predict the same measure at Time 2. The hypothesized baseline model demonstrated an acceptable fit, χ2(7) = 18.22, p = .011, CFI = .97, GFI = .95, RMSEA = .08. The baseline model again contained several non-­significant paths. These paths were removed to create the most parsimonious model, which demonstrated an acceptable fit, χ2(14) = 23.29, p = .056, CFI = .97, GFI = .95, RMSEA = .07. There was no difference between the baseline model and the most parsimonious model, ∆χ2(7) = 5.07, p = .651. Based on the most parsimonious model, two alternative models were then hypothesized to examine the concurrent direct prediction from Time 1 Chinese phonological awareness to Time 1 English reading and from Time 1 Chinese compound awareness to Time 1 English reading, each with an added path. These two models were then compared to the most parsimonious model. The model with the added path from Time 1 Chinese phonological awareness to Time 1 English word reading had a significantly better fit than the most parsimonious model, ∆χ2(1) = 9.94, p = .002, and the path was also significant, β = .31, p = .009. Overall, this model demonstrated a very good fit, χ2(13) = 13.35, p = .421, CFI = 1.00, GFI = .97, RMSEA = .02. Similarly, two alternative models were hypothesized to examine the longitudinal direct prediction from Time 1 Chinese phonological awareness to Time 2 English reading and from Time 1 Chinese compound awareness to Time 2 English reading. When compared to the most parsimonious model, neither model fit the data significantly better, ∆χ2 s (1) ≤ 2.39, p ≥ .122. The added longitudinal paths were not significant either, βs ≤ .13, ps ≥ .122. Therefore, our final model was the model with an additional path from Time 1 Chinese phonological awareness to Time 1 English word reading (see Figure 4). T1 EPA .28*

.67*** .22*

.31**

T1 CPA .48**

.59*** T1 CCA

T1 ECA

.43** .16*

T1 EWR .70***

T2 EWR

Figure 4.  The Chinese-to-English Final model, showing the standardized coefficients T1 = Time 1. T2 = Time 2. CPA = Chinese Phonological Awareness; CCA = Chinese Compound Awareness; EPA = English Phonological Awareness; ECA = English Compound Awareness; EWR = English Word Reading. Verbal and nonverbal abilities were also controlled in the model but eliminated in this simplified presentation. ***p < .001; **p < .01; *p < .05. N = 91. χ2(13) = 13.35, p = .421, CFI = 1.00, GFI = .97, RMSEA = .02.

 Yang Cathy Luo, Xi Chen & Esther Geva

Table 4.  Standardized direct, indirect, and total effects for the Chinese-to-English model Concurrent Direct effects

Indirect effects

Total effects

β

β

R2

β

Longitudinal Direct effects

Total effects

β

β

R2

β

T1 English Word Reading

Indirect effects

T2 English Word Reading

NR

–

–

.03

.03

–

–

.02

.02

VA

–

–

.04

.04

–

–

.02

.02

T1 EWR

–

–

–

–

.70***

.49

–

.70***

T1 EPA

.43**

.19

–

.43**

–

–

.30**

.30**

T1 ECA

.16*

–

.16*

–

–

.11

.11

T1 CPA

.31**

.10

.31**

.62**

–

–

.43**

.43**

T1 CCA

–

–

.07

.07

–

–

.05

.05

T1 English Phonological Awareness Nonverbal

–

–

–

–

VA

–

–

–

–

T1 CPA

.67***

.45

–

.67***

T1 CCA

–

–

–

–

T1 English Compound Awareness Nonverbal

.17*

.03

–

.17*

VA

.16*

.03

–

.26**

T1 CPA

.12*

.05

–

.12*

T1 CCA

. 42**

.18

–

.42**

Note. ***p < .001; **p < .01; *p < .05. NR = Nonverbal Reasoning; EPA = English Phonological Awareness; ECA = English Compound Awareness; CPA = Chinese Phonological Awareness; CCA = Chinese Compound Awareness. N = 91

Table 4 provides the standardized parameter estimates for the final model with direct, indirect, and total effects. Chinese phonological awareness predicted ­English word reading concurrently after controlling for nonverbal reasoning, verbal ability, and compound awareness in both Chinese and English, β = .67, p < .001. Such effects included both a direct relationship, β = .31, p < .01 (9% of the variance explained), and an indirect relationship, β = .31, p < .01. The indirect impact was through the effect of Time 1 Chinese phonological awareness on Time  1 English phonological awareness, β = .67, p < .001, which then contributed to English word reading, β = .43, p < .01. Chinese phonological awareness at Time 1 did not directly contribute variance to English word reading at Time 2,



Cross-linguistic transfer of phonological and morphological awareness 

instead, this ­contribution was completely mediated through autoregressive effects. Chinese compound awareness had neither direct nor indirect cross-linguistic effect on Time 1 or Time 2 English word reading.

5.  Discussion The present study was designed to examine cross-linguistic transfer of phonological awareness and morphological awareness at two levels, the construct level and the reading level. We investigated whether phonological awareness and morphological awareness measured in one language are related to the same constructs measured in another language in Chinese-English bilinguals. Moreover, we assessed the cross-linguistic effects of the two constructs on reading concurrently and one year later. The data were analyzed with comprehensive path models that included phonological awareness, morphological awareness, and word reading in both Chinese and English. Our results demonstrated cross-linguistic transfer of phonological awareness and morphological awareness at the construct level and cross-linguistic transfer of phonological awareness to reading concurrently. These results are discussed in detail below. 5.1  Cross-linguistic effects of phonological awareness on word-level reading Our study provides strong support for cross-linguistic transfer of phonological awareness between Chinese and English. There was consistent evidence for transfer at the construct level – Chinese phonological awareness was predictive of ­English phonological awareness (45% of the variance) after controlling for nonverbal reasoning, vocabulary, and morphological awareness in both languages; likewise, English phonological awareness was uniquely predictive of Chinese phonological awareness after the same controls (27% of the variance). While previous studies (e.g. Chow et al. 2005; Gottardo et al. 2001, 2006; Wang et al. 2005; Luk & Bialystok 2008) have reported cross-linguistic correlations of phonological awareness in Chinese-English bilinguals, our study is among the first to demonstrate a unique mutual prediction of phonological awareness after controlling for key reading related variables, especially morphological awareness. Our findings suggest that for Chinese-English bilinguals, phonological skill in one language facilitates the development of the same construct in the other language and this facilitation goes in both directions. Thus, children’s ability to reflect on and manipulate phonological units can be applied cross-linguistically between Chinese and English. In addition to transfer of phonological awareness at the construct level, we examined transfer of phonological awareness to word reading. Our results showed

 Yang Cathy Luo, Xi Chen & Esther Geva

that Chinese phonological awareness had a significant direct effect on concurrent English word reading. As mentioned earlier, most previous studies on transfer of phonological awareness only controlled for phonological awareness in one language (Cheung et al. 2010; Chow et al. 2005; Gottardo et al. 2001; Keung & Ho 2009; Pan et al. 2011; Wang et al. 2006, 2009). Our study adopted a more stringent test, controlling for phonological awareness and morphological awareness in both languages. Since phonological awareness and morphological awareness are interrelated metalinguistic skills, it is theoretically important to test whether there are unique cross-linguistic effects of phonological awareness after partialling out shared variance with morphological awareness. Our study extends previous findings (e.g. Chow et al. 2005; Gottardo et al. 2001, 2006; Wang et al. 2005) by demonstrating that Chinese phonological awareness is uniquely related to English word reading in Chinese-English bilinguals. Our study is one of a few that examined cross-linguistic effects of phonological awareness on word reading longitudinally. There was an indirect effect with Time 1 Chinese phonological awareness predicting Time 2 English word reading (β = .30, p < .01) through the mediation of the autoregressive effect of Time 1 ­English word reading. However, no direct effects from Time 1 phonological awareness to Time 2 word reading were found. Our results were different from those of Chow et al. (2005) and Pan et al. (2011) who observed direct transfer of phonological awareness longitudinally. The discrepancy is likely due to the difference in the control variables used. Notably, we had more control variables in our models than either previous study. We controlled for the autoregressor as well as metalinguistic skills in both languages at Time 1, whereas one or several of these control variables were absent in Chow et al. (2005) and Pan et al. (2011). Since few studies have investigated longitudinal effects in cross-linguistic transfer, more research is needed in this area. Our study produced interesting results regarding the direction of phonological transfer to word reading. Specifically, Chinese phonological awareness uniquely predicted English word reading, but no unique prediction was found from English phonological awareness to Chinese character reading. These results are consistent with those of most of the previous studies (Gottardo et al. 2006; Wang et al. 2005). The direction of transfer may be in part determined by the cognitive demands of reading in the language of the outcome variable. Phonological awareness plays a more important role in reading English, a language written with an alphabetic script, than in Chinese, a language using a logographic script (McBride-Chang et al. 2005). As a result, phonological awareness is more likely to transfer from Chinese to support English reading. However, Keung and Ho (2009), in a study involving Cantonese-speaking Hong Kong children, found transfer from English phonological awareness to Chinese character reading. It is not clear to us why this



Cross-linguistic transfer of phonological and morphological awareness 

study demonstrated a different pattern in terms of the direction of transfer. Future research needs to explore factors that influence the direction of transfer. To summarize, our study addressed the important question of how crosslinguistic transfer of phonological awareness occurs by revealing transfer at two different levels, at the construct level through the association between Chinese and English phonological awareness and at the reading outcome level through the association between Chinese phonological awareness and English word reading. It is likely that transfer at the construct level forms the foundation of transfer at the reading level. Chinese and English share three aspects of phonological awareness, syllable, onset-rhyme, and phoneme.1 Thus, sensitivity toward these common phonological units can be applied across the two languages. Our findings on cross-linguistic transfer of phonological awareness between Chinese and English, two typologically distant languages, confirm and extend previous research and further strengthen the notion that phonological awareness is a language-universal construct. On the other hand, our findings highlight differences between Chinese and English reading. Reading English demands more phonological awareness than reading Chinese. As a result, transfer occurs from Chinese phonological awareness to English reading but not the other way around. 5.2  Cross-linguistic effects of compound awareness on word-level reading Similar to phonological awareness, transfer of morphological awareness between Chinese and English was also found at the construct level. Specifically, English compound awareness explained 21% of the variance in Chinese compound awareness, after controlling for nonverbal reasoning, vocabulary, and phonological awareness in both languages. Similarly, Chinese compound awareness uniquely explained 18% of the variance in English compound awareness after the same controls. Consistent with previous research (Pasquarella et al. 2011; Cheung et al. 2010; Zhang et al. 2010), these findings suggest that compound awareness was related across Chinese and English and a heightened sensitivity to compound structures in one language facilitates similar sensitivity in the other language. As we mentioned earlier, Chinese and English have similar compounding features. The transfer of compound awareness at the construct level is likely based on these shared features. Our study did not detect direct cross-linguistic effects of morphological awareness on word reading between Chinese and English, concurrently or longitudinally. Notably, previous research produced at best weak evidence for transfer

.  Chinese is a tonal language, so phonological awareness consists of an additional aspect, tone awareness, in Chinese.

 Yang Cathy Luo, Xi Chen & Esther Geva

of compound awareness to word reading in Chinese-English bilinguals. With the exception of Cheung et al. (2010), none of the previous studies observed crosslinguistic transfer from Chinese compound awareness to English word reading. As previously noted, we believe that an important factor that influences crosslinguistic transfer is the demand of reading in the language of the outcome variable. Given that English has far fewer compound words than inflected or derived words, compound awareness plays a less important role in reading English than inflectional awareness or derivational awareness. Accordingly, Chinese compound awareness is unlikely to have an effect on English word reading. Cheung et al. (2010), the only study that observed cross-linguistic transfer from Chinese compound awareness to English word reading, was conducted in Hong Kong, a society where Chinese is the dominant language. It is possible that linguistic context (in terms of the dominant language of the society) has an impact on cross-linguistic transfer. This issue needs to be explored in future research. With respect to cross-linguistic transfer from English compound awareness to Chinese character reading, our findings were consistent with those of Wang et al. (2006) and Pasquarella et al. (2011) who showed that English compound awareness did not contribute to Chinese character reading after partialling out the effects of other reading related variables. By contrast, such findings were inconsistent with those of Wang et al. (2009) who reported that English compound awareness explained unique variance in Chinese character reading, even though we did observe a significant indirect transfer effect from English compound awareness to Chinese character reading, mediated through Chinese compound awareness. The discrepancy in the studies may be attributed to the different control measures included. For example, both our study and Wang et al. (2006) controlled for multiple aspects of Chinese phonological awareness (i.e. both Chinese onset and rhyme awareness in Wang et al. 2006 and syllable, onset, and rhyme awareness in our study), whereas Wang et al. (2009) only controlled for Chinese onset awareness. Thus, it is possible that controlling for multiple aspects of Chinese phonological awareness increased the shared variance between the phonological and morphological measures, which in turn reduced the unique contribution of morphological awareness to reading. 6.  Conclusion To conclude, our study examined cross-linguistic transfer of phonological awareness and morphological awareness at the construct level and at the reading level in Chinese-English bilinguals. With respect to phonological awareness, we observed evidence of transfer at both levels, between parallel measures



Cross-linguistic transfer of phonological and morphological awareness 

in Chinese and ­English, and from Chinese phonological awareness to English word reading. The fact that transfer occurs at both levels between two typologically distant languages suggests that phonological awareness is a language universal construct and a strong cross-linguistic predictor of word reading success in bilingual children. For morphological awareness, although there was also evidence of transfer at the construct level, transfer at the reading level was more limited with only an indirect effect from English compound awareness to Chinese character reading. Notably, previous studies produced mixed results on transfer of compound awareness to word reading between Chinese and E ­ nglish. Our finding, together with those of several previous studies (Wang et al. 2006; Pasquarella et al. 2011), seem to suggest that morphological awareness is a more language specific construct. Despite the fact that Chinese and English have similar compounding structures, transfer of compound awareness to reading between the two languages is limited and not observed consistently. Finally, the present study demonstrates that an important factor determining the direction of cross-linguistic transfer is the demand of reading in the language of the outcome variable. Relatively speaking, reading Chinese requires more compound awareness whereas reading English requires more phonological awareness. As a result, Chinese reading is more likely to associate with compound awareness and English reading is more likely to associate with phonological awareness, both within and across languages. The present study has several limitations that need to be addressed in further research. First, there are some concerns about the research methodology. The sample size was relatively small for path analyses investigating multivariate relationships. Although there was sufficient statistical power in our study, the reported relationships should be replicated with a larger sample in the future. Another concern is that our Chinese character reading measure consisted of single characters only. While compound awareness has been consistently shown to contribute to character reading in previous research (e.g. Chen, Hao, Geva, Zhu & Shu 2009), it may play an even larger role in reading compound words. Thus, future research should include Chinese reading measures that contain compound items. Second, our test of cross-linguistic transfer was based on correlations. As a result, we could not disentangle skills developed in each of the languages and their respective effects on reading cross-linguistically. As Koda (2008) pointed out, cross-language transfer in biliteracy development is an ever-changing and dynamic process. It could well be that skills develop in the two languages simultaneously for bilingual children. Nevertheless, it is still important to find out whether skills developed and measured in one language causally affect reading in the other. Future research may consider adopting an intervention design to establish causality. To expand the findings of Zhang et al. (2010) demonstrating

 Yang Cathy Luo, Xi Chen & Esther Geva

that Chinese-English bilinguals who received morphological awareness training in either Chinese or English showed more advanced morphological awareness in the other language than those who did not, future studies need to establish that receiving morphological awareness training in one language also leads to improvement of reading abilities in the other language. Despite the limitations, our findings have important implications on literacy development and education of bilingual children. Generally speaking, the evidence of cross-linguistic transfer suggests that for bilingual children, linguistic and cognitive resources established in one language can be accessible and ­beneficial for acquisition of literacy skills in the other language. Our findings show that this is also the case for Chinese-English bilinguals despite the fact that their two languages are typologically distant. Importantly, learning Chinese does not jeopardize but in fact facilitates the development of English literacy skills for ­Chinese-English bilinguals educated in an environment where English is the societal language. Thus, policy makers, teachers, and parents should encourage and support literacy development in both languages.

Acknowledgement This study was supported by a Social Sciences and Humanities Research Council of Canada (SSHRC) standard research grant to Xi Chen and a SSHRC Doctoral Fellowship to Yang Cathy Luo.

References Adams, Marilyn J. (1990). Beginning to read: Thinking and learning about print. Cambridge, MA: MIT Press. Benczes, Reka (2006). Creative compounding in English: The semantics of metaphorical and metonymical noun-noun combinations. Amsterdam: John Benjamins.  doi: 10.1075/hcp.19 Bradley, Lynette & Peter E. Bryant (1983). Categorizing sounds and learning to read: A causal connection. Nature 301(5899): 419–421.  doi: 10.1038/301419a0 Bryant, Peter E., Morag MacLean, Lynette Bradley & John Crossland (1990). Rhyme and alliteration, phoneme detection, and learning to read. Developmental Psychology 26(3): 429–438..  doi: 10.1037/0012-1649.26.3.429 Carlisle, Joanne F. (1995). Morphological awareness and early reading achievement. In L ­ aurie B. Feldman (ed.), Morphological aspects of language processing, 189–209. Hillsdale, NJ: Erlbaum. Carlisle, Joanne F. & Diana M. Nomanbhoy (1993). Phonological and morphological awareness in first graders. Applied Psycholinguistics 14(2): 177–195.  doi: 10.1017/S0142716400009541 Carlisle, Joanne F. (2003). Morphology matters in learning to read: A commentary. Reading Psychology 24: 291–322.  doi: 10.1080/02702710390227369



Cross-linguistic transfer of phonological and morphological awareness 

Chen, Xi, Meiling Hao, Esther Geva, Jin Zhu & Hua Shu (2009). The role of compound awareness in Chinese children’s vocabulary acquisition and character reading. Reading and ­Writing 22: 615–631.  doi: 10.1007/s11145-008-9127-9 Cheung, Him, Kevin K.H. Chung, Simpson W.L. Wong, Catherine McBride-Chang, Trevor B. Penney & Connie S.-H. Ho (2010). Speech perception, metalinguistic awareness, reading, and vocabulary in Chinese-English bilingual children. Journal of Educational Psychology 102(2): 367–380.  doi: 10.1037/a0017850 Chow, Bonnie W.-Y. Catherine McBride-Chang & Stephan Burgess (2005). Phonological processing skills and early reading abilities in Hong Kong Chinese kindergarteners learning to read English as a second language. Journal of Educational Psychology 97: 81–87.

doi: 10.1037/0022-0663.97.1.81

Cisero, Cheryl A. & James M. Royer (1995). The development and cross-language transfer of phonological awareness. Contemporary Educational Psychology 20(3): 275–303.

doi: 10.1006/ceps.1995.1018

Comeau, Liane, Pierre Cormier, Éric Grandmaison & Diane Lacroix (1999). A longitudinal study of phonological processing skills in children learning to read in a second language. Journal of Educational Psychology 91(1): 29–43.  doi: 10.1037/0022-0663.91.1.29 Deacon, S. Hélène (2012). Sounds, letters and meanings: The independent influences of phonological, morphological and orthographic skills on early word reading accuracy. Journal of Research in Reading 35(4): 456–475.  doi: 10.1111/j.1467-9817.2010.01496.x Deacon, S. Hélène & John R. Kirby (2004). Morphological awareness: Just ‘more phonological’? The roles of morphological and phonological awareness in reading development. Applied Psycholinguistics 25(02): 223–238.  doi: 10.1017/S0142716404001110 Dunn, Lloyd M. & Douglas M. Dunn (1997). Peabody picture vocabulary test (3rd edition). ­Circle Pines, MN: American Guidance Service. Durgunoğlu, Aydin Y., William E. Nagy & Barbara J. Hancin-Bhatt (1993). Cross-language transfer of phonological awareness. Journal of Educational Psychology 85(3): 453–465.

doi: 10.1037/0022-0663.85.3.453

Elementary Education Teaching and Research Center, Beijing Education and Science Institute (1996). Elementary school Chinese textbooks. Beijing, China: Beijing Publishers. Gottardo, Alexandra, Penny Chiappe, Bernice Yan, Linda Siegel & Yan Gu (2006). Relationships between first and second language phonological processing skills and reading in ­Chinese-English speakers living in English-speaking contexts. Educational Psychology 26(3): 367–393.  doi: 10.1080/01443410500341098 Gottardo, Alexandra, Bernice Yan, Linda Siegel & Lesly Wade-Woolley (2001). Factors related to English reading performance in children with Chinese as a first language: More evidence of cross-language transfer of phonological processing. Journal of Educational Psychology 93(3): 530–542.  doi: 10.1037/0022-0663.93.3.530 Ho, Connie S.-H. & Peter Bryant (1997). Phonological skills are important in learning to read Chinese. Developmental Psychology 33(6): 946–951.  doi: 10.1037/0012-1649.33.6.946 Hu, Chieh-Fang & Hugh W. Catts (1998). The role of phonological processing in early reading ability: What we can learn from Chinese. Scientific Studies of Reading 2(1): 55–79.

doi: 10.1207/s1532799xssr0201_3

Katamba, Francis (1993). Morphology. Houndmills, London: MacMillan.

doi: 10.1007/978-1-349-22851-5

Keung, Yuen-Ching & Connie S.-H. Ho (2009). Transfer of reading-related cognitive skills in learning to read Chinese (L1) and English (L2) among Chinese elementary school ­children. Contemporary Educational Psychology 34(2): 103–112.  doi: 10.1016/j.cedpsych.2008.11.001

 Yang Cathy Luo, Xi Chen & Esther Geva Koda, Keiko. (2008). Impacts of prior literacy experience on second-language learning to read. In Keiko Koda & Annette M. Zehler (eds.), Learning to read across languages: Cross-­ linguistic relationships in first- and second-language literacy development, 68–96. Mahwah, NJ: Routledge. Kuo, Li-Jen & Richard C. Anderson (2006). Morphological awareness and learning to read: A cross-language perspective. Educational Psychologist 41(3): 161–180.

doi: 10.1207/s15326985ep4103_3

Lafrance, Adèle & Alexandra Gottardo (2005). A longitudinal study of phonological processing skills and reading in bilingual children. Applied Psycholinguistics 26(4): 559–578.

doi: 10.1017/S0142716405050307

Luk, Gigi & Ellen Bialystok (2008). Common and distinct cognitive bases for reading in ­English-Cantonese bilinguals. Applied Psycholinguistics 29: 269–289.  doi: 10.1017/S0142716407080125 Mahony, Diana, Maria Singson & Virginia Mann (2000). Reading ability and sensitivity to morphological relations. Reading and Writing 12(3): 191–218.  doi: 10.1023/A:1008136012492 McBride-Chang, Catherine (2004). Children’s literacy development. London: Edward Arnold/ Oxford Press. McBride-Chang, Catherine, Jeung-Ryeul Cho, Hongyun Liu, Richard K. Wagner, Hua Shu, Aibao Zhou, Cecilia S.-M. Cheuk & Andrea Muse (2005). Changing models across cultures: Associations of phonological awareness and morphological structure awareness with vocabulary and word recognition in second graders from Beijing, Hong Kong, Korea, and the United States. Journal of Experimental Child Psychology 92(2): 140–160.

doi: 10.1016/j.jecp.2005.03.009

McBride-Chang, Catherine & Connie S.-H. Ho (2000). Naming speed and phonological awareness in Chinese children: Relations to reading skills. Journal of Psychology in Chinese ­Societies 1(1): 93–108. McBride-Chang, Catherine, Hua Shu, Aibao Zhou, Chun P. Wat & Richard K. Wagner (2003). Morphological awareness uniquely predicts young children’s Chinese character recognition. Journal of Educational Psychology 95(4): 743–751.  doi: 10.1037/0022-0663.95.4.743 Packard, Jerome L. (2000). The morphology of Chinese: A linguistic and cognitive approach. ­Cambridge: Cambridge University Press.  doi: 10.1017/CBO9780511486821 Pan, Jinger, Catherine McBride-Chang, Hua Shu, Yuping Zhang & Hong Li (2011). What is in the naming? A 5-year longitudinal study of early rapid naming and phonological sensitivity in relation to subsequent reading skills in both native Chinese and English as a second language. Journal of Educational Psychology 103(4): 897–908.  doi: 10.1037/a0024344 Pasquarella, Adrian, Xi Chen, Katie Lam, Yang C. Luo & Gloria Ramirez (2011). Cross-language transfer of morphological awareness in Chinese-English bilinguals. Journal of Research in Reading 34(1): 23–42.  doi: 10.1111/j.1467-9817.2010.01484.x Raven, John C. (1958). Standard progressive matrices: SPM. London, UK: H. K. Lewis. Rubin, Donald B. (1987). Multiple imputation for nonresponse in surveys. New York: John Wiley & Sons.  doi: 10.1002/9780470316696 Schafer, Joseph L. (1997). Analysis of incomplete multivariate data. London, UK: Chapman & Hall.  doi: 10.1201/9781439821862 Shu, Hua, Catherine McBride-Chang, Sina Wu & Hongyun Liu (2006). Understanding Chinese developmental dyslexia: Morphological awareness as a core cognitive construct. Journal of Educational Psychology 98(1): 122–133.  doi: 10.1037/0022-0663.98.1.122



Cross-linguistic transfer of phonological and morphological awareness 

Singson, Maria, Diana Mahony & Virginia Mann (2000). The relation between reading ability and morphological skills: Evidence from derivational suffixes. Reading and Writing 12(3): 219–252.  doi: 10.1023/A:1008196330239 Siok, Wai-Ting & Paul Fletcher (2001). The role of phonological awareness and visual-­ orthographic skills in Chinese reading acquisition. Developmental Psychology 37(6): 886–899.  doi: 10.1037/0012-1649.37.6.886 Stanovich, Keith E. (1986). Matthew effects in reading: Some consequences of individual differences in the acquisition of literacy. Reading Research Quarterly 21(4): 360–407.

doi: 10.1598/RRQ.21.4.1

Stanovich, Keith E. (1992). Speculations on the causes and consequences of individual differences in early reading acquisition. In Philip. B. Gough, Linnea C. Ehri & Rebecca Treiman (eds.), Reading acquisition, 307–342. Hillsdale, NJ: Lawrence Erlbaum Associates. Tabachnick, Barbara G. & Linda S. Fidell (2006). Using multivariate statistics (5th edition). ­Boston: Allyn & Bacon. Tong, Xiuli & Catherine McBride-Chang (2010). Chinese-English biscriptal reading: Cognitive component skills across orthographies. Reading and Writing 23: 293–310.

doi: 10.1007/s11145-009-9211-9

Wagner, Richard K. & Joseph K. Torgesen (1987). The nature of phonological processing and its causal role in the acquisition of reading skills. Psychological bulletin 101(2): 192–212.

doi: 10.1037/0033-2909.101.2.192

Wagner, Richard K., Joseph K. Torgesen & Carol A. Rashotte (1994). Development of readingrelated phonological processing abilities: New evidence of bidirectional causality from a latent variable longitudinal study. Developmental Psychology 30(1): 73–87.

doi: 10.1037/0012-1649.30.1.73

Wagner, Richard K., Joseph K. Torgesen & Carol A. Rashotte (1999). Comprehensive test of ­phonological processing. Austin, TX: PRO-ED. Wang, Min, Chengxi Cheng & Shi-Wei Chen (2006). Contribution of morphological awareness to Chinese-English biliteracy acquisition. Journal of Educational Psychology 98(3): 542–553.  doi: 10.1037/0022-0663.98.3.542 Wang, Min, Charles A. Perfetti & Ying Liu (2005). Chinese-English biliteracy acquisition: Crosslanguage and writing system transfer. Cognition 97(1): 67–88.

doi: 10.1016/j.cognition.2004.10.001

Wang, Ming, Chen Yang & Chengxi Cheng (2009). The contributions of phonology, orthography, and morphology in Chinese-English biliteracy acquisition. Applied Psycholinguistics 30(2): 291–314.  doi: 10.1017/S0142716409090122 Woodcock, Richard W., Kevin S. McGrew & Nancy Mather (2001). Woodcock-Johnson III tests of achievement. Itasca, IL: Riverside Publishing. Zhang, Dongbo (2012). Linguistic distance effect on cross-linguistic transfer of morphological awareness. Applied Psycholinguistics 34(5): 917–942.  doi: 10.1017/S0142716412000070 Zhang, Jie, Richard C. Anderson, Hong Li, Qi Dong, Xinchun Wu & Yan Zhang (2010). Crosslanguage transfer of insight into the structure of compound words. Reading and Writing 23(3): 311–336.  doi: 10.1007/s11145-009-9205-7

The effects of bilingual education on the English language and literacy outcomes of Chinese-speaking children Kathleen Hipfner-Boucher, Katie Lam & Xi Chen University of Toronto, Canada

To evaluate the effects of bilingual education on minority-language children’s English language and literacy outcomes, we compared grade 1 Chinese-speaking Canadian children enrolled in three different instructional programs (French Immersion, Chinese-English Paired Bilingual, English-only). ANCOVA results revealed that the French immersion children outperformed the other two groups on measures of English phonological awareness and word reading and that the bilingual groups were comparable to monolingual English norms on a test of receptive vocabulary. Multiple regression analyses were conducted to examine cross-language transfer of skills. French morphological awareness explained unique variance in English word reading and vocabulary for the French immersion group. For the other two groups, Chinese phonological awareness was significantly related to English word reading. Our results suggest that instruction in French or Chinese does not delay the development of early English language and literacy skills for Chinese-speaking children, as the children may be able to leverage skills from their other language to facilitate their English learning. Keywords:  Bilingual education; French immersion; cross-language transfer

1.  Introduction The cultural and linguistic composition of classrooms in North American schools is highly diverse (August & Shanahan 2006; National Centre for Education Statistics 2009). For example, recent Canadian census data revealed that one-fifth of the population spoke a language other than English or French, the two official languages in which publicly funded education is guaranteed (Statistics Canada 2012). Mandarin figures among the top ten minority languages spoken by Canadian elementary school children and is the sole language spoken in the home by the vast majority of the Chinese community (Statistics Canada 2012). The acquisition of language and literacy skills in the majority language contributes ­substantially

doi 10.1075/bct.89.06hip © 2016 John Benjamins Publishing Company

 Kathleen Hipfner-Boucher, Katie Lam & Xi Chen

to the academic achievement of minority-language children (August & Shanahan 2006). With this in mind, the present study was designed to compare the E ­ nglish language and literacy outcomes of Mandarin speaking, grade-one children in ­English-only, paired bilingual (Mandarin-English), and French immersion programs, which offer varying levels of exposure to English-language instruction. A second goal of the study was to examine the role of cross-language transfer in the reading outcomes of bilingual and multilingual Chinese children. 1. 1  Educational programs for minority-language children Two broad categories of educational programming are currently available to ­minority-language students attending North American schools (Francis, Lesaux & August 2006). The first of these is English-only programs which provide contentarea instruction exclusively in English. English-only instruction for minority-­ language children is predicated on the assumption that instruction in a language other than English interferes with English language development (Rossell & Baker 1996; Slavin & Cheung 2005). Most minority-language children in the United States and Canada are schooled in English-only programs (Goldenberg 2008; Markus 2012). Of the three groups of children participating in the current study, one received ­English-only instruction. Outside of their regular school hours, they attended a publicly funded, school-based Mandarin heritage language program where they received Mandarin language and literacy instruction for 2.5 hours every week. The second broad category of programs available to minority-language students attending North American schools is that of bilingual education programs. Genesee (2004) defines bilingual programming as education that aims to promote bilingual competence through delivery of curriculum in each of the students’ two languages for significant portions of the school day. Among the bilingual options, paired bilingual programs offer minority-language children instruction in their home language and in English for part of each day or on alternating days (Slavin & Cheung 2005). One of the three groups of children who participated in the present study was receiving paired bilingual instruction. One-half of their school day was conducted entirely in Mandarin, the other half in English. The efficacy of bilingual education programs has been demonstrated empirically. Most notably, a number of comprehensive reviews of the research literature provide compelling evidence of the positive effect of bilingual instruction in ­promoting both minority and majority language and literacy outcomes of minoritylanguage children (August & Shanahan 2006; Cheung & Slavin 2012; Genesee, Lindholm-Leary, Saunders & Christian 2006; Greene 1998; Rolstad, Mahoney & Glass 2005; Slavin & Cheung 2005; Willig 1985; but see Rossell & Baker 1996). Taken together, these reviews support a basic finding: that at a minimum, learning



The effects of bilingual education on English outcomes 

to read in the home language promotes modest but meaningful gains in literacy achievement in the second language and that its effects are reliable across the elementary and secondary school years (Goldenberg 2008). However, little research has focused specifcally on Chinese-speaking children schooled in North America. It remains unclear whether similar results would be found among these children. A third bilingual programming option open to minority-language children within the Canadian education system is French immersion, in which non-native speakers of French are instructed in French. The goal of French immersion is additive bilingualism, the development of French proficiency while maintaining native proficiency in the majority English first language (Wesche 2002). French immersion program models differ in terms of age of intake and proportion of instruction conducted in French. Early total immersion begins in kindergarten or grade 1. All academic content material is taught exclusively in French until grade 4, when English language arts instruction is initiated. The final group of children involved in the current study was enrolled in an early French immersion program. The school offered full-day French language education beginning in grade 1. As a result, French language and literacy instruction commenced simultaneously for these children. The French and English language and literacy outcomes of French immersion students have been the focus of a number of studies. Not surprisingly, the vast majority attest to the effectiveness of early French immersion in supporting the development of French oral and written skills in majority language students (for a review, see Genesee & Jared 2008). With respect to English language and literacy outcomes, research indicates an initial lag in performance on tests of English vocabulary and reading comprehension among early immersion students relative to monolingual English-speaking children in English-only programs. However, the gap is closed within one to two years following the initiation of English language arts instruction and French immersion students perform at least as well as students in English-only programs in the upper elementary grades (Genesee 2004; Swain & Lapkin 1982; Turnbull, Hart & Lapkin 2003). There has been a steady increase in the number of minority-language students in early French immersion in recent years (Sinay 2010). Minority-language children now comprise 32% of the student population in early French immersion in Canada’s largest district school board (Sinay 2010). This increase comes in spite of limited research investigating performance outcomes in English and French among these students. The most extensive examination of minority-language children in French immersion was conducted by Swain and colleagues in the late 1980’s and early 1990’s. In a series of studies, they demonstrated that minoritylanguage children in middle immersion classrooms who had developed literacy skills in their home language performed at the same level on French language and

 Kathleen Hipfner-Boucher, Katie Lam & Xi Chen

literacy measures as their English-speaking peers (Bild & Swain 1989; Swain & Lapkin 1991; Swain, Lapkin, Rowen & Hart 1990). However, the studies focused only on middle immersion students (i.e. students who began French immersion in the middle elementary grades) and did not examine English literacy skills. More recently, Bérubé and Marinova-Todd (2012) found that on average, minority-­ language early French immersion children tested in grade 4 performed at the level of their English monolingual peers in English word and pseudo-word reading. Again, this study focused on children in middle immersion. Research examining the outcomes of minority-language children in the initial stages of early French immersion programs is lacking. 1.2  Cross-language transfer of phonological and morphological awareness Bilingual education is premised on theories of bilingualism and language transfer that highlight the importance first-language skills play in supporting the development of second language competence (Uchikoshi & Maniates 2010). Success in bilingual education programs has largely been attributed to cross-language transfer (Riches & Genesee 2006). The current study examines the cross-language transfer of phonological and morphological awareness, two domains of metalinguistic knowledge that have been shown to be important for word reading and vocabulary. Phonological awareness refers to the ability to attend to, isolate, and manipulate the sound structure of oral language (Torgesen, Wagner & Rashotte 1994). Research supports the notion of a bi-directional, causal relationship between phonological awareness and word reading (Adams 1990; Snow, Burns & G ­ riffin 1998) as well as a bi-directional relationship between phonological awareness and vocabulary (McBride-Chang et al. 2005; Metsala 1999; Walley, Metsala & Garlock 2003). A substantial body of work has shown that phonological skills acquired in one language account for a significant portion of the variance in reading outcomes in a second language (e.g. Arab-Moghaddam & Sénéchal 2001; Durgunoğlu, Nagy & Hancin-Bhatt 1993; Lafrance & Gottardo 2005). A number of studies have focused specifically on cross-language transfer of phonological awareness in English-dominant French immersion children (Bruck & Genesee 1995; ­Caravolas  & Bruck 1993; Comeau, Cormier, Grandmaison & Lacroix 1999; Jared, Cormier, Levy & Wade-Woolley 2011; MacCoubrey, Wade-­Woolley, Klinger & Kirby 2004). Comeau and colleagues (1999) found that French and English phonological awareness predicted word reading both within and across languages in the elementary grades. MacCoubrey et al. (2004) reported that French word reading in grade 2 was predicted by performance on English phonological measures administered in grade 1. More recently, Jared et al. (2011)



The effects of bilingual education on English outcomes 

found that k­ indergarten English phonological awareness was a significant predictor of French word identification in Grades 1 to 3. Neither the MacCoubrey et al. (2004) nor Jared et al. (2011) studies considered cross-language transfer of skills from French to English. Research is needed examining back transfer in minoritylanguage children in early French immersion. Moreover, none of the studies cited above investigated cross-language transfer of phonological awareness to vocabulary, despite evidence of an association between the two in the L1 research. Cross-language transfer of phonological awareness has also been shown to operate in Chinese-English bilinguals (Chen, Xu, Nguyen, Hong & Wang 2010; Gottardo, Siegel, Yan & Wade-Woolley 2001; Marinova-Todd, Zhao & Bernhardt 2010). For example, Gottardo et al. (2001) found that among Canadian children who were native speakers of Cantonese, performance on a Chinese rhyme detection task was correlated with English rhyme detection and phoneme deletion and with English word reading. However, the sample included children from grades 1 to 8, so it is not clear whether transfer occurred for the youngest children in the study. McBride-Chang and colleagues (2006) demonstrated that Chinese phoneme onset awareness predicted a significant portion of the variance in ­English word knowledge among a sample of emergent Cantonese-English bilingual kindergarten children. Taken together, these studies indicate that phonological awareness is subject to cross-language transfer. Morphological awareness is another metalinguistic skill that is attracting attention among researchers. Morphological awareness refers to the ability to reflect on and manipulate morphemes and to use word formation rules to produce and understand multimorphemic words (Kuo & Anderson 2006). Three types of morphology are common across languages: inflectional, derivational, and compound. Inflections yield different grammatical forms of a word without altering its meaning or part of speech, e.g. read – reads. Derivations involve forming new words through the application of prefixes and suffixes that usually alter the meaning of the root word or change the word-class to which it belongs, e.g. beauty – beautiful. Lastly, compounding involves the combination of two or more words in forming new words, e.g. class + room = classroom. Most multimorphemic words in English are composed of a root and one or more inflectional and/ or derivational affixes, whereas compounding is the main morphological process in Chinese (Arcodia 2007). It is also important to note that in comparison to English, Chinese has a more productive and transparent compound morphology (Zhang et al. 2012). Substantial research has demonstrated significant within-language associations between morphological awareness and reading development (e.g. Carlisle 2000; Casalis & Louis-Alexandre 2000; Ku & Anderson 2003). More recently, there is increasing evidence of cross-language transfer of morphological awareness ­(Deacon, Wade-Woolley & Kirby 2007; Pasquarella et al. 2011; Ramírez, Chen,

 Kathleen Hipfner-Boucher, Katie Lam & Xi Chen

Geva & Kiefer 2010). Transfer is thought to be conditioned by shared morphological features (Koda 2005). In studies of alphabetic languages in which inflectional and derivational structures are similar (i.e. English and French, English and ­Spanish; Deacon et al. 2007; Ramírez et al. 2010, 2013), there is preliminary evidence supporting the cross-language transfer of inflectional awareness and derivational awareness to word reading. Ramírez et al. (2010, 2013) demonstrated that Spanish derivational awareness contributed directly to English word reading among fourth and seventh grade Spanish-speaking English language learners. Deacon et al. (2007) reported that English inflectional awareness measured among French immersion children in grade 1 predicted French word reading concurrently and in each of the subsequent two grades. We know of no studies investigating transfer of derivational awareness among emerging English-French bilingual children. Cross-language transfer of morphological awareness has also been shown to operate in Chinese-English bilinguals. Cross-language transfer of English compound awareness to Chinese vocabulary and word reading has been reported. By contrast, the same studies have largely failed to find transfer of Chinese compound awareness to English outcomes (Pasquarella et al. 2011; Wang et al. 2006, 2009), suggesting that the direction of morphological awareness transfer may be influenced by the morphological structure of the language of the outcome variable (Pasquarella et al. 2011). An exception was a study conducted by Chung and Ho (2010) examining children living in a Chinese-dominant society who were learning English as a second language. They found that Chinese compound awareness predicted English word reading. Research is needed to clarify the role of crosslanguage transfer of compound awareness from Chinese to English. 1.3  The current study The current study is an exploratory study designed to compare the English language and literacy outcomes of minority-language Chinese children in three different educational programs varying in terms of the amount of English language instruction they offered in the first year of elementary school. Specifically, we compared vocabulary and early reading outcomes in three groups of Chinese children: one group schooled in an English-only program, one in a paired ­Mandarin-English program, and one in an early French immersion program. A secondary goal was to examine cross-language patterns of transfer of phonological and m ­ orphological awareness to word reading and vocabulary in each of the three groups. Specifically, hierarchical linear regression was used to examine the role French phonological and morphological awareness played in predicting ­English word reading and vocabulary among the children in French immersion and the role Chinese



The effects of bilingual education on English outcomes 

phonological and morphological awareness played in predicting English word reading and vocabulary among the children in the paired bilingual and Englishonly programs.

2.  Method 2.1  Participants Participants of the study included 75 Chinese-speaking Canadian children who were enrolled in three different instructional programs in a large metropolitan area. The first group (hereafter, the ‘French Immersion’ group) consisted of 22 grade 1 students (50% males, mean age 6 years 10 months) enrolled in a Canadian French immersion program. At the time of testing, the children had been receiving formal instruction in French for approximately eight months. The second group (the ‘Paired Bilingual’ group) comprised 20 students (40% males, mean age 6 years 10 months) enrolled in a Mandarin language transition program. This group was instructed in Mandarin for one half of the day and in English for the other half at a local public school. The final group (the ‘English-only’ group) included 33 children (57% males, mean age 6 years 7 months) enrolled in a mainstream English program, where they received full-day instruction in English. In addition, these children attended publicly-funded, school-based Chinese heritage language classes for an average of 2.5 hours a week where they received instruction in both oral language and literacy skills. The French immersion group and the English-only group were part of larger studies that examined children’s biliteracy development in French and English and in Chinese and English, respectively. Demographic information for the three groups was collected through a family questionnaire completed by the parents at the time of the initial recruitment. Within the French Immersion group, all children but one were born in Canada. The age of immigration of the child born outside of Canada was 11 months. 35% of the children in the Paired Bilingual group were born in Canada. The average age of immigration of the children born outside of Canada was 2 years and 8 months (SD = 20.16 months). 33% of the children from the English-only group were born in Canada. For those born outside of Canada, the average age of immigration was 10 months (SD = 18.24 months). At the time of the present study, all children from the French Immersion group and the English-only group reported speaking Chinese (Cantonese or Mandarin) at home to varying extents. The average level of maternal education for the French Immersion and the Paired Bilingual groups was a professional degree; the average

 Kathleen Hipfner-Boucher, Katie Lam & Xi Chen

level of maternal education for the English-only group was a university degree. Maternal education was used as a covariate in the regression analyses. 2.2  Measures Each of the three groups received a comprehensive battery of language and ­literacy measures in English and another language (French for the French I­mmersion group and Chinese for Paired Bilingual and English-only groups) in their ­respective projects. Only measures that are relevant to the purposes of the present study are described here. These measures include English word reading, English vocabulary, and English phonological awareness for all three groups. To investigate cross-­language transfer, we included phonological awareness and ­morphological awareness tasks in French for the French Immersion group and parallel tasks in Mandarin Chinese for the Paired Bilingual and the English-only groups. English word reading: English decoding skills were measured using the LetterWord Identification subtest from the Woodcock Language Proficiency Battery (WLPB) (Woodcock 1984). This test required children to identify 14 letters and to read 62 words of increasing difficulty. The test was discontinued if the child read 6 consecutive words incorrectly. The score was the total number of words read correctly. English receptive vocabulary: To measure students’ English receptive vocabulary in the French Immersion and Paired Bilingual groups, we administered the Peabody Picture Vocabulary Test, Fourth Edition, Form A (PPVT-IV A) (Dunn & Dunn 2007). In this task, words are presented orally and students are asked to choose from four choices the picture that best represents the word. Testing was discontinued when children failed more than 8 items within a set of 12 words. For the English-only group, a shortened version of the Peabody Picture Vocabulary Test, Third Edition, Form III A (PPVT-III A) (Dunn & Dunn 1997) was administered. To maintain the same progression of item difficulty as the original task, every third item from the original task was selected to create the shortened version, for a total of 60 items. No ceiling rule was implemented. The modifications were made to allow for group administration. The experimenter read each item twice and the children selected the picture that described the word heard. Phonological awareness: English phonological awareness was measured using the Comprehensive Test of Phonological Processing (CTOPP) Elision subtest (Wagner, Torgesen & Rashotte 1999). The French phonological awareness task was an experimental task designed by the researchers using the same test structure as the CTOPP Elision subtest (Wagner et al. 1999). For the first three items, children were asked to delete a syllable from a two syllable compound word (e.g. surtout without sur is tout) or two syllable word (e.g. poisson without son is pois).



The effects of bilingual education on English outcomes 

For the remaining items, children were asked to delete a single phoneme from the initial (e.g. fleur without /f/ is leur), medial (e.g. mardi without /d/ is mari), or final (mouche without /∫/ is mou) position. Six practice items with feedback were administered to ensure that the children had understood the task. Testing discontinued when the child failed three consecutive items. The Chinese phonological awareness task was a deletion task designed by the researchers and included both real and pseudo syllables. Children were required to delete a syllable from a multi-syllabic word or a phoneme in the initial or final position of a syllable. For example, children were asked to delete the syllable /kong3/ from the two-syllable pseudo word /xun1kong3/ and to delete /s/ from the real syllable /se4/. Six practice items were given to verify that the children had understood the task. Morphological awareness: The French morphological awareness task was adapted from Carlisle (2000) and tapped children’s derivational awareness. For half the items in this task, children were orally presented with a root word and a sentence with a word missing and were asked to produce a derived form of the root word to complete the sentence. For example, Ecrire: J’aime ton écriture ‘Write: I like your writing’. For the other half of the items, children were presented with a derived word and were asked to produce its root form to complete a sentence. For example, Confort: Ma nouvelle chaise est très confortable. ‘Comfort: My new chair is very comfortable’. Children were given three practice trials with feedback to verify that they had understood the task. The Chinese morphological awareness task was a compound awareness task adapted from McBride-Chang et al. (2005). The experimenter provided the definition of an animal or an object that was already familiar to the child, then asked the child to create a name for an imaginary animal or object that bore some resemblance to it. For example, 斑马是身上有斑纹的一种马, 那么身上有斑纹的牛我们叫 什么? ‘Striped horse (Zebra) is a kind of horse with stripes on the body. What should we call a cow with stripes on the body?’ (斑牛striped cow). To reduce the effect of oral vocabulary on performance, the definitions were familiar to children and the answers were pseudo-words. 2.3  Procedures For all the groups, research assistants who were proficient in the testing language administered the language measures to the students individually in a quiet and vacant classroom at the schools. Task administration was divided into two sessions, one with English measures and the other with Chinese (for the Paired Bilingual and the English-only groups) or French (for the French Immersion group) measures. Sessions were counterbalanced for order of language across students.

 Kathleen Hipfner-Boucher, Katie Lam & Xi Chen

3.  Results The means (proportions correct) and standard deviations of the measures for each group are presented in Table 1. Mean standard scores for English phonological awareness, receptive vocabulary, and word reading are also reported in the table where available. As indicated by the mean standard scores, the English phonological awareness and word reading scores of all three groups were comparable to the normative sample of monolingual English speakers. Standard scores also suggested that children in the French Immersion and the Paired Bilingual groups approximated those of the normative sample with respect to their receptive vocabulary. Skewness and kurtosis values fell within the acceptable range (i.e. statistic/SE < ± 3.09) for all measures with the exception of the Chinese phonological awareness task for both the Paired Bilingual and English-only groups. Following guidelines set out in Tabachnick and Fidell (2006), a square root transformation was performed to bring skewness and kurtosis values within the acceptable range for this task. All further analyses were performed with the transformed variable. 3.1  Comparing performance levels We conducted two separate one-way analyses of covariance (ANCOVA) to compare the performance levels among the three groups of children on the English phonological awareness and word reading tasks. For both ANCOVAs, ­Instructional Program (French Immersion, Paired Bilingual, English-only) was entered as the independent variable in the analysis and maternal education and time in Canada were entered as covariates. The ANCOVA for English phonological awareness was significant, F(2, 67) = 3.73, MSE = 26.21, p = .029, η2partial = .10. The mean scores on the English phonological awareness task adjusted for maternal education and time in Canada were M = 13.62, M = 12.79, and M = 9.92 for the French Immersion group, Paired Bilingual group, and English-only group, respectively. Post-hoc tests were conducted to evaluate pairwise differences among these adjusted means. Based on the Least Significant Difference (LSD) procedure, the French Immersion group’s adjusted mean score was significantly higher than that of the English-only group, p = .013. On the other hand, no significant differences were found between the French Immersion and the Paired Bilingual groups, p = .630; the difference between the Paired Bilingual and the English-only groups approached significance (p = .069). The ANCOVA conducted for the English word reading task was also significant, F(2, 67) = 3.96, MSE = 91.03, p = .024, η2partial =.11. The mean scores on the English word reading task adjusted for maternal education and time in

The effects of bilingual education on English outcomes  Table 1. Means (proportions correct) and standard deviations of measures French immersion

Paired bilingual

SD

M

Maternal education

5.05

1.00

5.20

.89

4.87

.90

Time in Canada (in years)

6.55

0.55

5.16

1.79

6.10

1.34

125.56

18.79

126.56

16.79

–

Matrix Analogies Test (MAT) standard score

SD

English-only

M

ɑ

M

SD

–

English Measures Phonological awareness (20) Phonological awareness standard score Word reading (76) Word reading standard score Vocabulary (228; 60)

.61 (shortened form)

Vocabulary standard score

0.71

0.26

0.61

0.27

0.48

0.27

14.50

3.23

12.63

3.47

11.77

3.43

0.61

0.11

0.50

0.11

0.49

0.16

128.86

10.00

118.21

10.81

120.10

17.51

0.65

0.10

0.60

0.15

0.56

0.08

103.00

12.08

96.47

15.83

–

–

French Measures Phonological awareness (20)

.92

0.60

0.24

–

–

–

–

Morphological (derivational) awareness (12)

.75

0.48

0.18

–

–

–

–

Phonological awareness (24)

.91

–

–

0.82

0.21

0.74

0.25

Morphological (compound) awareness (15)

.84

–

–

0.37

0.33

0.52

0.21

Chinese Measures

Note: For the English vocabulary task, there were 228 items in the PPVT-4 A that was administered to the French Immersion and Paired Bilingual groups and 60 items in the shortened version of the PPVT-III A that was administered to the English-only group.

 Kathleen Hipfner-Boucher, Katie Lam & Xi Chen

­Canada were M = 44.74, M = 39.28, and M = 37.13 for the French Immersion group, Paired Bilingual group, and English-only group, respectively. Post-hoc tests based on the LSD procedure using the adjusted mean scores indicated that the French Immersion group performed significantly better than the Englishonly group (p = .007). The difference between the French Immersion and the Paired Bilingual group approached significance (p = .091). The Paired Bilingual and the English-only groups did not differ significantly in their performance, p = .461. To compare the performance levels on English vocabulary between the French Immersion and the Paired Bilingual groups, a one-way ANCOVA was conducted. Since the English-only group received a different receptive vocabulary task, the group was excluded from this analysis. Thus, Instructional Program (French Immersion, Paired Bilingual) was entered as the independent variable in the analysis and maternal education and time in Canada were entered as covariates. Results indicated that the ANCOVA was not significant, F(1, 38) = 0.65, MSE = 538.097, p = .425, η2partial =.02. Thus, the two groups were comparable with respect to their English vocabulary skills when maternal education and their length of stay in Canada were taken into account. 3.2  C  ross-language relations of phonological and morphological awareness to word reading and vocabulary Intercorrelations among the literacy and language measures administered to the French Immersion group are displayed in Table 2. The correlations among measures administered to the Paired Bilingual and the English-only groups are reported in Table 3. For the French Immersion group, French phonological awareness was significantly correlated with English phonological awareness; conversely, it was only marginally correlated with English word reading and receptive v­ ocabulary. French morphological awareness was significantly associated with the two ­English measures (i.e. word reading and vocabulary). For both Paired B ­ ilingual and ­English-only groups, the correlation between Chinese phonological awareness and English word reading was significant. By contrast, Chinese m ­ orphological awareness had limited relations with English word reading. For both groups, ­neither Chinese phonological awareness nor morphological ­awareness was significantly correlated with English vocabulary. Hierarchical regression analyses were conducted to examine cross-language transfer of phonological awareness and morphological awareness in the three groups of children. For the French Immersion group, we examined transfer from French phonological and morphological awareness to English word reading and vocabulary. For the Paired Bilingual and English-only groups, we focused



The effects of bilingual education on English outcomes 

Table 2.  Pearson correlation matrix of variables for French Immersion group  1

 2

 3

 4

 5

 6

1.  Time in Canada

 –

2.  Maternal education

.03

 –

3.  Eng. phonological awareness

.28

.26

 –

4.  Fr. phonological awareness

.36~

.11

.55**

 –

.24

.36~

.20

.22

 – .55**

 –

.51*

.48*

5.  Fr. morphological awareness 6.  Eng. word reading

.26

.35

.57**

.37~

7.  Eng. receptive vocabulary

.54**

.15

.47*

.41~

~p

< .10, *p < .05, **p < .01, ***p < .001

Table 3.  Pearson correlation matrix of variables for Paired Bilingual group and ­English-only group   1

  2

  3

 4

  5

 6

 7

 –

.03

.30

.30

.10

.40*

.62**

–.13

 –

.45*

.37*

.52**

.44*

.18

3.  Eng. phonological awareness

.24

–.10

 –

.61***

.38*

.69***

.37*

4.  Ch. phonological awareness

–.08

–.22

.43~

 –

.50**

.56**

.23

.15

 .37~

.04

.22

   –

.23

.11

.12

–.13

.63**

.57*

–.11

   –

.39*

–.06

–.23

.46*

.26

–.26

.36

   –

1.  Time in Canada 2.  Maternal education

5.  Ch. morphological awareness 6.  Eng. word reading 7.  Eng. receptive vocabulary

Note: The lower-left to the diagonal is the correlation matrix for children in the Paired Bilingual group; the upper-right to the diagonal is the correlation matrix for children in the English-only group. ~p < .10, *p < .05, **p < .01, ***p < .001

on transfer from Chinese phonological and morphological awareness to English outcomes. For each group, a regression was performed with maternal education entered in the first step, phonological awareness in the second step, followed by morphological awareness in the last step. The analyses conducted for the three groups are presented in Table 4. For the French Immersion group, French phonological awareness did not make a significant contribution to either English word reading or vocabulary. By contrast, when French morphological awareness was entered in the last step, it accounted for a significant amount of unique variance in both English word reading (15.5%) and vocabulary (17.8%). Final beta weights also suggest that French morphological awareness is a unique predictor of English word reading and vocabulary.

 Kathleen Hipfner-Boucher, Katie Lam & Xi Chen

Table 4.  Hierarchical linear regressions predicting English reading measures Outcome measure Eng. word reading Step and predictors

  ∆R2

   β

Eng. vocabulary ∆R2

  β

French Immersion group 1.  Maternal education

.125

.172

.022

-.048 .313 .461*

2.  Fr. Phonological awareness

.110

.258

.154~

3.  Fr. Morphological awareness

.155*

.429*

.178*

Paired Bilingual group 1.  Maternal education

.075

2.  Ch. Phonological awareness

.277*

3.  Ch. Morphological awareness

.089

–.025

.035

-.026

.048

.312

.065

-.284

.369~

.031

.105

.516*

.032

.195

.000

-.006

.642** –.333

English-only group 1.  Maternal education

.215*

2.  Ch. Phonological awareness

.170*

3.  Ch. Morphological awareness

.021

~p

–.188

< .10, *p < .05, **p < .01

Across the Paired Bilingual and English-only groups, Chinese phonological awareness made a significant contribution to English word reading, explaining 27.7% and 17% of additional variance in the Paired Bilingual and E ­ nglish-only group, respectively. Final beta weights suggest that Chinese phonological a­ wareness was a unique predictor of English word reading for both groups. Conversely, ­Chinese morphological awareness did not account for a significant proportion of unique variance in English word reading for the two groups. For the Paired ­Bilingual group, Chinese phonological awareness and morphological awareness only accounted for small amounts of variance in English vocabulary that were not statistically significant. Similar results were found for the English-only group.

4.  Discussion In this exploratory study, we examined the impact of varying levels of English language instruction on the English language and literacy outcomes of m ­ inority­language children. Specifically, we compared the phonological awareness and word reading skills of native Mandarin-speaking children in French ­Immersion, in a paired Mandarin-English bilingual program, and in an English-only program



The effects of bilingual education on English outcomes 

in the first year of elementary school. We found that after controlling for mother’s education and length of residency in Canada, the French Immersion group performed significantly better than the English-only group on both measures. There were no differences between the French Immersion and Paired Bilingual groups. When the performance of the two bilingual education groups was compared to population norms on tests of receptive vocabulary, both groups performed at a level comparable to the monolingual normative sample. Furthermore, we examined the role of cross-language transfer in supporting word reading and vocabulary in the three groups of children. Among the French Immersion students, French morphological awareness was a significant predictor of English word reading and English receptive vocabulary. Among the children in the paired bilingual and English-only programs, Chinese phonological awareness predicted word reading only. The convergent findings comparing the performance of the L1 Chinese children in the French immersion and paired bilingual programs to the English-only group on the one hand and to monolingual population norms on the other suggest that bilingual education does not delay the development of English language and literacy skills. Although the French Immersion and Paired Bilingual groups were receiving instruction in a language other than English for 100% and 50% of their school day, respectively, their English vocabulary, phonological awareness, and word reading skills were developing on a par with both comparison groups. These results are consistent with findings reported elsewhere, indicating that bilingual education does not negatively impact on English academic outcomes (e.g. August & Shanahan 2006; Genesee, Lindholm-Leary, Saunders & Christian 2006). Interestingly, our findings seem to be inconsistent with the finding in the French immersion literature of a lag in English vocabulary skills among early French immersion Anglophone children relative to their English-only peers in the years before English language arts instruction is initiated (e.g. Swain & Lapkin 1982). Assessed by PPVT, the English vocabulary scores of the minority-language Chinese children in our study were comparable to an English monolingual normative sample. This discrepancy may be the result of sampling bias. The children who participated in our study were of high socio-economic status (SES) (as indicated by mean maternal education levels corresponding to a professional degree) whereas the normative sample would likely have been stratified across various levels of SES. A second possible explanation relates to the measures used in the original studies. The authors used a comprehensive English measure that assessed literacy skills and mathematical problem solving. The performance of the French Immersion group on the mathematics component of the measure may have been compromised by the children’s limited understanding of the English instructions, resulting in an underestimation of their abilities (Lapkin, personal

 Kathleen Hipfner-Boucher, Katie Lam & Xi Chen

c­ ommunication, April 26, 2013). Alternatively, it may be the case that the original result, reported over 30 years ago, reflects a trend that is no longer current. Either way, our finding points to the need for renewed studies evaluating the effects of early French immersion programs on majority-language and minority-language students’ language and literacy outcomes. Cross-language transfer of skill from the L1 to the L2 is widely believed to be one of the factors that undergirds L2 performance in minority-language children. Our findings related to morphological and phonological awareness in French and in Mandarin largely supported that view. With respect to cross-language transfer of morphological awareness from French (L2) to English (L1), our finding that French derivational awareness accounted for a significant portion of the variance in English vocabulary appears to be novel. This relationship may be explained by the fact that derived words in French and English share a number of morphological features. Cross-language transfer of French derivational awareness to English vocabulary may be supported by children’s understanding that derived words are comprised of roots and affixes in both languages. Similarly, our results with respect to word reading indicated that French morphological awareness supported the development of skill in word reading in French immersion children. This finding is consistent with previous research (Deacon et al. 2007). However, we have extended the Deacon et al. (2007) findings in two important ways. First, Deacon et al. (2007) measured French morphological awareness using an inflectional awareness task that probed knowledge of past-tense use. By contrast, our French derivational awareness task comprised a broad range of derivational morphemes. To our knowledge, our study is the first to demonstrate transfer of derivational awareness from French to English. Second, whereas the participants of Deacon et al. (2007) were English monolingual children in middle immersion, our study focused on minority-language children in early French immersion. These children had reduced amounts of exposure to English because Mandarin was the primary language at home and French was the language of instruction at school. As such, transfer from French to English is particularly important for minority-language children as it facilitates their acquisition of English language and literacy skills. Our study produced mixed results with respect to cross-language transfer of French phonological awareness. Surprisingly, French phonological awareness failed to predict significant variance in English vocabulary or in English word reading. Our results appear to be inconsistent with previous research involving majority language children in French immersion (Comeau et al. 1999). It is important to note, however, that French phonological awareness accounted for a substantial, albeit statistically insignificant, 11% of the variance in English word reading and 15% of the variance in vocabulary. The study’s small sample size may have reduced



The effects of bilingual education on English outcomes 

the power to detect a significant finding. Moreover, we cannot exclude the possibility that the experimental French phonological task used in the present study was too difficult for young children with limited French proficiency. Given these limitations, transfer of phonological awareness in young French immersion children should be further investigated by future research. The most significant finding related to cross-language transfer from Chinese to English in the Bilingual Paired and English-only groups was the finding that Chinese phonological awareness, as measured by a deletion task, significantly predicted English word reading. This finding is consistent with previous findings (e.g. Gottardo et al. 2001; Pasquarella et al. 2011). Contrary to McBride-Chang et al. (2005), however, we did not find that Chinese phonological awareness accounted for variance in English vocabulary. The discrepancy in findings may be due to differences in measurement. The Chinese phonological awareness measure used in this study tapped syllable awareness as well as initial and final phoneme awareness. McBride-Chang and colleagues (2005) found that phoneme onset awareness alone was associated with English vocabulary knowledge. Further research is needed to determine the specific phonological awareness skills in one language that support vocabulary learning in another. Consistent with several other studies (Pasquarella et al. 2011; Wang et al. 2006, 2009), we found no evidence of transfer of Chinese morphological awareness, as measured by a task assessing compound awareness, to either English vocabulary or English word reading. As stated above, Chinese has a more productive and transparent compound morphology than English. An estimated 16% of words in English are transparent compounds, i.e. words whose meanings are directly predictable on the basis of the constituents (Anglin 1993). Since transparent compounds are relatively infrequent in English, we speculate that skills gained by children in decomposing transparent Chinese compounds may not be useful in helping them access English compound words. Thus, our results align with models positing that the extent of transfer of morphological awareness is influenced by the linguistic and orthographic proximities of the languages involved (Koda 2005). The typological distance between English and Chinese may preclude transfer of morphological awareness from Chinese compound awareness to English reading outcomes. Our study has several limitations. The first relates to the study’s cross-­sectional design. Because of its concurrent nature, our study cannot address the issue of causal relations between the metalinguistic skills we examined and language and literacy development. Longitudinal follow-up is necessary to determine causality. In a similar vein, without parallel French/ English and Chinese/ English measures, we are unable to investigate the possibility of a bidirectional, cross-linguistic transfer effect. Second, our sample size was small, restricting the number of ­control

 Kathleen Hipfner-Boucher, Katie Lam & Xi Chen

variables we could include in the regression equations as well as the power to generate significant results. Future research is needed to replicate our findings using a larger sample of children. Furthermore, our study focused exclusively on ­minority-language Chinese speakers from a privileged background in which the average level of maternal education was either an undergraduate university degree or a professional degree. While this narrowly defined sample could be argued to represent one of the strengths of the study, it is also a limiting factor since it remains unclear whether our results generalize to a more socio-economically and linguistically diverse population. Despite the limitations, our findings provide preliminary evidence that early French immersion is a viable option for minority-language children, including Chinese-speaking children whose home language is typologically distant from French. In recent years, early French immersion programs have been attracting an increasingly linguistically diverse student population. To date, little data was available documenting the educational outcomes of these children in either English or French. Our study presents a first step in investigating the English language and literacy outcomes of minority-language children in early French immersion. This information is essential for parents to make informed decisions about the educational options open to their children and for educators to determine the early predictors of later school success. Similarly, our study demonstrates the potential benefits of paired bilingual programs. Publicly-funded paired bilingual programs are rare in the Canadian school system and the program we examined is one of only three that are currently available in the province of Ontario. The vast majority of minority-language children are schooled in English-only programs. The present study offers preliminary findings that argue favourably for the expansion of paired bilingual programs. Our results replicate those of previous research mostly conducted in the United States among Hispanic children and extend them to Chinese-speaking children, another important minority population in North America. Previous research has shown that bilingual programs offer benefits that go beyond academic success. Minority children in bilingual programs tend to develop higher self-esteem and experience a better sense of identity and belonging as compared to their peers in English-only programs (Christian 1996; Lambert & Cazabon 1994). Future studies should examine whether that is also the case in paired bilingual programs in Canada. Finally, our results provide theoretical and educational implications for crosslanguage transfer. Our study contributes to theory of cross-language transfer by demonstrating that transfer of morphological awareness is conditioned by shared morphological structures between L1 and L2. Specifically, English reading outcomes are related to French derivational awareness, but not to Chinese compound



The effects of bilingual education on English outcomes 

awareness, because English has a large number of derived words and relatively few transparent compound words (Anglin 1993). In addition, our finding that phonological awareness transfers between Chinese and English, in combination with previous studies conducted in this area (e.g. McBride-Chang et al. 2006), supports the notion that phonological awareness is a language general construct (Anthony  & Francis 2005). With respect to educational implications, our study demonstrates that cross-language transfer positively affects the English development of minority-language children who are enrolled in French immersion or Chinese-English paired bilingual programs. Therefore, teachers should adopt classroom practices that facilitate cross-language transfer. Vocabulary instruction that focuses explicitly on word formation, for example, has the potential to enhance minority-language children’s performance in oral language, reading, and writing both within and across languages. Of course, intervention research evaluating the efficacy of such practices is needed to confirm the usefulness of these methods in supporting cross-language transfer of skills.

References Adams, Marilyn J. (1990). Beginning to read: Thinking and learning about print. Boston: MIT. Anglin, Jeremy M. (1993). Vocabulary development: A morphological analysis. Monographs of the Society for Research in Child Development 58(10, Serial No. 238): 1–165. Anthony, Jason L. & David J. Francis (2005). Development of phonological awareness. Current Directions in Psychological Science 14: 255–259.  doi: 10.1111/j.0963-7214.2005.00376.x Arab-Moghaddam, Narges & Monique Sénéchal (2001). Orthographic and phonological processing skills in reading and spelling in Persian/English bilinguals. International Journal of Behavioral Development 25: 140–147.  doi: 10.1080/01650250042000320 Arcodia, Giorgio F. (2007). Chinese: A language of compound words? In Fabio Montermini, Gilles Boyé & Nabil Hathout (eds.), Selected Proceedings of the 5th Décembrettes: Morphology in Toulouse, 79–90. Somerville, MA: Cascadilla Proceedings Project. August, Diane & Timothy Shanahan (2006). Developing literacy in second-language learners: Report of the National Literacy Panel on language minority children and youth. Mahwah, NJ: Lawrence Erlbaum. Bérubé, Denis & Stefka Marinova-Todd (2012). The development of language and reading skills in the second and third languages of multilingual children in French Immersion. International Journal of Multilingualism 9: 272–293.  doi: 10.1080/14790718.2011.631708 Bild, Eva-Rebecca & Merrill Swain (1989). Minority language students in a French immersion program: Their French proficiency. Journal of Multilingual and Multicultural Development 10: 255–274.  doi: 10.1080/01434632.1989.9994377 Bruck, Maggie & Fred Genesee (1995). Phonological awareness in young second language learners. Journal of Child Language 22: 307–324.  doi: 10.1017/S0305000900009806 Caravolas, Marketa & Maggie Bruck (1993). The effect of oral and written language input on children’s phonological awareness: A cross-linguistic study. Journal of Experimental Child Psychology 55: 1–30.  doi: 10.1006/jecp.1993.1001

 Kathleen Hipfner-Boucher, Katie Lam & Xi Chen Carlisle, Joanne F. (2000). Awareness of the structure and meaning of morphologically complex words: Impact on reading. Reading and Writing 12: 169–190.  doi: 10.1023/A:1008131926604 Casalis, Séverine & Marie-France Louis-Alexandre (2000). Morphological analysis, phonological analysis and learning to read French: A longitudinal study. Reading and Writing 12: 303–335.  doi: 10.1023/A:1008177205648 Chen, Xi, Tien-Kim Nguyen, Guanglei Hong, Fen Xu & Yun Wang (2010). Effects of crosslanguage transfer on first-language phonological awareness and literacy skills in Chinese children receiving English instruction. Journal of Educational Psychology 102: 712–728.  doi: 10.1037/a0018802 Cheung, Alan C.K. & Robert E. Slavin (2012). Effective reading programs for Spanish-dominant English language learners (ELLs) in the elementary grades: A synthesis of research. Review of Educational Research 82: 351–395.  doi: 10.3102/0034654312465472 Chung, Keven Kien Hoa & Connie Suk-Han Ho (2010). Dyslexia in Chinese Language: an overview of research and practice. Australian Journal of Learning Difficulties 15: 213–224.

doi: 10.1080/19404158.2010.495830

Christian, Donna (1996). Two-way immersion education: Students learning through two languages. The Modern Language Journal 80: 66–76.  doi: 10.1111/j.1540-4781.1996.tb01139.x Comeau, Liane, Pierre Cormier, Éric Grandmaison & Diane Lacroix (1999). A longitudinal study of phonological processing in children learning to read in a second language. Journal of Educational Psychology 91: 29–43.  doi: 10.1037/0022-0663.91.1.29 Deacon, S. Hélène, Lesly Wade-Woolley & John Kirby (2007). Cross-over: The role of morphological awareness in French immersion children’s reading. Developmental Psychology 43: 732–746.  doi: 10.1037/0012-1649.43.3.732 Dunn, Lloyd M. & Douglas M. Dunn (1997). Peabody picture vocabulary test (3rd edition). ­Circle Pines, MN: American Guidance Service. Dunn, Lloyd M. & Douglas M. Dunn (2007). Peabody picture vocabulary test (4th edition). ­Circle Pines, MN: American Guidance Service. Durgunoğlu, Aydin Y., William E. Nagy & Barbara J. Hancin-Bhatt (1993). Cross-language transfer of phonological awareness. Journal of Educational Psychology 85: 453–465.

doi: 10.1037/0022-0663.85.3.453

Francis, David J., Nonie Lesaux & Diane August (2006). Language of instruction. In Diane August & Timothy Shanahan (eds.), Developing literacy in second-language learners, ­365–413. Mahwah, NJ: Lawrence Erlbaum. Genesee, Fred (2004). What do we know about bilingual education for majority language students? In Tej K. Bhatia & William C. Ritchie (eds.), Handbook of bilingualism and multiculturalism, 547–576. Malden, MA: Blackwell. Genesee, Fred & Debra Jared (2008). Literacy development in early French immersion programs. Canadian Psychology 49: 140–147.  doi: 10.1037/0708-5591.49.2.140 Genesee, Fred, Kathryn Lindholm-Leary, William Saunders & Donna Christian (2006). ­Educating English language learners: A synthesis of research evidence. New York: Cambridge University Press.  doi: 10.1017/CBO9780511499913 Goldenberg, Claude (2008). Teaching English language learners: What the research does – and does not – say. American Educator 32: 8–44. Gottardo, Alexandra, Bernice Yan, Linda S. Siegel & Lesly Wade-Woolley (2001). Factors related to English reading performance in children with Chinese as a first language: More evidence of cross-language transfer of phonological processing. Journal of Educational Psychology 93: 530–542.  doi: 10.1037/0022-0663.93.3.530



The effects of bilingual education on English outcomes 

Greene, Jay P. (1998). A meta-analysis of the Rossell & Baker review of bilingual education research. Bilingual Research Journal 21: 103–122.  doi: 10.1080/15235882.1997.10668656 Jared, Debra, Betty-Ann Levy, Pierre Cormier & Lesly Wade-Woolley (2011). Early predictors of biliteracy development in children in French immersion: A 4-year longitudinal study. Journal of Educational Psychology 10: 119–139.  doi: 10.1037/a0021284 Koda, Keiko (2005). Learning to read across writing systems: Transfer, metalinguistic awareness, and second language reading development. In Vivian J. Cook (ed.), Second language writing systems, 311–334. Clevedon, UK: Multilingual Matters. Ku, Yu-Min & Richard C. Anderson (2003). Development of morphological awareness in ­Chinese and English. Reading and Writing 16: 399–422.  doi: 10.1023/A:1024227231216 Kuo, L.-J. & Anderson, R.C. (2006). Morphological awareness and learning to read: A ­cross-language perspective. Educational Psychologist 41: 161–180.  doi: 10.1207/s15326985ep4103_3 Lafrance, Adèle & Alexandra Gottardo (2005). A longitudinal study of phonological processing skills and reading in bilingual children. Applied Psycholinguistics 26: 559–578.

doi: 10.1017/S0142716405050307

Lambert, Wally E. & Mary Cazabon (1994). Students’ views of the Amigos program (Research Rep. No. 11). Santa Cruz, CA and Washington, DC: National Center for Research on ­Cultural Diversity and Second Language Learning. MacCoubrey, Sharon J. Lesly Wade-Woolley, Don Klinger & John R. Kirby (2004). Early identification of at-risk L2 readers. Canadian Modern Language Review 61: 11–28.

doi: 10.3138/cmlr.61.1.11

Marinova-Todd, Stefka, Jing Zhao & May Bernhardt (2010). Phonological awareness skills in the two languages of Mandarin-English bilingual children. Clinical Linguistics and Phonetics 24: 387–400.  doi: 10.3109/02699200903532508 Markus, Paula (2012). English language learners: Policy and school support in Ontario, Canada. Retrieved from http://www.colorincolorado.org/article/50577/ McBride-Chang, Catherine, Jeung-Ryeul Cho, Hongyun Liu, Richard K. Wagner, Hua Shu, Aibao Zhou, Cecilia S–M. Cheuk & Andrea Muse (2005). Changing models across cultures: Associations of phonological awareness and morphological structure awareness with vocabulary and word recognition in second graders from Beijing, Hong Kong, Korea and the United States. Journal of Experimental Child Psychology 92: 140–160.

doi: 10.1016/j.jecp.2005.03.009

McBride-Chang, Catherine, Him Cheung, Wing Yin Bonnie Chow, C.S.-L. Chow & L. Choi (2006). Metalinguistic skills and vocabulary knowledge in Chinese (L1) and English (L2). Reading and Writing 19: 695–716.  doi: 10.1007/s11145-005-5742-x Metsala, Jamie L. (1999). Young children’s phonological awareness and nonword repetition as a function of vocabulary development. Journal of Educational Psychology 91: 3–19.

doi: 10.1037/0022-0663.91.1.3

National Centre for Education Statistics (2009). Fast facts: English language learners. Retrieved from http://nces.ed.gov/fastfacts/display.asp?id=96 Pasquarella, Adrian, Xi Chen, Katie Lam, Yang C. Luo, Y.C. & Gloria Ramírez (2011). Crosslanguage transfer of morphological awareness in Chinese-English bilinguals. Journal of Research in Reading 34: 23–42.  doi: 10.1111/j.1467-9817.2010.01484.x Ramírez, Gloria, Xi Chen, Esther Geva & Heidi Kiefer (2010). Morphological awareness in Spanish-English bilingual children: Within and cross-language effects on word reading. Reading and Writing 23: 337–358.  doi: 10.1007/s11145-009-9203-9

 Kathleen Hipfner-Boucher, Katie Lam & Xi Chen Ramírez, Xi Chen & Adrian Pasquarella (2013). Cross-Linguistic transfer of morphological awareness in Spanish-Speaking English Language Leaners: the facilitating effect of cograte knowledge. Topics in Language Disorders 33: 73–92.  doi: 10.1097/TLD.0b013e318280f55a Riches, Caroline & Fred Genesee (2006). Literacy: Crosslinguistic and crossmodal issues. In Fred Genesee, Kathryn Lindholm-Leary, William Saunders & Donna Christian (eds.), Educating English language learners: A synthesis of research evidence, 64–108. New York: Cambridge University Press.  doi: 10.1017/cbo9780511499913.004 Rolstad, Kellie, Kate Mahoney & Gene V. Glass (2005). The big picture: A meta- analysis of program effectiveness research on English language learners. Educational Policy 19: 572–594.  doi: 10.1177/0895904805278067 Rossell, Christine H. & Keith Baker (1996). The educational effectiveness of bilingual education. Research in the Teaching of English 30: 7–69. Sinay, Erhan (2010). Programs of choice in the TDSB: Characteristics of students in French immersion, alternative schools, and other specialized schools and programs. Retrieved from http://www.tdsb.on.ca/wwwdocuments/about_us/external_research_application/docs/ ProgramsOfChoiceStudentCharacteristics.pdf Slavin, Robert E. & Alan Cheung (2005). A synthesis of research on language of reading instruction for English Language Learners. Review of Educational Research 75: 247–284.

doi: 10.3102/00346543075002247

Snow, Catherine E., M. Susan Burns & Peg Griffin (1998). Preventing reading difficulties in young children. Washington, D.C.: National Academy Press. Statistics Canada (2012). Linguistic characteristics of Canadians, Language, 2011. census. (Catalogue number 98-314-X2011001). Retrieved from http://www12.statcan.gc.ca/­ ­ census-recensement/2011/as-sa/98-314-x/98-314-x2011001-eng.cfm Swain, Merrill & Sharon Lapkin (1982). Evaluating bilingual education: A Canadian case study. Clevedon: Multilingual Matters. Swain, Merrill & Sharon Lapkin (1991). Additive bilingualism and French immersion e­ ducation: The roles of language proficiency and literacy. In Allan G. Reynolds & Wallace E. Lambert (eds.), Bilingualism, multiculturalism, and second language learning, 203–216. Hillsdale, NJ: Lawrence Erlbaum. Swain, Merrill, Sharon Lapkin, Norman Rowen & Doug Hart (1990). The role of mother tongue literacy in third language learning. Language, Culture and Curriculum 3: 65–81.

doi: 10.1080/07908319009525073

Tabachnick, Barbara G. & Linda S. Fidell (2006). Using multivariate statistics (5th edition). ­Boston: Allyn & Bacon. Torgesen, Joseph K., Richard K. Wagner & Carol A. Rashotte (1994). Longitudinal studies of phonological processing and reading. Journal of Learning Disabilities 10: 27–34.

doi: 10.1177/002221947701000108

Turnbull, Miles, Doug Hart & Sharon Lapkin (2003). Grade 6 French immersion students’ performance on large-scale reading, writing, and mathematics tests: Building explanations. The Alberta Journal of Educational Research 49: 6–23. Uchikoshi, Yuuko & Helen Maniates (2010). How does billingual instruction enhance English achievement? A mixed-methods study of Cantonese-speaking and Spanish-speaking bilingual classrooms. Bilingual Research Journal 33: 364–385.  doi: 10.1080/15235882.2010.525294 Wagner, Richard K., Joseph K. Torgesen & Carol A. Rashotte (1999). Comprehensive test of phonological processing. Austin, TX: PRO-ED.



The effects of bilingual education on English outcomes 

Walley, Amanda C., Jamie L. Metsala & Victoria M. Garlock (2003). Spoken vocabulary growth: its role in the development of phoneme awareness and early reading ability. Reading and Writing 16: 5–20.  doi: 10.1023/A:1021789804977 Wang, Min, Chengxi Cheng & Shi-Wei Chen (2006). Contribution of morphological awareness to Chinese-English biliteracy acquisition. Journal of Educational Psychology 98(3): 542–553.  doi: 10.1037/0022-0663.98.3.542 Wang, Ming, Chen Yang & Chengxi Cheng (2009). The contributions of phonology, orthography, and morphology in Chinese-English biliteracy acquisition. Applied Psycholinguistics 30(2): 291–314.  doi: 10.1017/S0142716409090122 Wesche, Marjorie B. (2002). Early French immersion: How has the original Canadian model stood the test of time? In Petra Burmeister, Thorsten Piske & Andreas Rohde (eds.), An integrated view of language development. Papers in honor of Henning Wode, 357–379. Trier: Wissenschaftlicher Verlag Trier. Willig, Ann C. (1985). A meta-analysis of selected studies on the effectiveness of bilingual education. Review of Educational Research 55: 269–317.  doi: 10.3102/00346543055003269 Woodcock, Richard W. (1984). Woodcock language proficiency battery. Allen, TX: DLM. Zhang, Jie, Richard C. Anderson, Qiuying Wang, Jerome Packard, Xinchun Wu, Shan Tang & Xiaoling Ke (2012). Insight into the structure of compound words among speakers of ­Chinese and English. Applied Psycholinguistics 33: 753–779.  doi: 10.1017/S0142716411000555

The role of L1 and L2 reading on L1 preservation and positive cross-linguistic transfer among sequential bilinguals Elena Zaretsky University of Massachusetts, Boston, USA

Previous research shows strong evidence of positive cross-linguistic transfer as a factor of L1 literacy. Moreover, research shows that L1 literacy supports the preservation of L1 language skills while learning L2, as oral and written languages are highly interdependent. Two theoretical frameworks can be instrumental in accounting for cross-linguistic transfer: The constructive analysis hypothesis suggests that the closer L1 and L2 are typologically, the more cross-linguistic transfer should be observed. The interdependence hypothesis postulates that bilingual transfer is also supported by L1 proficiency. These hypotheses were tested in oral language experiment, comparing typologically different languages (L1 Russian and L2 English) with different orthographic depth and language-specific characteristics. The study examines the use of morphosyntactic and lexical structures through story retelling in L1 and L2 by 13 Russian-English bilingual children (Mage = 8;2) who were readers in their L1 as well as in their L2. The results indicate a strong effect of L1 reading skills on children’s ability to use appropriate lexical and syntactic processes in their L1 and L2 narratives as well as on the reduction of L1 grammatical errors, one of the most sensitive aspects of Russian inflectional morphology. These results support the interdependence hypothesis of cross-linguistic transfer. Keywords:  bilingual transfer; literacy; morphosyntax; lexicon; language-specific structures

1.  Introduction In a multicultural, multi-linguistic world, more and more students are learning languages different from the language used at home. Some research suggests that these children may be better served by being labeled just “second language learners” or L2 learners, (Durgunoğlu 2002: 189–204), despite the fact that they often become true bilinguals. While learning the majority language, or language of

doi 10.1075/bct.89.07zar © 2016 John Benjamins Publishing Company

 Elena Zaretsky

school and increasingly important language of social interaction, these children continue to maintain their native language (L1) for the purpose of interacting with family members who may not be proficient in the majority language. There is an abundance of evidence surrounding the issue of cross-linguistic transfer that may help L2 learners to acquire that language. While structural similarities between two languages may greatly facilitate acquisition and use of L2 (Odlin 1989), crosslinguistic transfer is seen even in the presence of radically different L1 and L2 orthographies (Cho & McBride-Chang 2005: 564–571). Studies examining cross-linguistic transfer often focus on phonological awareness as an example of the general factor related to reading skills across ­languages (Deacon, Wade-Woolley & Kirby 2009: 215–229; Durgunoğlu, Nagy & Hancin-Bhatt 1993: 453–465; Cho & McBride-Chang 2005: 564–571; for a systematic review see Lesaux, Geva, Siegel, Koda & Shanahan 2008: 75–122). The orthographic processing, on the other hand, is a language-specific factor, as it taps representation of a word commonly found in a given language (Abu-Rabia 2001: 437–455; Cunningham, Perry & Stanovich 2001: 549–568). It follows that cross-linguistic transfer should be more evident in the production of L2 language learners when the native language and target language are typologically related (Cenoz 2001: 8–20; 2003: 103–116). For example, although Spanish is a direct descendant from Latin and English is a Germanic language, the latter has been heavily influenced by Latin; both languages share words originating from Latin (Chacón Beltràn 2006: 29–39) and both languages use Latin script. Syntactic processing by children who learn a second language is another area that is important to consider when addressing the issue of cross-linguistic transfer. Syntactic awareness is a metalinguistic skill and supports awareness of the “internal grammatical structure of sentences” (Durgunoğlu 2002: 194). The concept of ‘internal grammatical structure’ reflects the knowledge of spoken language. The influence of syntactic awareness on literacy development is seen in helping the reader notice when a word may not fit the representation of the text during decoding and listening comprehension (Tunmer & Chapman 2001). Syntactic awareness also helps the reader to use sentence context to supplement possible incomplete visual and phonological information encountered while reading unfamiliar words (Durgunoğlu 2002: 189–204). Moreover, research on Spanish-English students shows common metalinguistic awareness when the students were asked to identify and correct syntactic errors such as tense, inflection, and word order (Durgunoğlu, Mir & Ariño-Martin 2002: 81–100). The results of Durgunoğlu et al. (2002: 81–100) indicate that children who were able to correctly analyze syntactic structure of the sentences in one language were more likely to do the same in another language, showing analytic accuracy in language processing.



The role of L1 and L2 reading on L1 preservation 

Oral language proficiency is undoubtedly one of the core domains of language and literacy, as it relates to language comprehension and use. Research suggests that proficiency in the stronger language of the second language learner, such as knowledge of vocabulary and morphosyntax, may increase proficiency in a weaker language as a result of positive transfer (Durgunoğlu 2002: 189–204). Oral language is also a very complex domain, encompassing a number of subskills, vocabulary and morphosyntax among them. However, attaining a high level of lexical and morphosyntactic knowledge as well as comprehension in L2 may rely on cognitive factors in general and language-specific factors in particular (Geva & Siegel 2000: 1–30; Proctor, Carlo, August & Snow 2006: 159–169). 1.1  Factors influencing L2 acquisition and L1 preservation 1.1.1  Theoretical frameworks Two theoretical frameworks can be referenced to account for cross-linguistic transfer: the contrastive analysis hypothesis (Lado 1964) and the interdependence hypothesis (Cummins 2000). The first one, which gained significant notoriety in the field of applied linguistics, suggests that L2 errors will occur when L1 and L2 structures differ from one another. Second language learners bring to the task a fully assembled set of lexical and grammatical features of L1, which may not correspond to L2 and therefore require some reconfiguration of L2 representations of lexical and grammatical features (Lardiere 2009: 173–227). As mentioned above, typologically close languages should provide L2 learners with substantial support for cross-linguistic transfer and may reduce the number of errors in L2, while typologically distant languages would produce multiple errors, as postulated by contrastive analysis hypothesis. The second theoretical framework, the interdependence hypothesis (­Cummins 2000), suggests that L1 development can influence and facilitate the development of L2. The interdependence hypothesis acknowledges that not all aspects of L1 facilitate L2 development equally; facilitation rather depends on the contextual support required for communication. In this view, exchanging shared information will allow L2 learners to find lexical and grammatical constituents that directly convey the meaning. When the contextual support is reduced, e.g. communicative partners do not share common knowledge, the use of L2 requires more precise knowledge of the lexicon and syntactic structures. This process becomes increasingly important during school years for learning academic subjects (see Cummins 2000 for full account). Cummins (2000) also suggests that sufficient competence in L1 and L2 oral language will facilitate cross-linguistic transfer, although the precise level of L1 and L2 proficiency was never specified. However, given the emphasis on language proficiency in L1 and L2, the interdependence hypothesis may

 Elena Zaretsky

provide a solid background for understanding cross-linguistic transfer between typologically distant languages, the use of morphosyntactic constructions, and lexical constituents in L1 and L2, as well as address the issue of L1 preservation. The literature on L1 maintenance points to the relationship between the age of L2 onset and the subsequent proficiency in L1, suggesting that a later start of L2 acquisition (between 9–12 years of age) results in better long-term L1 proficiency (Jia, Aaronson & Wu 2002: 599–622; Jia & Aaronson 2003: 131–161; Kohnert, Bates & Hernandez 1999: 1400–1413; McElree; Jia & Litvak 2000: 229–254; YeniKomshian, Flege & Liu 2000: 131–149). As stated above, the interdependence hypothesis is also sensitive to L1 language proficiency, thus enabling cross-­ linguistic transfer and having a positive effect on L2 development. However, L1 proficiency may be also affected by the socioeconomic status (SES), suggesting that children from higher SES families may have an additional support provided by the family in acquiring L1 proficiency (Cummins 2004: 424–429; Melby-Lervåg & Lervåg 2011: 114–135). 1.1.2  The role of literacy in L1 preservation Cross-linguistic transfer has been shown to play an important role in L2 acquisition in general and reading acquisition in particular (Leikin, Schwartz & Share 2010: 269–292). Therefore, it is important to examine the reverse role of L1 and L2 literacy in preservation of the ambient language during the process of learning and gradual complete immersion in L2. In particular, comparing the role of L1 literacy in a language different from L2 in orthographic depth, morphological complexity, and typology in L2 achievements and L1 preservation may provide additional insight into the nature of language transfer. From a neurolinguistic point of view, reading allows an increased opportunity to learn language-specific grammatical structures by providing additional input (visual) to reinforce the auditory perception of a given language. It follows that the memory for specific linguistic structures may be considerably strengthened due to the additional access to language, as a reader learns orthographic representations of oral language structures (Köpke 2007: 9–39). Zaretsky and Bar-Shalom (2010: 401–415) examined the issue of L1 (Russian) attrition in children who spoke English as L2 through a Grammatical Judgment task. The authors found that children who acquired reading skills in Russian (L1) were better at the perception of correct grammatical structures in that language and had a reduced number of grammatical and lexical perceptual errors. Moreover, an adult in the control group who spoke Russian fluently but was not an L1 reader made the same number of grammatical errors as the youngest participant in the study, also a nonreader. These results support previous finding that reading skills in the ambient language reinforce the auditory perception of that language (Köpke 2007: 9–39).



The role of L1 and L2 reading on L1 preservation 

1.2  Structural differences between Russian (L1) and English (L2) The study of Russian as L1 and English as L2 provides an interesting comparative background in addressing the issue of cross-linguistic transfer and the roles that L1 and L2 literacy play in L1 preservation and L2 achievements. The two languages are typologically different: Although both languages use alphabets for orthographic representations, Russian employs the Cyrillic alphabet while English uses the Latin alphabet. Russian is also a highly inflected language with “a rather elegant orthographic system: on the one hand, quite complex and hierarchical, and on the other hand, organized around a dominant principle and therefore, sufficiently regular and predictable, even though the number of exceptions is high” (Kerek & Niemi 2009: 3). In addition, Russian and English are representatives of shallow and deep orthographies, respectively. 1.2.1  Language specific differences in oral and written modalities Major European languages in the Germanic group, e.g. English, German, Dutch, are considered analytic languages, where the meaning is expressed more by grammatical constructions rather than by morphology. For example, English uses particles and word order to show subject-object relationship. In comparison, Russian, as a representative of the Eastern Slavic group, is considered a synthetic language, with an extensive variety of affixes and internal modifications of roots, resulting in a high morpheme-per-word ratio. It also incorporates several challenging grammatical features. Learners must acquire case, gender, and number inflection on nouns, adjectives, and pronouns, or principles of declension and non-adjacent dependencies, e.g. the relationship between prepositions and noun case markers. The following examples illustrate the differences in case markers for feminine, masculine, and plural nouns: 〈к реке〉 k reke ‘to the river’ – dative, fem.; 〈к столу〉 k stolu ‘to the table’ – dative, masc.; 〈два дома〉 dva doma ‘two houses’ – plural, nominative. Agreement between the noun/ pronoun and the verb or the process of conjugation are also challenging for ‘heritage’ speakers of L1, i.e. individuals who had incomplete acquisition of their native language, as well as for L2 learners of Russian. Polinsky (2006: 1–64) points out that Russian-English bilinguals, especially ones who have been using L2 as their main language of interaction, use a reduced two-case system (nominative, accusative) and consistently omit verb agreement markers in their speech. Russian verbs have only three basic tenses (present, past, and future), but this seeming simpleness is complicated by aspectual distinctions (perfective and imperfective) that can only be used in the past and future tense. Word order in Russian is rather flexible and changing the general Subject-Verb-Object (SVO) order of simple statements does not result in a change of meaning. Isurin (2005: 1115–1130) suggests six possible word orders: SVO, SOV, VOS, VSO, OVS, and OSV.

 Elena Zaretsky

As far as the written language system is concerned, Russian is considered a shallow orthography with relative regularity of phoneme-grapheme mapping. In general, the written system follows alphabetic rules, although some irregularities exist and should be taught explicitly. Those irregularities are the results of stress and co-articulation patterns (e.g. certain sounds are not pronounced in oral language (weak position); however, choices for writing become clear when a related word can be found with a phoneme in question in strong position). In addition, there is a soft sign 〈ь〉 that does not have phonemic value but is required in written language to denominate a ‘soft’ consonant in the final position of a word (e.g. 〈апрель〉 ‘April’) (Kerek & Niemi 2009: 1–21). In contrast, English has a complex verb tense system and very strict SVO order, while lacking the complexity in its inflectional system. English is also an example of a deep orthography with one-to-many, many-to-one and manyto-many spelling patterns. Thus, Russian as L1 presents a compelling case for addressing the issue of cross-linguistic transfer. 1.3  Present study The main purpose of the present study was to identify the influence of L1 and L2 literacy on the preservation of L1 morphosyntactic and lexical abilities as well as on proficient use of L2. The study was designed to examine cross-linguistic transfer in oral language skills among children learning English (L2), whose L1 is Russian and who were either born in the US or arrived well before the suggested cut-off period between 9 and 12 years of age for L1 stability. In addition, all children participating in this study were from a moderately high SES background. Based on the abovementioned language-specific differences between Russian and English and in view of the interdependence hypothesis, the following research questions were addressed: 1. Do children use similar processes related to the representation of lexicon and morphosyntactic structures in their L1 and L2 during oral narratives? 2. Is there an interaction between L1 and L2 reading, systematic use of L1, length of formal exposure to L2, and use of the language-specific lexicon and morphosyntactic structures in both languages as part of the cross-linguistic transfer? 3. Does reading ability in L1 support L1 retention and promote L2 knowledge? In accordance with the interdependence hypothesis that proposes L1 and L2 proficiency as a significant influential factor in cross linguistic transfer, we postulate that continuous support of L1 (e.g. use of L1 oral language and acquiring L1 reading skills) concurrent with the acquisition of L2 oral and written language will result in equal ability to use L1 and L2 grammatical and lexical constituents despite L1 and L2 typological distance.



The role of L1 and L2 reading on L1 preservation 

2.  Method 2.1  Participants 2.1.1  Inclusion criteria and description of the sample The present study was concerned with L1 preservation and L2 language transfer as a factor of reading in both languages among sequential bilinguals. The participants were chosen based on the following inclusion criteria: (1) the participants were active users of their L1, (2) were L1 readers and (3) followed regular school curriculum in L2 without the support of bilingual services with average or above average achievements in English Language Arts (reading, writing and reading comprehension). Initial interviews with the families identified a pool of thirteen children (seven females, six males) between 5;8 and 10;11 years of age (M = 8;2, SD = 1;7). All children had Russian as their L1 and English as L2. The families were from moderately high SES before arriving in the United States and they achieved the same status in the US. All parents of the participants were minimally college graduates. Maternal education was at the level of Baccalaureate (n = 8) and Masters (n = 5) degrees. Due to the small sample, we provide a detailed description of the length of uninterrupted L1 and formal exposure to L2 among the participants. Five children (two males and three females) were born in the US, the rest of the children were brought to the country between 1;0 and 6;5 years of age. The age of initial formal exposure to L2 (English) ranged from 2;0 to 6;7 years of age, which effectively makes it the duration of uninterrupted exposure to L1 (M = 5.0, SD = 1.1). The duration of formal exposure to L2 lasted between 2;0 and 7;7 years at the time of data collection (M = 3.4, SD = 2.3). Although there are differences in the length of exclusive use of L1 and exposure to L2, the common characteristic of all but one of the participants was the involvement in group activities conducted in Russian, i.e. reading/listening groups, theatre productions, art, music, and dance lessons as well as formal L1 lessons. (See Table 1 for individual information on each subject’s age, birth place, length of formal exposure to L2, and the extent of L1 use). The beginning of formal exposure to L2 was identified as the time the child entered the day care center or a school, and for a specific period of time during the day every interaction was conducted in L2 only. Casual exposure to L2 inevitably happens in child’s daily life, e.g. television programs, interaction with English-speaking peers on the playground, and sometimes with older siblings. At the time of the study, all children (except for the 5;8 year old male who was homeschooled in L1) were attending local public schools, participating in all aspects of the general school curricula in L2 without the support of bilingual services. Available school records (school reports provided by parents) indicated that all children excelled in language arts, i.e. oral and written language and

 Elena Zaretsky

Table 1.  Bilingual group description (n = 13, Mage = 8;2, SD = 1;7) at the time of the study Part.

Age/ Gender

Birth place/Age of arrival

L1 only/ y;m

Formal L2/grade

1.

5;8/m

USA born

5;8

Home schooled

2.

6;0/f*

Russia/1;0

4;4

3.

6;0/f*

Russia/1;0

4.

7;6/m

5.

Use of L1

L1 read

L2 read

Home (exclusive), formal L1 lessons/ group activities

yes

no

1;10/1

Home (exclusive), formal L1 lessons/ group activities

yes

yes

4;4

1;10/1

Home (exclusive), formal L1 lessons/ group activities

yes

yes

USA born

6;7

1;11/1

Home (exclusive), formal L1 lessons/ group activities

yes

yes

8;0/f

Russia/2;0

4;0

4;0/3

Home (exclusive), formal L1 lessons

yes

yes

6.

8;0/m

USA born

5;1

2;11/3

Home, not strictly observed

yes

yes

7.

8;8/m

Russia/6;5

6;5

1;3/4

Home (exclusive)

yes

yes

8.

9;0/f

USA born

5;4

3;8/4

Home (exclusive), formal L1 lessons

yes

yes

9.

9;7/f

USA born

2;0

7;7/5

Inconsistent L1/ code-switching

no

yes

10.

10;7/f*

USA born

5;7

4;9/5

Home (exclusive), formal L1 lessons

yes

yes

11.

10;7/f*

USA born

5;7

4;9/5

Home (exclusive), formal L1 lessons

yes

yes

12.

10;10

Russia/3;0

5;1

5;9/6

Home (exclusive)

yes

yes

13.

10;11

Russia/3;0

5;2

5;9/6

Home (mixed; older siblings)

yes

yes

*fraternal twins

reading comprehension. At the same time, all children (except for the 9;7 year old female) were readers in L1 as a result of family efforts and formal L1 activities. Children in Russia, as a rule, acquire basic reading skills by the end of first grade, roughly at the age of 7 years, but often begin to read already at the age four of five years. Only one of the participants attended school in Russia (1st grade) before arriving in the US at the age of 6;5, all others (except for 9;7 year old female) acquired reading skills age appropriately, despite the fact that they did not attend school in Russia.



The role of L1 and L2 reading on L1 preservation 

As for the age differences within the group of participants, the sample represents a steady yearly increase in age. Therefore, these differences do not constitute a problem for the nature of this study. To the contrary, the age differences among the participants, as well as the differences in the length of their uninterrupted exposure to L1 and formal exposure to L2, allows a closer look at the interaction between the length of uninterrupted L1, length of formal exposure to L2, and knowledge of language-specific structures. The inclusion of two participants who did not exactly fit the established criteria for the study was justified as it allowed a closer examination of the effect of L1 reading in L1 maintenance and crosslinguistic transfer to L2. A group of twelve monolingual (Russian) children (7 females and 5 males) from Moscow served as controls for the L1 preservation comparison. This group was used in previous research by the same author. All children were between 6;2 and 11;6 years of age (M = 8.46, SD = 1.8), were readers in their native language, attended public schools, were from similar SES background, and represented the same increases in age as their bilingual counterparts. 2.2.2  Materials and procedures As stated above, all children were recruited for the study based on specific criteria: they had to show L1 oral language proficiency as well as L1 reading skills. The L1 oral language skills were assessed informally through conversation with the researcher during the initial meeting on a topic of child’s interest. At that meeting, the family members were asked to fill out a detailed questionnaire regarding L1 use at home and recreationally. To assess reading skills, all children were asked to read a passage from a story (‘Philipok’, by L. Tolstoy, considered to be ‘juvenile fiction’) and then answer comprehension questions. As stated above, all but one of the participants were readers in their L1 and had no difficulties comprehending the reading material. As the experimental measure, all children were asked to listen to and retell the folk tale ‘Two Unhappy Friends’ in both languages (Appendix A). The story comprises simple and syntactically complex sentences with descriptive, locative, and temporal information, as well as instances of direct speech. Due to the language differences, the total number of sentences was different for the English and Russian versions. To equalize the narratives in both languages, the story line was divided into 16 sequences, each carrying important information related to the location, time, and the actions taken place within the story. Aside from the initial interview and the assessment of reading skills, each child was seen in two sessions, separated by a month, to hear and retell the story, either in Russian (L1) or English (L2). The story was read to the child only once by a native speaker of either language (L1 or L2) and the order of presentation of

 Elena Zaretsky

the English and Russian versions was randomized. The ensuing narratives were recorded and transcribed for the analysis. Children in the control group were read only the Russian version of the story and they were asked to retell it. These narratives were also recorded and transcribed for analysis. 2.2.2.1  Coding system for syntactic structure.  Three specific syntactic processes of story retelling were identified in children’s narratives. 1. Formal transformation: The original or source sentence (SS) and the retold sentence (RS) have the same structure, meaning, and lexicon; variations may occur only at the word level. Formal transformations may rely on the memory function, i.e. cognitive skills, as part of the complexity of the oral language domain. For example: (1) (SS) On the bank of a wide river there stood a little house. (RS) Once there stood a little house…on the bank of the river. (2) (SS) a starushka vse eto vremja sidela v dome. (RS) V tо zhе sаmое vrеmja, starushka sidela doma. ‘All this time the old woman was sitting at home.’

(6;0, L2)

(10;6, L1)

2. Restructuring: SS and RS have the same meaning and lexicon, but differ in syntactic structure, which may include changes in preposition, adding or substituting constituent without introducing new information. Therefore, restructuring requires a well-developed knowledge of syntactic structures in the given language. For example: (3) (SS) …come and help me (RS) It asked the dog to help (4) (SS) delat’ vsju rabotu, kotoruju delela ljagushka (RS) delat’ vce chto delala ljagushka ‘do all the work that… – do everything the frog did’

(10;11 L2)

(8;0, L1)

3. Semantic paraphrasing: The meaning of the sentence is retained in the RS, but lexicon and syntactic structure may differ from the SS. Semantic paraphrases rely on comprehension as well as on substantial knowledge of syntactic structures and a well developed lexicon. For example: (5) (SS) made her work from morning till night (RS) She was being slaved

(7;6, L2)



The role of L1 and L2 reading on L1 preservation 

(6) (SS) zastavljala rabotat’ s utra do nochi (RS) hotela chtobi ona vse delala ‘made her work from morning till’ night – wanted her to do everything’  (10;10, L1)

2.2.2.2  Coding system for lexical processes.  Two lexical processes were identified in children’s retelling: 1. Analytic reformulation: Analytic reformulation involves defining a word (or word combination) either relative to derivational possibilities in a given language or with other words. Therefore, analytic reformulation may be grammatically dependent, as it may require a change in syntactic structure of the sentence, i.e. (7) (SS) vzgljanula na lunu (RS) posmotrela naverh I uvidela lunu ‘glanced at the moon – looked up and saw the moon’ (8) (SS) got impatient (RS) she was tired of waiting

(10;6, L1) (8;0, L2)

2. Synthetic reformulation: Use of one word to describe complex predicates. Therefore, synthetic reformulation may be more lexical-dependent and requires extensive vocabulary, i.e. (9) (SS) prizhavshis’ drug k drugu (RS) spinoj k spine ‘sitting next to each other – back to back’ (10) (SS) during all this time (RS) meanwhile

(7;6, L1) (10;6, L2)

All narratives were examined for different types of errors, i.e. grammatical (conjugation) and lexical (declension). 3.  Results 3.1  Descriptive analysis Descriptive statistics and post hoc analyses from the experimental and control groups’ L1 narratives regarding the use of lexical and morphosyntactic processes, lexical and grammatical errors, as well as the actual length of narratives are presented in Table 2. As the data indicate, bilingual and monolingual children used similar processes to retell the story in L1. Within the syntactic processes, bilingual children used significantly more Restructuring and Semantic paraphrases as

 Elena Zaretsky

well as more Synthetic reformulations, as part of the lexical processes, suggesting that they use the available lexicon and knowledge of grammar in L1to convey the message. There were no differences in the number of produced grammatical and lexical errors. Table 2.  Comparative data and post hoc analysis for Russian narratives (bilingual and monolingual participants) Task

Bilinguals

Monolinguals

M

SD

Range

M

SD

Range

t

df

 P

13.85

2.4

8–16

13.16

2.12

8–15

.74

23

Ns

Lexical

1.69

Gramm

2.69

1.5

0–6

1.6

0.9

0–3

1.16

23

Ns

6.6

0–24

2.1

2.3

0–7

.73

23

Ns

9.15

6.2

1–19

5.3

3.02

1–17

1.9

23

.06*

12.1

4.2

7–22

8.3

3.98

1–13

2.28

23

.03

6.8

3.8

1–13

2.72

1.6

1–5

3.62

23

.001

Analytic

1.8

1.6

0–6

2

0.7

1–3

1.72

23

Ns

Synthetic

1.15

1.3

0–4

1

0

0

2.56

23

.01

Seq. Errors

Syntactic processes FormTrans Restruct SemParaph Lexical processes

Note: Seq = Total number of produced sequences in narrative; FormTrans = Formal transformation (verbatim); Restruct = Restructuring, change in syntactic structure, no new information; SemParaph = Semantic paraphrasing, change in syntactic structure and lexicon, same meaning; Analytic = describing a word with a word combination; Synthetic = describing word combination with one word *= approaching significance

Table 3 represents the descriptive data regarding the performance on tasks associated with lexical and morphosyntactic processes used by the bilingual group of participants in their L1 and L2 narratives. Post hoc analyses were used to address possible similarities/ differences in the L1 and L2 use of morphosyntactic and lexical processes as well as in the number of retold sequences. The results show that the only significant differences were seen in Analytic reformulations, i.e. a process that changes the syntactic structure through lexical constituents (t (12) = -2.624, p = .02). Moreover, the total number of the produced sequences in both languages were significantly correlated (r = .696, p = .008), as were Semantic paraphrases, i.e. use of the lexical and grammatical knowledge (r = .787, p = .001), and Synthetic reformulations, i.e. lexical process that requires extensive vocabulary (r = .665, p  =  .013), suggesting parallel processes in L1 and L2. The differences between lexical and grammatical errors in L1 and L2 were non-significant (Table 3).



The role of L1 and L2 reading on L1 preservation 

Table 3.  Descriptive data and post hoc analysis for L1 and L2 narratives Task

Russian M

Seq.

SD Range

13.85 2.4 8–16

English M

SD

13.62 2.4

Range

t

df

p

9–16

.443

12 ns

r

P

.696 .008

Errors Lexical

1.69 1.5 0–6

1.1

.9

0–3

–1.2

12 ns

–.16

Ns

Gramm

2.69 6.6 0–24

1.1

1.1

0–3

.964

9 ns

–.404 Ns

10.2

5.3

2–20

.714

12 ns

.563 .04 –.012 Ns

Syntactic processes FormTrans Restruct

9.15 6.2 1–19 12.1

4.2 7–22

11.2

3.7

5–17

–.591

12 ns

6.8

3.8 1–13

6.3

4.4

0–16

–.704

12 ns

.787 .001

Analytic

1.8

1.6 0–6

12 .022

.019 Ns

Synthetic

1.15 1.3 0–4

12 ns

.665 .013

SemParaph Lexical processes

.62 1.1

.65 1.3

0–2 0–4

–2.62 –.267

Note: Seq = Total number of produced sequences in narrative; FormTrans = Formal transformation (verbatim); Restruct = Restructuring, change in syntactic structure, no new information; SemParaph = Semantic paraphrasing, change in syntactic structure and lexicon, same meaning; Analytic = describing a word with a word combination; Synthetic = describing word combination with one word

3.2  Correlation analysis The data was examined to identify possible correlations between the length of uninterrupted exposure to the L1 before onset of L2, use of L1 in social situations, length of L2, reading in L1 and L2, grammatical errors, and the use of lexical and syntactic processes in narratives. The p values for all correlations were adjusted to reflect a Bonferroni correlation (alpha level = .05) for most reliable results. 3.2.1  Syntactic and lexical processes correlations First, we looked at the correlations between reading in L1 and L2, length of uninterrupted exposure to L1, use of L1, use of syntactic processes, instances of grammatical errors, and instances of code-switching. The length of uninterrupted exposure to L1 was significantly correlated with L2 reading (r = -.774**, p = .002), number of sequences in L1 (r = .563*, p = .04), L1 use (r = -608*, p = .03), and the number of sequences in L2 (r = .696**, p = .008). Reading in L1 was significantly correlated with L1 use (r = .968**, p =