Professional Documents
Culture Documents
Development of Reading in Grades K-2 in Spanish-Speaking English-Language Learners
Development of Reading in Grades K-2 in Spanish-Speaking English-Language Learners
Copyright
C 2004, The Division for Learning Disabilities of the Council for Exceptional Children
Development of English- and Spanish-reading skills was explored in a sample of 251 Spanish-
speaking English-language learners from kindergarten through Grade 2. Word identification and
reading comprehension developed at a normal rate based on monolingual norms for Spanish-
and English-speaking children, but English oral language lagged significantly behind. Four
categories of predictor variables were obtained in Spanish in kindergarten and in English in
first grade: print knowledge, expressive language (as measured by vocabulary and sentence
repetition tasks), phonological awareness, and rapid automatic naming (RAN). Longitudi-
nal regression analyses indicated a modest amount of cross-language transfer from Spanish
to English. Hierarchical regression analyses indicated that developing English-language skills
(particularly phonological awareness and RAN) mediated the contribution of Spanish-language
variables to later reading. Further analyses revealed stronger within- than cross-language as-
sociations of expressive language with later reading, suggesting that some variables function
cross-linguistically, and others within a particular language. Results suggest that some of the
cognitive factors underlying reading disabilities in monolingual children (e.g., phonological
awareness and RAN) may be important to an understanding of reading difficulties in bilingual
children.
Children who learn to read in a language different from the sion (see Adams, 1990; Perfetti, 1985; Rayner et al., 2002). In
one spoken at home face an enormous challenge. Research addition, expressive and receptive language can be expected
attention to the developmental issues facing these children to contribute increasingly to reading comprehension as chil-
has increased in recent years (August & Hakuta, 1997). The dren master printed-word recognition and move on to more
present study is concerned with English-language learners difficult texts in mid to late elementary school (Hoover &
in the United States. For these children, learning to read in Gough, 1990; Perfetti, 1985).
English is vital to their academic success. Little is known There is considerable evidence that phonological process-
about the cognitive mechanisms underlying reading acqui- ing is one of the major cognitive determinants of the devel-
sition in this group. The general thrust of the present re- opment of word-reading skills in the early phases of learning
search is to apply what is known about cognitive processes in to read (Goswami & Bryant, 1990; Share & Stanovich, 1995;
reading within a monolingual context to a biliteracy context. Wagner & Torgesen, 1987). Wagner and Torgesen (1987) dis-
The present study was conducted with a sample of 251 chil- tinguished between three types of phonological skill. Phono-
dren participating in an early-transition bilingual Spanish- logical awareness is a set of linguistic and metalinguistic
English program. The specific purpose of the study was to skills involving sensitivity to the sound structure of spoken
explore associations between theoretically important cogni- words. Phonological access in lexical memory represents the
tive skills assessed in Spanish or English in Grades K–1 with efficiency with which phonological representations of letters,
reading skill in both English and Spanish in Grade 2. Both digits, and words are accessed. Phonological processing in
cross-language and within-language relationships between verbal working memory is the retention of verbal information
early Spanish-language skills and later English- and Spanish- for a short period of time (as in digit, word, or sentence recall).
reading skills were explored. Developmental data on reading The evidence for longitudinal correlational relationships be-
in a representative sample of biliterate children will provide tween the three aspects of phonological awareness and read-
a foundation for the investigation of reading disabilities in ing is well documented in several different language com-
such a population. munities (e.g., Bruck, Genesee, & Caravolas, 1997; Comeau,
A great deal of knowledge about cognitive processes in Cormier, Grandmaison, & Lacroix, 1999; de Jong & van der
reading acquisition has accumulated in the past 25 years. A Leij, 2002; Lundberg, Olofsson, & Wall, 1980; Manis, Sei-
clear insight emerging from this research is that accurate and denberg, & Doi, 1999; Sprenger-Charolles, Siegel, & Bonnet,
efficient recognition of individual words is of crucial impor- 1998; Wagner et al., 1997; Wimmer, 1993; Wolf, Pfeil, Lotz,
tance to the development of fluency and reading comprehen- & Biddle, 1994).
Other processes beyond the phonological domain
Requests for reprints should be sent to Frank R. Manis, Department of
(although perhaps correlated with phonological skill) are
Psychology, University of Southern California, Los Angeles, CA 90089- important in reading acquisition. Scarborough (1998) ana-
1061. Electronic inquiries may be sent to manis@usc.edu. lyzed 61 prediction studies of young readers in English. The
MANIS, LINDSEY, AND BAILEY: SPANISH-SPEAKING ENGLISH-LANGUAGE LEARNERS 215
criterion for inclusion was that a minimum of one year in- It makes sense that certain kinds of knowledge obtained
tervened between initial and follow-up measures. A repeated early in reading development (such as phonological aware-
finding was that once a child had begun formal literacy train- ness) would play a fundamental role in the acquisition of
ing, the best predictor of future reading was the child’s current reading skill in both L1 and L2, and hence would show cross-
level of skill with printed letters and words. For preliterate linguistic correlations. However, vocabulary knowledge may
children, a measure of letter naming by itself accounted for be specific to languages, as the ability to retrieve word mean-
an average of 35 percent of the variance in later reading. As- ings in the course of reading a passage is dependent on famil-
sessment of the child’s functional knowledge of print (Clay, iarity with that specific language. Processing of grammatical
1979) was nearly as good a predictor, with a mean correla- structure may also show language-specific relationships to
tion of 0.53 with later reading. Of the cognitive variables, the reading, to the extent that L1 and L2 differ in their gram-
next best predictors were confrontation naming, full scale IQ, matical structure. Durgunoglu et al. (1993) did not find a
verbal IQ, broad language skill, verbal memory for sentences relationship between oral-language proficiency (as measured
or stories, and phonological awareness (r values about 0.40 by a pre-LAS test) and word- and pseudoword-reading skill in
to 0.45), followed by rapid serial naming, digit or word span, Spanish or English. However, they pointed out that a stronger
and receptive vocabulary (r values of 0.30 to 0.40). These relationship might be expected between oral language and
studies indicate that phonological awareness, lexical access, reading comprehension.
and verbal working memory are part of a group of actively de- Bialystok and Hakuta (1994) argue that bilingual speak-
veloping skills associated with early reading progress. Stud- ers have certain language representations in common (such
ies employing these predictors in combination are rare, but as metalinguistic understandings) and others that are spe-
Scanlon and Vellutino (1996) found that a combination of the cific to a language (such as idiomatic meanings). Durgunoglu
above predictors accounted for 49 percent of the variance in (2002) suggests certain reading concepts and strategies may
Grade 1 word reading. Letter naming alone accounted for the operate across languages, but there may be other con-
majority of variance. cepts and strategies that are specific to languages. This
To summarize this brief review of the reading acquisition means that developing competence in L2 should show a
literature, phonological skills are of undeniable theoretical complex relationship with prior competence in L1. One
importance in reading development. However, it is also im- such relationship would be mediational. For example, in
portant to consider other factors, most notably measures of the present case, English linguistic competence may in-
letter knowledge, print awareness, and oral-language skill in creasingly mediate the influence of Spanish competence on
order to achieve a more complete picture of the predictive English-reading ability, particularly in a typical early-exit
factors in early reading achievement. Applying this insight bilingual program like the one employed with the present
to the biliterate context, it is of both theoretical and practical sample where the major emphasis is on early transition to
importance to assess the relationship of these skills to reading L2 reading skill. For example, children with high phono-
in both L1 and L2. logical awareness in Spanish can be expected to develop
Although research on transfer of reading-related skills phonological awareness more quickly than other children in
from one language to another has not been extensive (see English, reflecting not only metalinguistic insights about on-
Cummins, 1979 for a review of earlier research), there is sets, rimes, phonemic units, and so forth, but also knowl-
growing evidence for cross-language transfer of phonolog- edge of shared spelling-sound correspondences (such as the
ical awareness (August, Calderon, & Carlo, 2001; Cisero pronunciation of most consonants). A measure of phoneme
& Royer, 1995; Comeau et al., 1999; Durgunoglu, Nagy, & awareness in English might be expected to mediate the
Hancin-Bhatt, 1993; Lindsey, Manis, & Bailey, 2003), single- cross-linguistic influence of Spanish phoneme awareness on
word reading errors and fluency (August et al., 2001; Da English reading.
Fontoura & Siegel, 1995; Geva, Wade-Woolley, & Shany, A second kind of relationship would be that develop-
1997; Kendall, Lajeunesse, Chmilar, Shapson, & Shapson, ing English competence would account for unique variance
1987; Lambert & Tucker, 1972; Lindsey et al., 2003), and in later reading, over and above that explained by cross-
reading comprehension (Escamilla, 1987; Jimenéz Gonzalez language transfer. Some part of this may be language-specific
& Haro Garcia, 1996; Lindsey et al., 2003). knowledge, such as vocabulary items, orthographic patterns,
What evidence is there that variables beyond phonological and so forth, as Dugunoglu (2002) indicates. However, an-
awareness and word recognition show cross-language trans- other part may simply reflect the further development of
fer? Analyses of the sample of children in the current study at basic skills common to literacy in both languages, such as
an earlier point in time (Grades K–1) by Lindsey et al. (2003) phonological awareness. Phonological awareness, particu-
contrasted the predictive power of Spanish-language mea- larly at the level of the phoneme, is thought to be influ-
sures of phonological awareness with other variables. Lindsey enced by the child’s developing knowledge of words (Ehri,
et al. (2003) found that letter knowledge, print concepts, sen- 1998; Goswami & Bryant, 1990; Morais, Cary, Alegria, &
tence memory, and rapid automatic naming speed showed Bertelson, 1979). As knowledge of English words and their
correlations with English-reading skill (word identification spelling-sound correspondences develops, phoneme aware-
and reading comprehension) of similar magnitude to corre- ness should develop within English. A measure of phono-
sponding correlations of phonological awareness. However, logical awareness in English might therefore be expected to
a measure of expressive vocabulary showed stronger within- account for some common and some unique variance as com-
language correlations with later reading skill, particularly for pared to an earlier measure of phonological awareness in
reading comprehension. Spanish.
216 SPECIAL SERIES: READING RISK AND INTERVENTION FOR YOUNG ELs
251, an attrition rate of 5.7 percent per year. Children who or sounded alike at the end. The words were presented in sets
were lost from the sample (N = 52) were not statistically dif- of three pictures, which were named by the experimenter,
ferent on any of the kindergarten measures from those who and the child was asked to point to the two items that rhymed
remained. (e.g., salero, basurero, sandia). Internal consistency reliabil-
The children were individually tested in a quiet room dur- ity calculated at Time 2 was 0.97.
ing regular school hours. The kindergarten test battery took
approximately 45 minutes to administer. The Grade 1 and 2
batteries took 50–90 minutes. Rapid Automatized Naming (RAN)—Objects
Picture Vocabulary (Woodcock All the letters of the Spanish alphabet were presented in ran-
& Muñoz-Sandoval, 1995) dom order, first in uppercase and then in lowercase. Spanish
has 30 letters, counting the double letters (e.g., ll ), but one
This task was a subtest of the Woodcock Language Profi- uppercase letter (RR) was omitted because it never occurs at
ciency Battery (WLPB) that measures the ability to pro- the beginning of a word in Spanish, making 59 the maximum
duce the names of pictured objects. Split-half reliability from score on this task. Children were asked to give the letter name
norms was 0.68 at age six. (Letter Names) and its sound (Letter Sounds) for all lowercase
and uppercase letters. Because the entire population of pos-
sible items was administered, it was not necessary to calcu-
Memory for Sentences (Woodcock late reliability, as reliability is only needed when an estimate
& Muñoz-Sandoval, 1995) of the true ( population) score is being made. The task was
given in kindergarten only. Only the letter-name data were
In this subtest of the WLPB, children listened to and repeated analyzed.
phrases and sentences increasing in length and grammatical
complexity. Split-half reliability from norms was 0.88 at age
six. Concepts about Print
Sound Categorization The child was first shown letters and then words increasing
in difficulty and asked to name them. There were 12 letters
This task was adapted from the CTOPP (Wagner et al., 1998). and 46 words. Split-half reliability at age six from norms was
The child was asked to say which two of three words rhymed 0.95.
218 SPECIAL SERIES: READING RISK AND INTERVENTION FOR YOUNG ELs
Spanish Passage Comprehension (Woodcock & Sound Matching ( Wagner et al., 1998) (Grade 1)
Muñoz-Sandoval, 1995) (Kindergarten, Grade 1,
and Grade 2) This task was taken from the CTOPP and had the same format
as our Spanish-language task given in kindergarten but had
The child was given a series of short printed passages in different items. Internal consistency reliability (Cronbach’s
Spanish that gradually increased in length, complexity, and alpha) from norms was 0.93 at age six.
vocabulary level. The child had to fill in a word or phrase
that was missing. Split-half reliability from norms was 0.89
Phoneme Elision ( Wagner et al., 1998) (Grade 1 Only)
at age six.
Norms for all Woodcock-Muñoz tests in Spanish were
In this task from the CTOPP, the child was presented a word
obtained from predominantly monolingual Spanish-speaking
orally and asked to repeat it. The child was then asked to say
children in the United States and Mexico as well as Spain and
the word with a target syllable/phoneme deleted. Split-half
several Latin American countries. The items were Rasch-
reliability from norms was 0.92 (age six).
calibrated and compared across languages to make the
Spanish and English versions of the same test comparable
in difficulty. Rapid Automatic Naming (RAN)—Objects and Digits
( Wagner et al., 1998) (Grade 1 Only)
Tasks Administered in English These two measures were from the CTOPP. RAN–Objects
(First and Second Grades) had the same format as the Spanish version of the task but
used the standard items in the test and required the child to
Letter Knowledge (First Grade) respond in English. In RAN–Digits, numbers were randomly
presented in four rows of nine. Children named the numbers
All the letters of the alphabet were presented in random order, as rapidly as possible in English from left to right, row by row,
first in uppercase and then in lowercase. Children provided until the last item was named. Total time and accuracy were
their names and sounds. Separate scores for letter names and recorded. Test-retest reliability for two equivalent forms of
sounds were obtained. One letter was inadvertently adminis- each task (letters or digits) obtained from the manual was 0.81
tered twice in lowercase (g) and was scored only once. The for RAN–Objects and 0.75 for RAN–Digits (time scores) at
maximum possible score was 52. Because all possible items age seven.
were administered, reliability was not calculated. Only letter-
naming scores were analyzed in the present study.
RESULTS
Letter-Word Identification ( Woodcock & Johnson, Preliminary Analyses
1989) (Grades 1 and 2)
Mean percentile scores are shown in Table 1 for tasks on
This was the English version of the test already described. which normative data was available. It is apparent that the
Split-half reliability from norms at age six was 0.96. children’s reading scores in English and Spanish and their
oral-language scores in Spanish were within the average
Passage Comprehension ( Woodcock & Johnson,
1989) (Grades 1 and 2) TABLE 1
Descriptive Information for the Normative Tests
This is the English version of the Spanish measure described
previously. Split-half reliability from norms was 0.94 at age Minimum Maximum
Mean Score Score
six.
Measure Percentile SD Achieved Achieved
1 2 3 4 5 6 7 8
range across the three years of testing, but their oral-language English. The data in Table 4 show significant and moderately
scores in English were below average, with the exception of sized correlations between Spanish measures in kindergarten
the phonological-awareness tasks, where they scored within and parallel English measures in first grade. In addition,
the average range. The findings suggest the children were Spanish measures in kindergarten correlated significantly
able to transfer phonological awareness and word-decoding with English-reading measures in second grade. More com-
skills from Spanish to English. However, they were develop- plete data on correlations in kindergarten and first grade
ing much more slowly in the domain of English vocabulary (e.g., measures of phonological awareness) were reported
and memory for sentences. Scores in English Passage Com- in Lindsey et al. (2003). To determine the relative impor-
prehension were only slightly lower overall than those for tance of each variable in cross-language associations with
Spanish Passage Comprehension, suggesting that this com- reading achievement, three sets of hierarchical regressions
prehension task measured primarily word-identification skills were conducted. The variables were entered in a fixed order.
at this point in development. In other words, lagging English- Print knowledge was entered on the first step, phonological
oral-language-comprehension skills have not begun to hold awareness on the second step, RAN on the third step, and
down English-reading comprehension yet in first and second expressive language on the last step. Entering print knowl-
grade. It is possible that they will in later grades, if the chil- edge on the first step serves to control (to some extent) for
dren do not begin to make more progress in vocabulary and previous literacy experience and training. Entering phono-
other oral-language skills. logical awareness on the second step allowed us to determine
Descriptive data (means and correlations) are reported for whether it accounted for any variance in later reading con-
the Spanish variables in kindergarten in Table 2 and for the trolling for prior reading knowledge. RAN was entered third
English (and two Spanish) variables at first grade in Table 3. to see if it accounted for a different portion of the variability
It can be seen that the variables were for the most part inter- from phonological awareness. Expressive language was en-
correlated within a time period. tered last for purposes of comparing the strength of this vari-
Because the number of variables across the three test in- able’s unique contribution to later reading within and across
tervals was fairly large, the data were reduced prior to con- languages.
ducting longitudinal analyses by combining pairs of vari-
ables (where they existed). The two tasks making up the pair
were weighted equally. This resulted in composite scores Spanish (Kindergarten) to English (Grade 2)
for Spanish Print Knowledge (Letter Naming and Con-
cepts about Print), Spanish Phonological Awareness (Sound In the first analysis, the outcome variable was English Letter-
Matching and Sound Categorization), English Phonological Word Identification at Grade 2 and the predictors were the
Awareness (Phoneme Elision and Sound Matching), Span- four kindergarten measures in Spanish. The results are shown
ish and English Expressive Language (Picture Vocabulary in the top half of Table 5. Results of the hierarchical regression
and Memory for Sentences), and English RAN Time (RAN– are shown as R2 values and change in R2 and the significance
Objects and RAN–Digits). All other variables represented in of the change at each step is indicated. The final column of the
the longitudinal analyses were represented by a single task. table shows the beta values for the final simultaneous regres-
Table 4 provides the correlations from kindergarten to Grade sion, and the significance of these beta weights. In this first
1 and 2 for each variable. analysis, the four predictor variables accounted for 26.3 per-
cent of the variance in second grade Letter-Word Identifi-
Cross-Language Predictions cation, and each variable contributed significant variance to
Letter-Word Identification at each step in the hierarchical re-
The first hypothesis was that key cognitive variables would gression, and in the final equation. However, print knowledge
show cross-language relationships between Spanish and was clearly the strongest contributor.
220 SPECIAL SERIES: READING RISK AND INTERVENTION FOR YOUNG ELs
TABLE 3
Intercorrelations in First Grade
1 2 3 4 5 6 7 8 9 10 11
1. English Letter —
Naming
2. English Sound 0.48∗∗∗ —
Matching
3. English 0.54∗∗ 0.55∗∗∗ —
Phoneme
Deletion
4. English log −0.42∗∗∗ −0.37∗∗∗ −0.42∗∗∗ —
RAN–Object
time
5. English log −0.48∗∗∗ −0.21∗∗ −0.31∗∗∗ 0.61∗∗ —
RAN–Digit
time
6. English Picture 0.52∗∗∗ 0.39∗∗∗ 0.43∗∗∗ −0.37∗∗∗ −0.29∗∗∗ —
Vocabulary
7. English 0.46∗∗∗ 0.48∗∗∗ 0.53∗∗∗ −0.32∗∗∗ −0.16∗ 0.53∗∗∗ —
Memory for
Sent.
8. English 0.63∗∗∗ 0.51∗∗∗ 0.69∗∗∗ −0.57∗∗∗ −0.54∗∗∗ 0.47∗∗∗ 0.46∗∗∗ —
Letter-Word
Ident.
English Passage 0.51∗∗∗ 0.49∗∗∗ 0.65∗∗∗ −0.44∗∗∗ −0.34∗∗∗ 0.50∗∗∗ 0.56∗∗∗ 0.69∗∗∗ —
Comp.
Spanish 0.39∗∗∗ 0.41∗∗∗ 0.50∗∗∗ −0.29∗∗∗ −0.32∗∗∗ 0.11 0.22∗∗∗ 0.64∗∗∗ 0.47∗∗∗ —
Letter-Word
Ident.
Spanish Passage 0.38∗∗∗ 0.47∗∗∗ 0.60∗∗∗ −0.33∗∗∗ −0.25∗∗∗ 0.21∗∗ 0.34∗∗∗ 0.62∗∗∗ 0.62∗∗∗ 0.77∗∗∗ —
Comp.
Means 47.3 14.5 10.1 1.7 1.5 18.5 32.6 25.7 10.0 40.2 14.5
SD 10.0 4.6 6.0 0.17 0.15 3.9 5.0 9.3 5.9 16.9 8.7
Maximum 52 20 20 2.3 2.3 58 59 58 43 58 43
possible
∗p < 0.05; ∗∗ p < 0.01; ∗∗∗ p < 0.001.
In the second analysis (top half of Table 6), the outcome First-Grade English to Second-Grade
variable was English Passage Comprehension. Print knowl- Spanish Prediction
edge, phonological awareness, and expressive language were
significant at each step of the hierarchical regression, but The third cross-linguistic analysis examined the contribu-
phonological awareness dropped out in the final equation. tion of the four English predictor variables in first grade
Altogether, the four variables accounted for 19.8 percent of to Spanish Passage Comprehension at Grade 2 (Table 7).
the variance, and print knowledge was once again the most Print knowledge and phonological awareness were signifi-
powerful predictor. cant contributors at their point of entry in the hierarchical
TABLE 4
Correlation of Four Key Composite Variables from Kindergarten (in Spanish) to First Grade and to Reading Achievement in Second Grade
Kindergarten Variable Change Final R2 —First First Grade Variable Change Final R2 —These
(Spanish) R2 in R2 Four Variables (English) R2 in R2 Four Variables
1. Print Knowledge 0.188 0.188∗∗∗ 0.261∗∗∗ 1. Print Knowledge 0.103 0.103∗∗∗ 0.087
2. Phon. Awareness 0.220 0.032∗∗ 0.143∗ 2. Phon. Awareness 0.194 0.091∗∗∗ 0.395∗∗∗
3. Log RAN–Objects 0.248 0.028∗∗ −0.170∗∗ 3. Log RAN Time 0.203 0.009 −0.112
4. Expr. Language 0.263 0.015∗ 0.139∗ 4. Expr. Language 0.208 0.005 −0.095
∗p < 0.05; ∗∗ p < 0.01; ∗∗∗ p < 0.001.
Kindergarten Variable Change Final R2 —All
(Spanish) R2 in R2 Eight Variables
Continuation of Hierarchical Regression—Adding the First-Grade single predictor. In contrast, only phonological awareness
English Predictors was significant for the English to Spanish prediction. These
1. Print Knowledge 0.188 0.188∗∗∗ 0.036 findings suggest that the amount of exposure to printed ma-
2. Phon. Awareness 0.220 0.032∗∗ 0.076 terials in Spanish is the primary predictor of later English-
3. Log RAN–Objects 0.248 0.028∗∗ −0.023 reading skills. In contrast, children’s ability to acquire English
4. Expr. Language 0.263 0.015∗ 0.064 phonological awareness seems to be most closely associated
First-Grade Variable (English) with later Spanish-reading ability, suggesting that in a lan-
5. Print Knowledge 0.408 0.145∗∗∗ 0.104 guage with very consistent spelling-sound correspondences,
6. Phon. Awareness 0.530 0.122∗∗∗ 0.390∗∗∗ phonological skills are a key predictor of reading in this
7. Log RAN Time 0.574 0.034∗∗∗ −0.258∗∗∗
population.
8. Expr. Language 0.578 0.004 0.089
∗p < 0.05; ∗∗ p < 0.01; ∗∗∗ p < 0.001.
Comparing Cross- and Within-
Language Predictions
variables in kindergarten accounted for a total of 28.5 per- Turning to English Passage Comprehension at Grade 2,
cent of the variance in later Spanish Passage Comprehension. English letter-name knowledge accounted for 7.6 percent
Expressive language was the strongest predictor in the final entered on the fifth step, followed by 11.6 percent for En-
equation. A different picture emerged for English to English glish phonological awareness, 1.5 percent for English RAN
(Grades 1 to 2). The four English predictors accounted for a time, and 1.5 percent for English expressive language (see
total of 40.1 percent of the variance in English Passage Com- Table 6, bottom half ). Altogether, 42.0 percent of the vari-
prehension. Although expressive language was significant, it ance was accounted for by the Spanish and English variables
accounted for only 2 percent unique variance. Phonological combined. Once again, none of the Spanish variables were
awareness was the strongest predictor. significant in the final regression equation, but three of the
Consistent with the second hypothesis, expressive lan- English variables were (phonological awareness, RAN, and
guage in Spanish at kindergarten was a stronger predictor expressive language). Commonality analyses indicated that
of second-grade Spanish reading comprehension (accounting the four English variables accounted for 22.2 percent unique
for 11.2 percent unique variance) than second-grade English- variance, the four Spanish variables 1.9 percent unique vari-
reading comprehension (accounting for 2 percent unique vari- ance, and common variance was 17.9 percent. Once again the
ance, from Table 6). For English expressive language at first English variables nearly completely mediated the relationship
grade, there was less of a difference (2 percent in Table 8 vs. of Spanish variables to English Passage Comprehension.
0.5 percent in Table 7). The results are consistent with the
second hypothesis, but indicate that within-language effects
are stronger for Spanish. DISCUSSION
one measure of expressive language. Our data support this their independent predictive power, both within- and cross-
hypothesis for Spanish Passage Comprehension (compare linguistically.
Tables 7 and 8). The difference favored the within-language One interpretation of the findings for the English variables
predictions for English, but was smaller in magnitude (com- is that cross-language transfer accounts for about half of the
pare Tables 6 and 8). The fact that children’s performance variability in reading achievement in L2 in second grade.
was generally low on the English expressive language tasks The other half appears to be related to developing skills in
in first grade (below the 13th percentile, see Table 1), may English. A possible practical implication of this finding is
account for their weakness as predictors of later reading that assessment strategies will need to be dynamic (Lundberg,
achievement. A limitation of the present study is that re- 2002). Children at risk for poor reading might tentatively be
ceptive language generally progresses more quickly than ex- identified based on their performance on a screening battery
pressive language in second-language learners (Bialystok, in L1. Previously reported analyses of our first-grade data
2001). It is possible that receptive language measures indicated that about 68 percent of poor readers (depending
would show a more robust relationship with later reading on the outcome measure) could be correctly classified us-
comprehension. ing the present set of kindergarten variables (Lindsey et al.,
A final set of questions concerned the contribution of 2003). However, classification accuracy might improve sig-
developing English-language skills to English reading, in- nificantly if the first-grade English measures were utilized
dependent of the correlation of earlier Spanish skills. The in combination with, or even in place of, the kindergarten
results were striking. When entered after the four Spanish Spanish measures. Our findings may help to point the way
language predictors, the four parallel measures in English toward development of screening measures for reading dis-
reduced the contribution of the Spanish variables to non- ability in biliterate children. Nevertheless, the results of our
significance. The English measures accounted for substantial study are limited to a particular time, place, and curriculum,
unique variance (22.2 percent for Passage Comprehension and a broader picture of biliterate development will be nec-
and 31.5 percent for Letter-Word Identification). In addition, essary before firm recommendations about practice can be
they shared substantial amounts of variance with the earlier made.
Spanish measures. The strongest English-language predictor To summarize, the present study replicated and ex-
across both reading measures was unquestionably phonologi- tended the findings of several previous studies focusing
cal awareness, followed by RAN. Expressive Language made on cross-linguistic prediction of reading skills in bilin-
a significant contribution only for Passage Comprehension. gual children. Print knowledge, phonological awareness,
In comparison to the English-language measures, all four and rapid naming correlated cross-linguistically with later
of the Spanish measures accounted for less than 2 percent reading achievement. Expressive language, by contrast, was
unique variance. Virtually all the variability previously ac- found to show a stronger within- than across-language
counted for by the Spanish measures, considered alone (e.g., relationship to later reading. The most novel result was
as shown in the upper part of Tables 5 and 6), was shared in the finding that English-predictor measures mediated the
common with the English measures. One cautionary note on contribution of Spanish variables to the prediction of later
interpreting these results is that the English predictors were English reading. In addition, English variables accounted for
obtained closer in time than the Spanish predictors to the unique variance beyond that attributable to cross-linguistic
English outcome measures. As the study continues, it will associations.
be possible to compare the contribution of the two sets of
predictors over an equal interval of time (K to second grade
vs. first to third grade). REFERENCES
There are several implications of the last set of results.
What is important about the present research design is that August, D. A., Calderon, M., & Carlo, M. (2001). The transfer of skills from
this is the first study (to our knowledge) to examine the per- Spanish to English. NABE News, March/April, 11–12, 42.
formance of parallel predictors of reading in Spanish and August, D. A., & Hakuta, K. (1997). Improving schooling for language-
minority children: A research agenda. Washington, DC: National
English. This means that the additional variability accounted Academy Press.
for by the English tasks is not due to the fact that they measure Bialystok, E. (2001). Bilingualism in development: Language, literacy &
different skills, but the fact that the development of English cognition. Cambridge, UK: Cambridge University Press.
literacy and language skills over a one-year period is an im- Bialystok, E., & Hakuta, K. (1994). In other words. New York: Basic Books.
Bruck, M., Genesee, F., & Caravolas, M. (1997). A cross-linguistic study
portant predictor of future success in English reading. The- of early literacy acquisition. In B. Blachman (Ed.), Dyslexia and its
oretical accounts of the development of biliteracy will need treatment (pp. 45–62). Mahwah, NJ: Erlbaum.
to factor in the extent to which developing English-language Cisero, C. A., & Royer, J. M. (1995). The development of cross-language
skills mediate the earlier influence of L1 skills as opposed to transfer of phonological awareness. Contemporary Educational Psy-
accounting for unique variability. The extent of unique and chology, 20, 275–303.
Clay, M. (1979). The early detection of reading difficulties: A diagnostic
common variance for the two languages will need to be stud- survey with recovery procedures. Portsmouth, NH: Heinemann.
ied for a larger set of measures, at different time points, and Comeau, L., Cormier, P., Grandmaison, E., & Lacroix, D. (1999). A lon-
in different samples. For example, a different set of relation- gitudinal study of phonological processing skills in children learning
ships might be predicted for early-exit bilingual programs as to read in a second language. Journal of Educational Psychology, 91,
29–43.
compared to other programs. In a dual-language program, Cummins, J. (1979). Linguistic interdependence and the educational de-
where children are theoretically encouraged to develop both velopment of bilingual children. Review of Educational Research, 49,
L1 and L2 literacy, both L1 and L2 predictors might maintain 222–251.
224 SPECIAL SERIES: READING RISK AND INTERVENTION FOR YOUNG ELs
de Jong, P. F., & van der Leij, A. (2002). Effects of phonological abilities naming and the longitudinal prediction of reading subskills in first and
and linguistic comprehension on the development of reading. Scientific second graders. Scientific Studies of Reading, 3, 129–157.
Studies of Reading, 6, 51–77. Morais, J., Cary, L., Alegria, J., & Bertelson, P. (1979). Does awareness of
Durgunoglu, A. Y. (2002). Cross-linguistic transfer in literacy development speech as a sequence of phonemes arise spontaneously? Cognition, 7,
and implications for language learners. Annals of Dyslexia, 52, 189– 323–331.
204. Neuhaus Education Center. (1980). Alphabetic phonics. Bellaire, TX:
Durgunoglu, A. Y., Nagy, W. E., & Hancin-Bhatt, B. J. (1993). Cross- Neuhaus Education Center.
language transfer of phonological awareness. Journal of Educational Perfetti, C. A. (1985). Reading ability. New York: Oxford University Press.
Psychology, 85, 453–465. Scanlon, D. M., & Vellutino, F. R. (1996). Prerequisite skills, early instruc-
Ehri, L. C. (1998). Grapheme-phoneme knowledge is essential for learning tion, and success in first-grade reading: Selected results from a lon-
to read words in English. In J. L. Metsala & L. C. Ehri (Eds.), Word gitudinal study. Mental Retardation and Developmental Disabilities
recognition in beginning literacy (pp. 3–40). Mahwah, NJ: Erlbaum. Research Review, 2, 54–63.
Escamilla, K. (1987). The relationship of native language reading achieve- Scarborough, H. S. (1998). Early identification of children at risk for reading
ment and oral English proficiency to future achievement in reading disabilities: Phonological awareness and some other promising predic-
English as a second language. Unpublished doctoral dissertation, Uni- tors. In B. Shapiro, P. Accardo, & A. Capute, A. (Eds.), Specific reading
versity of California. Los Angeles, CA. disability: A view of the spectrum (pp. 77–121). Timonium, MD: York
Geva, E., Wade-Woolley, L., & Shany, M. (1997). Development of reading Press.
efficiency in first and second language. Scientific Studies of Reading, Share, D. L., & Stanovich, K. E. (1995). Cognitive processes in early read-
1, 119–144. ing development: Accommodating individual differences into a model
Goswami, U., & Bryant, P. (1990). Phonological skills and learning to read. of acquisition. Issues in Education: Contributions From Educational
Hove, UK: Erlbaum. Psychology, 1, 1–57.
Gottardo, A., Yan, B., Siegel, L. S., & Wade-Woolley, L. (2001). Fac- Sprenger-Charolles, L., Siegel, L. S., & Bonnet, P. (1998). Reading and
tors related to English reading performance in children with Chi- spelling acquisition in French: The role of phonological mediation and
nese as a first language: More evidence of cross-language transfer of orthographic factors. Journal of Experimental Child Psychology, 68,
phonological processing. Journal of Educational Psychology, 93, 530– 134–165.
542. Wagner, R. K., & Torgesen, J. K. (1987). The nature of phonological process-
Hagan, E. C. (1997). Esperanza: A Spanish multisensory structured language ing and its causal role in the acquisition of reading skills. Psychological
program. Unpublished program materials. Brownsville, TX. Bulletin, 101, 192–212.
Hoover, W. A., & Gough, P. B. (1990). The simple view of reading. Reading Wagner, R. K., Torgesen, J. K., & Rashotte, C. A. (1998). Comprehensive
and Writing: An Interdisciplinary Journal, 2, 127–160. Test of Phonological Processing (C-TOPP). Austin, TX: Pro-Ed.
Jiménez Gonzalez, J. E., & Haro Garcı́a, C. R. (1996). The reading strate- Wagner, R. K., Torgesen, J. K., Rashotte, C. A., Hecht, S., Barker, T. A.,
gies of bilingual Latina/o students who are successful English read- & Burgess, S. R. (1997). Changing relations between phonological
ers: Opportunities and obstacles. Reading Research Quarterly, 31, 90– processing abilities and word-level reading as children develop from
112. beginning to fluent readers: A 5-year longitudinal study. Developmental
Kendall, J. R., Lajeunesse, G., Chmilar, P., Shapson, L. R., & Shapson, Psychology, 33, 468–479.
S. M. (1987). English reading skills of French immersion students in Wimmer, H. (1993). Characteristics of developmental dyslexia in a regular
kindergarten and grades 1 and 2. Reading Research Quarterly, 22, 135– writing system. Applied Psycholinguistics, 14, 1–33.
159. Wolf, M., & Bowers, P. G. (1999). The double-deficit hypothesis for the
Lambert, W. E., & Tucker, G. R. (1972). Bilingual education of children: developmental dyslexias. Journal of Educational Psychology, 91, 415–
The St. Lambert experiment. Rowley, MA: Newbury House. 438.
Lindsey, K. A., Manis, F. R., & Bailey, C. E. (2003). Prediction of first- Wolf, M., Pfeil, C., Lotz, R., & Biddle, K. (1994). Towards a more uni-
grade reading in Spanish-speaking English-language learners. Journal versal understanding of the developmental dyslexias: The contribu-
of Educational Psychology, 95, 482–494. tion of orthographic factors. In V. W. Berninger (Ed.), The varieties
Lundberg, I. (2002). Second language learning and reading with the addi- of orthographic knowledge. I.: Theoretical and developmental issues
tional load of dyslexia. Annals of Dyslexia, 52, 165–187. (pp. 137–171). Dordrecht, the Netherlands: Kluwer Academic.
Lundberg, I., Olofsson, A., & Wall, S. (1980). Reading and spelling skills in Woodcock, R. W., & Johnson, M. B. (1989). Woodcock-Johnson Psycho-
the first school years from phonemic awareness skills in kindergarten. Educational Battery Revised. Chicago, IL: Riverside.
Scandinavian Journal of Psychology, 21, 159–173. Woodcock, R. W., & Munoz-Sandoval, A. F. (1995). Woodcock Language
Manis, F. R., Seidenberg, M. S., & Doi, L. M. (1999). See Dick RAN: Rapid Proficiency Battery—Revised, Spanish form. Chicago, IL: Riverside.
Frank Manis, Professor, Department of Psychology, University of Southern California. Dr. Manis earned his Ph.D. in Child
Psychology in 1981 at the University of Minnesota. His current research interests are reading and learning disabilities, literacy
development in bilingual children, and neuropsychological bases of cognitive development.
Kim Lindsey, Research Assistant Professor, Department of Psychology, University of Southern California. Dr. Lindsey earned
her Ph.D. in Psychology in 2003 at the University of Southern California. Her current research interests are bilingualism and
biliteracy, and the manifestation of dyslexia across languages.
Caroline Bailey, Postdoctoral Research Associate, Children’s Hospital of Oakland. Dr. Bailey received her Ph.D. in 2004 at the
University of Southern California. Her current research interests are prediction, assessment, and treatment of reading disabilities
and the neuropsychological foundations of childhood cognitive disorders.