Reading in English as a First or Second Language: The Case of Grade 3 Spanish, Portuguese, and...

17
Learning Disabilities Research & Practice, 26(2), 67–83 C 2011 The Division for Learning Disabilities of the Council for Exceptional Children Reading in English as a First or Second Language: The Case of Grade 3 Spanish, Portuguese, and English Speakers Amy Grant and Alexandra Gottardo Wilfrid Laurier University Esther Geva University of Toronto This study compared variables related to reading ability in Grade 3 students learning English as a first language (L1) and second language (L2). The students learning English as an L2 came from diverse backgrounds, with different levels of bilingualism in Spanish and English or Portuguese and English before they entered school. Both within-group and between-group differences emerged in comparing Spanish children from two cohorts, and in comparing the Spanish group to the Portuguese and English groups. Models predicting reading comprehension found differences with respect to the contribution of receptive vocabulary, decoding, and print exposure in the L1 and L2 groups, depending on the L2 students’ bilingual language status and language acquisition experiences. Additionally, print exposure was more highly related to comprehension in the lowest performing L2 groups. Implications and future directions are discussed. Research and policy debates in the field of bilingual reading development have focused on the cognitive advantages and cognitive disadvantages of learning more than one language, and more specifically, learning to read in English as a second language (L2) (Bialystok, 2007). Additionally, studies have focused on comparing English as a first language (L1) or English L2 groups in their acquisition and performance on cognitive components related to reading (Bialystok, Majumder, & Martin, 2003; Jongejan, Verhoeven, & Siegel, 2007; Lesaux, Rupp, & Siegel, 2007). Comparisons of cognitive–linguistic skills in students from different language backgrounds have revealed conflicting results depending on the skill measured. Some studies have found that L2 students continue to lag behind English L1 learners, specifically in areas such as reading comprehension and vocabulary (Aarts & Verhoeven, 1999; Droop & Verhoeven, 1998; Verhoeven, 1990, 2000). While other researchers have found that English Language learners (ELLs) can surpass L1 students in certain aspects of reading ability, specifically rapid naming, word-level reading, and spelling (Bialystok et al., 2003; Bruck & Genesee, 1995; Campbell & Sais, 1995; Caravolas & Bruck, 1993; Chiappe, Glaeser, & Ferko, 2007; Geva, Yaghoub-Zadeh, & Schuster, 2000; Lesaux & Siegel, 2003; Schwartz, Geva, Share, & Leikin, 2007). In this study, comparisons are made among three groups of English L2 learners and English L1 speakers on measures of vocabulary, decoding, reading comprehension, and print exposure. Another research question relevant to L2 development is not whether there are cognitive advantages to bilingualism, Requests for reprints should be sent to Amy Grant, Department of Psychology, Wilfrid Laurier University, Waterloo, Ontario, N2V3C5. Elec- tronic enquiries may be sent to [email protected]. but which cognitive processing skills predict reading ability in bilinguals. This study examines the relationships among variables that predict reading comprehension and compares these relationships across groups of students with differing language acquisition backgrounds. ELLs from two different L1 backgrounds, Spanish and Portuguese, will be compared to a group of English L1 students. Specifically, the groups include ELLs who learned English and their L1 simultane- ously or learned their L1 and L2 sequentially. The sequential bilinguals belong to two subgroups, balanced bilinguals for whom their L1 and L2 are approximately equal or children for whom their L1 is dominant. It is expected that the groups will differ in terms of the relationships among variables and in terms of the variables most highly related to reading comprehension. THE EFFECTS OF L1 AND L2 STATUS When comparing different groups of learners, it is neces- sary to define the degree and type of bilingual experience of each group. Thus, in studies comparing children from different L1 backgrounds, the “type” of bilingualism that is used to classify the children must be, but seldom is, taken into account. Though it may be preferential to see children on a continuum with respect to their language status, re- searchers and theorists have delineated discrete categories by which to classify bilinguals. In its simplest form, bilin- gualism is defined as “knowing” two languages (Valdez & Figueora, 1994). Simultaneous bilingualism occurs when two languages are acquired at birth or prior to one year of age (De Houwer, 2005), whereas sequential bilingualism is said to occur when one language is acquired prior to another

Transcript of Reading in English as a First or Second Language: The Case of Grade 3 Spanish, Portuguese, and...

Learning Disabilities Research & Practice, 26(2), 67–83C©2011 The Division for Learning Disabilities of the Council for Exceptional Children

Reading in English as a First or Second Language: The Case of Grade 3Spanish, Portuguese, and English Speakers

Amy Grant and Alexandra GottardoWilfrid Laurier University

Esther GevaUniversity of Toronto

This study compared variables related to reading ability in Grade 3 students learning Englishas a first language (L1) and second language (L2). The students learning English as an L2came from diverse backgrounds, with different levels of bilingualism in Spanish and Englishor Portuguese and English before they entered school. Both within-group and between-groupdifferences emerged in comparing Spanish children from two cohorts, and in comparing theSpanish group to the Portuguese and English groups. Models predicting reading comprehensionfound differences with respect to the contribution of receptive vocabulary, decoding, and printexposure in the L1 and L2 groups, depending on the L2 students’ bilingual language statusand language acquisition experiences. Additionally, print exposure was more highly relatedto comprehension in the lowest performing L2 groups. Implications and future directions arediscussed.

Research and policy debates in the field of bilingual readingdevelopment have focused on the cognitive advantages andcognitive disadvantages of learning more than one language,and more specifically, learning to read in English as a secondlanguage (L2) (Bialystok, 2007). Additionally, studies havefocused on comparing English as a first language (L1) orEnglish L2 groups in their acquisition and performanceon cognitive components related to reading (Bialystok,Majumder, & Martin, 2003; Jongejan, Verhoeven, &Siegel, 2007; Lesaux, Rupp, & Siegel, 2007). Comparisonsof cognitive–linguistic skills in students from differentlanguage backgrounds have revealed conflicting resultsdepending on the skill measured. Some studies have foundthat L2 students continue to lag behind English L1 learners,specifically in areas such as reading comprehension andvocabulary (Aarts & Verhoeven, 1999; Droop & Verhoeven,1998; Verhoeven, 1990, 2000). While other researchers havefound that English Language learners (ELLs) can surpassL1 students in certain aspects of reading ability, specificallyrapid naming, word-level reading, and spelling (Bialystok etal., 2003; Bruck & Genesee, 1995; Campbell & Sais, 1995;Caravolas & Bruck, 1993; Chiappe, Glaeser, & Ferko, 2007;Geva, Yaghoub-Zadeh, & Schuster, 2000; Lesaux & Siegel,2003; Schwartz, Geva, Share, & Leikin, 2007). In this study,comparisons are made among three groups of English L2learners and English L1 speakers on measures of vocabulary,decoding, reading comprehension, and print exposure.

Another research question relevant to L2 development isnot whether there are cognitive advantages to bilingualism,

Requests for reprints should be sent to Amy Grant, Department ofPsychology, Wilfrid Laurier University, Waterloo, Ontario, N2V3C5. Elec-tronic enquiries may be sent to [email protected].

but which cognitive processing skills predict reading abilityin bilinguals. This study examines the relationships amongvariables that predict reading comprehension and comparesthese relationships across groups of students with differinglanguage acquisition backgrounds. ELLs from two differentL1 backgrounds, Spanish and Portuguese, will be comparedto a group of English L1 students. Specifically, the groupsinclude ELLs who learned English and their L1 simultane-ously or learned their L1 and L2 sequentially. The sequentialbilinguals belong to two subgroups, balanced bilinguals forwhom their L1 and L2 are approximately equal or childrenfor whom their L1 is dominant. It is expected that the groupswill differ in terms of the relationships among variablesand in terms of the variables most highly related to readingcomprehension.

THE EFFECTS OF L1 AND L2 STATUS

When comparing different groups of learners, it is neces-sary to define the degree and type of bilingual experienceof each group. Thus, in studies comparing children fromdifferent L1 backgrounds, the “type” of bilingualism that isused to classify the children must be, but seldom is, takeninto account. Though it may be preferential to see childrenon a continuum with respect to their language status, re-searchers and theorists have delineated discrete categoriesby which to classify bilinguals. In its simplest form, bilin-gualism is defined as “knowing” two languages (Valdez &Figueora, 1994). Simultaneous bilingualism occurs when twolanguages are acquired at birth or prior to one year of age(De Houwer, 2005), whereas sequential bilingualism is saidto occur when one language is acquired prior to another

68 GRANT, GOTTARDO, AND GEVA: READING IN ENGLISH AS A FIRST OR SECOND LANGUAGE

language (Flege, 1992). Research has shown that the pro-cessing and organization of oral language in the brain, acrossthe learner’s two languages, differs depending upon the ageof acquisition (Weber-Fox & Neville, 1996).

Sequential bilinguals can either be dominant in their L1and have lower skill in their L2, or can be balanced in their L1and L2 ability. Those who are dominant in their L1 (languagemost often spoken at home), are those who speak primarilyone language until after late childhood (8–11 years), and thenbegin to learn English as an L2, or they are recent immigrantswho are in the early stages of acquiring their L2 (Valdez &Figueora, 1994). Those who are balanced bilinguals, how-ever, are those who begin to acquire English as an L2 beforethe age of 11, and are equally proficient in both their L1 andL2. The students in this sample who learned their L1 priorto their L2 all learned English in early elementary school.Therefore, they are likely to be sequential bilinguals.

Not only does acquiring a second language result in differ-ences in L1 or L2 skill, but also results in structural organiza-tion differences of the brain. Individuals who are dominant intheir L1 tend to organize their languages in nonoverlappingstructural areas of the brain, whereas those who are referredto as balanced bilinguals or simultaneous bilinguals, tend tohave overlapping areas of brain activation in their L1 and L2(Neville & Bavelier, 1998). For example, sequential bilin-guals who have learned one language and then another (theSpanish group in this study) would have their two languagesstored in two separate areas of the brain, which tends to re-sult in less mixing of words from the two languages onespeaks. For simultaneous or balanced bilinguals who haveacquired both languages at the same time (the Portuguesegroup in this study), the two languages they speak are storedin the same general area of the brain. This can result in moremixing of words from these two languages because retrievaloccurs from overlapping brain areas (Dehane et al., 1997;Kim, Relkin, Lee, & Hirsch, 1997). The specific areas of ac-tivation, and the specific functions that are affected by thesedifferences in bilingual classification are more complex thanpresented, however, the basic knowledge that there are coredifferences with respect to brain organization dependent onwhen language is learned, and that there are behavioral differ-ences in language output are most relevant to this study (seeNeville & Bavelier, 1998; Scherag, Demuth, Rosler, Neville,& Roder, 2004; and Weber-Fox & Neville, 1996 for moredetailed descriptions of these differences).

In the context of this study, the Spanish L2 students willbe defined as sequential bilinguals, because they acquiredSpanish as their L1 at home, and then began learning En-glish once they entered the school system. On the other hand,the students in the Portuguese group are simultaneous bilin-guals, who were exposed to both English and Portugueseat home from birth. This difference in language acquisitionoccurred in the Portuguese families because the parents ofthe Portuguese- and English-speaking children were morelikely to be second-generation immigrants. Most of theseparents had attended some schooling in Canada, althoughhigh school dropout rates are high for Portuguese speakers.The parents of the children in the Spanish groups were morelikely to be first-generation immigrants who had just recentlymoved to Canada and had completed all their education in

Spanish in their country of origin. The degree of exposureto English for sequential bilinguals might also be particu-larly relevant (Flege, 1992). It is important to consider whenthese students began learning English before school, as thestudents involved in the study all began schooling in Englishin at least Grade 1. This means that by Grade 3, the studentshad from 2 to 9 years of experience learning English. Thisalso means that the Spanish students had relatively greaterproficiency in their L1 than their L2 at the time of schoolentry, whereas Portuguese students possibly had equal pro-ficiency in their L2 (English) and their L1 at the time ofschool entry. However, the Portuguese-English students hadexperienced language loss in Portuguese and were Englishdominant by third grade. It is expected that the groups willdiffer in terms of the relationships among variables based onlanguage experience and language acquisition patterns.

SKILLS RELATED TO READINGCOMPREHENSION IN AN L2

The major component skills related to reading compre-hension are decoding and oral language proficiency (Cain,Oakhill, & Bryant, 2000; Gough & Tunmer, 1986). Vocab-ulary knowledge is often used as a proxy for general orallanguage skill (Proctor, Carlo, August, & Snow, 2005). Manystudies have found that L2 learners develop their vocabularyin English much more slowly than they develop word read-ing skills in English, and that English vocabulary knowledgeremains below that of their native English-speaking peers fora long period of time (August, Carlo, Dressler, & Snow,2005; Cummins, 1991; Farnia & Geva, in press; Proctoret al., 2005). Both the depth of vocabulary, which includes therichness of the representation, and the breadth of vocabulary,usually defined as the number of words known, influencesL2 vocabulary acquisition. Although L2 students are weak inboth areas of vocabulary acquisition, they tend to have moredifficulty with depth of vocabulary knowledge (e.g., Feld-man & Healy, 1998; Ordonez, Carlo, Snow, & McLaughlin,2002). The acquisition of vocabulary knowledge is impor-tant due to the relationship it has with other components ofreading development (Ouellette, 2006). Vocabulary mediatesperformance on other linguistic skills such as grammaticaland morphological knowledge (Chall, 1987). When vocab-ulary knowledge is sufficiently delayed, it disrupts readingcomprehension based on the proportion of unknown wordsin text (Carver, 1994). Insufficient vocabulary knowledgealso impedes growth in reading comprehension outcomes(Ouellette, 2006; Proctor, Carlo, August, & Snow, 2006).

A better understanding of differences in vocabulary de-velopment in ELLs in relation to reading comprehension canalso assist in identifying students with learning difficultiesdue to limited L2 proficiency versus students with underly-ing disabilities (Durgunoglu, 2002). For example, educatorsare often interested in whether a low-achieving ELL has alearning disability or a problem related to low levels of L2acquisition. ELLs with low oral-language proficiency aremore likely to be misclassified as having a learning disability(Limbos & Geva, 2001). Thus, researchers have suggestedthat assessment of ELLs should be separated at the level of

LEARNING DISABILITIES RESEARCH 69

oral language versus written language so that discrepanciesbetween students’ abilities in their L1 and L2 do not lead tomisidentification (Geva, 1999).

The other major component skill of reading compre-hension, decoding, requires sufficient skill in phonologicalprocessing. Therefore, skill in phonological awareness isone of the best predictors of word reading ability withinthe first years of school for L1 and L2 students (Gottardo,Collins, Baciu, & Gebotys, 2008; Lindsey, Manis, & Bailey,2003; Manis, Lindsey, & Bailey, 2004; Wagner, Torgesen, &Rashotte, 1994). This effect holds true, even after controllingfor receptive vocabulary, intelligence, memory, and socialclass in English L1 students (Anthony & Lonigan, 2004;Bryant, MacLean, Bradley, & Crossland, 1990; Wagneret al., 1994; Whitehurst & Lonigan, 1998). Another keyphonological processing skill related to reading is lexicalaccess. This skill is typically measured through rapid namingtasks, and is associated with the fluency of retrieval ofverbal labels, which is a skill highly related to word readingproficiency (Wagner & Torgesen, 1987; Wolf, 1997). Rapidnaming is one skill in which L2 children tend to perform atan equal or higher level than L1 children in the early stages ofreading acquisition despite lower performance in other areasof reading (e.g., Chiappe & Siegel, 1999; Chiappe, Siegel,& Gottardo, 2002; Chiappe, Siegel, & Wade-Woolley, 2002;Geva and Siegel, 2000; Lesaux & Siegel, 2003).

Despite L2 students showing deficits in early vocabularyknowledge, the acquisition of word reading skills has beenfound to be approximately equal between L1 and L2 chil-dren (Jongejan et al., 2007), given appropriate instruction(Gersten & Baker, 2000). Jongejan et al. found that stu-dents learning Dutch as an L2 performed lower on somemeasures including phonological awareness and vocabularyknowledge, though they performed equally well on worddecoding tasks. Similar patterns of performance have beenfound in other studies (Aarts & Verhoeven, 1999; Geva &Yaghoub-Zadeh, 2006; Jean & Geva, 2009; Lesaux & Siegel,2003). Differences with respect to students’ L1 were foundto contribute to differences in phonological awareness andword reading skills based on the regularity and consistencyof the orthographic conventions of languages, specificallytheir grapheme-phoneme correspondences (Bialystok, Luk,& Kwan, 2005; Ziegler & Goswami, 2005). In the contextof this study, English has the least consistent orthographyand Spanish has the most consistent orthography. Portugueseorthography is in the middle—as it is highly consistent withsome sounds represented by accent markers rather than con-ventional graphemes. If children have acquired proficiencyin skills related to word reading, such as phonological aware-ness and naming speed, other oral language skills do not con-tribute as much to differences in word reading skills, evenin L2 students (e.g., Castles & Coltheart, 2004; D’Angiulli,Siegel, & Serra, 2001; Durgunoglu, Nagy, & Hancin-Bhatt,1993; Geva & Siegel, 2000; Geva et al., 2000).

Though much is known with regard to emergent literacyskills and their development in an L2, less is known about thedevelopment of reading comprehension in ELLs. Phonologi-cal awareness predicts reading comprehension in both L1 andL2 beginning readers (Lafrance & Gottardo, 2005; Lesauxet al., 2007; Proctor et al., 2005; Tannenbaum, Torgesen &

Wagner, 2006). Thus, are these relationships among languageand reading proficiency in older ELLs similar to relationshipsfound with monolinguals (Lindsey et al., 2003; Manis et al.,2004; Nakamoto, Lindsey, & Manis, 2007; Nakamoto, Lind-sey, & Manis, 2008)? Whether or not phonological awarenessremains an important predictor of reading ability in older stu-dents who are ELLs, or whether other factors contribute moreto the differences in L2 reading ability in this age group of stu-dents is still unknown. For example, vocabulary knowledgeand oral language proficiency play unique roles in readingcomprehension development and performance in English L1and L2 learners (August et al., 2005; Biemiller, 2003; Got-tardo & Mueller, 2009; Proctor et al., 2005; Verhoeven & VanLeeuwe, 2008). That is, vocabulary predicts reading compre-hension ability both concurrently and over time (Verhoeven& Van Leeuwe, 2008). The relationship between vocabu-lary knowledge and reading comprehension though, is notconsistent over development. According to the Simple Viewof Reading, reading comprehension in younger children ismore highly related to decoding, whereas in older children,who are assumed to have mastered basic decoding skills,vocabulary knowledge is more important for reading com-prehension (Catts, Hogan, & Adolf, 2005; Gough & Tunmer,1986; Hoover & Gough, 1990).

Another important factor related to the development ofreading comprehension, is print exposure, or extracurricularreading in school-aged children. Measures of print expo-sure are used to estimate how much a student reads out-side of school. These measures, used in previous studies byStanovich and colleagues, have been checked for convergentvalidity with diary studies (Allen, Cipielewski, & Stanovich,1992) and predict reading behavior in natural settings (West,Stanovich, & Mitchell, 1993). These measures also predictgrowth in receptive vocabulary and reading comprehension(Echols, West, Stanovich, & Zehr, 1996). Additionally, earlyreading ability measured in first grade predicts the amountthat students choose to read for pleasure in 11th grade, asmeasured through print exposure (Cunningham & Stanovich,1997). Conversely, Cunningham and Stanovich found thatprint exposure measured concurrently with vocabulary andreading comprehension, also explained differences in theseskills in the 11th grade. These findings indicate the impor-tance of extracurricular reading for monolingual speakersfrom middle-class backgrounds. The specific relationshipprint exposure has to reading in L2 students is unknownbut is presumed to play a similar, if not more important, rolein acquiring literacy skills in English for ELLs. Print expo-sure would provide another avenue for exposure to Englishin ELLs.

COGNITIVE ABILITY AND READING SKILL

Common underlying cognitive abilities (e.g., phonologicalawareness, memory, and speed of processing) are related tothe acquisition of reading skills in an L1 or an L2 (e.g.,Geva & Wade-Woolley, 1998). Therefore, three highly in-terrelated factors in reading acquisition were currently ex-amined: vocabulary, reading comprehension, and print expo-sure. L2 learners with a higher verbal ability have been found

70 GRANT, GOTTARDO, AND GEVA: READING IN ENGLISH AS A FIRST OR SECOND LANGUAGE

to have a higher vocabulary in both their L1 and L2, and tendto learn to read more easily (Bialystok, 2007; Carlisle, Bee-man, Davis, & Spharim, 1999). Print exposure, as previouslymentioned, has been shown to be related to both vocabularyand reading comprehension. Specifically, individuals withhigher exposure to print are more likely to have higher read-ing comprehension and vocabulary skill, and vice versa. L1students with higher reading ability are more likely to en-gage in extracurricular reading (Cunningham & Stanovich,1997; Grant, Wilson, & Gottardo, 2007; Stanovich, 1986;Stanovich, Cunningham, & West, 1998). To the best of theauthors’ knowledge, there has been no published literatureto date that has examined the influence of extracurricularreading on students learning to read in English as an L2.

Though many studies have looked at how skills in onelanguage predict the development of skills in an L2, or howability in one skill predicts ability in another, few studies haveexamined how these interrelations compare in different L1 orL2 groups learning the same language. This study exploresreading comprehension in Spanish and Portuguese ELLs,and native English speakers. The ELLs belong to groupswho began to acquire English as an L2 at different periodsof time, in addition to having differential exposure to theirnative language. These groups of ELLs were compared toeach other and to native English speakers with respect torelations among cognitive–linguistic measures and readingcomprehension. This study had three main research goals,which were:

(1) to compare the reading-related skills (i.e., vocabulary,decoding, lexical access, reading comprehension) ofsimultaneous and sequential bilinguals to each other,and to a group of English L1 students,

(2) to examine the relationship among these cognitive–linguistic measures, known to be related to readingcomprehension, across the three groups of learners,and

(3) to examine whether there are differences in the extentto which decoding, vocabulary, and print exposureare related to reading comprehension across the threegroups.

METHOD

Participants

Students who participated in this study were recruited fromurban and metropolitan areas in Canada. They participatedin a longitudinal research study beginning as early as juniorkindergarten (at the age of four) and lasting up until fourthgrade. However, only data involving participants who wereparticipating in third grade will be reported in this study. Thechildren were recruited to participate over several academicyears, due to difficulty in obtaining consent from the L2groups. The recruitment process continued until a sufficientnumber of students had consented to participate. Each yearof recruitment represented one wave of data. Thus, for thestudents in each language group, details on the number ofchildren recruited in each wave (i.e., cohort) of testing areincluded. Before the cohorts were collapsed into one group,their scores on all measures used in the assessment batterywere examined. If cohort differences were found, the cohortswere not collapsed and data were analyzed separately forthose groups. For further clarification on the breakdown ofthe three language groups and related information on the SESand generational status of families in each recruitment area,please see Table 1.

The groups differed not only based on L1, but also basedon L2 acquisition background. The Spanish L2 students areclassified as sequential bilinguals, having acquired Spanishas their L1 at home, and English once they entered the schoolsystem. The Portuguese students are classified as simultane-ous bilinguals, exposed to both English and Portuguese athome from birth or within their first year. One other ma-jor difference between these L2 groups was that the Span-ish children tended to come from families who were first-generation immigrants to Canada, whereas the Portuguesechildren tended to come from second-generation families.All of the L2 students began learning English in at leastGrade 1. Therefore, students had at least 3 years of formalschooling in English while in third grade, but had from 2(Grade 1 and Grade 2) to 9 years (birth to Grade 3) of expe-rience learning English dependent upon whether they began

TABLE 1Language Group Characteristics and Corresponding Testing Waves

HighestMedian Educational LevelYears of Income of % of

% of % of Experience GeographicLanguage Type of 1st-Generation 2nd-Generation Learning School High Bachelor Waves of N ofGroup Bilingual Immigrants Immigrants English Areas School Level Recruitment Each Wave

Spanish Sequential 69.5 22.0 2–5 $47,975 27.0 11.9 1st wave Grade 3 in 2006 262nd wave Grade 3 in 2007 22

Portuguese Simultaneous 29.0 21.2 Up to 9 $64,746 29.7 8.7 1st wave Grade 3 in 2005 82nd wave Grade 3 in 2006 53rd wave Grade 3 in 2007 134th wave Grade 3 in 2008 6

English Monolingual – – $64,746 29.7 8.7 1st wave Grade 3 in 2006 182nd wave Grade 3 in 2007 11

LEARNING DISABILITIES RESEARCH 71

TABLE 2Performance on All Measures by Group—Spanish Speakers (Broken up into Cohort 1 and Cohort 2), Portuguese Speakers, and Native English

Speakers

1 2 3 4

Spanish-Cohort 1 Spanish-Cohort 2 Portuguese English(N = 26) (N = 22) (N = 32) (N = 29)

Group* Mean SD Mean SD Mean SD Mean SD Differences∗

PPVTa 97.92 15.571 113.80 22.04 117.94 14.94 132.24 14.15 4 > 1, 2, 3; 2, 3 > 1PPVTb 85.11 0.94 98.33 17.35 98.47 10.85 111.69 11.20Word IDa 51.46 18.121 66.04 13.99 63.00 12.17 65.72 7.89 2, 3, 4 > 1Word IDb 91.04 8.61 108.93 18.37 102.16 15.12 107.59 10.30Word Att.a 22.08 10.861 27.89 9.07 27.91 10.21 30.52 6.80 4 > 1Word Att.b 92.08 7.88 105.04 15.55 102.06 18.14 108.72 14.43RAN-letters 43.34 15.08 38.40 5.92 35.55 7.26 38.96 5.49RAN-digits 39.24 9.34 36.38 7.02 34.43 7.79 37.52 6.20PAa 6.54 3.86 7.78 4.07 9.41 3.19 10.35 2.06 4 > 1, 2; 3 > 1Neale Ratea 57.15 26.611 71.56 27.54 70.42 20.21 78.33 19.42 4 >1Neale Rateb 98.04 4.00 104.85 12.91 103.74 10.51 109.48 8.71Neale Acc.a 39.73 22.041 51.71 19.98 47.87 17.43 53.10 15.36Neale Acc.b 92.50 3.70 101.00 13.32 96.16 12.15 101.69 10.34Neale RCa 11.58 7.28 18.89 9.09 19.36 7.56 22.59 7.33 2, 3, 4 > 1Neale RCb 85.27 11.83 97.44 14.60 96.39 12.40 103.10 10.81MATa 25.04 6.69 32.19 10.55 33.69 8.80 39.83 8.73 4 > 1, 2, 3; 2, 3 > 1MATb 94.56 7.33 102.22 12.30 101.34 9.67 109.41 9.29TRT-z score −0.79 1.40 0.08 0.70 0.27 .64 .40 .74 3, 4 > 1

Note: a = raw score; b = standard score; ∗1 = Spanish, Cohort 1; 2 = Spanish Cohort 2; 3 = Portuguese; 4 = English (differences are based on raw scores).PPVT = Peabody Picture Vocabulary Test; Word ID = word reading: Word Att. = pseudoword reading (word attack); RAN = rapid automatized naming;PA = phonological awareness (maximum score of 12); Neale Acc. = accuracy; Neale RC = reading comprehension; MAT = Matrix Analogies Test.

learning English at home or not. However, the specific num-ber of years each student had learning English was not alwaysavailable. The Spanish students had relatively greater profi-ciency in their L1 than their L2 at the time of school entry,whereas Portuguese students possibly had equal proficiencyin their L2 (English) and their L1 at the time of school entry.Both the language acquisition status of the students as wellas the number of years of English experience is a factor ininterpreting the results.

A total of 48 Spanish-speaking ELLs (N = 28 females)were tested. These students were recruited as part of the largerlongitudinal research study and were recruited in two waves.The first wave of students was in Grade 3 in 2006 (M = 106.0months, N = 12 males, 14 females), whereas the second waveof students was in Grade 3 in 2007 (M = 104.8 months, N =8 males, 14 females). Due to group differences (see Table 2)the L2 speaker’s data were analyzed separately for each waveof recruitment, with the first wave hereafter being referredto as the first cohort and the second wave of testing beingreferred to as the second cohort. The majority of Spanish-speaking residents in this urban area are from the Caribbeanand Central/South America. A large proportion of residentscome from: Cuba, Dominican Republic, Trinidad and To-bago, West Indies, Costa Rica, Colombia, Ecuador, Chile, ElSalvador, and also Mexico (Statistics Canada, 2006). Anec-dotally, the majority of children from this group reportedspeaking Spanish at home to their parents and often to theirbrothers and sisters. Additionally, many of the children re-ported being of Spanish origin and when prompted abouttheir language background in conversation, reported their

country of origin and stories of their parents’ immigration toCanada.

Thirty-two Portuguese-speaking bilingual learners (M =107.9 months, N = 25 females) were also recruited to par-ticipate through several waves of testing, with students beingin Grade 3 in four different waves—2005 (n = 8), 2006(n = 5), 2007 (n = 13), and 2008 (n = 6). The data forthese students were collapsed across cohorts as no cohortdifferences emerged among these four groups. The majorityof Portuguese-speaking people in this area of Canada emi-grated from the Azores region of Portugal, which is knownancestrally for its predominantly fishing background as ameans of full-time labor. Thus, the majority of this popu-lation at the time of immigration was not highly educated.One of the largest periods of immigration from Portugal toCanada was between 1960 and 1971 (Pier 21, 2010). Anec-dotal evidence from the Portuguese children tested was thatfew children actually identified with the Portuguese cultureas they only spoke Portuguese “a little bit” at home, and totheir grandparents.

Twenty-nine Grade 3 students (M = 105.1 months, N =14 females) who spoke English as an L1 were also recruited.The inclusion of the English L1 students allowed for a com-parison between the L2 students and the English L1 children.These students were recruited in two waves, the first waveof students was in Grade 3 in 2006 (n = 18), and the sec-ond wave of students was in Grade 3 in 2007 (n = 11).The data for these students were also collapsed across co-horts as no significant differences emerged between the twosamples.

72 GRANT, GOTTARDO, AND GEVA: READING IN ENGLISH AS A FIRST OR SECOND LANGUAGE

Although all reasonable efforts were taken to ensure groupcomparability, some differences among groups are still pos-sible despite efforts to recruit children from similar areas.In the area of Canada where the data were collected, manychildren from similar language backgrounds reside in thesame areas and go to the same schools. Thus, children withineach language group were more likely to be from within thesame geographic area than children from different languagegroups. There were two main areas from which children wererecruited in this study. All of the Portuguese L2 and EnglishL1 students came from the same six schools and were in thesame classrooms as one another. The Spanish L2 studentswere from a metropolitan area close to the area where thePortuguese and English students were recruited, but attendeddifferent schools (also a total of six schools). Additionally,students with many other L1s attended these schools but werenot recruited for the study.

Despite living in different communities, children fromdifferent language groups were recruited from communitieswith a very similar socioeconomic status (SES) accordingto data on median income and educational level from Statis-tics Canada (2006). The data available for the communi-ties in which these children resided showed that generallyacross these areas, approximately 30 percent of parents hada high school degree, 8 percent to 10 percent had a univer-sity bachelor-level degree, and the median income was ap-proximately $45,000–$65,000 per family before taxes. Thus,although the SES of these children was similar according tothe two most popular indicators of SES (income and edu-cation; Ensminger & Fothergill, 2003; Entwisle & Astone,1994), other differences not measured through home liter-acy questionnaires are possible. SES is not the only indicatorof home environment, other variables such as the numberof single-family detached houses, the number of individualsliving in high-rise apartment buildings, the number of indi-viduals living in one household, are possible contributors topositive and negative child outcomes. Neighborhood cohe-sion defined as a “person’s perception of their neighboursand the sense of support they feel from these neighbours”and other similar variables could be another possible in-dicator of child outcome that is not typically measured oraccounted for in research looking at children from differentgeographic areas (The Chief Public Health Officer’s Report,2009).

Due to differences that could have been present as a re-sult of nonmeasured variables, an analysis was carried outto examine if there were school effects related to readingperformance. A multivariate analysis of covariance, control-ling for language group, showed that there was no significantrelationship between school and performance on any of themeasures used in this study (ps > .06).

Materials

Five standardized and two nonstandardized measures, all ad-ministered in English, were used in this study. The tests in-cluded measures of oral language proficiency, measures ofreading, print exposure, and comprehension skills that arerelated to reading.

Oral Language Proficiency

Peabody Picture Vocabulary Test (PPVT-III ; Dunn & Dunn,1997). Children’s receptive vocabulary was measured usingForm B of the PPVT-III. Each participant was presented withfour pictures at a time and pointed to the picture he thoughtrepresented the word being presented orally. The test becameincreasingly difficult as the participant progressed. Reportedreliability for age eight is .95, and age nine is .96 (Dunn &Dunn, 1997).

Reading Measures

Woodcock Reading Mastery Test-Revised (WRMT-R; Wood-cock, 1987). The Word Identification subtest involved read-ing as many words as possible before reaching a ceiling of sixconsecutive errors. The words began at an easy level (e.g.,is, you, and) and got progressively harder throughout thetask (e.g., torpedo, amazement). The Word Attack subtest, ameasure of pseudoword reading, also involved items that gotprogressively more difficult throughout the task (e.g., dee, ift,zirdn’t, gnouthe). Reported reliabilities from the norms forEnglish-speaking children for the pseudoword reading taskat age nine are .91, and are .94 for the word reading task(Woodcock, 1987).

Neale Analysis of Reading Ability (Neale, 1989). Thistest involved reading aloud a series of short passages, forwhich open-ended comprehension questions followed. Threetypes of scores were available. For each passage childrenwere timed, which provided a measure of reading fluency(reading rate score). Errors made in reading text were alsorecorded in order to analyze word reading accuracy (accuracyscore). Reading comprehension scores were based on thenumber of questions answered correctly for each passageread (comprehension score). Reported reliabilities for theaccuracy score from the norms for English-speaking childrenat age 8–9; 11 are .90, for the comprehension score are .85,and for the reading rate score are .71 (Neale, 1989).

Phonological Processing

Phonological awareness. This experimental measure in-volved three sections. However, only the phoneme deletionsection of the test was administered to third grade students.This subtest involved deleting phonemes within the onsetor rime (e.g., say “bip” without /p/). Students had reachedceiling on the syllable deletion and onset deletion sections inprevious years. The maximum score on each subtest was 12.Cronbach’s alpha for this test was .70 for the last 12 itemsalone.

Lexical Access

Rapid Automatized Naming (RAN ; Wagner, Torgesen, &Rashotte, 1999). Two tests from the Comprehensive Test ofPhonological Processing were used, which involved readinga list of numbers (digits) or letters, presented in rows, as fast

LEARNING DISABILITIES RESEARCH 73

as possible. Each subtest (letters, and digits) involved twotrials, for which the total time taken to read each trial wasrecorded in seconds. Reported reliabilities from the normsfor English-speaking children at age eight are .80 for thedigit naming, and are .72 for the letter naming task (Wagner,Torgesen, & Rashotte, 1999).

Cognitive Factor: Nonverbal Reasoning

Matrix Analogies Test (MAT-Expanded Form; Naglieri,1989). This test involved four subtests related to nonver-bal reasoning and problem solving. The first was PatternCompletion, whereby the participant was asked to choosefrom one of the six pictures below in order to complete thepattern. The other three tests were: Reasoning by Analogy,Serial Reasoning, and Spatial Visualization, which involvedsimilar procedures to the first subtest and also involved rea-soning about visual patterns. Reported reliabilities from thenorms for English-speaking children at age eight are .93, andat age nine are .92 (Naglieri, 1989).

Print Exposure

Title Recognition Test (TRT; Stanovich & West, 1989). TheTRT listed both real titles of popular children’s books, anditems that were not real book titles, the latter of which actedas foils to detect and prevent guessing. Participants wereinstructed to check off the names of the books that theyrecognized to be real titles, and to not guess names they didnot recognize as being “real.” The test involved a series of40 titles, 25 of which were “real” titles, and 15 that served asfoils. The reliability based on the percentage of correct itemschecked was .88 (Cronbach’s alpha).

Scoring on this task was carried out in a manner consis-tent with past research. This method involved a correctedpercentage that took into account the amount of guessing onthe questionnaire. This corrected percentage was based on thediscrimination index from the two-high threshold model ofrecognition performance (Snowgrass & Corwin, 1988). Thisdiscrimination index involved taking the percentage of cor-rect items minus the percentage of incorrect items (foils) onthe task. This is based on analyses carried out by Stanovichand colleagues involving use of these questionnaires. Thiscorrected percentage was then converted to a z-score, whichwas used in all subsequent analyses.

Procedure

The series of standardized and experimental measures wereadministered to students in two to three individual testingsessions over a maximum period of 2 weeks. The total test-ing time was approximately three to three and a half hours,as it included several other tasks, the results of which willnot be reported in this study. The tasks varied slightly in theorder that they were presented as each student was involvedin several individual testing sessions and one group testingsession in which they completed the TRT questionnaire. Thepresentation of these tasks was not rigid, but generally in-

volved giving the tests a predetermined order that varied taskdifficulty and response requirements (i.e., oral vs. written).The order typically proceeded in the following way: receptivevocabulary, word and nonword reading, test of phonologicalawareness, reading comprehension test, rapid naming of let-ters, rapid naming of digits, nonverbal reasoning, and thenthe TRT questionnaire.

RESULTS

The analysis of the data proceeded in three main steps. First,the mean standard scores on the primary measures of readingwere examined in order to compare the mean performanceon these tests between the L1 and L2 groups. It is impor-tant to interpret these results with some caution as norms formany standardized measures are based on English-speakingpopulations that come from middle-class homes. Althoughsome of the standardized measures were normed on repre-sentative samples of the population at the time of their pub-lication, many individuals from lower SES groups and thosewith lower educational levels were still underrepresented. Forthis reason, although standardized scores are reported in themeans tables, only raw scores are used for analyses includingregressions and comparisons of means between groups.

The L2 learners were subdivided into groups to exam-ine whether there were differences among the groups due tothe language acquisition background of each group (sequen-tial or simultaneous bilinguals). The data for the Spanishspeakers were analyzed by cohort because group differenceswere found in performance. The second step of the anal-yses involved examining the correlations among the mainvariables in these four groups. Last, a series of hierarchicalregression analyses were carried out in order to further un-derstand the specific relationships among several cognitivecomponents and print exposure variables, and reading com-prehension. Prior to conducting the analyses, distributionswere examined for outliers, and due to this, one participantwas excluded from further analyses. This participant was notexcluded due to deficits in reading skill, but due to the re-quirement that participants obtain scores broadly within theaverage range on the measure of intelligence used in thisstudy. This student scored below the 10th percentile on theMatrix Analogies Test.

Comparison of Mean Scores

Before the analyses proceeded, cohorts were compared todetermine whether or not there were differences betweenstudents who completed the testing in different years of datacollection. Significant differences emerged on some of thestandardized measures between the two cohorts of Span-ish L2 students—thus all further analyses for this languagegroup proceeded separately for each cohort (see Table 2 fordifferences). No significant differences emerged between thedifferent cohorts of students within the English or Portuguesegroups, thus these data were collapsed across cohorts and an-alyzed as one group for these languages.

Additionally, the data were analyzed for gender differ-ences for descriptive purposes before further analyses were

74 GRANT, GOTTARDO, AND GEVA: READING IN ENGLISH AS A FIRST OR SECOND LANGUAGE

continued. One significant gender difference emerged in thesecond Spanish cohort. Males significantly outperformed fe-males on the test of reading comprehension, F(1, 20) = 6.38,p < .05. There were no outliers for either group, and the rangeof scores obtained on this test was approximately equal. Thisgroup was the only group for which significant gender differ-ences emerged, though a great deal of caution must be takenin interpreting the results due to the small sample size ofeach gender within groups (as low as eight participants in agroup). Due to these methodological issues, no further anal-yses based on gender were carried out. This initial analysiswas conducted for descriptive purposes and the results couldbe considered in future studies where sufficient sample sizespermit a reliable analysis between genders.

A one-way Analysis of Variance was used to compare thetwo Spanish groups—the Portuguese group and the Englishgroup on the variables of interest (see Table 2). Levene’s Testfor Homogeneity of Variance was carried out to determinewhether or not the variance was approximately equal amongthese four groups. Four tests, specifically, word reading, wordattack, phonological awareness, and the title recognition test,showed significant differences in the variance between thesamples for the measures used. Due to these general vio-lations in homogeneity of variance, Dunnett’s T3 post hoccomparison test for unequal variances and unequal samplesizes was used to analyze the differences between all groups.

Some similar relationships emerged in comparing thesegroups across tests. Though the patterns were not always sta-tistically significant, the Portuguese group generally outper-formed the first and second Spanish cohort, while the secondSpanish cohort typically outperformed the first Spanish co-hort. The English group outperformed all three groups onmany measures and had significantly higher scores than thefirst Spanish cohort on most measures (see Table 2). The firstSpanish cohort had significantly lower scores than the otherthree groups on vocabulary, word reading, and nonverbalreasoning, and significantly lower scores than the Portugueseand English groups on phonological awareness. The Englishgroup outperformed all the L2 groups on vocabulary andnonverbal reasoning. No significant differences were foundamong the groups on rapid naming of letters and digits.

Some differences also emerged on measures of readingcomprehension and print exposure. As previous research hasdemonstrated, some caution must be taken when examin-ing differences on measures of reading comprehension. Re-search has found developmental differences in test perfor-mance (Keenan, Betjemann, & Olson, 2008), and a differ-ential relationship between measures of oral language anddecoding for multiple tests of reading comprehension (Cut-ting & Scarborough, 2006; Keenan et al., 2008; Nation &Snowling, 1997). Performance differences were found on theNeale reading comprehension score with the second Spanishcohort, the Portuguese group, and the English group outper-forming the first Spanish cohort. Interestingly, standardizedz-scores on the print exposure measure, the TRT, showed thatthe Portuguese and English groups were outperforming thefirst cohort of Spanish speakers. The second cohort of Span-ish speakers had scores in between the first cohort of Spanishspeakers and the Portuguese speakers. This pattern matchesthe patterns exhibited in the majority of the data and will beexamined further in subsequent analyses.

Despite the demonstrated differences among these fourgroups on standardized measures of reading, they were allgenerally performing within the average range (at or slightlyabove the mean) on standardized measures. The lowest scorewas one standard deviation below the mean for the recep-tive vocabulary scores of the first Spanish cohort. Althoughthe L2 students were performing within the average range,both the Spanish cohorts and the Portuguese cohort scoredsignificantly lower on receptive vocabulary than the Englishcomparison group. These low scores are consistent with otherresearch demonstrating lower receptive vocabulary scores forL2 students despite average to above average performanceon measures of word reading skill (e.g., August et al., 2005;Schwartz et al., 2007).

Correlations among Measures

There are a large number of comparisons that can be made be-tween the intercorrelations among variables (see Tables 2 and3). Some comparisons will be highlighted in the context of

TABLE 3Intercorrelations among Variables with Cohort 1 of Spanish-Speaking Participants Below the Diagonal and Cohort 2 of Spanish- Speaking

Participants Above the Diagonal

Variables 1 2 3 4 5 6 7 8 9 10 11

1. PPVT – .579∗∗ .554∗∗ −.070 −.175 .419 .684∗∗∗ .765∗∗∗ .619∗∗ .461∗ .601∗∗2. Word ID .388 – .824∗∗∗ −.342 −.237 .607∗∗ .846∗∗∗ .733∗∗∗ .497∗∗ .466∗ .541∗∗3. Word Att. .312 .881∗∗∗ – −.299 −.234 .792∗∗∗ .785∗∗∗ .733∗∗∗ .346 .583∗∗ .508∗4. RAN-letters −.232 −.500∗∗ −.468∗ – .866∗∗∗ −.361 −.501∗∗ −.438∗ −.170 −.045 −.1995. RAN digits −.204 −.521∗∗ −.514∗∗ .847∗∗ – −.309 −.490∗ −.390 −.239 .071 −.2736. PA .439∗ .530∗∗ .604∗∗∗ −.324 −.350 – .756∗∗∗ .573∗∗ .294 .369 .3957. Neale Acc. .354 .939∗∗∗ .833∗∗∗ −.526∗∗ −.591∗∗ .484∗ – .813∗∗ .444∗ .430∗ .586∗∗8. Neale RC .478∗ .875∗∗∗ .768∗∗∗ −.477∗ −.532∗∗ .346 .923∗∗∗ – .499∗ .499∗∗ .662∗∗9. Neale Rate .325 .618∗∗∗ .505∗∗ −.572∗∗ −.543∗∗ .242 .524∗∗ .491∗ – .181 .607∗∗10. MAT .347 .723∗∗∗ .604∗∗∗ −.423∗ −.274 .303 .579∗∗ .648∗∗∗ .646∗∗∗ – .28111. TRT z-score .254 .480∗ .182 −.709∗∗∗ −.532∗∗ .480∗ .583∗∗ .553∗∗ .296 .370 –

∗ p < .05, ∗∗ p < .01, ∗∗∗ p < .001.

LEARNING DISABILITIES RESEARCH 75

TABLE 4Intercorrelations among Variables with Portuguese-Speaking Participants Below the Diagonal and English-Speaking Participants Above the

Diagonal

Variables 1 2 3 4 5 6 7 8 9 10 11

1. PPVT – .268 .040 −.262 −.283 .202 .387∗ .475∗∗ .597∗∗∗ .368∗ .508∗2. Word ID .372∗ – .610∗∗ −.416∗ −.582∗∗ .479∗∗ .870∗∗∗ .747∗∗∗ .275 .148 −.0633. Word Att. .225 .860∗∗ – −.429∗ −.447∗ .551∗∗ .613∗∗∗ .387∗ .136 .023 .0274. RAN-letters −.095 −.260 −.088 – .797∗∗∗ −.315 −.368∗ −.455∗ −.464∗ −.171 .1395. RAN-digits −.270 −.512∗∗ −.408∗ .621∗∗∗ – −.256 −.413∗ −.429∗ −.431∗ −.143 .1486. PA .295 .716∗∗ .713∗∗∗ −.135 −.210 – .445∗ .457∗ .145 .053 .1437. Neale Acc. .395∗ .833∗∗ .865∗∗∗ −.259 −.568∗∗∗ .661∗∗∗ – .813∗∗∗ .249 .141 .1268. Neale RC .466∗∗ .688∗∗ .784∗∗∗ −.153 −.366∗ .746∗∗∗ .817∗∗∗ – .257 .306 .2299. Neale Rate .215 .486∗∗ .523∗∗ −.305 −.465∗∗ .398∗ .561∗∗∗ .517∗∗ – .119 .11610. MAT .429∗ .435∗ .398∗ −.095 −.234 .454∗ .493∗ .642∗∗∗ .307 – .22211. TRT z-score .063 .197 .146 −.154 −.164 .150 .276 .263 .267 .703∗∗∗ –

∗ p < .05, ∗∗ p < .01, ∗∗∗ p < .001.

this study. First, some interesting comparisons can be madebetween the two cohorts of Spanish speakers—who havevastly different data despite the fact that they were recruited1 year apart from each other in the same school system, com-ing from the same areas and even attending the same schools.For the first cohort of Spanish students, receptive vocabularywas related to phonological awareness and reading compre-hension. Whereas for the second cohort of Spanish speakers,receptive vocabulary was highly correlated with measuresof word and pseudoword reading, reading comprehension,nonverbal reasoning and print exposure. This pattern is in-teresting given the lower performance of the first cohort onmany measures of reading (see Table 3). For the first Spanishcohort, reading comprehension was related to decoding mea-sures. For the second Spanish cohort, reading comprehensionwas related to decoding and vocabulary. For the Portugueseand English L1 students, reading comprehension was alsorelated to decoding and vocabulary (see Table 4).

Nonverbal reasoning is designed to test general intellec-tual ability, and as such is used to ensure that children arewithin the average ability range. There is inconclusive evi-dence as to whether nonverbal reasoning is related to generalreading ability in L1s, though research suggests a lack ofrelationship between IQ and reading ability (e.g., Jimenez,Siegel, O’Shanahan, & Ford, 2009). One would hypothesizeconsistent relationships between nonverbal reasoning andother reading-related variables in both L2 and L1 groups,however, this did not hold true for all groups in this study(see Tables 2 and 3). Nonverbal reasoning was related toword reading in the first Spanish cohort, while in the secondSpanish cohort, nonverbal ability was also found to be relatedto receptive vocabulary, word and pseudoword reading, andreading comprehension. In the Portuguese group, nonverbalreasoning was related to receptive vocabulary, phonologicalawareness, and reading comprehension. However, in the En-glish group, nonverbal ability was only related to receptivevocabulary (see Jimenez et al., 2009 for similar conclusionswith native Spanish speakers reading in their L1).

Overall, print exposure was related to word reading, wordreading efficiency and reading comprehension in both Span-ish cohorts, and was also significantly related to rapid naming

and phonological awareness in the first cohort. Print exposurewas not related to measures of reading in both the Portugueseand the English groups but was related to nonverbal reason-ing and receptive vocabulary, respectively.

Hierarchical Regression Analyses

Relationships among cognitive–linguistic variables andreading comprehension. A series of hierarchical regres-sion analyses were carried out in order to further under-stand the relationship between reading comprehension andthe cognitive–linguistic variables for the different L1 or L2groups. The first hierarchical regression analysis examinedthe relationships among nonverbal reasoning, decoding, andreceptive vocabulary, and reading comprehension, as theseare key variables known to be related to reading compre-hension (see Table 5 for β values).1 Nonverbal reasoning,followed by receptive vocabulary and word reading were en-tered as predictors of reading comprehension.

In the first Spanish cohort, nonverbal reasoning wasa significant predictor of reading comprehension, F(1,22) = 19.55, p < .001, R2 = .48. The final modelalso included all three variables, however, only wordreading added unique variance to reading comprehensionwhile receptive vocabulary did not, F(3, 19) = 21.99,p < .001, R2 = .78. This pattern may have occurred becausethis group had much lower scores on vocabulary, which maynot have helped these students to comprehend the materialthey read. For the second Spanish cohort, nonverbal reason-ing once again significantly predicted variance in readingcomprehension, F(1, 20) = 6.62, p = .02, R2 = .25. Forthe final model, only receptive vocabulary and word readingwere associated with unique variance in reading comprehen-sion, F(3, 18) = 15.13, p < .001, R2 = .72. The Portuguesegroup demonstrated similar results, where nonverbal rea-soning was significant as the first step, F(1, 22) = 10.74,p < .01, R2 = .22, but in the final model, only receptive

1Due to the small sample size, the number of variables entered in eachregression equation was limited.

76 GRANT, GOTTARDO, AND GEVA: READING IN ENGLISH AS A FIRST OR SECOND LANGUAGE

TABLE 5Hierarchical Multiple Regression Analysis Predicting Reading Comprehension

Spanish-Cohort 1 Spanish-Cohort 2 Portuguese English

Std. Std. Std. Std.Predictors R2 R2 � Beta t-test R2 R2� Beta t-test R2 R2 � Beta t-test R2 R2 � Beta t-test

Step 1: .482∗∗∗ .249∗ .222∗ .093MAT .694 4.42∗∗∗ .499 2.57∗ .471 2.45∗ .306 1.67

Step 2: .514∗∗∗ .032 .612∗∗∗ .363 .347∗ .125 .245∗ .152MAT .617 3.63∗∗ .185 1.14 .316 1.60 .151 .83PPVT .194 1.14 .680 4.22∗∗∗ .386 1.96 .419 2.29∗

Step 3: .776∗∗∗ .294 .716∗∗∗ .104 .698∗∗ .351 .650∗∗∗ .405MAT .027 0.15 .082 .555 .132 .93 .11 .89PPVT .160 1.35 .489 3.06∗ .299 2.15∗ .256 1.95Word ID .793 4.72∗∗∗ .411 2.57∗∗ .637 4.71∗∗∗ .661 5.37∗∗∗

∗ p < .05, ∗∗ p < .01, ∗∗∗ p < .001.

vocabulary and word reading were unique statistical pre-dictors of reading comprehension, F(3, 20) = 17.08, p< .001, R2 = .69. The English group showed differ-ent results where nonverbal reasoning was not a signifi-cant predictor in any step of the model, F(1, 27) = 2.78,p = .11, R2 = .09. Only word reading predicted unique vari-ance in reading comprehension for the third and final step ofthe model, F(3, 25) = 15.46, p < .001, R2 = .65.

Relationships between print exposure and reading com-prehension. The final set of hierarchical multiple regres-sion analyses looked at the relationship between print expo-sure and reading comprehension. Three different hierarchicalmodels were tested in which receptive vocabulary (Model 1),word reading (Model 2), and nonverbal reasoning (Model 3)were each entered in the first step for each separate model,with print exposure entered subsequently in the second stepin all three models (see Table 6 for β values). The regressionswere carried out in this manner to examine the unique contri-bution of print exposure to the prediction of reading compre-hension above vocabulary, decoding, or nonverbal reasoning,which were all highly related to reading. Thus, is the amountof time that a student chooses to read for pleasure related totheir skill in comprehension over and above receptive vocab-ulary, decoding, or general cognitive ability? The results ofthis series of regression analyses are only displayed for thetwo Spanish cohorts due to the nonsignificant relationshipbetween print exposure and the variables of interest in thePortuguese- and English-speaking groups. The one excep-tion to this finding was for the English group in relation toword reading. The results of this analysis are reported.

The first set of regression analyses looked at receptive vo-cabulary and the TRT in relation to reading comprehension.For the first Spanish cohort, receptive vocabulary signifi-cantly predicted reading comprehension, F(1, 22) = 5.39, p= .03, R2 = .19, but once print exposure was added in thesecond step, the unique contribution of vocabulary becamenonsignificant, F(2, 21) = 7.12, p < .01, R2 = .40, whileprint exposure added unique variance. For the second Span-ish cohort, receptive vocabulary predicted unique variancein both the first, F(1, 20) = 28.22, p < .001, R2 = .58, andsecond model, F(2, 19) = 17.58, p < .001, R2 = .65, whilethe contribution of print exposure to the prediction of readingcomprehension did not add unique variance to the model.

The second set of regression analyses looked at theprediction of reading comprehension through decoding andprint exposure. For the first Spanish cohort, word readingexplained unique variance, F(1, 22) = 68.45, p < .001,R2 = .76, though print exposure did not contribute uniquelyto the final model, F(2, 21) = 37.40, p < .001, R2 = .78.For the second Spanish cohort, however, word reading didcontribute uniquely in both the first step of the model,F(1, 20) = 23.16, p < .001, R2 = .54, and the completemodel, F(2, 19) = 16.61, p < .001, R2 = .64. For the finalmodel, print exposure explained unique variance in readingcomprehension over and above word reading for this group.As mentioned previously, this was the only model in whichprint exposure added unique variance to the prediction ofreading comprehension in the English group. In the firststep, word reading (β = .75) significantly predicted readingcomprehension, F(1, 27) = 33.99, p < .001, R2 = .56. Forthe complete model, both word reading and print exposuresignificantly predicted comprehension (β = .76, β = .28,respectively), F(2, 26) = 22.47, p < .001, R2 = .63.

Finally, the third set of regression analyses was run in orderto determine whether print exposure could explain additionalvariance in comprehension over and above cognitive abilityas measured through nonverbal reasoning in this study. Inboth cohorts, print exposure explained unique variance overand above the contribution of nonverbal reasoning. For thefirst Spanish cohort, the final model predicted slightly morevariance, F(2, 20) = 14.72, p < .001, R2 = .59, with bothfactors contributing significantly to the model with nonver-bal reasoning predicting a large amount of variance. Similarresults were found for the second Spanish cohort, F(2, 19) =11.35, p < .001, R2 = .54, with print exposure contributingunique variance. Thus, print exposure seems to be explain-ing additional variance in reading comprehension depend-ing upon the group measured and the contribution of othercognitive–linguistic variables.

DISCUSSION

The inclusion of more than one language group in this re-search study provided unique insights into the differences

LEARNING DISABILITIES RESEARCH 77

TABLE 6Hierarchical Multiple Regression—Does Print Exposure Add to the Prediction of Reading Comprehension?

Spanish-Cohort 1 Spanish-Cohort 2

Predictors R2 � R2 Std. Beta t-test R2 � R2 Std. Beta t-test

Model 1 Step 1: .197∗ . .585∗∗∗PPVT .443 2.23∗ .765 5.31∗∗∗

Step 2: .404∗∗ .208 .649∗∗∗ .064PPVT .324 1.86 .575 3.38∗∗TRT z-score .471 2.71∗ .316 1.86

Model 2 Step 1: .757∗∗∗ .537∗∗∗Word ID .870 8.27∗∗∗ .733 4.81∗∗∗

Step 2: .781∗∗∗ .024 .636∗∗∗ .100Word ID .785 6.74∗∗∗ .529 3.22∗∗TRT z-score .177 1.52 .375 2.28∗

Model 3 Step 1: .482∗∗∗ .249∗MAT .694 4.42∗∗∗ .499 2.57∗

Step 2: .595∗∗∗ .113 .544∗∗∗ .295MAT .560 3.66∗∗ .340 2.10∗TRT z-score .362 2.37∗ .566 3.51∗∗

∗ p < .05, ∗∗ p < .01, ∗∗∗ p < .001.

in reading comprehension among ELLs with different L1sas well as an English L1 group. The language acquisitionbackground of L2 students being studied needs to be takeninto account, as evidenced by significant group differencesbetween simultaneous and sequential bilinguals in this study.It also speaks to the generalizability of the results of studieswhere multiple L1 groups are collapsed and English L1 stu-dents are the comparison group. Not only do the results ofthese studies potentially mask important group differences,but they fail to examine within-group and between-group dif-ferences in ELLs alone. Research thus far in the L2 literaturehas produced divergent results as to whether L2 students areperforming at, below, or above the level of L1 students on var-ious academic tasks. One of the reasons for these differencesis reflected in this study. In the Spanish as an L1 group, dif-ferences emerged in performance between students recruitedin elementary school 1 year apart. These students were fromthe same urban area, the same schools, and there was no spe-cific school effect on performance. Thus, the diversity evenwithin the same language group demonstrates the limitationsof basing models of reading acquisition for L2 learners on asingle sample of students.

The consistent differences that emerged between the twoSpanish cohorts in this study were unexpected. There aremany potential reasons as to why these differences arose. Per-haps the countries from which students’ parents immigratedwere different from one year to the next, or more of the stu-dents in the second cohort had greater exposure to English athome before they entered school. Also, many families whoimmigrate to Canada from other countries have advancedtraining in their home country, which is one of the factorsCanadian immigration policy evaluates when selecting indi-viduals to enter the country (www.cic.gc.ca/ english/skilled).Thus, although families may be living in lower SES areas,their educational levels may be higher than what is typicallyexpected in those areas due to underemployment, unemploy-ment, and lower wages once they settle in Canada. Therefore,

it is possible that the educational background of the familiesin the second cohort was higher than those in the first, despiteliving in the same areas and attending the same schools asthose families in the first Spanish cohort. Given that thesedifferences were not expected within language groups, morein-depth data on the families were not collected at the timethe study began. Also, given the difficulty in recruitment, andthe time taken to translate questionnaires into the parent’s L1,it was difficult to obtain thorough background informationfrom all families involved. Future studies should seek toascertain as much background information on families andchildren as possible in order to explain potential confoundsand differences within a sample.

Despite differences between groups in performance, it isimportant to emphasize that L2 learners in this study wereperforming well within the average range on standardizedtests based on norms expected for native English speak-ers. The measure for which L2 students performed in theslightly below average range, was English receptive vocab-ulary, where the first Spanish cohort had average scores onestandard deviation below the mean. Significant differencesemerged between not only the two Spanish samples, but alsobetween the Spanish, Portuguese, and English groups in re-lation to the specific measures related to reading comprehen-sion. The pattern of results indicated that the English grouptypically demonstrated the highest scores, and consistentlyoutperformed the first Spanish cohort (the lowest performinggroup) on many cognitive–linguistic measures. In comparingthe English to the Portuguese group, however, the major dif-ference that emerged between these two groups was that theEnglish group had higher vocabulary scores. This finding isconsistent with the literature showing that L2 students tendto lag behind L1 students in vocabulary (e.g., Farnia & Geva,in press; Proctor et al., 2005), though they often performon par with L1 students on other academic tasks includingphonological awareness (e.g., Bruck & Genesee, 1995) andrapid naming (e.g., Lesaux & Siegel, 2003).

78 GRANT, GOTTARDO, AND GEVA: READING IN ENGLISH AS A FIRST OR SECOND LANGUAGE

Students who began acquiring English simultaneous totheir L1 (the Portuguese students) had higher scores on manylinguistic variables in comparison to the Spanish studentswho acquired their L2 sequentially and had greater profi-ciency in their L1 (Spanish) at the time of school entry. Oneconfound these group comparisons present is the number ofyears students had been learning English. This factor thoughwas difficult to control for and partial out from the effectsof language acquisition background in this study, given thatsimultaneous bilinguals would by definition have more ex-perience with the L2 than sequential bilinguals of the samechronological age. Obtaining sufficient sample sizes and stu-dents from specific language backgrounds proved difficultgiven the very diverse urban area from which the studentsin this study were recruited. Future studies should consideracquiring this information from students and/or parents at theoutset of the study in order to ensure group comparability andto have additional information on the linguistic backgroundof students being studied.

In this study, the level of vocabulary knowledge in eachgroup was related to each group’s rank order performanceon the test of reading comprehension. Adams (1990) sug-gests that children with higher levels of oral proficiency havehigher comprehension skill than children with less elabo-rated vocabularies. Work with monolingual English speakersshows that oral language skills explain an increasing amountof variance in reading comprehension as children get olderand presumably gain reading skill (Carver, 1997; Catts etal., 2005). Significant differences were found between theL1 group and the three L2 groups in vocabulary knowledge.This finding highlights the importance of better understand-ing L2 children’s vocabulary development, and illustratesthat L2 children continue to lag behind in their vocabularydevelopment over time compared to L1 norms.

Vocabulary development is also an issue to be consideredwhen evaluating ELLs who may be considered “at-risk” forreading disabilities due to their poorer oral language (i.e., vo-cabulary) skills instead of due to inherent problems learning(Limbos & Geva, 2001). In the United States, research hasshown a consistent overrepresentation of minority students,especially Hispanic students, as having learning disabilities(Office of Special Education Programs, 2002). As a result,Geva (1999) emphasizes that assessments should be sepa-rated at the level of oral and written language. Assessmentof word reading skills can be conducted early, when the chil-dren have a few years of English experience. Since the de-velopment of higher-level oral language skills can often takeseveral years to develop (Cummins, 1991), the language ac-quisition background of the students being assessed shouldalso be considered. Students learning English sequentiallymay need additional years of English experience before theycan be accurately assessed for oral language proficiency andreading comprehension.

Several recent studies in vocabulary training have shownpromising effects that early intervention can have for L2 stu-dents (Biemiller & Boote, 2006; Townsend & Collins, 2009).When ELLs enter the school system, they tend to be treated asa homogeneous group, who all need train to learn academicand perhaps also everyday English. However, the specific ar-eas in which each ELL needs training may differ depending

upon their language acquisition background. Given the dif-ferences demonstrated in this study, both practitioners andeducators should take into account multiple factors whenconsidering the type and amount of training these studentsneed.

Though vocabulary in these L2 groups lagged behind thedevelopment of other cognitive–linguistic skills in compari-son to the English L1 group, it played a differential role inpredicting reading comprehension across groups. Dependingupon the group measured, either vocabulary or word read-ing predicted more unique variance in reading comprehen-sion. For the first Spanish cohort with the lowest vocabularyscores, word reading was related to reading comprehensionwhile vocabulary knowledge was not. Though word read-ing was more highly related to reading comprehension in thesecond Spanish cohort and the Portuguese group, vocabularyalso played a significant role. Perhaps due to the first Span-ish cohort’s lower receptive vocabulary scores, vocabularywas less related to reading skill and did not play as muchof a role in explaining reading comprehension ability in thisgroup. Vocabulary knowledge was likely too low to aid inreading comprehension in this group. This model, similar tothe Simple View of Reading, shows the relative influenceand importance of decoding and vocabulary in different L2groups.

What is particularly interesting in this model is that non-verbal reasoning was significantly related to reading compre-hension in all three L2 groups, while it was not in the Englishgroup. Although the literature on the relationship betweenIQ and reading in a first language is somewhat inconclusive,nonverbal reasoning might have a different relationship toreading in an L2. Nonverbal reasoning has been found toaccount for variance in general reading ability (Francis et al.,2006; Jimenez et al., 2009; Naglieri, 1996), and may suggestthe importance of general reasoning ability in allowing oneto learn to read in an L2. Also, relevant to models of read-ing comprehension is the finding that receptive vocabularyand word reading made differing contributions to the predic-tion of reading comprehension in all three L2 groups and inthe English L1 group. Because each of these groups had adifferent level of reading skill and English oral language pro-ficiency, relationships of variables to reading comprehensionskills are not identical for all groups in this study.

There has been discussion as to what variables should beincluded when investigating the prediction of reading com-prehension in ELLs (e.g., Kirby & Savage, 2008; Ouelette &Beers, 2009). Kirby and Savage argue that though the Sim-ple View of Reading is a good base model for understand-ing comprehension, further clarification and specification isneeded with respect to the variables used to proxy “linguisticcomprehension” and “decoding.” More specifically, linguis-tic comprehension comprises variables such as vocabulary,syntax, inferencing, and the construction of mental schemas,though it is typically measured through only one variablein studies testing this model. Additionally, decoding thoughtypically measured through word reading can also includepseudoword reading and phonological awareness. Thus, inL2 students it may be particularly important to consider mul-tiple indicators of specific skills (e.g., measuring word read-ing and phonological awareness as a proxy for decoding) in

LEARNING DISABILITIES RESEARCH 79

models of reading comprehension due to differential perfor-mance of L2s on components of reading included in modelssuch as the Simple View.

Rapid naming was the only task for which there wereno differences among the four groups in this study, despitethe large differences that consistently emerged between theEnglish group and the first Spanish cohort on other tasks.This cognitive advantage for bilinguals on rapid naming tasksseems to be a finding that is emerging in the literature (e.g.,Chiappe & Siegel, 1999; Chiappe et al., 2002; Chiappe et al.,2002; Geva et al., 2000; Lesaux & Siegel, 2003). Bialystokand colleagues have found other cognitive advantages forbilingual or L2 individuals on tasks measuring lexical access,inhibition, or measuring controlled attention such as throughthe Simon Task (Bialystok, 2006). Researchers exploring therole of rapid naming have examined the impact of responsetime and response inhibition on performance for RAN tasksin native speakers (Savage, Pillay, & Melidona, 2007). Inexploring cognitive differences in young L2 learners, it isimportant to consider using measures that tap potential areasof strength in ELLs in order to gain a more accurate pictureof their cognitive and linguistic skills.

This finding is also particularly relevant in diagnosingELLs with learning disabilities. As mentioned, difficultiesexist in accurate assessment of ELLs due to challenges indifferentiating language-learning difficulties that result fromlearning an L2 and learning difficulties based on a disabil-ity (Durgunoglu, 2002). Since low oral language proficiencyis often related to the misclassification of L2 students ashaving a learning disability (Limbos & Geva, 2001), othercomponent skills (i.e., rapid naming), which are not lowerin L2s may give psychologists and other educational profes-sionals better means to accurately assess the English skillsof ELLs. Additionally, rapid naming has been a predictorof at-risk status in L1 learners and could possibly be ap-plied to L2 learners (Wolf, 1997). Previous research withpoor comprehenders has also shown that there were no dif-ferences in the cognitive profiles of at-risk readers in L1 andL2 groups (Lesaux, Lipka, & Siegel, 2006). Lesaux et al.(2006) also found that phonological awareness as a predic-tor of reading comprehension, acted differently dependingon the language status and reading profile of students (i.e.,good comprehenders, poor comprehenders, or average com-prehenders). Further research with larger samples of ELLswill help determine the specific effects of language status,language acquisition background (sequential or simultane-ous L2) and general cognitive processing in diagnosis andassessment of learning disabilities.

Few studies to date have examined print exposure ingroups other than typically developing monolingual Englishspeakers. Though there are inherent issues in using print ex-posure measures such as potential floor and ceiling effects,and more specifically, the potential issue of cultural differ-ences in reading preferences, this study’s findings showedthat differences can emerge in print exposure in different L2groups. The issue of cultural differences in print exposure isrelevant. Before formal learning begins, children learn aboutprint through storybook reading, or shared reading with aparent or caregiver. Storybook exposure is related to the de-velopment of language skills and exposes children to written

material and vocabulary, regardless of whether it is in theirL1 or L2 (Senechal, 2006). Storybook exposure is also re-lated to the frequency with which children report reading forpleasure even after parental education and alphabet knowl-edge were controlled in an L2 group (Senechal, 2006). Thus,early exposure to print is especially important in engagingchildren in print in the later years. Differences might existamong children who are born in Canada and those who arenot, in terms of exposure to English forms of print throughexperiences with immediate and extended family, media, andcommunity resources. Children who are learning English asan L2 at home simultaneously to their L1 (i.e., the Portuguesegroup in this study) may have more exposure to English printthan those learning their L2 sequentially to their L1, typicallyonce school begins (i.e., the Spanish group in this study). Itis also possible that although there are differences in expo-sure to English forms of print before school begins, much ofthe exposure to written material may take place once children(especially those from lower SES backgrounds) begin formalschooling.

Print exposure showed differential relationships with read-ing comprehension depending upon the group examined. Inthe first Spanish cohort, print exposure predicted readingcomprehension above and beyond the effect of receptive vo-cabulary, whereas in the second Spanish cohort, print expo-sure predicted variance after word reading, or nonverbal rea-soning. In the Portuguese group, print exposure did not addunique variance to reading comprehension beyond vocabu-lary knowledge, word reading, or nonverbal reasoning andin the English group print exposure only added variance in amodel with word reading. Due to the pattern of performanceexhibited for these groups with the first Spanish cohort per-forming the most poorly on most of the tasks and the Englishgroup performing the best on most of the tasks, it may be thatprint exposure plays a different role in lower ability L2 groupsthan previously found in studies looking at L1 students frommiddle socioeconomic status backgrounds (e.g., Cunning-ham & Stanovich, 1997). The relatively low scores on printexposure for the Portuguese and English group suggest thatthese working-class children did not engage in out-of-schoolreading. For the lowest performing L2 groups, extracurric-ular reading appears to be more highly related to readingperformance. Since these students are struggling in reading,they may need more exposure to print to gain skill in compre-hension than typically achieving students. Students who aregood readers and have good basic skills (e.g., phonologicalawareness, vocabulary) may be able to make the expectedgrade-level progress in reading with little extra effort—orlittle extra print exposure at home. For students with low L2scores, reading comprehension likely requires more time andeffort both in-school and through out-of-school reading tomake progress.

Print exposure as measured through this study is only,however, one measure of home literacy. In order to fully ex-plore home literacy practices, a questionnaire asking a moreexpansive array of questions must be used. There are poten-tially many different sources of rich linguistic experiencesin the homes of children who come from different culturalbackgrounds (i.e., those who practice oral traditions). Futureresearch should examine and develop other measures of print

80 GRANT, GOTTARDO, AND GEVA: READING IN ENGLISH AS A FIRST OR SECOND LANGUAGE

exposure that might be less culturally or linguistically biased,and that can tap other sources of print and literacy-related ac-tivities that take place for children outside of school.

Stanovich’s (1986) theoretical framework of Matthew Ef-fects, which states that those rich in reading skills becomericher while those poor in reading skills typically remainpoorer is relevant to the above finding. It has been suggestedthat those with reading disabilities tend to read less, partlybecause of insufficient decoding and fluency skills (McBride-Chang, Manis, Seidenberg, Custodio, & Doi, 1993). Thesame might be applicable to young L2 learners who demon-strate lower decoding and comprehension skills due to lim-ited exposure to the English language and to English print.Increased practice in reading might lead to better readingand reading-related skills in poorer readers whether it is dueto learning difficulties or weak language skills as a result oflearning English as an L2. A sample of university students,diagnosed as having a reading disability in childhood, werefound to have comparable vocabulary and comprehensionskills in comparison to a control group without a history ofreading difficulties (Grant et al., 2007). They also had com-parable exposure to print, which might have assisted themin compensating for their previously diagnosed reading dif-ficulties.

This study highlights a pattern of differences that emergedamong these L2 groups partly due to the age at which Englishwas first acquired, specifically whether the L2 was acquiredsimultaneous or sequential to the L1, and also home supportfor school and literacy activities based on SES. The Por-tuguese group in this study, as previously mentioned, wasbilingual in the sense that they began learning English froma young age in addition to learning Portuguese concurrently.The Spanish groups, however, began learning English muchlater. A child from one of the Spanish groups was morelikely to have emigrated from a Spanish-speaking country,and was less likely to have parents who spoke English athome. Thus, the support for early English language learningand for assistance in English for completing academic workwas likely much lower for students from Spanish-speakingbackgrounds. If the language-learning environment and ageat which students began learning English was more similarbetween these two groups, many differences that emergedin this study might not have existed. It is difficult to pin-point whether some differences in performance may still beattributed to orthographic differences or other experientialdifferences in the children’s L1 given the differing language-learning backgrounds of the students.

This study revealed relatively large differences betweenstudents with the same L1 background, from the same ge-ographical area, and from the same schools. This finding,though surprising, highlights the large within-group differ-ences that can emerge for ELLs. The population of studentswho are ELL is a very heterogeneous one, due to the range ofcountries from which children emigrate within the Canadiancontext. These differences within the ELL group also high-light the importance of extracurricular reading, which mightlargely be influenced by home literacy environment.

Past research has highlighted large differences betweenchildren from different SES and ethnic backgrounds with re-spect to variables such as the number of books in the home,

and the richness of language experience children receive be-fore they even enter the school system (e.g., Bradley, Cor-wyn, McAdoo, & Garcia-Coll, 2001; Hart & Risley, 1995).One of the factors investigated in this study, print exposure,was related to the development of reading comprehension.However, this relationship was dependent upon the languageacquisition background of the group, suggesting differentinfluences of extracurricular reading on reading comprehen-sion. Print exposure may also be a factor that acts as a proxyfor children’s home environmental influences, which may beespecially relevant in L2 groups. These children often strug-gle to catch up to their L1 peers with respect to vocabularyknowledge, which is a vital component of reading compre-hension in L1 and L2 groups.

This study found differences between groups who haddiffering linguistic backgrounds, and who were classifiedas different types of bilinguals, having either acquired theirtwo languages simultaneously or sequentially. Though therewere apparent differences in this study with respect to thelanguage acquisition backgrounds of these students, there aremany other contextual factors that are undoubtedly related tothe results of this research. As research continues in the L2field, researchers should consider these other factors and whatinfluence these factors have on the successful acquisition ofreading skills in an L2. More specifically, we need to knowmore about the home environment of children in general,and of ELLs in particular, before we make conclusions aboutgroup differences and the potential causal routes of thesedifferences.

ACKNOWLEDGMENTS

The authors of this article would like to thank the teach-ers, principals, parents, and students who participated in thisstudy. They would also like to thank everyone who was in-volved in the assessment teams over the duration of the study.This research was supported by a grant to Alexandra Gottardoand Esther Geva from the Social Sciences and HumanitiesResearch Council of Canada and the Canadian Language andLiteracy Research Network. The authors would also like toacknowledge the support of Wilfrid Laurier University andthe University of Toronto.

REFERENCES

Aarts, R., & Verhoeven, L. (1999). Literacy attainment in second languagesubmersion context. Applied Psycholinguistics, 20, 377–393.

Adams, M. (1990). Beginning to read: Thinking and learning about print.Cambridge, MA: MIT Press.

Allen, L., Cipielewski, J., & Stanovich, K. (1992). Multiple indicators ofchildren’s reading habits and attitudes: Construct validity and cognitivecorrelates. Journal of Educational Psychology, 84, 489–503.

Anthony, J. L., & Lonigan, C. J. (2004). The nature of phonological aware-ness: Converging evidence from four studies of preschool and earlygrade school children. Journal of Educational Psychology, 96, 43–55.

August, D., Carlo, M., Dressler, C., & Snow, C. (2005). The critical roleof vocabulary development for English language learners. LearningDisabilities Research & Practice, 20, 50–57.

Bialystok, E. (2006). Effect of bilingualism and computer video game ex-perience on the Simon Task. Canadian Journal of Experimental Psy-chology, 60, 68–79.

LEARNING DISABILITIES RESEARCH 81

Bialystok, E. (2007). Acquisition of literacy in bilingual children: A frame-work for research. Language Learning, 57, 45–77.

Bialystok, E., Luk, G., & Kwan, E. (2005). Bilingualism, biliteracy, andlearning to read: Interactions among languages and writing systems.Scientific Studies of Reading, 9, 43–61.

Bialystok, E., Majumder, S., & Martin, M. (2003). Developing phonologicalawareness: Is there a bilingual advantage? Applied Psycholinguistics,24, 27–44.

Biemiller, A. (2003). Vocabulary: Needed if more children are to read well.Reading Psychology, 24, 315–327.

Biemiller, A., & Boote, C. (2006). An effective method for building meaningvocabulary in primary grades. Journal of Educational Psychology, 98,44–62.

Bradley, R. H., Corwyn, R. F., McAdoo, H. P., & Garcia-Coll, C. (2001).The home environment of children in the United States Part 1: Vari-ations by age, ethnicity, and poverty status. Child Development, 72,1844–1867.

Bruck, M., & Genesee, F. (1995). Phonological awareness in young secondlanguage learners. Journal of Child Language, 22, 307–324.

Bryant, P. E., MacLean, M., Bradley, L. L., & Crossland, J. (1990). Rhymeand alliteration, phoneme detection, and learning to read. Developmen-tal Psychology, 26, 429–438.

Cain, K., Oakhill, J., & Bryant, P. (2000). Phonological skills and compre-hension failure: A test of the phonological processing deficit hypothesis.Reading and Writing: An Interdisciplinary Journal, 13, 31–56.

Campbell, R., & Sais, E. (1995). Accelerated metalinguistic (phonologi-cal) awareness in bilingual children. British Journal of DevelopmentalPsychology, 13, 61–68.

Caravolas, M., & Bruck, M. (1993). The effect of oral and written languageinput on children’s phonological awareness: A cross-linguistic study.Journal of Experimental Child Psychology, 55, 1–30.

Carlisle, J., Beeman, M., Davis, L., & Spharim, G. (1999). Relationship ofmetalinguistic capabilities and reading achievement for children whoare becoming bilingual. Applied Psycholinguistics, 19, 115–131.

Carver, R. P. (1994). Percentage of unknown vocabulary words in text as afunction of the relative difficulty of the text: Implications for instruc-tion. Journal of Reading Behavior, 26, 413–437.

Carver, R. P. (1997). Reading for one second, one minute, or one year fromthe perspective of rauding theory. Scientific Studies of Reading, 1,3–43.

Castles, A., & Coltheart, M. (2004). Is there a causal link from phonologicalawareness to success in learning to read? Cognition, 91, 71–111.

Catts, H. W., Hogan, T. P., & Adlof, S. M. (2005). Developing changesin reading and reading disabilities. In H. W. Catts, & A. G. Kahmi(Eds), The connections between language and reading disabilities (pp.50–71). Mahwah, NJ: Lawrence Erlbaum Associates.

Chall, J. (1987). Two vocabularies for reading: Recognition and meaning. InM. McKeown, & M. Curtis (Eds.), The nature of vocabulary acquisition(pp. 7–17). Hillsdale, NJ: Erlbaum.

Chiappe, P., Glaeser, B., & Ferko, D. (2007). Speech perception, vocabulary,and the development of reading skills in English among Korean- andEnglish-speaking children. Journal of Educational Psychology, 99,154–166.

Chiappe, P., & Siegel, L. (1999). Phonological awareness and reading ac-quisition in English-and Punjabi-speaking Canadian children. Journalof Educational Psychology, 91, 20–28.

Chiappe, P., Siegel, L., & Gottardo, A. (2002). Reading related skills ofkindergarteners from diverse linguistic backgrounds. Applied Psy-cholinguistics, 23, 95–116.

Chiappe, P., Siegel, L., & Wade-Woolley, L. (2002). Linguistic diversityand the development of reading skills: A longitudinal study. ScientificStudies of Reading, 6, 369–400.

Cummins, J. (1991). Interdependence of first- and second-language profi-ciency in bilingual children. In E. Bialystok (Ed.), Language process-ing in bilingual children (pp. 70–89). New York: Cambridge UniversityPress.

Cunningham, A., & Stanovich, K. (1997). Early reading acquisition and itsrelation to reading experience and ability 10 years later. DevelopmentalPsychology, 33, 934–945.

Cutting, L., & Scarborough, H. (2006). Prediction of reading comprehen-sion: Relative contributions of word recognition, language proficiency,and other cognitive skills depend on how comprehension is measured.Scientific Studies of Reading, 10, 277–299.

D’Angiulli, A., Siegel, L, & Serra, E. (2001). The development of readingin English and Italian in bilingual children. Applied Psycholinguistics,22, 479–507.

Dehane, S., Dupoux, E., Mehler, J., Cohen, L., Paulesu, E., Perani, D., etal. (1997). Anatomical variability in the cortical representation of firstand second languages. Neuroreport, 8, 3809–3815.

DeHouwer, A. (2005). Early bilingual acquisition: Focus on morphosyntaxand the separate development hypothesis. In J. F. Kroll, & A. M. deGroot (Eds.), Handbook of bilingualism: Psycholinguistic approaches(pp. 30–48). New York: Oxford University Press.

Droop, M., & Verhoeven, L. (1998). Language proficiency and reading abil-ity in first- and second- language learners. Reading Research Quarterly,38, 78–103.

Dunn, L., & Dunn, L. (1997). The peabody-picture vocabulary test-III .Circle Pines, MN: American Guidance Service.

Durgunoglu, A. Y. (2002). Cross-linguistic transfer in literacy developmentand implications for language learners. Annals of Dyslexia, 52, 189–204.

Durgunoglu, A., Nagy, W., & Hancin-Bhatt, B. (1993). Cross-languagetransfer of phonological awareness. Journal of Educational Psychol-ogy, 85, 453–465.

Echols, L., West, R., Stanovich, K., & Zehr, K. (1996). Using children’s liter-acy activities to predict growth in verbal cognitive skills: A longitudinalinvestigation. Journal of Educational Psychology, 88, 296–304.

Ensminger, M. E., & Fothergill, K. (2003). A decade of measuring SES:What it tells us and where to go from here. In M. H. Bornstein, & R. H.Bradley (Eds.), Socioeconomic status, parenting and child development(pp. 13–28). Mahwah, NJ: Lawrence Erlbaum Associates.

Entwisle, D. R., & Astone, N. M. (1994). Some practical guidelines formeasuring youth’s race/ethnicity and socioeconomic status. Child De-velopment, 65, 1521–1540.

Farnia, F., & Geva, E. (in press). Cognitive correlates of vocabulary growthin English language learners. Applied Psycholinguistics.

Feldman, A., & Healy, A. (1998). Effect of first language phonological con-figuration on lexical acquisition in a second language. In A. Healy, &L. Bourne (Eds.), Foreign language learning: Psycholinguistic studieson training and retention (pp. 57–76). Mahwah, NJ: Lawrence ErlbaumAssociates.

Flege, J. E. (1992). Speech learning in a second language. In C. Ferguson, L.Menn, & C. Stoel Gammon (Eds.), Phonological development: Models,research, and implications (pp. 565–604). Timonium, MD: York.

Francis, D. J., Snow, C. E., August, D., Carlson, C. D., Miller, J., & Iglesias,A. (2006). Measures of reading comprehension: A latent variable anal-ysis of the diagnostic assessment of reading comprehension. ScientificStudies of Reading, 10, 301–322.

Gersten, R., & Baker, S. (2000). What we know about effective instructionalpractices for English language learners. Exceptional Children, 66, 454–470.

Geva, E. (1999). Issues in the development of second language reading: Im-plications for instruction and assessment. In T. Nunes (Ed.), Learningto read (pp. 343–367). New York: Springer.

Geva, E., & Siegel, L. (2000). Orthographic and cognitive factors in the con-current development of basic reading skills in two languages. Readingand Writing: An Interdisciplinary Journal, 12, 1–30.

Geva, E., & Wade-Woolley, L. (1998). Component processes in becomingEnglish-Hebrew biliterate. In A. Y. Durgunoglu & L. Verhoeven (Eds.),Literacy development in a multilingual context. Cross cultural perspec-tives (pp. 85–110). Mahwah: NJ: Lawrence Erlbaum Associates.

Geva, E., & Yaghoub-Zadeh, Z. (2006). Reading efficiency in native Englishspeaking and ESL children: The role of oral proficiency and underly-ing cognitive-linguistic processes. Scientific Studies of Reading, 10,31–57.

Geva, E., Yaghoub-Zadeh, Z., & Schuster, B. (2000). Understanding indi-vidual differences in word recognition skills of ESL children. Annalsof Dyslexia, 50, 123–153.

Gottardo, A., Collins, P., Baciu, I., & Gebotys, R. (2008). Predictors ofgrade 2 word reading and vocabulary learning from grade 1 variablesin Spanish-speaking children: Similarities and differences. LearningDisabilities Research & Practice, 23, 11–24.

Gottardo, A., & Mueller, J. (2009). Are first and second language factorsrelated in predicting L2 reading comprehension? A study of Spanishspeaking children acquiring English as a second language from first tosecond grade. Journal of Educational Psychology, 101, 330–344.

82 GRANT, GOTTARDO, AND GEVA: READING IN ENGLISH AS A FIRST OR SECOND LANGUAGE

Gough, P. B., & Tunmer, W. (1986). Decoding, reading, and reading disabil-ity. Remedial and Special Education, 7, 6–10.

Grant, A., Wilson, A., & Gottardo, A. (2007). Using measures of printto predict reading skills in college students with and without readingdisabilities. Exceptionality Education Canada, 7, 175–193.

Hart, B. A., & Risley, T. R. (1995). Meaningful differences in the every-day experience of young American children. Baltimore, MD: Paul H.Brookes Publishing.

Hoover, W., & Gough, P. (1990). The simple view of reading. Reading andWriting: An Interdisciplinary Journal, 2, 127–160.

Jean, M., & Geva, E. (2009). The development of vocabulary in Englishas a second language children and its role in word recognition ability.Applied Psycholinguistics, 30, 153–185.

Jimenez, J. E., Siegel, L., O’Shanahan, I., & Ford, L. (2009). The relativeroles of IQ and cognitive processes in reading disability. EducationalPsychology, 29, 27–43.

Jongejan, W., Verhoeven, L., & Siegel, L. S. (2007). Predictors of readingand spelling abilities in first- and second-language learners. Journal ofEducational Psychology, 99, 835–851.

Keenan, J. M., Betjemann, R. S., & Olson, R. K. (2008). Reading compre-hension tests vary in the skills they assess: Differential dependence ondecoding and oral comprehension. Scientific Studies of Reading, 12,281–300.

Kim, K. H., Relkin, N. R., Lee, K. M., & Hirsch, J. (1997). Distinct corticalareas associated with native and second languages. Nature, 388, 171–174.

Kirby, J. R., & Savage, R. S. (2008). Can the simple view deal with thecomplexities of reading? Literacy, 42, 75–82.

Lafrance, A., & Gottardo, A. (2005). A longitudinal study of phonologicalprocessing skills and reading in bilingual children. Applied Psycholin-guistics, 26, 559–758.

Lesaux, N., Lipka, O., & Siegel, L. (2006). Investigating cognitive andlinguistic abilities that influence the reading comprehension skills ofchildren from diverse linguistic backgrounds. Reading and Writing,19, 99–131.

Lesaux, N., Rupp, A. A., & Siegel, L. S. (2007). Growth in readingskills of children from diverse linguistic backgrounds: Findings froma 5-year longitudinal study. Journal of Educational Psychology, 99,821–834.

Lesaux, N., & Siegel, L. (2003). The development of reading in childrenwho speak English as a second language. Developmental Psychology,6, 1005–1019.

Limbos, M., & Geva, E. (2001). Accuracy of teacher assessments of second-language students at-risk for reading disability. Journal of LearningDisabilities, 34, 136–151.

Lindsey, K. A., Manis, F. R., & Bailey, C. L. (2003). Prediction of first-grade reading in Spanish-speaking English-language learners. Journalof Educational Psychology, 95, 482–494.

Manis, F. R., Lindsey, K. A., & Bailey, C. E. (2004). Development ofreading in grades K-2 in Spanish-speaking English-language learners.Learning Disabilities Research & Practice, 19, 214–224.

McBride-Chang, C., Manis, F., Seidenberg, M., Custodio, R., & Doi, L.(1993). Print exposure as a predictor of word reading and reading com-prehension in disabled and nondisabled readers. Journal of EducationalPsychology, 85, 230–238.

Naglieri, J. (1989). Matrix analogies test – expanded form. San Antonio,TX: Psychological Corporation.

Naglieri, J. (1996). An examination of the relationship between intelligenceand reading achievement using the MAT-SF and MAST. Journal ofPsychoeducational Assessment, 14, 65–69.

Nakamoto, J., Lindsey, K. A., & Manis, F. R. (2007). A longitudinal analysisof English language learners’ word decoding and reading comprehen-sion. Reading and Writing, 20, 691–719.

Nakamoto, J., Lindsey, K. A., & Manis, F. R. (2008). A cross-linguisticinvestigation of English language learners’ reading comprehen-sion in English and Spanish. Scientific Studies of Reading, 12,351–371.

Nation, K., & Snowling, M. (1997). Assessing reading difficulties: Thevalidity and utility of current measures of reading skill. British Journalof Educational Psychology, 67, 359–370.

Neale, M. (1989). Neale analysis of reading ability. Windsor: NFER-Nelson.Neville, H. J., & Bavelier, D. (1998). Neural organization and plasticity of

language. Current Opinion in Neurobiology, 8, 254–258.

Office of Special Education Programs. (2002). Twenty-fourth annual reportto Congress on the implementation of the Individuals with DisabilitiesAct. Washington, DC: U.S. Department of Education.

Ordonez, C., Carlo, M., Snow, C., & McLaughlin, B. (2002). Depth andbreadth of vocabulary in two-languages: Which vocabulary skills trans-fer? Journal of Educational Psychology, 94, 719–728.

Ouellette, G. P. (2006). What’s meaning got to do with it: The role ofvocabulary in word reading and reading comprehension. Journal ofEducational Psychology, 98, 554–566.

Ouellette, G., & Beers, A. (2009). A not-so-simple view of reading: How oralvocabulary and visual-word recognition complicate the story. Reading& Writing, 23, 189–208.

Pier 21. (2010). 1928 to 1971 immigration statistics. RetrievedSeptember 1, 2010, from http://www.pier21.ca/wp-content/uploads/files/research_1928to1971_immigration statistics.pdf.

Proctor, P., Carlo, M., August, D., & Snow, C. (2005). NativeSpanish-speaking children learning to read in English: Toward amodel of comprehension. Journal of Educational Psychology, 97,246–256.

Proctor, P., Carlo, M., August, D., & Snow, C. (2006). The intriguing role ofSpanish language vocabulary knowledge in predicting English readingcomprehension. Journal of Educational Psychology, 98, 159–169.

Savage, R., Pillay, V., & Melidona, S. (2007). Deconstructing rapid autom-atized naming: Component processes and the prediction of readingdifficulties. Learning and Individual Differences, 17, 129–147.

Scherag, A., Demuth, L., Rosler, F., Neville, H. J., & Roder, B. (2004). Theeffects of late acquisition of L2 and the consequences of immigrationon L1 for semantic and morpho-syntactic language aspects. Cognition,93, B97–B108.

Schwartz, M., Geva, E., Share, D., & Leikin, M. (2007). Learning to read inEnglish as a third language: The cross-linguistic transfer of phonolog-ical processing skills. Written Language and Literacy, 10, 25–52.

Senechal, M. (2006). Testing the Home Literacy Model: Parent involvementin kindergarten is differentially related to grade 4 reading comprehen-sion, fluency, spelling, and reading for pleasure. Scientific Studies ofReading, 10, 59–87.

Snowgrass, J., & Corwin, J. (1988). Pragmatics of measuring recognitionmemory: Applications to dementia and amnesia. Journal of Experi-mental Psychology: General, 117, 34–50.

Stanovich, K. E. (1986). Matthew effects in reading: Some consequences ofindividual differences in the acquisition of literacy. Reading ResearchQuarterly, 21, 360–407.

Stanovich, K. E., Cunningham, A., & West, R. (1998). Literacy experiencesand the shaping of cognition. In S. G. Paris, & H. M. Wellman (Eds.),Global prospects for education: Development, culture, and schooling(pp. 253–288). Washington, DC: American Psychological Association.

Stanovich, K. E., & West, R. (1989). Exposure to print and orthographicprocessing. Reading Research Quarterly, 24, 402–433.

Statistics Canada (2006). 2006 Census. Retreived July 28, 2010, fromhttp://www12.statcan.ca/census-recensement/2006/dp-pd/prof/92–597/index.cfm?lang = E.

Tannenbaum, K. R., Torgesen, J. K., & Wagner, R. K. (2006). Relationshipsbetween word knowledge and reading comprehension in third-gradechildren. Scientific Studies of Reading, 10, 381–398.

The Chief Public Health Officer’s Report. (2009). The Chief Public HealthOfficer’s report on the state of public health in Canada 2009: Growingup well – priorities for a healthy future. Retreived August 19, 2010,from http://publichealth.gc.ca/CPHOreport.

Townsend, D., & Collins, P. (2009). Academic vocabulary and middle schoolEnglish learners: An intervention study. Reading & Writing, 22, 993–1019.

Valdez, G., & Figueora, R. A. (1994). Bilingual and testing: A special caseof bias. Norwood, NJ: Ablex Publishing.

Verhoeven, L. (1990). Acquisition of reading in a second language. ReadingResearch Quarterly, 25, 90–114.

Verhoeven, L. (2000). Components in early second language reading andspelling. Scientific Studies of Reading, 4, 313–330

Verhoeven, L., & Van Leeuwe, J. (2008). Prediction of the developmentof reading comprehension: A longitudinal study. Applied CognitivePsychology, 22, 407–423.

Wagner, R. K., & Torgesen, J. K. (1987). The nature of phonological process-ing and its causal role in the acquisition of reading skills. PsychologicalBulletin, 2, 192–212.

LEARNING DISABILITIES RESEARCH 83

Wagner, R. K., Torgesen, J. K., & Rashotte, C. A. (1994). Developmentof reading-related phonological processing abilities: New evidence ofbidirectional causality from a latent variable longitudinal study. Devel-opmental Psychology, 30, 73–87.

Wagner, R. K., Torgesen, J. K., & Rashotte, C. A. (1999). Comprehensivetest of phonological processing. Austin, TX: Pro-ED.

Weber-Fox, C., & Neville, H. J. (1996). Maturational constraints on func-tional specializations for language processing: ERP and behaviouralevidence in bilingual speakers. Journal of Cognitive Neuroscience, 8,231–256.

West, R., Stanovich, K., & Mitchell, H. (1993). Reading in the real worldand its correlates. Reading Research Quarterly, 28, 34–50.

Whitehurst, G. J., & Lonigan, C. J. (1998). Child development and emergentliteracy. Child Development, 69, 848–872.

Wolf, M. (1997). A provisional, integrative account of phonological andnaming-speed deficits in dyslexia: Implications for diagnosis and in-tervention. In B. Blachman (Ed.), Foundations of reading acquisitionand dyslexia: Implications for early intervention (pp. 67–92). Mahwah,NJ: Lawrence Erlbaum Associates.

Woodcock, R. (1987). Woodcock Reading Mastery Test, Revised. CirclePines, MN: American Guidance Service.

Ziegler, J., & Goswami, U. (2005). Reading acquisition, developmentaldyslexia, and skilled reading across languages: A psycholinguisticgrain size theory. Psychological Bulletin, 131, 3–29.

About the Authors

Amy Grant is a PhD candidate in Developmental and Social Psychology at Wilfrid Laurier University. She earned her B.Sc.in Psychology at Mount Allison University in Sackville, NB. She is interested in reading preferences and cognitive factors thatinfluence successful reading development in typical and atypical populations.

Alexandra Gottardo is an associate professor in Developmental Psychology at Wilfrid Laurier University in Canada. Sheearned her PhD in Educational Psychology at the Ontario Institute for Studies in Education of the University of Toronto and herMasters in speech-language pathology. She is interested in the links between reading and oral language, specifically phonologyand syntax, in second language learners and students with reading disabilities.

Esther Geva, PhD University of Toronto; C. Psych. Esther Geva is Professor at the Ontario Institute for Studies in Education,University of Toronto. Her research, publications, policy work, and teaching focus on developmental issues and best practicesconcerning language and literacy skills in normally developing children and youth and children with learning disabilities comingfrom various immigrant and minority backgrounds.