Upload
marcia-invernizzi
View
214
Download
2
Embed Size (px)
Citation preview
Diversity among Spanish-speaking English languagelearners: profiles of early literacy skills in kindergarten
Karen L. Ford • Sonia Q. Cabell •
Timothy R. Konold • Marcia Invernizzi •
Lauren B. Gartland
Published online: 12 July 2012
� Springer Science+Business Media B.V. 2012
Abstract This study explored heterogeneity in literacy development among 2,300
Hispanic children receiving English as a Second Language (ESL) services at the
start of kindergarten. Two research questions guided this work: (1) Do Spanish-
speaking English language learners receiving ESL services in the fall of kinder-
garten demonstrate homogeneous early literacy skills, or are there distinct patterns
of achievement across measures of phonological awareness, alphabet knowledge,
and orthography? and (2) if there are distinct profiles, to what extent do they predict
literacy achievement at the end of kindergarten and the beginning of first grade?
Using cluster analysis, the authors identified four distinct literacy profiles derived
from fall kindergarten measures of phonological awareness, alphabet knowledge,
and phonetic spelling. These profiles were found to be associated with literacy
outcomes in spring of kindergarten and fall of first grade. The two profiles that were
associated with greater success on later measures of concept of word in text, letter
sound knowledge, word reading, and spelling were the two that included stronger
performance on orthographic skills (i.e., alphabet knowledge and phonetic spelling).
These findings demonstrated that there is heterogeneity among Hispanic ESL stu-
dents at kindergarten entry and suggested that literacy instruction must be differ-
entiated from the very beginning in order to meet students’ individual needs. The
findings also suggested that orthographic skills should be assessed and taught early
on. While phonological awareness may be a necessary precursor to reading, pho-
nological awareness in the absence of orthographic skills may not be sufficient.
Keywords English language learners � Spanish–English bilinguals � Reading
development � Literacy profiles � Cluster analysis
K. L. Ford (&) � S. Q. Cabell � T. R. Konold � M. Invernizzi � L. B. Gartland
Curry School of Education, University of Virginia, P.O. Box 800785, Charlottesville,
VA 22908-8785, USA
e-mail: [email protected]
123
Read Writ (2013) 26:889–912
DOI 10.1007/s11145-012-9397-0
Introduction
Much attention has been given to the achievement gap that exists between Hispanic
English language learners (ELLs) and their native English-speaking peers. Hispanic
ELL students as a group have consistently scored lower than native English
speakers on the National Assessment of Educational Progress (NAEP) (National
Center for Education Statistics [NCES], 2010), and they have the highest school
dropout rate of any group in the nation (NCES, 2007). This disparity in achievement
is well documented and presents a very real concern for all who are dedicated to
helping these children succeed in school. Nevertheless, this emphasis on the
achievement gap sometimes results in Hispanic ELL students being regarded as a
homogeneous at-risk group, characterized simply by limited English proficiency. It
is likely, however, that Hispanic ELL students exhibit the same variance in content
area skills, including early literacy skills, as native English-speaking students. If
such variance exists, revealing and understanding it could lead to improved literacy
instruction and better learning outcomes for Hispanic ELL students.
There is little doubt that ELL students’ English language proficiency impacts
their reading development. Research has clearly demonstrated the relationship
between first-language (L1) oral language development and emergent literacy skills,
such as print knowledge and phonological awareness, as well as the relationship
between language skills and reading comprehension in elementary school and
beyond (Storch & Whitehurst, 2002). Moreover, delays in language development in
kindergarten have been shown to be associated with later difficulties in both word
reading and reading comprehension (Catts, Fey, Tomblin, & Zang, 2002).
Studies have also demonstrated a relationship between English oral language
development and both word reading (Gottardo, 2002; Quiroga, 2002) and reading
comprehension (Geva, 2006) among ELL students learning to read in English.
Typically, however, English proficiency has been found to account for a relatively
small percentage of the variance in word reading (Geva, 2006). As with English
L1 speakers, oral language likely makes its contribution early on as emergent
literacy skills are being developed, and thus by the beginning of formal schooling,
its effect on word reading is indirect (i.e., through such skills as phonological
awareness) rather than direct (Storch & Whitehurst, 2002). In a longitudinal study
that followed 626 English speaking children from preschool through fourth grade,
Storch and Whitehurst (2002) found that there was a strong relationship between
oral language and early literacy skills in preschool; however, in kindergarten
through second grade, the effect of oral language was found to be mediated by
skills such as phonological awareness and print knowledge. By third and fourth
grade, language came into prominence once again because of its direct relationship
with reading comprehension. These findings are consistent with research with
English language learners. Although the relationship between English language
proficiency and early word reading is usually not found to be robust, multiple
language components, including vocabulary skills, syntactical knowledge, and
listening comprehension, have all been shown to be associated with reading
comprehension among English language learners of elementary-school age
through adulthood (Geva, 2006).
890 K. L. Ford et al.
123
Metsala and Walley’s (1998) Lexical Restructuring Model suggests a unique
relationship between English oral language development and reading. According to
this model, in early childhood, when children’s oral lexicons are relatively small,
their lexical representations are largely holistic and based on units larger than
phonemes, such as syllables. As children learn more words, however, representa-
tions become more fine-grained, which allows them to distinguish between the
increasing number of words in their lexicons and to develop the phonemic
awareness necessary for beginning reading. It stands to reason that limited English
vocabulary might cause ELL students to experience an initial delay in developing
the finer lexical representations that support early reading in English. Once ELL
children start school, however, they tend to develop age-appropriate social English
relatively quickly (Cummins, 1979). Since texts for beginning readers by design
include only vocabulary that would be in the lexicon of a typical kindergartner or
first grader, it may be that many ELL students are able to develop sufficient
vocabulary to support early reading by the time they begin reading instruction. By
third grade, however, when comprehension of more sophisticated texts places
greater language demands on the reader, ELL students often begin to experience
difficulty.
Thus, although oral language skills are clearly foundational for children’s early
literacy learning, evidence suggests that language ability pays bigger dividends to
reading later in the developmental process when the characteristics of texts place
greater demands on the reader for comprehension (Lindsey, Manis, & Bailey, 2003;
Storch & Whitehurst, 2002). Moreover, although early language and literacy skills
are interrelated (Catts & Kamhi, 2005; Storch & Whitehurst, 2002), children require
different kinds of instruction to support development in these two areas. For
example, the instruction provided to support children’s language development is
very different from that designed to improve early decoding skills, and children who
may require intensive literacy support may not always have the poorest language
skills (e.g., Cabell, Justice, Konold, & McGinty, 2011; Lesaux & Geva, 2006).
Currently, however, decisions regarding early literacy instruction for ELL students
are often guided by students’ English language proficiency rather than by their
specific literacy needs. In many cases, ELL students receive most or all of their
literacy instruction in their English as a Second Language (ESL) classes. In ESL,
instruction is entirely in English. Some schools follow a ‘‘pull-out’’ model, in which
students leave the mainstream classroom to work with an ESL teacher in small groups.
Other schools use a ‘‘push-in’’ model, in which the ESL teacher joins the mainstream
classroom to support ELL students as they participate in classroom activities with
other students. In a study that examined instructional practices among 164 ESL
teachers in 16 school divisions across the Commonwealth of Virginia (Bowerman,
2007), ESL teachers reported among their daily routines activities related to guided
reading, reading comprehension, writing, spelling, reading fluency, decoding, and
read-alouds, in addition to oral language and vocabulary. Literacy instruction was
most often reported as undifferentiated and offered in whole-group formats.
Although placement in ESL services varies from state to state, it is almost
uniformly based on tests of English oral proficiency (Ragan & Lesaux, 2006). Thus,
literacy instruction may be differentiated by language proficiency level without
Early literacy ELL profiles 891
123
considering strengths and weaknesses in foundational literacy skills. This practice
could very well place these students at additional risk for reading difficulties if there
are significant differences in literacy development among Hispanic ELL students
who are classified at the same level of English language proficiency. To design
instruction that will effectively meet students’ needs, teachers must understand their
progress toward developing specific foundational literacy skills. The first step is
determining whether there is, indeed, the same level of heterogeneity among
Hispanic ELL students as there is among monolingual English speakers.
Research has demonstrated this heterogeneity both in overall literacy develop-
ment and in development of specific literacy skills in the English-speaking
population in kindergarten and first grade (Connor, Morrison, & Katch, 2004;
Kaplan & Walpole, 2005). Research has also shown differential effects for type of
instruction, based on specific student literacy profiles (Connor et al., 2004). For
example, Kaplan and Walpole (2005) employed latent transition analysis to define
five latent classes, representing various combinations of skills related to alphabet
knowledge, phonological awareness, and word reading. They were then able to
determine the likelihood that children in each of these latent classes would transition
from lower to higher developmental literacy stages during three specific time
periods (i.e., fall kindergarten to spring kindergarten, spring kindergarten to fall first
grade, and fall first grade to spring first grade). Connor et al. (2004) also
demonstrated differences in fall literacy profiles among English-speaking first
graders and further found that matching instructional activities to these profiles
resulted in greater achievement in decoding skills over the first-grade year. It is
likely that this same scenario exists among Hispanic ELL students, even at
kindergarten entry, due to such factors as English language proficiency and
preschool attendance. As discussed above, English language proficiency has a
strong relationship to emergent literacy development during the preschool years.
Research has also demonstrated differential benefits of preschool attendance in
general on early literacy skills, especially for Hispanic students (Huang, Invernizzi,
& Drake, 2012). The benefits to early literacy are most pronounced at kindergarten
entry and appear to persist through first grade. If Hispanic ELL students indeed
prove to be a heterogeneous group, even at kindergarten entry, the practice of
delivering early literacy instruction in ESL classes that are grouped solely according
to English language proficiency may be misguided.
Foundational literacy skills
Research supports the importance of phonological awareness, alphabet knowledge,
and orthographic knowledge as predictors of English literacy development, both for
native speakers (Catts, Fey, Zhang, & Tomblin, 2001; Scarborough, 1998, 2000)
and for ELL students (Chiappe, Siegel, & Wade-Woolley, 2002; Lesaux, Koda,
Siegel, & Shanahan, 2006). Phonological awareness may be defined as the ability to
attend to and reflect on the sounds of oral language, such as syllables, rhymes, and
onsets. Early phonological awareness skills include the ability to identify words that
rhyme or begin with the same sound. The latter, often referred to as phoneme
892 K. L. Ford et al.
123
awareness, a subset of phonological awareness, is particularly predictive of early
literacy development (Hulme et al., 2002). Alphabet knowledge includes both
alphabet recognition (i.e., the ability to identify by name the letters of the alphabet)
and letter-sound knowledge (i.e., the ability to identify the sounds that particular
letters make). Orthographic knowledge refers to knowledge about words in their
written form and includes the application of grapheme-phoneme correspondences as
well as an awareness of allowable and unallowable spelling patterns within a
language (e.g., qu is an acceptable letter combination in English, but qg is not).
Orthographic knowledge also includes awareness of morphemes or meaning units in
semantically related words (e.g., hid, hide, hidden). Children who have developed
skills in these three domains by the end of kindergarten, either in English or in
another alphabetic language, are likely to experience success in learning to read in
English (Durgunoglu, Nagy, & Hancin-Bhatt, 1993; Hammill, 2004; Lesaux &
Siegel, 2003; Scarborough, 1998).
Although research has demonstrated that phonological awareness, alphabet
knowledge, and orthographic knowledge are inter-related (Schatschneider, Fletcher,
Francis, Carlson, & Foorman, 2004; Warley, Landrum, Invernizzi, & Justice, 2005),
the relative importance of skills within these domains tends to change once formal
literacy instruction begins. As children start to develop a more sophisticated
understanding of written language, knowledge of the writing system quickly
replaces sound-related skills as the most accurate predictor of later reading
achievement (Hammill, 2004; Scarborough, 1998). Research has shown, for
example, that from fall to spring of the kindergarten year, print-related skills replace
phonological awareness skills as more robust predictors of reading achievement one
and two years later (Morris, Bloodgood, & Perney, 2003b; Warley et al., 2005).
Nevertheless, the combination of phonological awareness, alphabet knowledge,
and orthographic knowledge provides a more ecological and comprehensive view of
literacy achievement than any single measure alone because acquiring the
alphabetic principal requires the melding of all three. The alphabetic principle,
defined as the understanding that written words are composed of letters that
represent speech sounds and that the systematic relationship between letters and
sounds can be used to retrieve the pronunciation of printed words, is generally
accepted as a universal prerequisite for learning to read an alphabetic orthography
(Liberman, Shankweiler, & Liberman, 1989). Thus, even though individual
variables in each of these domains are accurate predictors of reading achievement,
constructs that include multiple variables considered together tend to be more
powerful in predicting reading outcomes than any single variable alone (Morris,
Bloodgood, Lomax, & Perney, 2003a; Scarborough, 1998). Schatschneider et al.
(2004), for example, demonstrated that three variables (phonological awareness,
letter sound knowledge, and rapid automatized naming of letters), measured in fall
or spring of kindergarten, together accounted for more variance in first- and second-
grade reading outcomes than any single variable alone.
In a similar study with Spanish-speaking ELL students, Manis, Lindsey, and
Bailey (2004) found that four variables (print knowledge, expressive language,
phonological awareness, and rapid naming of objects), measured in Spanish at the
end of kindergarten, together accounted for 26.3 % of the variance in second-grade
Early literacy ELL profiles 893
123
letter-word identification in English. Print knowledge, which included knowledge
about print concepts and letter names and sounds, was the strongest single predictor,
explaining 19 % of the variance in letter-word identification; however, each of the
other variables also explained a statistically significant portion of the variance. Taken
together, research on the importance of phonological awareness, alphabet knowledge,
and orthographic knowledge as predictors of successful English literacy development
for native speakers and Hispanic ELL students suggests that multiple measures are
better than single measures and that the predictive power of these measures shifts
gradually over time from phonological awareness to print-related skills.
Empirical profiling of English language learners’ early literacy skills
The existing research base on English literacy development among Spanish-
speaking ELL students has primarily focused on variable-centered approaches, such
as those described above. Variable-centered approaches to data analysis focus on
relationships among variables with the goal of predicting outcomes. Although these
approaches provide valuable information about the types of early literacy skills that
predict success for ELL students, they can potentially mask important subgroups
within the population that could be expected to be more or less at risk for reading
difficulty (Konold & Pianta, 2005). In contrast, person-centered approaches to data
analysis focus on relationships among individuals with the goal of identifying
categories in which members are similar to each other but different from others in
other categories. Person-centered approaches are also useful in analyzing longitu-
dinal data to explore heterogeneity in developmental trajectories (Muthen &
Muthen, 2000). Despite the potential usefulness of person-oriented approaches,
there are few, if any, published person-centered studies that examine profiles of
early literacy skills among Spanish-speaking ELL students entering kindergarten.
In this study, we use cluster analysis to explore the heterogeneous nature of ELL
students’ literacy development and to identify profiles of children that display
similar patterns of strengths and weaknesses across multiple constructs of early
literacy: phonological awareness, alphabet knowledge, and orthographic knowl-
edge. Unlike procedures that reflect variation in single variables, cluster analysis
accounts for the full network of relationships among multiple variables (Sternberg,
1984) and can provide greater insight into early literacy development. The use of
cluster analysis could be an important methodological and conceptual advance in
research on second-language literacy because it describes within-group variability
rather than treating ELL students as a homogeneous group.
Purpose of the present study
The purpose of this study was to examine the within-group individual differences of a
Spanish-speaking ELL population. We use cluster analysis to identify profiles across
children’s early literacy skills and explore the extent to which these profiles predict
the achievement of early literacy milestones at the end of kindergarten and the
beginning of first grade. We address two major questions: (1) Do Spanish-speaking
894 K. L. Ford et al.
123
ELL students receiving English as a Second Language (ESL) services in the fall of
kindergarten demonstrate homogeneous early literacy skills, or are there distinct
patterns of achievement across measures of phonological awareness, alphabet
knowledge, and orthography? and (2) If there are distinct profiles, to what extent do
they predict early literacy milestones at the end of kindergarten and the beginning of
first grade? Based on research on native English-speaking children’s varying abilities
at kindergarten entry (Dickinson & Neuman, 2006; Snow, Burns, & Griffin, 1998), we
hypothesized that several different core profiles would emerge, reflecting different
patterns of early literacy development. Further we hypothesized that differential
profiles would be weakly related to demographic characteristics such as age, gender,
and socio-economic status (SES) and more strongly related to preschool attendance
and English oral proficiency.
We cross-validated the measures used to establish the fall profiles by incorporating
both internal and external validity into the analytic method. Internal validity was
addressed by demonstrating the tightness of individual clusters (internal cohesion)
and the distance between clusters (external isolation). External validity was addressed
by comparing children within clusters to children in other clusters on early literacy
measures external to the derivation of the profiles. The first milestone used to validate
the fall kindergarten profiles at the end of kindergarten was PALS-K Concept of Word
in Text (COWT). COWT has been described as a ‘‘pivotal event’’ in the development
of early reading that requires the application of beginning sound awareness, automatic
recognition of letters and letter-sounds at the beginning of word boundaries, and the
ability to coordinate these skills to finger point accurately to words in a printed text
representing a memorized rhyme (Flanigan, 2007). Some experts argue that COWT is
the closest approximation to reading of any of the early literacy skills (Morris et al.,
2003). The milestone used to validate the fall kindergarten profiles in the fall of first
grade was the PALS 1–3 first-grade composite score, made up of word recognition of
beginning first-grade words, letter sounds, and a qualitative spelling assessment that
inventories specific phonics features associated with successful beginning reading,
including: beginning and ending consonants, consonant digraphs and blends, short
vowels, etc. The first-grade composite does not include contextual reading or
comprehension since many first graders are unable to read at the very beginning of the
year. The National Center for Response to Intervention rates the evidence for
classification accuracy, reliability, and validity of the PALS 1–3 as ‘‘convincing’’ with
moderately-high ratings for generalizability based on these composite scores
(http://www.rti4success.org/screeningTools/). By the end of first grade, when more
children have become readers, contextual reading and reading comprehension would
be more important.
Method
Participants
This study sampled children from a longitudinal archival database representing the
literacy scores of 98 % of the total population of children in Grades K-3 in the state
Early literacy ELL profiles 895
123
of Virginia (n = 79,169). Children were included based on the following criteria,
using data gathered during the 2007–2008 school year: (a) They were designated
Hispanic1 by their parents during kindergarten registration, (b) they received ESL
services in kindergarten,2 (c) they had no diagnosed disabilities, and (d) they had a
complete data set for fall of their kindergarten year.
Participants were 2,351 children (49 % females, 51 % males) attending
kindergarten in 436 Virginia public schools. The average age in the fall of
kindergarten was 69 months (SD = 4.6). Based on data entered into a secure,
statewide Internet database (http://pals.virginia.edu) and verified by the Virginia
Department of Education, very few children in the sample (6 %) received supple-
mental reading support (e.g., Title I services) in addition to their classroom
instruction, and 48 % of children had attended preschool the previous year (30 %
unreported).
General procedures
Children were administered the Phonological Awareness Literacy Screening (PALS)
for kindergarten (PALS-K, Invernizzi, Juel, Swank, & Meier, 2007a) in the fall and
spring of kindergarten, and PALS for grades 1–3 in the fall of first grade (PALS 1–3,
Invernizzi, Meier, & Juel, 2007a). During a four-week testing window, the
assessment was administered by classroom teachers in a familiar, quiet setting.
Total time for the assessment was approximately 30–40 min per child. PALSassessments do not require extensive training for teachers. Training videos, online
streaming footage related to each subtask, and clear, accessible teachers’ manuals
are provided, and previous studies have documented a high degree of inter rater
reliability when teachers administer the assessment (Invernizzi, Justice, Landrum, &
Booker, 2004). An online score entry reporting system allows teachers to enter
children’s scores into an encrypted, password-protected database via the Internet.
The reliability of online score entry is regularly checked against a randomly selected
sample of original, hand-scored protocols (see Invernizzi, Swank, Juel, & Meier,
2007b).
Measures
The ELL profiles were derived from five PALS-K fall measures, representing three
constructs: (1) phonological awareness, (2) alphabet knowledge, and (3) ortho-
graphic knowledge. Two additional measures were used to determine the external
and predictive validity of the profiles. The technical adequacy of both PALS
1 Hispanic is one of nine ethnicities available for parents to choose at kindergarten registration. The
Commonwealth of Virginia does not collect information on the country of origin, but recently published
U.S. census data indicate that 22 % of Hispanics in Virginia identify themselves as Mexican, 13 % as
Puerto Rican, 3 % as Cuban, and 63 % as ‘‘other Spanish/Hispanic/Latino (www.census.gov/population/
www/cen2000/briefs.html).’’2 At the time of this study, assignment to ESL services was based primarily on scores on the Stanford
Test of English Proficiency (SELP), in some cases supplemented by additional assessments chosen by
individual schools and/or school divisions.
896 K. L. Ford et al.
123
instruments has been demonstrated through piloting, field testing, and ongoing
statistical analyses of assessment scores in the PALS database. Inter-rater reliability
is reported to be stable for all tasks (r = .96–.99). Content validity was supported
through the use of advisory and review panels, and criterion-related validity was
deemed adequate through examination of both concurrent and predictive validity.
Internal consistency (Cronbach’s alpha) for Hispanic children ranged from .82 to
.89 on PALS-K tasks across 6 years (Invernizzi et al., 2007b) and from .82 to .84 on
PALS literacy composite scores across 2 years (Invernizzi, Meier, & Juel, 2007b).
Concurrent and predictive validity for PALS-K has been established using the
Stanford-9 and the Stanford-10, and the National Center on Response to
Intervention (2010) rates the disaggregated reliability, validity, and classification
accuracy for diverse populations as ‘‘convincing’’ (http://www.rti4success.org/
screeningTools).
Cluster identification variables (fall of kindergarten measures)
Phonological awareness and orthographic knowledge
The PALS-K Beginning Sound Awareness and Rhyme Awareness tasks measured
phonological awareness in English. For the Beginning Sound Awareness task,
children were presented with a target picture (e.g., fish) next to a row of three
additional pictures (e.g., mop, bell, fence). The teacher named each picture as
children compared it to the target. Children identified the picture that shared the
same beginning sound as the target (maximum score = 10). No initial sounds were
used that do not also exist in Spanish. The PALS-K Rhyme Awareness task followed
a similar administrative procedure on a different day. Children were shown a target
picture (e.g., sun) next to a row of three additional pictures (e.g., mop, run, ball) and
were asked to identify the picture that rhymed with the target (maximum
score = 10). The PALS-K Phonetic Spelling task measured children’s early
attempts at phonetic spelling, which encompasses both orthographic knowledge
and phonological awareness. For this task, children spelled five words with a
consonant–vowel-consonant (CVC) pattern. The teacher modeled writing the
sounds heard in the word mat and then asked children to spell fan, pet, rug, sit, and
mop. Points were awarded for providing phonetically acceptable representations of
each sound in a word (i.e., beginning, middle, and ending sounds), an aspect of
scoring that captures phoneme segmentation in addition to phoneme-grapheme
correspondences. Another point was given for each word spelled conventionally
(maximum score = 20), an aspect of scoring that relates to awareness of the CVC
structure of short vowels in English orthography.
Alphabet knowledge
Alphabet knowledge was measured using the Lower-Case Alphabet Recognition
task and the Letter Sounds task. For the Lower-Case Alphabet Recognition task, the
child pointed to and named each of the 26 letters of the alphabet, presented in
random order on a single 8- by 11-inch page (maximum score = 26).
Early literacy ELL profiles 897
123
For the Letter Sounds task, the child produced the sounds of 23 upper-case letters
and 3 digraphs (sh, th, ch), presented in random order on a single 8- by 11-inch page
(maximum score = 26). The child was encouraged to provide the most common
sounds for ambiguous letters (i.e., the lax, or short, sounds for vowels and the hard
sounds for the letters c and g). Three letters were not included in the task: The letter
M was omitted because it was used as an example, and X and Q were eliminated
since neither can be pronounced in isolation.
Given that the scaling of variables entering a clustering algorithm can influence
their relative contributions in the final solution, all PALS-K measures were
standardized to the same T-score metric (M = 50; SD = 10) prior to analyses.
External/predictive validity variables
Concept of word in text
The individually administered PALS-K Concept of Word in Text task requires
children to integrate their knowledge of phonological awareness, alphabet
knowledge, and orthographic knowledge as they match spoken language to the
printed words in a memorized text. Using picture support, the teacher first helped
the child memorize a short rhyme (e.g., Little Bo Peep). Once the child could
accurately recite the rhyme from memory, the teacher presented the printed version
of the rhyme and modeled finger point reading by pointing to each word while
reciting it. The teacher then engaged the child in choral and echo reading of the text
to further demonstrate the process of fingerpoint reading. In the assessment phase of
the task, the child was asked to ‘‘read’’ the rhyme while pointing to each word in the
text (the Pointing subtask). One point was awarded for each line of text in which the
child accurately pointed to each word while reciting the poem (maximum score = 5
points). Next, the teacher pointed to 10 target words within the text and asked the
child to identify them (the Word Identification in Context subtask; maximum
score = 10). Finally, the child was shown a list of 10 words from the rhyme,
presented in random order, and was asked to identify each one (the Word List
subtask; maximum score = 10). Scores from the three subtasks were then summed
to create the COWT total score (maximum total score = 25).
First-grade literacy composite
The first-grade literacy composite is made up of three tasks from PALS 1–3:
Word Recognition, Spelling, and Letter Sounds (see description above). For the
Word Recognition task, children were asked to read a list of twenty words from
a beginning-first-grade word list (maximum score = 20). Children were encour-
aged to read the words without stopping to ‘‘sound out.’’ The first-grade Spelling
task differed from the kindergarten Phonetic Spelling task in that children spelled
a list of words carefully selected to represent phonics features usually learned in
first grade (e.g., beginning and ending consonants, blends and digraphs, medial
short vowels, and the silent e long vowel marker). In a group setting, the teacher
called out each word, along with a sentence using the word in context. Children
898 K. L. Ford et al.
123
received one point for each correctly spelled phonics feature plus an additional
point if they spelled the entire word correctly (maximum score = 44). To create
the literacy composite score used in the current study, scores from the Word
Recognition in Isolation, Spelling, and Letter Sounds tasks were summed
(maximum score = 90).
Results
Table 1 presents descriptive statistics related to early literacy skills, namely
phonological awareness, alphabet, and orthographic knowledge, for the ELL sample
in the current study, as well as for a statewide sample consisting of all children who
were administered PALS tasks in Virginia during the three targeted time periods
(i.e., fall of kindergarten, spring of kindergarten, and fall of first grade). The
Spanish-speaking ELL students in our sample collectively scored lower across all
measures of early literacy when compared with the average performance reported
for children in Virginia. In the fall of their kindergarten year, the ELL students, on
average, named approximately 12 lowercase letters, produced seven letter sounds,
and demonstrated above-chance phonological awareness ability. In addition,
spelling assessment scores indicated that the average child in the sample was using
only beginning sounds to spell CVC words. However, the large standard deviations
and wide ranges across measures indicate that there was much variability in the
early literacy skills within the sample. For example, children’s phonetic spelling
ability in the average range (i.e., within 1 SD of the mean) varied from children not
representing any sounds in words to children spelling with both beginning and
ending sounds. Moreover, the actual range indicated that some children in the
sample were able to correctly spell all CVC words on the assessment in the fall of
kindergarten.
Table 1 Raw score descriptive statistics: ELL sample and statewide sample
Variable ELL samplea Statewide sampleb
M SD Range M SD Range
1. Beginning sounds 5.5 3.0 0–10 7.4 2.8 0–10
2. Rhyme awareness 5.9 2.8 0–10 8.0 2.7 0–10
3. Letter-name knowledge 12.5 9.2 0–26 19.0 7.7 0–26
4. Letter-sound knowledge 6.7 7.2 0–26 12.5 7.9 0–26
5. Phonetic spelling 3.5 4.4 0–20 7.3 6.0 0–20
6. Concept of word in text 17.8 7.1 0–25 20.4 6.0 0–25
7. Literacy composite 50.1 15.5 0–89 57.1 15.78 0–90
The statewide sample represents all Virginia children who were administered PALS tasks in each of the
three targeted time periods (i.e., fall of K, spring of K, and fall of 1st grade)a n = 2,300 for variables 1–5; 2,143 for variable 6; and 1,549 for variable 7; b N = 77,193 for variables
1–5; 77,617 for variable 6; and 60,999 for variable 7
Early literacy ELL profiles 899
123
Table 2 provides the correlations among the Fall-K, Spring-K, and Fall-1st grade
variables. The Fall-K measures used to form the clusters were moderately to highly
correlated (r = .43–.79). These measures were also moderately correlated with
Spring-K Concept of Word in Text and the Fall-1st grade literacy composite
(r = .33–.48), with the following exception. Fall-K letter-name knowledge was
highly correlated with Spring-K and Fall-1st scores (r = .54 and .52, respectively).
Cluster analysis
The clustering strategy we adopted was similar to the one used elsewhere for
identifying normative profiles (Glutting, McDermott, & Konold, 1997; Konold &
Pianta, 2005), as detailed in McDermott (1998). This procedure involved three
steps. In the first step, the total sample (n = 2,351) was randomly divided into four
approximately equal sub samples and Ward’s (1963) hierarchical-agglomerative
procedure was performed on a Euclidean distance matrix that is sensitive to level,
shape, and scatter. Decisions regarding the number of clusters to retain within each
of the four samples were based on a number of indices: pseudo-F (Calinski &
Harabasz, 1985), pseudo t2 (Duda & Hart, 1973), and R2. This step also utilized a
‘‘trim’’ procedure that removed a maximum of 2 % of the outlier cases from
consideration in the analysis (McDermott, 1998). Consequently, 51 cases were
eliminated, resulting in 2,300 children assigned to cluster membership.
Information from the clusters identified in Step 1 was pooled to form an overall
similarity matrix that was used for Step 2. Thus, Step 2 clustering began with a
proximity matrix whose diagonal elements held error sums of squares (ESS) statistic
values for respective Step 1 clusters, with off-diagonal elements corresponding to
potential ESSs for merging each pair of first-stage clusters. Ward’s method was
employed on the resulting similarity matrix from Step 1 to assess the extent to
which cluster profiles from sub samples of the data matched those found for the total
sample (i.e., replication). Each of the aforementioned statistical indices was again
considered when determining how many clusters to retain at Step 2. Steps 1 and 2
led to the identification of four clusters. All clusters yielded replication rates of
Table 2 Intercorrelations: cluster identification and external/predictive validity variables
1 2 3 4 5 6 7
1. Beginning soundsa _ .58* .53* .57* .58* .44* .40*
2. Rhyme awarenessa _ .44* .45* .46* .40* .33*
3. Letter-name knowledgea _ .79* .64* .54* .52*
4. Letter-sound
knowledgea_ .73* .48* .48*
5. Phonetic spellinga _ .42* .43*
6. Concept of word in textb _ .70*
7. Literacy compositec
a Fall kindergarten (n = 2,300); b spring kindergarten (n = 2,143); c fall 1st grade (n = 1,549)
* p = \.01
900 K. L. Ford et al.
123
100 % (see Table 3). These replication rates indicate that each of the four Step-2
clusters was also identified in each of the four subsamples of Step 1.
Group centroids from the Step-2 solution served as starting seeds for the stage
three iterative partitioning analysis conducted using K-means passes. This third step
was necessary because hierarchical-agglomerative procedures (Steps 1 and 2) do not
allow subjects to shift clusters after their original assignment, despite the fact that
they may fit better in a different profile later in the solution. By contrast, iterative
partitioning procedures allow subjects to migrate to neighboring clusters, following
Table 3 Early literacy mean score patterns (standard deviations), psychometric properties, and demo-
graphic variables of core clusters
Profile types
Profile 1 Profile 2 Profile 3 Profile 4
Kindergarten literacy variablesa
Beginning sound awareness 62 (4) 53 (7) 51 (6) 39 (6)
Rhyme awareness 60 (6) 53 (8) 52 (7) 40 (6)
Letter names 62 (3) 58 (5) 43 (5) 42 (6)
Letter sounds 66 (6) 54 (7) 43 (3) 42 (3)
Phonetic spelling 66 (9) 51 (6) 45 (4) 43 (3)
Psychometric properties
Prevalence 17.5 % 25.9 % 28.0 % 28.6 %
Independent replication across four blocks 100 % 100 % 100 % 100 %
Internal profile cohesion (H) 0.80 0.77 0.86 0.86
External isolation (Rp) -0.15 0.26 0.27 0.14
Demographic variables
Eligibility for FRPLb 46.5 % 47.5 % 46.2 % 48.2 %
Child genderb
Boys 47.6 % 51.8 % 52.8 % 51.3 %
Girls 52.4 % 48.2 % 47.2 % 48.7 %
Preschool attendanceb
Attended 72.5 % 63.3 % 43.0 % 35.8 %
Did not attend 10.7 % 14.1 % 24.8 % 31.7 %
Unknown 16.9 % 22.7 % 32.1 % 32.6 %
Child age in monthsc 71.1 (5.0) 69.7 (4.7) 68.9 (4.4) 68.1 (4.1)
English proficiency (fall 1st grade)b
Level 1 27.1 % 27.1 % 33.0 % 38.6 %
Level 2 71.1 % 71.6 % 65.6 % 61.0 %
Level 3 \1.0 % \1.0 % \1.0 % \1.0 %
Level 4 \1.0 % \1.0 % \1.0 % 0.0 %
a Means (and standard deviations) rounded to nearest whole number for ease of presentationb Free or reduced-price lunch (FRPL), child gender, preschool attendance, and English proficiency level
percentages within a given profile typec Means (and standard deviations) of child age by profile type
Early literacy ELL profiles 901
123
identification of the number of suspected clusters (Steps 1 and 2), and generally
result in tighter solutions.
Mean profile configurations for the resulting four-cluster solution are presented in
Table 3 and illustrated in Fig. 1. The four profiles represent the natural variation of
Spanish-speaking ELL students on early measures of phonological awareness,
alphabet knowledge, and phonetic spelling and are typical of what we would expect
in the beginning months of kindergarten. Table 3 also provides other psychometric
properties for each profile. The final cluster solution from Step 3 was required to
retain the dual properties of internal cohesion and external isolation (Aldenderfer &
Blashfield, 1984). Both internal cohesion and external isolation address the issue of
internal validity. Internal cohesion refers to the tightness of a cluster, or the
closeness of objects around the cluster centroid. External isolation refers to the
distance between clusters in multivariate space. Thus, subjects within a given cluster
should be similar to one another, whereas clusters composed of homogeneous
individuals should be distinct from one another. The average H coefficient (Tryon &
Bailey, 1970) across profiles satisfied a priori expectations for internal cluster
cohesion C.60 (Average H = .82), thereby providing evidence in support of
homogeneous within-cluster representation. In addition, the average Rp (Cattell
1949) across profiles also satisfied a priori expectation in support of external
isolation \.40 (Average Rp = .21). Descriptions of the four emergent clusters are
provided below.
Fig. 1 Early literacy profile configurations expressed in standardized T-scores. Profile 1: highest earlyliteracy skills (n = 403, prevalence = 17.5 %); profile 2: average phonological awareness and phoneticspelling, strength in alphabet knowledge (n = 596, prevalence = 25.9 %); profile 3: average phonologicalawareness, weakness in alphabet knowledge and phonetic spelling (n = 644, prevalence = 28 %); profile4: lowest early literacy skills (n = 657, prevalence = 28.6 %)
902 K. L. Ford et al.
123
Patterns of early literacy skills among Hispanic children receiving ESL services
Figure 1 depicts the cluster profiles, representing children’s patterns of performance
in the fall of kindergarten. The clusters were named based on their relative patterns
of early literacy strengths and weaknesses. Clusters were ordered from the highest to
lowest literacy ability.
Cluster 1: Highest early literacy skills (n = 403; prevalence = 17.5 %)
The children in Cluster 1 demonstrated strong literacy performance relative to their
peers, scoring approximately one to one and one-half standard deviations above the
sample mean across skills. On average, children in this cluster exhibited 90 %
accuracy on measures of phonological awareness. They produced 24 letter names
and 18 letter sounds in fall of kindergarten. A mean phonetic spelling score of 11
indicated that on average children spelled CVC words with acceptable phonetic
representations of beginning and ending sounds.
Cluster 2: Average phonological awareness and phonetic spelling, strengthin alphabet knowledge (n = 596; prevalence = 25.9 %)
Cluster 2 was characterized by average phonological awareness and phonetic
spelling, with alphabet knowledge approximately one-half of a standard deviation
above the mean. Children in this cluster performed with 60–70 % accuracy on
phonological awareness tasks and produced 20 letter names and 10 letter sounds in
fall of kindergarten. A mean phonetic spelling score of 4 indicated that, on average,
these children were likely spelling using beginning sounds only.
Cluster 3: Average phonological awareness, weakness in alphabet knowledgeand phonetic spelling (n = 644; prevalence = 28 %)
Children in Cluster 3 demonstrated average performance on phonological awareness
comparable to that of Cluster 2, scoring with 60-70 % accuracy. However, children
exhibited a weakness in alphabet knowledge (one-half to one standard deviation
below the mean), producing only six letter names and two letter sounds in fall of
kindergarten. Phonetic spelling was also weaker than that of children in Cluster 2,
with a mean score of 1, indicating that children did not consistently represent any
sounds while spelling words.
Cluster 4: Lowest early literacy skills (n = 657; prevalence = 28.6 %)
Children in Cluster 4 exhibited relatively low performance in the sample, with
scores on all measures approximately one standard deviation below the mean.
Children in this cluster scored, on average, below chance level on phonological
awareness tasks (i.e., 20–30 % accuracy). They produced five letter names and one
letter sound in fall of kindergarten. Phonetic spelling scores suggested that the
average child in the cluster did not represent any sounds while spelling words.
Early literacy ELL profiles 903
123
Demographic variables
Contrasts between the four clusters were evaluated in terms of differences in
compositions of school-level socioeconomic status (SES), as well as child gender,
preschool attendance, age, and English language proficiency. Children in the study
attended 436 different schools in 91 school divisions across the state of Virginia.
Table 3 reports school-level SES by cluster, calculated as the percentage of students
receiving free or reduced-price lunch (FRPL). Percentages ranged from 46.2 % (Cluster
3) to 48.2 % (Cluster 4). A one-way ANOVA showed that there were no statistically
significant differences in SES between clusters, F(3,2280) = 1.25, p = .29.
The categorical demographic variables of child gender and preschool attendance
were evaluated by contrasting the observed demographic proportions within each
cluster to what would be expected if these characteristics were proportionately
distributed across the four profiles in accordance with their representation in the
total sample. For example, the total sample following the trim procedure used in the
clustering algorithm (n = 2,300) included 51.2 % males and 48.8 % females.
Cluster 1 contained 17.5 % of the total sample (n = 403). Accordingly, we would
expect 51.2 % of the n = 403 to be male, and 48.8 % to be female. Violations of
these hypotheses were evaluated within each cluster. Type I error rates were
controlled through Bonferroni adjustments. Results indicated that gender did not
differ from expectancy across any of the four clusters (all ps [ .05/4 = .013) (see
Table 3). By contrast, the two strongest profiles (i.e., Clusters 1 and 2) contained
statistically more students who had attended preschool than would be expected, and
the two weaker profiles (i.e., Clusters 3 and 4) included statistically fewer than
expected proportions of children who attended preschool (all ps \ .05/4 = .013)
(see Table 3).
A one-way ANOVA showed statistically significant age differences across
clusters, F(3,2296) = 41.26, p \ .01, with the average age declining slightly in
order from Cluster 1 to Cluster 4 (see Table 3). Although Tukey tests revealed these
pair-wise comparisons to be statistically significant, the overall effect size for age
(x2 = .049) was relatively small and indicated that only 4.9 % of the between
cluster differences could be attributed to age.
English language proficiency was compared across clusters, based on a four-level
designation that was used statewide at the time of this study (Level 1 = lowest level
of oral and written proficiency; Level 4 = highest). Language proficiency levels
were provided by the students’ schools and were based on scores on the Stanford
English Language Proficiency Test (SELP, 2007) administered by school personnel
according to standard protocol. For the SELP assessment, p-values provided as a
measure of test difficulty ranged from .78 to .84 across the three primary forms. The
alpha for the entire primary test was .94, indicating acceptable internal consistency.
Criterion validity was established by correlating the SELP and the Stanford
Diagnostic Reading Test (SDRT) given to the same students; the correlations were
strong. The construct validity of the oral proficiency levels was established through
the use of a modified Angoff procedure (Abedi, 2007; Angoff, 1971).
Language proficiency data from fall of first grade were available for 57.5 % of
students in Cluster 1 (n = 232), 69.1 % in Cluster 2 (n = 412), 68.6 % in Cluster 3
904 K. L. Ford et al.
123
(n = 442), and 65.2 % in Cluster 4 (n = 429). In each of the four clusters, over
98 % of students with available data were designated as either Level 1 or Level 2 in
English proficiency (see Table 3). Chi-square results showed statistically significant
differences in language proficiency between clusters, v2(6) = 19.26, p \ . 01.
(Proficiency Levels 3 and 4 were combined for this analysis to avoid low cell
frequencies). The lowest performing cluster had a higher concentration of students
at the lowest English proficiency level and also had a lower concentration of
students at higher proficiency levels. Despite this difference, however, the overall
effect size for language was very small (x2 = .009), indicating that language
proficiency accounted for less than 1.0 % of the between cluster differences.
External/predictive validity of clusters
To explore the external and predictive validity of our four-cluster solution, we
compared the fall of kindergarten clusters to children’s literacy milestones at two
future time points: (a) spring of kindergarten and (b) fall of first grade.
Spring of kindergarten
Of the 2,300 children who were assigned to a cluster, scores on the Spring-K
Concept of Word in Text measure were available for 2,143 children. The
distribution of children among clusters in this subsample was similar to that in the
total sample: 18.2 % (n = 390), 26.4 % (n = 566), 27.5 % (n = 589), and 29.7 %
(n = 598), respectively. We also compared children for whom spring kindergarten
data were available (n = 2,143) with children for whom these data were unavailable
(n = 157) on the initial Fall-K clustering variables, as well as gender and age. There
were no significant differences between groups on gender, age, or Fall-K rhyme
awareness. Children with available Spring-K scores performed significantly higher
on Fall-K beginning sound awareness, letter names, letter sounds, and phonetic
spelling (all ps \ .01). However, these differences were not substantial; effect sizes
ranged from .23 to .33 (Cohen’s d).
Results of a one-way ANOVA revealed statistically significant differences among
clusters in Spring-K Concept of Word in Text (COWT), F(3,2139) = 272.95,
p \ .001 and demonstrated that 27.6 % of the variance in Spring-K literacy scores
could be attributed to between cluster differences (x2 = .276). Further, post hoc
Tukey tests indicated that all pair-wise comparisons between clusters were statistically
significant, all ps \ .001. As shown in Table 4, COWT scores at this time point
reflected the ordering in the fall clusters, with large effect sizes for differences between
children in Clusters 1 and 3, Clusters 1 and 4, and Clusters 2 and 4.
Fall of first grade
Of the 2,300 children who were assigned to a cluster in the fall of kindergarten, first
grade literacy composites were available for 1,549 children. The distribution of
these children among the four Fall-K clusters also roughly reflected the original
distribution: 17.5 % (n = 306), 25.6 % (n = 449), 28 % (n = 412), and 28.6 %
Early literacy ELL profiles 905
123
(n = 382), respectively. Children for whom data were available in fall of first grade
(n = 1,549) exhibited significantly better performance than children without
available data (n = 751) on all Fall-K clustering variables (all ps \ .001), but
effect sizes were small (d = .21–.37). There were no gender differences, and the
children with available data were slightly older (i.e., less than 1 month).
Results of a one-way ANOVA indicated statistically significant differences
among clusters in the first grade literacy composite, F (3,1545) = 156.84, p \ .001
and demonstrated that 23.2 % of the variance in first-grade literacy scores could be
attributed to between cluster differences (x2 = .232). Post hoc comparisons among
the four clusters revealed statistically significant pair-wise differences between all
groups (all ps \ .01). Effect sizes were similar to those from the spring kindergarten
time point, with differences between disparate clusters resulting in large effect sizes
(see Table 4). The difference in the literacy composite scores between the children
in Clusters 3 and 4 was less pronounced in fall of first grade (d = 0.22) than in
spring of kindergarten.
Upon closer examination of the component skills tested within the literacy
composite, significant differences between all clusters in the expected order were
evident for word recognition, spelling, and letter sounds (all ps \ .001), with the
exception of Clusters 3 and 4. Children in Cluster 3 significantly outperformed
children in Cluster 4 only in the area of spelling; there were no differences between
these clusters on word recognition and letter sounds (see Table 5).
Table 4 Spring of kindergarten and fall of first grade literacy means (standard deviations) by cluster and
results of post hoc tests
Cluster 1 Cluster 2 Cluster 3 Cluster 4 Post hoc Effect sizes (d)
M (SD) M (SD) M (SD) M (SD)
COWT (Spring-K) 22.88 20.54 16.64 12.88 1 [ 2, 3, 4 0.56, 1.13, 1.58
(3.22) (4.67) (6.62) (7.68) 2 [ 3, 4 0.68, 1.20
3 [ 4 0.52
LC (Fall-1st) 61.65 53.95 44.79 41.53 1 [ 2, 3, 4 0.64, 1.25, 1.46
(11.72) (12.22) (14.66) (15.25) 2 [ 3, 4 0.68, 0.90
3 [ 4 0.22
COWT Concept of word in text, LC literacy composite. All ‘[’ indicate statistically significant (p \ .01)
between group differences
Table 5 Fall of first grade subtest literacy means (standard deviations) by cluster
Cluster 1 Cluster 2 Cluster 3 Cluster 4
M (SD) M (SD) M (SD) M (SD)
Word recognition 17.40 (3.52) 15.21 (4.71) 11.69 (5.74) 10.81 (5.93)
Letter sounds 23.85 (2.34) 22.70 (2.83) 20.74 (4.46) 20.05 (5.01)
Spelling 20.40 (7.89) 16.01 (6.89) 12.36 (6.59) 10.67 (6.28)
All tasks are from the Phonological Awareness Literacy Screening for grades 1 through 3, maximum
scores (respectively) = 20, 26, 44
906 K. L. Ford et al.
123
Discussion
The results of this study demonstrate that Hispanic children receiving ESL services
at the start of kindergarten are not a homogeneous group when it comes to
foundational literacy skills. Further, the heterogeneity of their early literacy
development is associated with distinct patterns of strengths and weaknesses across
dimensions of phonological awareness, alphabet knowledge, and orthographic
knowledge. Children’s distinct literacy profiles in fall of kindergarten were also
found to be associated with later literacy milestones. Specifically, the two clusters
that were associated with greater success on later measures of concept of word in
text and a composite measure of letter sound knowledge, word reading, and
spelling, were the two that included stronger performance on alphabet knowledge
and phonetic spelling, the two tasks that measured orthographic knowledge.
The performance of the two middle clusters in this study clearly illustrates the
effect of higher versus lower orthographic skills. While children in both these
clusters had adequate phonological awareness skills in fall of kindergarten, they
differed significantly on orthographic skills. Children in Cluster 2, on average, had
stronger orthographic skills than children in Cluster 3, and Cluster 2 children also
performed significantly better on the spring kindergarten and fall first-grade
outcome measures. Cluster 3 children’s performance on the orthographic tasks was,
in fact, comparable to that of children in the lowest-performing cluster, Cluster 4,
and their performance on the spring kindergarten and fall first-grade measures was
also comparable to that of children in Cluster 4. Although Cluster 3 children
performed slightly higher than Cluster 4 children on the first-grade spelling task,
neither group was able to represent even half of the grade-level phonics/spelling
features tested. Thus, the orthographic skills of alphabet knowledge and phonetic
spelling were the skills most closely associated with success on the later literacy
milestones, suggesting that while phonological awareness may be a necessary
precursor to reading, phonological awareness in the absence of orthographic skills
may not be sufficient.
Our findings that Hispanic kindergarteners in ESL do not represent a
homogeneous group in terms of literacy development suggest that it may not be
appropriate to design literacy instruction based on English language proficiency
alone. As with native English speakers, best practice in early literacy instruction for
ELL students requires differentiation to meet individual students’ specific needs.
When students are grouped together for literacy instruction based on language
proficiency alone, teachers may not be aware of their students’ individual strengths
and weaknesses in literacy development. Since this study demonstrates that
foundational literacy skills are associated with important literacy milestones a year
later, it is important to assess ESL students’ literacy development early and to
differentiate instruction accordingly.
The results of the current study are consistent with previous research
demonstrating that while phonological awareness is an important precursor to
learning to read, it may not be sufficient in and of itself (Morris et al., 2003; Warley
et al., 2005), and, in fact, once children begin formal literacy instruction, skills
related to written language may become more highly related to later reading
Early literacy ELL profiles 907
123
achievement than are sound-related skills alone (Hammill, 2004; Scarborough,
1998). In the current study, it was the orthographic skills of alphabet knowledge and
phonetic spelling that were most closely associated with later literacy milestones.
Our results also support existing research showing that multiple variables,
considered together, can better predict later reading development than any single
variable, taken alone (Manis et al., 2004; Scarborough, 1998; Warley et al., 2005).
By looking at phonological awareness, alphabet knowledge, and phonetic spelling
together, we were able to identify distinct profiles related to early literacy
development in Hispanic ELL students. We then used those profiles to demonstrate
heterogeneity in literacy development among Hispanic ELL students, a finding that
is consistent with research demonstrating similar heterogeneity among English-
speaking children in kindergarten and first grade (Connor et al., 2004; Kaplan &
Walpole, 2005).
Our results also support existing research suggesting that while English language
proficiency is clearly associated with reading achievement, early literacy develop-
ment among ELL students should not be exclusively determined by English
language proficiency (Geva, 2006; Lesaux & Geva, 2006). Although Cluster 4 had a
higher concentration of students at the lowest English proficiency level and a lower
concentration of students at the higher levels, Cluster 4 students’ performance on
the literacy composite in first grade was, on average, very similar to that of students
in Cluster 3. Notably, language proficiency accounted for less than 1.0 % of the
between cluster differences. Clearly, oral language proficiency plays a role in
English literacy acquisition, but in this study, the early literacy predictor skills (e.g.,
beginning sound knowledge, letter sound knowledge, alphabet recognition, phonetic
spelling, and concept of word) made a greater contribution. This is consistent with
Storch and Whitehurst’s (2002) findings that oral language development has a
strong relationship to emergent literacy skills in preschool, but that relationship
weakens significantly in kindergarten and only becomes important again in third
grade, when reading comprehension becomes the focus of instruction. Our findings
are also consistent with Metsala and Walley’s (1998) Lexical Restructuring Model,
which suggests that once children’s lexicons grow to a sufficient size to require
attention to language at the phoneme level, they are able to use their developing
phonemic awareness to support early reading. It is likely that language proficiency
accounted for such a small percentage of the between cluster differences in the
current study because all of the children were beginning readers who were not yet
encountering the types of texts that would require highly developed language and
vocabulary to ensure comprehension.
The findings of this study could have implications for literacy instruction for
Hispanic ELL students. Because these students represent a heterogeneous group in
terms of early literacy skills, even at the start of kindergarten, our findings suggest
that literacy instruction should be differentiated from the very beginning in order to
meet their individual needs. As we have shown, phonological awareness is a
necessary but not sufficient precursor skill to reading, and some children arrive in
kindergarten already having developed adequate phonological awareness skills. The
middle two clusters, for example, were rather equal in phonological awareness skill
but differed substantially in orthographic skills (i.e., letter sound knowledge,
908 K. L. Ford et al.
123
alphabet recognition, phonetic spelling). Children who had stronger orthographic
skills in fall of kindergarten had better performance on the literacy composite in fall
of first grade. Thus, instruction that over-emphasizes phonological awareness at the
expense of orthographic skills in kindergarten may waste valuable time for many
children.
The results of this study do suggest, however, that orthographic skills need to be
assessed and taught early on. Fall kindergarten alphabet knowledge and phonetic
spelling were the factors that distinguished children who would have stronger versus
poorer performance on later outcome measures. Our finding that children in the
higher-performing clusters also had higher rates of preschool attendance may
suggest that those children had an earlier start on developing the orthographic skills
that would support their early literacy development. The fact that the children with
the poorest performance on orthographic skills in fall of kindergarten continued to
perform significantly lower on spelling, even in fall of first grade, further suggests
that ongoing instruction and assessment in written word knowledge would be
beneficial.
The goals of the current study were (a) to determine whether ELL students
receiving ESL services in the fall of kindergarten demonstrate distinct patterns of
achievement across measures of phonological awareness, alphabet knowledge, and
orthographic knowledge, and (b) to determine whether any distinct profiles that do
exist can predict early literacy milestones at the end of kindergarten and the
beginning of first grade. These goals were adequately addressed by this research;
however, certain limitations must be considered when interpreting the results. First
of all, information on child-level SES and home language use (i.e., the extent to
which English and/or Spanish were used in the home) were not available. It is
possible that variance in one or both of these factors could have affected cluster
membership and/or performance on the outcome measures. We also lacked
measures of both Spanish language proficiency and Spanish literacy development;
possible variation on these measures may have been associated with our English
literacy profiles, and controlling for these variables would have strengthened our
claim that Hispanic students are not a monolithic group. We also had limited
information about students’ English language proficiency. Having data beyond
simply their SELP levels might have allowed us to address the effect of oral
language more thoroughly. There was also no information available on the type of
instruction children received during the kindergarten year. Participants’ perfor-
mance on the milestone measures may have been affected by the literacy instruction
they received, either in their ESL classes or in their regular classrooms.
Nevertheless, the potential effects of instruction are likely negligible since our
participants attended over 400 schools that represented a range of urban, rural,
suburban, and SES demographics.
Further research is needed to determine whether the results of the current study
can be replicated with other samples of Hispanic ELL students and with native
English speakers. If future person-centered research results in the same profiles for
native English speakers as for ELL students, it might be advisable to consider
instructing ELL students with native English speakers who have similar literacy
needs rather than segregating students based on English language proficiency.
Early literacy ELL profiles 909
123
Future studies should also include Hispanic ELL students in spring of first grade and
beyond to determine whether the identified profiles are associated with contextual
reading and reading comprehension, once children are reading connected text.
Future research might also use Spanish assessment data along with English data to
explore possible cross-linguistic transfer of literacy skills.
To conclude, we have demonstrated that there is, indeed, heterogeneity in literacy
development among Hispanic ELL students receiving ESL services, even as early as
kindergarten entry. These children arrive at school with very different sets of skills
in place, and literacy instruction can only be successful if it is designed to address
each child’s individual needs. If we treat Hispanic ELL students as a homogeneous
group characterized primarily by a lack of English language proficiency, we may be
placing them at greater risk for developing reading difficulties.
References
Abedi, J. (Ed.). (2007). English language assessment in the nation: Current status and future directions.
Davis, CA: U.C. Davis School of Education.
Aldenderfer, M. S., & Blashfield, R. K. (1984). Cluster analysis. Newbury Park, CA: Sage.
Angoff, W. H. (1971). Scales, norms and equivalent scores. In R. L. Thorndike (Ed.), EducationalMeasurement (2nd ed.). Washington, D.C.: American Council on Education.
Bowerman, J. (2007). Literacy achievement for English language learners: Best practices. Unpublished
doctoral dissertation, University of Virginia, Charlottesville, VA.
Cabell, S. Q., Justice, L. M., Konold, T. R., & McGinty, A. S. (2011). Profiles of emergent literacy skills
among preschool children who are at risk for academic difficulties. Early Childhood ResearchQuarterly, 26, 1–14.
Calinski, T., & Harabasz, J. (1985). A dendrite method for cluster analysis. Communications in Statistics,3, 1–27.
Cattell, R. B. (1949). The dimensions of culture patterns by the factorization of national character.
Journal of Abnormal and Social Psychology, 44, 443–469.
Catts, H. W., Fey, M. E., Tomblin, J. B., & Zhang, X. (2002). A longitudinal investigation of reading
outcomes in children with language impairments. Journal of Speech, Language, and HearingResearch, 45, 1142–1157.
Catts, H. W., Fey, M. E., Zhang, X., & Tomblin, J. B. (2001). Estimating the risk of future reading
difficulties in kindergarten children: A research-based model and its clinical implementation.
Language, Speech, and Hearing Services in Schools, 32, 38–50.
Catts, H. W., & Kamhi, A. G. (2005). Language and reading disabilities (2nd ed.). Boston, MA: Pearson
Education.
Chiappe, P., Siegel, L. S., & Wade-Woolley, L. (2002). Linguistic diversity and the development of
reading skills: A longitudinal study. Scientific Studies of Reading, 6, 369–400.
Connor, C. M., Morrison, F. J., & Katch, L. E. (2004). Beyond the reading wars: Exploring the effect of
child-instruction interactions on growth in early reading. Scientific Studies of Reading, 8, 305–336.
Cummins, J. (1979). Cognitive/academic language proficiency, linguistic interdependence, the optimum
age question and some other matters. Working Papers on Bilingualism, 19, 121–129.
Dickinson, D. K., & Neuman, S. B. (Eds.). (2006). Handbook of early literacy research (Vol. 2). New
York, NY: Guilford Press.
Duda, R. O., & Hart, P. E. (1973). Pattern classification and scene analysis. New York: Wiley.
Durgunoglu, A. Y., Nagy, W. E., & Hancin-Bhatt, B. J. (1993). Cross-language transfer of phonological
awareness. Journal of Educational Psychology, 85, 453–465.
Flanigan, K. (2007). A concept of word in text: A pivotal event in early reading acquisition. Journal ofLiteracy Research, 39(1), 37–70.
910 K. L. Ford et al.
123
Geva, E. (2006). Second-language oral proficiency and second-language literacy. In D. August & T.
Shanahan (Eds.), Developing literacy in second language-learners: Report of the National LiteracyPanel on Language-Minority Children and Youth. Mahwah, NJ: Erlbaum.
Glutting, J. J., McDermott, P. A., & Konold, T. R. (1997). Ontology, structure and diagnostic benefits of a
normative subtest taxonomy from the WISC-III standardization sample. In D. P. Flanagan, J.
L. Genshaft, & P. L. Harrison (Eds.), Contemporary intellectual assessment: Theories, tests, andissues (pp. 349–372). New York, NY: Guilford.
Gottardo, A. (2002). The relationship between language and reading skills in bilingual Spanish-English
speakers. Topics in Language Disorders, 22(5), 46–70.
Hammill, D. D. (2004). What we know about correlates of reading. Exceptional Children, 70, 453–468.
Huang, F. L., Invernizzi, M. A., & Drake, E. A. (2012). The differentiated effects of preschool: Evidence
from Virginia. Early Childhood Research Quarterly, 27, 33–45.
Hulme, C., Hatcher, P., Nation, K., Brown, A., Adams, J., & Stuart, G. (2002). Phoneme awareness is a
better predictor of early reading skill than onset-rime awareness. Journal of Experimental ChildPsychology, 82, 2–28.
Invernizzi, M., Juel, C., Swank, L. K., & Meier, J. (2007a). Phonological awareness literacy screening(PALS): Kindergarten. Charlottesville, VA: University of Virginia.
Invernizzi, M., Juel, C., Swank, L. K., & Meier, J. (2007b). Phonological awareness literacy screening(PALS)-K technical reference. Charlottesville, VA: University of Virginia.
Invernizzi, M., Justice, L., Landrum, T., & Booker, K. (2004). Early literacy screening in kindergarten:
Widespread implementation in Virginia. Journal of Literacy Research, 36, 479–500.
Invernizzi, M., Meier, J., & Juel, C. (2007c). Phonological awareness literacy screening (PALS): Grades1–3. Charlottesville, VA: University of Virginia.
Invernizzi, M., Meier, J., & Juel, C. (2007d). Phonological awareness literacy screening (PALS) 1–3technical reference. Charlottesville, VA: University of Virginia.
Kaplan, D., & Walpole, S. (2005). A stage-sequential model of reading transitions: Evidence from the
Early Childhood Longitudinal Study. Journal of Educational Psychology, 97, 551–563.
Konold, T. R., & Pianta, B. C. (2005). Empirically-derived, person-oriented patterns of school readiness
in typically-developing children: Description and prediction to first-grade achievement. AppliedDevelopmental Science, 9, 174–187.
Lesaux, N. K., & Geva, E. (2006). Synthesis: Development of literacy in language-minority students. In
D. August & T. Shanahan (Eds.), Developing literacy in second-language learners: Report of theNational Literacy Panel on Language-Minority Children and Youth. Mahwah, NJ: Erlbaum.
Lesaux, N. K., Koda, K., Siegel, L. S., & Shanahan, T. (2006). Development of literacy. In D. August &
T. Shanahan (Eds.), Developing literacy in second language-learners: Report of the NationalLiteracy Panel on Language-Minority Children and Youth. Mahwah, NJ: Erlbaum.
Lesaux, N. K., & Siegel, L. S. (2003). The development of reading in children who speak English as a
second language. Developmental Psychology, 39, 1005–1019.
Liberman, I. Y., Shankweiler, D., & Liberman, A. M. (1989). The alphabetic principle and learning to
read. In D. Shankweiler & I. Y. Liberman (Eds.), Phonology and reading disability: Solving thereading puzzle (pp. 1–33). Ann Arbor, MI: University of Michigan Press.
Lindsey, K. A., Manis, F. R., & Bailey, C. E. (2003). Prediction of first-grade reading in Spanish-speaking
English-language learners. Journal of Educational Psychology, 95, 482–494.
Manis, F. R., Lindsey, K. A., & Bailey, C. E. (2004). Development of reading in grades K-2 in Spanish-
speaking English-language learners. Learning Disabilities Research & Practice, 19, 214–224.
McDermott, P. A. (1998). MEG: Megacluster analytic strategy for multistage hierarchical grouping with
relocations and replications. Educational and Psychological Measurement, 58, 677–686.
Metsala, J. L. & Walley, A. C. (1998). Spoken vocabulary growth and the segmental restructuring of
lexical representations: Precursors to phonemic awareness and early reading ability. In J. L. Metsala
& L. C. Ehri (Eds.), Word recognition in beginning literacy (pp. 89–120). Hillsdale, NJ: Erlbaum.
Morris, D., Bloodgood, J. W., Lomax, R. G., & Perney, J. (2003a). Developmental steps in learning to
read: A longitudinal study in kindergarten and first grade. Reading Research Quaterly, 38(3),
302–328.
Morris, D., Bloodgood, J., & Perney, J. (2003b). Kindergarten predictors of first- and second-grade
reading achievement. The Elementary School Journal, 104, 93–109.
Muthen, L. K., & Muthen, B. O. (2000). Mplus user’s guide (4th ed.). Los Angeles, CA: Muthen &
Muthen.
Early literacy ELL profiles 911
123
National Center for Education Statistics (2007). Dropout rates in the United States: 2005. Washington,
DC: U.S. Department of Education, Institute of Education Sciences.
National Center for Education Statistics. (2010). The nation’s report card: Reading 2009 (NCES2010–458). Washington, D.C.: Institute of Education Sciences, U.S. Department of Education.
National Center on Response to Intervention. (2010). Screening tools 1129 chart. Retrieved from
http://www.rti4success.org/screeningTools/
Quiroga, T., Lemos-Britton, Z., Mostafapour, E., Abbott, R. D., & Berninger, V. W. (2002). Phonological
awareness and beginning reading in Spanish-speaking ESL first graders: Research into practice.
Journal of School Psychology, 40, 85–109.
Ragan, A., & Lesaux, N. K. (2006). Federal, state, and district level English language learner program
entry and exit requirements: Effects on the education of language minority learners. EducationPolicy Analysis Archives, 14(20), 1–29.
Scarborough, H. S. (1998). Early identification of children at risk for reading disabilities: Phonological
awareness and some other promising predictors. In B. K. Shapiro, P. J. Accardo, & A. J. Capute
(Eds.), Specific reading disability: A view of the spectrum. Timonium, MD: York Press.
Scarborough, H. S. (2000, September). Predictive and causal links between language and literacydevelopment: Current knowledge and future direction. In: Paper presented at the workshop on
emergent and early literacy: Current status and research direction, Rockville, MD.
Schatschneider, C., Fletcher, J. M., Francis, D. J., Carlson, C. D., & Foorman, B. (2004). Kindergarten
prediction of reading skills: A longitudinal comparative analysis. Journal of EducationalPsychology, 96, 264–282.
Snow, C. E., Burns, M. S., & Griffin, P. (Eds.). (1998). Preventing reading difficulties in young children.
Washington, DC: National Academy Press.
Stanford English Language Proficiency Test (Virginia). (2007). San Antonio, TX: Harcourt Assessment,
Inc. and the Virginia Department of Education.
Sternberg, R. J. (1984). The Kaufman assessment battery for children: An information-processing
analysis and critique. Journal of Special Education, 18, 269–278.
Storch, S. A., & Whitehurst, G. J. (2002). Oral language and code-related precursors to reading: Evidence
from a longitudinal structural model. Developmental Psychology, 38, 934–947.
Tryon, R. C., & Bailey, D. E. (1970). Cluster analysis. New York, NY: McGraw-Hill.
Ward, J. H., Jr. (1963). Hierarchical grouping to optimize an objective function. American StatisticalAssociation Journal, 58, 236–244.
Warley, H., Landrum, T., Invernizzi, M., & Justice, L. (2005). Prediction of first grade reading
achievement: A comparison of kindergarten predictors. In B. Maloch, J. V. Hoffman, D.
L. Schallert, C. M. Fairbanks, & J. Worthy (Eds.), National reading conference yearbook (pp.
428–442). Oak Creek, WI: National Reading Conference.
912 K. L. Ford et al.
123