Journal of Research in Reading, ISSN 0141-0423
Volume 00, Issue 00, 2013, pp 1–24
DOI:10.1111/jrir.12009
English as a foreign language spelling:
comparisons between good
and poor spellers
Susie Russak
English Department and Support Center for Students with Learning Disabilities,
Beit Berl Academic College, Israel
Janina Kahn-Horwitz
Languages Department, Oranim Academic College of Education and Edmond
J. Safra Brain Research Center for the Study of Learning Disabilities,
University of Haifa, Israel
This study examined English as a foreign language (EFL) spelling development
amongst 233 fifth-grade, eighth-grade and 10th-grade Hebrew first-language speakers
to examine effects of English orthographic exposure on spelling. Good and poor
speller differences were examined regarding the acquisition of novel phonemes
(/æ/, /Λ/ and / /) and orthographic conventions (/ð/, /θ/, / / and silent ‘e’). Hebrew
measures included standardised spelling and orthographic and phonological tasks. Experimental English measures included real-word and pseudoword spellings, orthographic tasks and standardised spellings. Results showed significant differences in
spelling accuracy between good and poor spellers at all grades. Spelling accuracy
for most conventions did not improve after the eighth grade. Spellings of consonantal clusters, initial h and /ð/ differed between good and poor spellers in the fifth
grade only. Hebrew spelling was one of the strongest predictors of EFL real-word
and pseudoword spellings in both fifth and eighth grades. Implications for teaching
practice are discussed.
Practitioner points
What is already known about this topic
• The depth of the English orthography makes spelling acquisition an extended
process for first-language (L1) English speakers.
• Reading and spelling are reliant on similar cognitive and linguistic mechanisms.
• Different linguistic components impact on L1 and foreign-language spelling
accuracy at different stages.
Copyright © 2013 UKLA. Published by Blackwell Publishing, 9600 Garsington Road, Oxford OX4 2DQ,
UK and 350 Main Street, Malden, MA 02148, USA
2
RUSSAK and KAHN-HORWITZ
What this paper adds
• The gap between good and poor English as a foreign language (EFL) spellers
whose L1 is Hebrew is already evident at early stages of spelling acquisition,
and although good spellers show improvement in spelling accuracy up to the
eighth grade and poor spellers show continued improvement in spelling accuracy
across grades, the gap between good and poor spellers remains significant.
• Poor fifth-grade EFL spellers demonstrate difficulties with basic grapheme–phoneme
correspondence knowledge as well as minimal orthographic pattern knowledge.
• Proficiency with specific phonological and orthographic features of EFL
distinguishes between good and poor spellers in the fifth grade.
Implications for practice and/or policy
• Greater emphasis should be placed on explicit spelling instruction in EFL.
• As accurate spelling for consonantal clusters, initial h and /ð/ discriminate very
clearly between good and poor EFL spellers whose L1 is Hebrew in the initial
stages of EFL letter–sound knowledge acquisition, it would be beneficial to
include these features in assessment measures to identify struggling Hebrew
L1 EFL learners at early stages of EFL literacy acquisition.
Introduction
Spelling is an integral part of word knowledge. Accurate spelling in English is based on
layers of linguistic knowledge, which include phonological, orthographic, morphological
and semantic information pertaining to any particular word. Spelling provides a window
into knowledge of these various linguistic components. Just as accurate and fluent word
recognition is a necessary precursor to reading comprehension, knowledge of spelling conventions is a necessary precursor to written expression and further linguistic development.
Knowing how to spell a word accurately is dependent on the same underlying linguistic
representations involved in knowing how to read a word. First-language (L1) research
has shown that spelling and reading develop in a mutually dependent fashion for English
as an L1 (EL1) (Caravolas, Hulme & Snowling, 2001; Conrad, 2008; Ehri, 2000; Ellis
& Cataldo, 1990; Ritchey, 2008) as well as other languages (e.g., Leppanen, Niemi,
Aunola & Nurmi, 2006). Similar to EL1 reading–spelling connections, previous research
has shown a strong correlation between English as a foreign language (EFL) reading and
spelling (Fender, 2008; Kahn-Horwitz, Shimron & Sparks, 2005; Kahn-Horwitz, Sparks
& Goldstein, 2012). Thus, the development of accurate spelling in EFL is dependent on the
same underlying linguistic knowledge base as the development of EFL reading. Therefore,
acquiring EFL spelling skills should support and facilitate EFL reading skills. Whereas
there is extensive research on the development of spelling in EL1 (Arndt & Foorman,
2010; Caravolas et al., 2001; Ehri, 1991; Foorman & Petscher, 2010; Treiman & Cassar,
1997; Varnhagen, McCallum & Burstow, 1997), considerably less is known about the
linguistic processes involved in the acquisition of spelling in EFL (Figueredo, 2006).
The present study aimed to examine differences in spelling accuracy amongst good and
poor spellers at early, middle and later stages of literacy acquisition in EFL with a specific
focus on the development of phonological and orthographic representations for familiar as
well as unfamiliar (novel) patterns.
Copyright © 2013 UKLA
EFL GOOD AND POOR SPELLERS
3
EL1 and EFL spelling development
The ability to establish sound–letter correspondence knowledge is based on the ability to
perceive, discriminate and store speech sounds and then the ability to match these specific and
distinct speech sounds to the appropriate orthographic symbols. In EL1, beginning spellers
tend to rely more heavily on sound–letter correspondence knowledge, together with
phoneme awareness, in the absence of orthographic pattern knowledge (Caravolas et al.,
2001). Evidence for this has been found in EFL as well. Within the framework of a longitudinal study, Kahn-Horwitz et al. (2012) reported that knowledge of English letter names
and sounds predicted rudimentary English spelling after 1 year of EFL literacy instruction.
Accurate spelling for both EL1 and EFL is reliant on stable and reliable lexical representations. Stage models of spelling development for EL1 concur that after an initial stage of
phonetic writing where phonological knowledge is the main source of information,
beginning spellers need to learn orthographic conventions. This in turn enables them to
accurately represent more complex orthographic patterns (Davis & Bryant, 2006). This
is accomplished through repeated exposure and practice (Caravolas et al., 2001; Wang &
Geva, 2003). However, phonological and/or orthographic strategies are used to differing
degrees as a result of exposure to print (Lennox & Siegel, 1993). Ultimately, accurate
spelling relies on the use of multiple sources of phonological, orthographic and morphological linguistic information (Treiman & Bourassa, 2000).
L1 and EFL spelling connections
Theories such as the Linguistic Coding Differences Hypothesis (Sparks, 1995; Sparks &
Ganschow, 1993a, 1993b), the Central Processing Hypothesis (Geva & Siegel, 2000) and
the Interdependence Hypothesis (Cummins, 1979) provide a framework for understanding
the connection between underlying linguistic abilities measured in L1 and literacy abilities
in the target foreign language (FL). These theories suggest that linguistic abilities measured
in L1 should predict literacy outcomes in additional languages. In other words, if a learner
has strong phonological, orthographic, semantic and/or syntactic skills in L1, one would
expect to see similarly strong linguistic abilities in the target FL, whereas, if the learner has
weak phonological, orthographic, semantic and or syntactic skills in L1, these would be
expressed as similarly weak skills in the target FL. Supporting cross-linguistic evidence from
different L1 backgrounds to additional language spelling has been found for the following
languages: English (Sparks, Patton, Ganschow & Humbach, 2009), Arabic (Fender, 2008),
Spanish (Sun-Alperin & Wang, 2005), Welsh (James, Scholfield, Garrett & Griffiths, 1993)
and Chinese (Wang & Geva, 2003). In the case of EFL spelling development amongst Hebrew
L1 students, a longitudinal study conducted at the beginning, middle and end of EFL studies in
school found that Hebrew word attack and Hebrew spelling measured at the end of the fourth
grade accounted for most of the variance in explaining EFL spelling 6 years later. The word
attack measure in Hebrew represented a more phonological strategy contributing to English
spelling. EFL spelling at the end of the 12th grade was mainly explained by English spelling
measured at the end of the ninth grade, which demanded an orthographically based strategy
(Kahn-Horwitz et al., 2012). Thus, different language and literacy components, measured in
both L1 and EFL, impact on EFL spelling at different stages of development. The present study
expands on findings of this study by focusing on the development of EFL spelling abilities in
relation to specific phonological and orthographic features that are familiar as well as unfamiliar
(novel) to native Hebrew speakers from a cross-sectional perspective.
Copyright © 2013 UKLA
4
RUSSAK and KAHN-HORWITZ
The challenges of acquiring the English orthography
The English orthography is considered to be a deep orthography with a complex syllable
structure impacting on vowel pronunciation (Frost, 2005; Goswami, 1986, 1988;
Shankweiler & Fowler, 2004; Venezky, 1999) as well as inconsistent grapheme–phoneme
relationships for both reading and spelling (e.g., cut as opposed to put). As a result of the
complexity and depth of the English orthography, spelling acquisition in English is considered to be a lifelong process (Bruck, 1990; Kemp, 2009). This is in contrast with languages
that are represented by relatively more shallow orthographies, such as German and Finnish,
where accurate spelling might be expected much earlier (Leppanen et al., 2006; Ravid,
2001, 2011; Wimmer & Landerl, 1997).
English spellings, which reflect the unique orthographic profile of the language, can be
categorised into words with regular spellings (phonemically transparent words), as
opposed to words whose spelling relies on knowledge of word patterns, orthographic rules
and morphophonemic relationships, and finally opaque irregular words (sight words),
which can mostly be learned by rote (Fischer, Shankweiler & Liberman, 1985). Inherent
in this categorisation is attention to linguistic information and conventions that facilitate the
learning process, such as knowing when and why to double consonants, as in cut–cutting.
Categorising spellings according to orthographic properties can be instrumental in the construction of spelling instruction and intervention programmes. In the case of interventions,
programmes can be tailored on the basis of specific pupil needs as opposed to using generic
spelling lists. The present study aimed to examine the development of spelling accuracy in
this deep orthography amongst native Hebrew-speaking pupils.
The influence of novel linguistic and orthographic features on EFL spelling
The Hebrew orthography is a consonantal orthography that provides full phonological
information in its vowelled form, with vowels or diacritical marks appearing above, below
or within consonants (Ravid, 2011; Share & Levin, 1999; Shimron, 2006). The vowelled
form of the Hebrew orthography is thus considered to be an example of a shallow
orthography. Up to the third grade, Hebrew L1-speaking children read most texts in their
vowelled form (Share & Levin, 1999), but from the first grade, they mostly spell without
vowels (Navon & Shimron, 1984). From the third grade onwards, children are gradually
exposed to unvowelled texts, where they become dependent on the orthographic,
morphological, semantic and contextual cues to decode and spell without full phonological
information (Ravid, 2001; Share & Levin, 1999). The unvowelled version of the Hebrew
orthography would be considered to be an example of a deep orthography because
accurate decoding is dependent on contextual as well as morphological information to
compensate for the missing phonological information (Benuck & Peverly, 2004). The
Hebrew-speaking participants in the current study were all literate in both vowelled and
unvowelled Hebrew.
Based on differences between Hebrew and English, different linguistic and orthographic
English features are expected to be novel for Hebrew L1 speakers. Novel features in the
target language of additional language learners have been found to be particularly
challenging as a result of lack of experience with these features (Raynolds & Uhry,
2010; Russak & Saiegh-Haddad, 2011; Wang & Geva, 2003). Evidence supporting this
comes from numerous studies that examined the effect of novelty on additional language
and literacy acquisition. Kahn-Horwitz, Schwartz and Share (2011) found that children
Copyright © 2013 UKLA
EFL GOOD AND POOR SPELLERS
5
who were literate in Russian and Hebrew had greater success in spelling and decoding the
English short vowels as opposed to children who were literate in Hebrew only. The
Russian orthography fully represents vowel sounds as opposed to the unvowelled Hebrew
orthography, where vowels are not represented (Share & Levin, 1999). The Phonological
and Orthographic Proximity Hypothesis was proposed to explain and predict the impact of
L1 phonological and orthographic features on the literacy development of an additional
language (Kahn-Horwitz et al., 2011). This hypothesis is compatible with the Linguistic
Affiliation Constraint Hypothesis (Russak & Saiegh-Haddad, 2011; Saiegh-Haddad,
2003), which postulates that phonemes from the second language (L2) that do not exist
in the L1 phonemic inventory will be more difficult to process than phonemes that are
familiar from the L1 inventory. This difficulty was illustrated in a study of the impact of
novel phonemes on the phonological awareness abilities of a group of native Hebrewspeaking adults learning EFL with and without reading disabilities. Words that included
novel EFL phonemes were found to be more difficult to segment for both populations
tested (Russak & Saiegh-Haddad, 2011).
Evidence from other languages and literacies contributes further to our understanding of
how L1 phonology and orthography were found to have an impact on English as an
L2 (ESL) spelling. Wang and Geva (2003) found that Chinese L1 speakers at the start of
ESL literacy acquisition struggled with accurate spellings of particular English phonemes
that do not exist in the Chinese phonemic inventory. Similarly, Uhry and Raynolds (2011)
found a strong impact of Spanish (L1) phonemes on the English (L2) spellings of Spanishspeaking (L1) first graders. In the present cross-sectional study, the development of novel
linguistic and orthographic features was examined amongst Hebrew L1 spellers.
Differences between good and poor spellers in L1 and EFL
Similar to the interest in the differences between good and poor readers (Elbro, 1998;
Siegel, Share & Geva, 1995; Spear-Swerling & Sternberg, 1996), there has been a noticeable interest in the differences between good and poor spellers (Arndt & Foorman, 2010;
Cassar & Treiman, 2004; Van Bon & Uit De Haag, 1997). Differences between good
and poor spellers have been examined with regard to specific error types. A recent study
by Arndt and Foorman (2010) comparing typically developing with poor second-grade
spellers found that the poor spellers graphemically represented all phonemes of the dictated
words and showed quantitatively but not qualitatively more errors than their typically
developing peers. However, difficulty in spelling short and long vowel sounds was
observed amongst these poor spellers, whereas this was not observed amongst typically
developing children. This finding has been confirmed for spellers with dyslexia as well
(Bernstein, 2008; Moats, 1996; Sawyer, Wade & Kim, 1999). Errors with the accurate representation of vowel sounds in spelling seem to be the result of inaccurate phonological
representations, as evidenced in vowel substitutions of other vowels with many shared
articulatory features (Bernstein, 2008; Moats, 1996; Sawyer et al., 1999).
Further fine-grained analyses of spelling errors indicate that there is greater heterogeneity in the spelling errors of spellers with dyslexia (Moats, 1996), who tend to omit consonants when spelling consonantal clusters (Bruck & Treiman, 1990; Cassar, Treiman,
Moats, Pollo & Kessler, 2005; Moats, 1996; Van Bon & Uit De Haag, 1997). Regarding
spellings of orthographic conventions, it was found that children with dyslexia have more
difficulties with accurate spelling of the silent ‘e’, although these differences were not
Copyright © 2013 UKLA
6
RUSSAK and KAHN-HORWITZ
necessarily reflected in real-word spellings (Bourassa & Treiman, 2003). In addition,
certain high-frequency sight words such as ‘want’ and ‘because’ seem to be problematic
for spellers with dyslexia (Moats, 1996). In the present study, we were interested in examining the differences between good and poor EFL spellers in both L1 (Hebrew) and EFL
with specific attention to the effect of novel phonological and orthographic features.
Whereas the aforementioned studies focused on dyslexic spellers, the current study used
a method of fine-grained analysis to better understand the differences between good and
poor spellers.
EFL literacy instruction in Israel
EFL literacy instruction in Israel usually begins in the third grade after one or more
years of exposure to the oral language. EFL instruction takes place on an average of
3–4 hours per week. Spelling instruction begins with the introduction of letter names,
followed by sound–letter matching instruction for most of the letters. However, after this
initial stage of letter–sound matching, it is common for EFL teachers in Israel to move
directly to a more whole language-based approach (for a more complete description of this
approach, see Masterson & Apel, 2010; Schlagel, 2001). At this point, target content words
are studied for the purposes of spelling and vocabulary acquisition. Within this framework,
children are required to memorise lists of words for weekly dictation tests. There is little
explicit instruction regarding orthographic conventions, and this is reflected in the fact that
spelling receives little emphasis in official Israel Ministry of Education guidelines for
teachers regarding prefoundation literacy instruction (State of Israel Ministry of Education
English Inspectorate, 2009). Although spelling is mentioned as one of the 10 ‘musts’ for
teaching in prefoundation literacy instruction and spelling is stated as important for
supporting reading, there is no mention of explicit instruction of English orthographic
conventions. Teachers are told to ‘engage pupils in extensive spelling practice’ (State of
Israel Ministry of Education English Inspectorate, 2009, p. 9); however, they are not given
any guidelines as to what kind of practice will result in optimal outcomes. Whereas in
many EFL classrooms in Israel routine spelling tests are given throughout elementary
school, this practice tapers off during junior high school. By the time an EFL learner
reaches high school in Israel, there is virtually no emphasis placed on spelling instruction.
However, some high school EFL textbooks draw attention to morphological components
of the English language to facilitate vocabulary acquisition for reading comprehension
for the matriculation exam (Loney, 2005). In these same matriculation examinations,
spelling errors are hardly penalised. This perhaps best expresses the lack of recognition
of spelling being the link between the respective linguistic elements involved in knowing
a word.
The present study
In the present study, we were interested in reaching a deeper understanding of the development of accurate EFL spelling amongst L1 Hebrew-speaking good and poor spellers.
Within the framework of the current study, we were interested in examining the following
two questions:
1. What are the differences in accuracy levels between good and poor EFL spellers in the
fifth, eighth and 10th grades specifically regarding phonological and orthographic
Copyright © 2013 UKLA
EFL GOOD AND POOR SPELLERS
7
representations for novel as well as non-novel phonemes and orthographic patterns?
From previous findings, it was expected that good spellers would spell with higher
levels of accuracy than poor spellers. We hypothesised that increased exposure and
practice across grades would lead to more accurate phonological and orthographic representations for both novel and non-novel phonological and orthographic patterns.
However, we expected to see more difficulties with accurate representations for novel
as opposed to non-novel features.
2. To what extent does performance on orthographic and phonological choice tasks in L1
Hebrew and orthographic choice and sensitivity tasks in EFL account for EFL spelling?
From previous studies, we expected to find a stronger impact for phonological
measures on EFL spelling earlier on and a stronger impact for orthographic measures
later on.
Method
Participants
A total of 233 children participated in the study. They had all begun English studies in the
third grade and began formal literacy instruction in the fourth grade. Fifth, eighth and 10th
graders were targeted to represent the beginning, middle and advanced stages of literacy
acquisition. There were 81 fifth graders from two elementary schools, 72 eighth graders
from one regional middle school and 80 10th graders from one regional high school in
the centre of Israel. The schools are all considered middle–high socioeconomic background.
This is based on an index determined by the Israeli Central Bureau of Statistics (Central
Bureau of Statistics, 2011, http://www.cbs.gov.il/locals.htm). This index is based on demographic information, parent education, standard of living, employment and unemployment
data and percentage of the population in that particular area receiving government subsidies.
Initially, children were divided into good and poor spellers on the basis of their Hebrew L1
spelling scores. However, all tenth graders obtained ceiling results, and so the median score
of the standardised Wide-Range Achievement Test, a test that has been standardised on the
English L1 population, was used to divide the grades into good versus poor spellers. The
poor EFL speller group included children whose scores were !0.5 standard deviations
below the median score or lower, and the good speller group included children whose scores
were 0.5 standard deviations or more above the median score. The groups were divided this
way to clearly differentiate between good spellers, as opposed to spellers who are potentially at risk for EFL spelling difficulties and poor spellers, while removing a group of
spellers from the middle range around the median. There were 22 good spellers and 24 poor
spellers in the fifth grade, 19 good spellers and 20 poor spellers in the eighth grade and 25
good spellers and 29 poor spellers in the 10th grade.
Baseline measures: L1 Hebrew
Phonological choice (Shatil & Nevo, 2007). This timed-standardised Hebrew task involves
circling food items from a list of 104 misspelled words that can be phonologically decoded
.
but are orthographically incorrect, for example, lachmanya (Hebrew for bread roll)
The maximum correct score on this task is 29 for the fifth grade and 31 for the eighth and
Copyright © 2013 UKLA
8
RUSSAK and KAHN-HORWITZ
10th grades. The K-R20 internal consistency is .87 for the fifth grade, .87 for the seventh
grade and .89 for the ninth grade.
Orthographic choice (Shatil & Nevo, 2007). This timed-standardised Hebrew task
involves circling 42 animal words out of a list of 149 correctly spelled words for the
fifth grade and 43 animal words out of a list of 149 correctly spelled words for the
. The maximum
eighth and 10th grades, for example, kof (Hebrew for monkey)
correct score is 42 for the fifth grade and 43 for the eighth and 10th grades. The K-R20
internal consistency is .94 for the fifth grade, .93 for the seventh grade and .94 for the
ninth grade.
Spelling dictation (Shatil & Nevo, 2007). This standardised Hebrew task is a dictation of a
text comprising 35 words for the fifth grade and 42 words for the eighth and ninth grades.
One form was given to the fifth grade and another form to the eighth and 10th graders. The
K-R20 internal consistency is .95 for the fifth grade, .96 for the seventh grade and .96 for
the ninth grade. The Cronbach alpha value is .91 for the fifth grade, .89 for the eighth grade
and .95 for the 10th grade.
Baseline measures: EFL
Orthographic sensitivity. This task, which is an adaptation of a similar task used by
Cunningham, Perry and Stanovich (2001) as well as Stanovich and Siegel (1994), examines sensitivity towards orthographic conventions appearing in pseudowords in English.
The orthographic conventions include both legal letter sequences and legal letter positions
within words. Fifteen pseudoword pairs are presented, and children are required to circle
the word that looks most like a real English word in each pair (for a full description of
this task, see Russak, 2007). The task includes homophone–pseudoword pairs such as
sckap/skap, gake/gayk and qoast/quost. The Cronbach alpha is .63. The maximum score
for this task is 15.
Orthographic choice. This task, which is an adaptation of a similar task used by Cunningham
et al. (2001) as well as Stanovich and Siegel (1994), tests knowledge of orthographic patterns
in real words. Children are presented with 20 pairs of words where one of the pair is correctly
spelled and the other is not; however, all word pairs are homophonous, for example, place
versus plase, thum versus thumb and more versus moar. All of the real-word targets appear
in the list of the 2,000 most frequent words compiled from spoken and written text corpora
(http://www.lextutor.ca/vp/). Children are required to circle the correctly spelled word in
the pair (for a full description of this task, see Russak, 2007). The Cronbach alpha is .83.
The maximum score for this task is 20.
Spelling dictation. Words were dictated from the standardised Wide-Range Achievement
Test 3 (Wilkinson, 1993). However, raw scores were used as this test has not been
standardised for EFL populations.
Copyright © 2013 UKLA
EFL GOOD AND POOR SPELLERS
9
Experimental EFL spelling measures
Word spelling task. To create this task, we targeted phonological and orthographic properties
that are novel to the Hebrew L1 speaker. Five categories included the various items that made
up this task. First, three novel phonemes, /æ/, /Λ/ and / /, were included in eight words (e.g.,
hand, but and thin). Second, three orthographic conventions that comprise both novel phonological and orthographic properties, /ð/, /θ/ and / /, appeared in 11 words (e.g., this, with and
out). Phonologically, these phonemes do not exist in Hebrew, and orthographically, there are
no digraphs in Hebrew. The third category consisted of words that are orthographically novel
for Hebrew L1 speakers: the silent ‘e’ (split digraph) appeared in four words (e.g., time). The
fourth category included six sight words, which include letters that do not follow grapheme–
phoneme correspondences (e.g., people, want and does). The fifth category consisted of two
non-novel phonemes, /e/ and initial /h/, and consonant clusters (which exist in Hebrew in a
restricted manner and to a lesser degree), which appeared in 19 words (e.g., help and best).
Target words were chosen from the first 400 words in Brown Corpus (2009) and included
24 one-syllable words and four two-syllable words of which six were sight words. A total
of 28 words were dictated to the children (Appendix). Each word was read out loud. It was
then presented within the context of a meaningful sentence, and finally, the tester read it out
loud again. Children then wrote the target words. For example, the word ‘just’ includes two
targets, the Λ and the consonant cluster -st. The Cronbach alpha is .95.
Pseudoword spelling task. The same phonological and orthographic properties novel to the
Hebrew L1 speaker and targeted in the preceding experimental word spelling task were used
for creating pseudowords. Pseudowords were created by substituting one nontarget grapheme
in the beginning, middle or end of each real word with an orthographically legal grapheme
(e.g., nith instead of with). An attempt was made to limit the number of orthographically
ambiguous options that could be accepted. No substituted pseudowords were created for
the sight words, resulting in a list of 22 pseudowords (Appendix). Children were told that they
would hear a list of words with no meaning in English and they were to write them. Each
target pseudoword was read out twice, and children wrote what they heard. For example,
the pseudoword ‘thip’ includes two targets, the / / and the digraph /θ/. All orthographically
legal options for representing a phoneme were accepted. For example, for the target ‘mome’,
the alternate spelling ‘moam’ was also accepted. The Cronbach alpha is .88.
Procedure
Permission was obtained from the Israeli Ministry of Education Chief Scientist’s Office to
conduct the study. Towards the end of the school year, the first author conducted group
testing in the fifth-grade, eighth-grade and 10th-grade classrooms. The order of test administration was counterbalanced so that in each grade, half of the participants began with the
Hebrew section and half with the English section. The order of task administration within
each language section remained constant. Testing took one full 45-minute lesson.
Results
The aim of this research was to examine differences between good and poor EFL spellers
amongst L1 Hebrew-speaking children in the fifth, eighth and 10th grades with specific
Copyright © 2013 UKLA
10
RUSSAK and KAHN-HORWITZ
attention to accuracy in the spelling of novel phonemes and orthographic patterns. This
was tested by conducting ANOVAs, which enabled examination of EFL spelling accuracy
across the three grades and amongst good versus poor spellers. In addition, we were
interested in examining the connections between L1 underlying phonological, orthographic
and spelling ability and EFL spelling. Hierarchical regression analyses were conducted to
perform this.
Differences between good and poor EFL spellers across grades
Means and standard deviations were calculated for all tasks. Table 1 shows the means and
standard deviations for all tasks across grades amongst good and poor spellers. Scores are
reported as percentages except for the standardised Hebrew measures, which are reported
as z scores.
Standard deviations on the experimental real-word spelling task showed greater variance
amongst the poor spellers as opposed to the good spellers, particularly for the eighth and
10th grades. In other words, poor spellers demonstrated greater heterogeneity in their
EFL spelling ability than good spellers. Good spellers in the fifth grade scored 67% as
opposed to good spellers in the eighth and 10th grades who spelled with 93% accuracy
or above. Poor spellers in fifth grade showed dramatically poor spelling. Mean scores
for experimental pseudoword spelling showed generally very low scores across
grades. Here, 10th-grade good spellers scored 61%, whereas fifth-grade good spellers
scored 52%.
Main effects were found for grade and group in all four of the English measures: real-word
spelling (grade, F(2, 139) = 107.88, p < .001; group, F(1, 139) = 225.04, p < .001),
pseudoword spelling (grade, F(2, 139) = 17.09, p < .001; group, F(1, 139) = 111.77,
p < .001), orthographic sensitivity (grade, F(2, 139) = 15.83, p < .001; group, F(1, 139) =
10.17, p < .01) and orthographic choice (grade, F(2, 139) = 56.21, p < .001; group,
F(1, 139) = 30.67, p < .001). Main effects were found for grade and group in the Hebrew phonological choice task (grade, F(2, 139) = 17.51, p < .001; group, F(1, 139) = 20.56, p < .001).
Main effects were found for grade and group in the Hebrew spelling task as well
(grade, F(2, 139) = 8.91, p < .001; group, F(1, 139) = 7.26, p < .01). No main effects
were found for the Hebrew orthographic choice task.
Significant interactions were found between grade and group for English experimental
real-word spelling, F(2, 133) = 5.72, p < .01. Bonferroni post hoc tests showed that the
good spellers scored higher than the poor spellers at every grade. For both groups of
spellers, the scores in the fifth grade were significantly lower than the scores in the eighth
and 10th grades. For the good spellers, there was no significant difference between
spelling accuracy scores in the eighth and 10th grades, whereas for the poor spellers, there
was a significant improvement between the eighth and 10th grades.
An interaction was found between grade and group for English pseudoword spelling,
F(2, 133) = 3.79, p < .05. Bonferroni post hoc tests showed that the good spellers scored
higher than the poor spellers at every grade. However, observation of the means in Table 1
indicates generally depressed scores. For the poor spellers, the scores in the fifth grade were
significantly lower than the scores in the eighth and 10th grades, and despite the fact that there
was an improvement in accuracy scores between the eighth and 10th grades, this improvement was not statistically significant. For the good spellers, there was no significant difference
on scores between grades.
Copyright © 2013 UKLA
69.40 (18.59)
77.73 (10.20)
Orthographic sensitivity (0–15)
Orthographic choice (0–20)
0.95 (0.88)
0.12 (0.71)
0.03 (0.76)
Phonological choice
Orthographic choice
Spelling
Hebrew
66.56 (14.60)
51.65 (13.57)
Experimental real-word spelling (0–28)
N = 22
Experimental pseudoword spelling (0–22)
English
Variables (range)
Good
N = 24
Poor
!0.86 (1.48)
!0.02 (0.87)
0.29 (1.05)
57.08 (22.60)
55.28 (23.79)
16.48 (13.19)
22.47 (16.23)
Fifth graders
0.19 (0.59)
!0.09 (0.83)
!0.03 (1.40)
95.00 (4.71)
74.39 (10.95)
58.61 (13.63)
93.05 (7.85)
N = 19
Good
N = 20
Poor
!0.16 (1.45)
!0.63 (0.95)
!0.28 (1.05)
80.50 (15.38)
69.33 (12.12)
33.18 (17.70)
56.43 (17.60)
Eighth graders
0.38 (0.26)
0.08 (0.88)
0.16 (0.66)
96.60 (4.50)
83.20 (9.65)
60.55 (14.05)
96.86 (5.29)
N = 25
Good
N = 29
Poor
0.35 (0.23)
0.07 (1.08)
!0.61 (0.73)
90.35 (8.23)
76.78 (13.29)
41.54 (15.40)
71.31 (15.87)
10th graders
Table 1. Means and standard deviations (in parentheses) for English measures as percentage values and for Hebrew standardised measures as z scores.
EFL GOOD AND POOR SPELLERS
11
Copyright © 2013 UKLA
12
RUSSAK and KAHN-HORWITZ
An interaction was found for English orthographic word choice, F(2, 133) = 4.07, p < .05,
with significant differences between good and poor spellers in the fifth and eighth grades but
not in the 10th grade. In addition, within both groups, the differences between the fifth and
eighth grades and the differences between the fifth and 10th grades were significant, but the
differences in accuracy scores between the eighth and 10th grades were not statistically
significant for either group of spellers.
No interactions were found for the three Hebrew measures: phonological choice, orthographic choice and spelling.
Spelling development of novel phonemes and orthographic patterns
To better understand development of novel phoneme and orthographic pattern knowledge
amongst poor versus good spellers across grades, ANOVAs were performed with the target
novel phoneme or orthographic pattern as the within-subject factor and grade (fifth, eighth
and 10th grades) and group (good vs poor) as the between-subject factors. Table 2 shows
the means and standard deviations for all novel and non-novel linguistic features across
grades amongst good and poor spellers.
Main effects were found for group for every linguistic (phonological and orthographic) feature, with good spellers outperforming poor spellers (novel phonemes: / / – F(1, 139) = 19.96,
p < .001; /Λ/ – F(1, 139) = 62.32, p < .001; /æ/ – F(1, 139) = 35.78, p < .001; non-novel phonemes and patterns: /e/ – F(1, 139) = 42.66, p < .001; initial ‘h’ – F(1, 139) = 23.45, p < .001;
consonant clusters – F(1, 139) = 22.25, p < .001; novel orthographic patterns: silent ‘e’ –
F(1, 139) = 69.40, p < .001; sight words – F(1, 139) = 136.25, p < .001; novel phonemes
and novel orthographic patterns: /ð/ – F(1, 139) = 47.53, p < .001; /θ/ – F(1, 139) = 48.84,
p < .001; / / – F(1, 139) = 107.48, p < .001). Main effects for grade were found for every
linguistic feature (novel phonemes: / / – F(2, 139) = 3.26, p < .05; /Λ/ – F(2, 139) = 15.1,
p < .001; /æ/ – F(2, 139) = 9.94, p < .001; non-novel phonemes and patterns: /e/ –
F(2, 139) = 39.15, p < .001; initial ‘h’ – F (2, 139) = 5.37, p < .01; consonant clusters –
F(2, 139) = 8.83, p < .001; novel orthographic patterns: silent ‘e’ – F(2, 139) = 8.65,
p < .001; sight words – F(2, 139) = 104.15, p < .001; novel phonemes and novel orthographic patterns: /ð/ – F(2, 139) = 7.56, p < .01; /θ/ – F(2, 139) = 13.95, p < .001; / /
– F(2, 139) = 23.05, p < .001).
An interaction was found between grade and group for sight words, F(2, 133) = 3.67,
p < .05. Bonferroni post hoc tests showed that there was a significant difference between
good and poor spellers in all grades. For both the good and poor spellers, there was a significant development in spelling accuracy for sight words from the fifth to eighth grades.
Whereas the difference in spelling accuracy for sight words was not significant between
the eighth and 10th grades for the good spellers, the difference between scores for the
poor spelling group was significant, with higher scores in the 10th grade than in the
eighth grade.
An interaction between grade and group was found for the silent ‘e’, F(2, 133) = 6.11,
p < .01. Bonferroni post hoc tests showed that there was a significant difference between
good and poor spellers in all grades. For the good spellers, there was no significant
development in spelling ability for silent ‘e’ across grades. For the poor spellers, however,
there was a statistically significant development in spelling accuracy from the fifth to
eighth grades, and although accuracy scores continued to improve in the 10th grade, this
improvement was not statistically significant.
Copyright © 2013 UKLA
.64
æ
.83
Consonant clusters
85.61 (19.45)
64.55 (28.91)
51.82 (29.70)
.75
.77
.88
θ
92.53 (7.14)
41.67 (17.63)
ð
Novel phonemes within novel orthographic patterns
.79
Sight words (from 400 frequent words)
78.57 (14.46)
97.73 (5.28)
.85
.65
68.69 (22.13)
.84
Silent ‘e’
Novel orthographic patterns
Initial h
e
Non-novel phonemes
51.82 (25.94)
77.27 (22.15)
86.36 (20.01)
.60
N = 22
..53
a
Λ
Novel phonemes
Variables
Good
N = 24
Poor
7.08 (13.02)
17.50 (27.23)
38.89 (34.98)
67.26 (31.59)
12.50 (16.49)
35.12 (31.52)
67.50 (38.36)
31.02 (29.30)
41.67 (33.69)
13.33 (17.36)
55.21 (36.10)
Fifth graders
80.00 (23.57)
80.00 (23.09)
85.97 (16.91)
95.49 (5.93)
85.09 (15.61)
74.44 (14.74)
95.26 (7.72)
93.57 (8.54)
87.72 (14.53)
57.90 (19.88)
88.16 (17.42)
N = 19
Good
Poor
34.00 (27.61)
49.00 (39.72)
61.67 (26.55)
87.14 (15.65)
38.33 (18.81)
52.86 (21.29)
85.00 (20.13)
69.44 (28.36)
65.00 (25.88)
29.00 (29.36)
68.75 (25.49)
N = 20
Eighth graders
Table 2. Means and standard deviations (in parentheses) for novel and non-novel linguistic features as percentage value.
80.80 (19.98)
82.40 (18.55)
87.33 (16.16)
94.57 (6.28)
91.33 (11.90)
81.71 (14.59)
98.40 (4.73)
92.44 (10.00)
87.33 (15.43)
59.20 (17.78)
87.00 (21.80)
N = 25
Good
N = 29
Poor
41.04 (28.83)
57.93 (30.40)
74.14 (22.53)
89.90 (10.19)
62.64 (21.67)
64.04 (13.02)
91.72 (11.97)
80.84 (22.79)
72.41 (24.51)
38.62 (19.22)
80.17 (23.51)
10th graders
EFL GOOD AND POOR SPELLERS
13
Copyright © 2013 UKLA
14
RUSSAK and KAHN-HORWITZ
A significant interaction between grade and group was found for the phoneme /e/,
F(2, 133) = 4.40, p < .001. Bonferroni post hoc tests showed that the difference between good
and poor spellers was significant in the fifth and eighth grades but not in the 10th grade. For
both the good and poor spellers, there was a significant difference between accuracy scores in
the fifth and eighth grades, whereas there was no statistically significant development in
accuracy between the eighth and 10th grades.
An interaction was found between grade and group for consonant clusters, F(2, 133) = 5.74,
p < .01. Bonferroni post hoc tests showed that there was a significant difference between good
and poor spellers in the fifth grade only. For the good spellers, there was no significant development in spelling ability for consonant clusters. For the poor spellers, however, there was a
significant development in spelling accuracy for consonant clusters from the fifth to eighth
grades, whereas the difference in spelling accuracy for consonant clusters was not significant
between the eighth and 10th grades.
A similar interaction between grade and group was found for the initial h, F(2, 133) = 5.29,
p < .01. Bonferroni post hoc tests showed that once again, there was a significant difference
between good and poor spellers in the fifth grade only. For the good spellers, there
was no significant development in spelling ability for initial h. For the poor spellers,
however, there was significant development in spelling for initial h from the fifth to
eighth grades.
Yet another similar interaction between grade and group was found for the phoneme /ð/,
F(2, 227) = 6.54, p < .01. Bonferroni post hoc tests showed a significant difference
between good and poor spellers in the fifth grade only. For the good spellers, there was
no significant development in spelling ability for the phoneme /ð/. For the poor spellers,
however, there was a significant development in spelling accuracy from the fifth to eighth
grades. The difference in spelling accuracy was not significant between the eighth and
10th grades.
To better understand the different results found for /ð/ as opposed to /θ/, an ANOVA
was performed with /ð/ versus /θ/ as the within-subject measure and grade and group as
the between-subject factors. An interaction between grade and /ð/ versus /θ/ was found,
F(2, 133) = 4.76, p < .01. A Bonferroni post hoc analysis indicated that scores for the /ð/
were significantly higher than scores for /θ/ across each of the three grades. Whereas the
scores for /ð/ did not change significantly across grades, there was a significant improvement in accuracy scores for /θ/ between the fifth and eighth grades.
The contributions of L1, EFL and ability level to EFL spelling
The second question examined to what extent underlying orthographic and phonological
measures in L1, EFL and ability level (good vs poor spellers) account for EFL spelling
development. Hierarchical regression analyses were conducted in three stages. In the first
stage, the three Hebrew measures – Hebrew spelling, Hebrew orthographic choice and
Hebrew phonological choice – were entered to examine the contributions of L1 Hebrew
to EFL spelling. In the second stage, EFL orthographic sensitivity and EFL orthographic
choice were entered to see if English variables contribute additionally to the explained
variance. Finally, in the third stage, the ability level variable was entered. In what follows,
the data will be reported by grade, first focusing on predictors of EFL real-word spelling,
followed by predictors of EFL pseudoword spelling. Table 3 shows the findings for the
hierarchical regression analyses.
Copyright © 2013 UKLA
EFL GOOD AND POOR SPELLERS
15
Table 3. Hierarchical regression analysis: L1 Hebrew measures predicting EFL word and pseudoword
spellings.
Grade 5
Predictor
ΔR2
Grade 8
b
ΔR2
Grade 10
b
ΔR2
b
EFL experimental word spelling
Step 1
.22***
.31***
.27***
HS
.29*
HPC
.22
.15
HOC
.07
!.04
Step 2
.23***
.50***
.26***
HS
.15
.28**
HPC
.25*
.11
HOC
.04
EOS
.14
EOC
Step 3
.18***
!.04
.32**
!.02
!.80
.55***
.09***
.49***
.07***
.12
HPC
.20*
HOC
.04
EOS
.07
EOC
.28**
.41***
Group
.42***
.33***
n
.01
.06
HS
Total R2
.52***
!.06
.42***
.14***
!.03
.33**
!.06
.04
.25*
.02
!.08
.06
!.14
.42***
.31**
.58***
.66***
.52***
81
72
80
EFL experimental pseudoword spelling
Step 1
.19***
.27***
.09
HS
.37**
.49***
.07
HPC
.08
.16
.24*
HOC
Step 2
.05
.17***
.05
!.15
.11
.04
HS
.24*
.39**
.06
HPC
.11
.17
.16
HOC
.01
.10
EOS
.21*
!.14
EOC
.30**
Step 3
.15***
.12
!.15
.18
.10**
.24
.17***
HS
.21*
.44***
.04
HPC
.05
.10
.05
HOC
.01
!.17
.17
(Continues)
Copyright © 2013 UKLA
16
RUSSAK and KAHN-HORWITZ
Table 3. (Continued)
Grade 5
Predictor
ΔR
2
Grade 8
b
ΔR
2
Grade 10
b
EOS
.14
.12
EOC
.16
.03
Group
Total R2
n
.44***
ΔR
2
b
!.24*
.13
.36**
.46***
.51***
.42**
.30***
81
72
80
HOC = Hebrew orthographic choice, HS = Hebrew spelling, HPC = Hebrew phonological choice, EOS = English
orthographic sensitivity, EOC = English orthographic choice, Group = strong versus weak spellers.
*p < .05, **p < .01, ***p < .001.
With regard to EFL real-word spelling at the end of the fifth grade, Hebrew spelling was
found to explain 22% of the variance, F(3, 77) = 7.27, p < .001. Adding EFL orthographic
sensitivity and orthographic knowledge tasks resulted in an additional 23% of explained
variance, F(2, 75) = 15.26, p < .001. When spelling ability was added, an additional 14%
of the variance, F(1, 74) = 24.00, p < .001, was explained.
Similarly, with regard to EFL pseudoword spelling at the end of the fifth grade, Hebrew
spelling accounted for 19% of the variance in EFL pseudoword spelling, F(3, 77) = 5.95,
p < .001. Adding EFL orthographic sensitivity and orthographic choice tasks resulted in
an additional 17% of the explained variance, F(2, 75) = 10.07, p < .001. When spelling
ability was added, an additional 15% of the variance, F(1, 74) = 22.20, p < .001 was
explained.
At the end of the eighth grade, Hebrew spelling explained 31% of the variance in
EFL word spelling, F(3, 66) = 9.67, p < .001. Adding EFL orthographic sensitivity and
orthographic choice tasks explained an additional 26% of the variance, F(2, 64) = 19.45;
p < .001. When spelling ability was added, an additional 9% of the variance, F(1, 63) =
15.75, p < .001, was explained.
With regard to EFL pseudoword spelling at the end of the eighth grade, Hebrew spelling
explained 27% of the variance, F(3, 66) = 7.97, p < .001. Adding EFL orthographic
sensitivity and orthographic choice tasks did not contribute additional variance. However,
when spelling ability was added to the Hebrew and English tasks, an additional 10% of the
variance was explained, F(1, 63) = 10.61, p < .01.
Hebrew phonological choice accounted for 27% of the variance in EFL word spelling at
the end of the 10th grade, F(3, 76) = 9.38, p < .001. Adding EFL orthographic sensitivity
and orthographic choice tasks contributed an additional 18% variance, F(2,
74) = 11.67, p < .001. When spelling ability was added, an additional 7% of the variance,
F(1, 73) = 11.32, p < .001, was explained.
Whereas Hebrew variables explained 9% of the variance in EFL pseudoword spelling at
the end of the 10th grade, this contribution was not significant, F(3, 76) = 2.54, p = ns.
When the EFL orthographic sensitivity and orthographic choice tasks were added to
the Hebrew tasks, an additional 4% of the variance was explained, which was also
nonsignificant, F(2, 74) = 1.86, p = ns. However, when spelling ability was added to
the Hebrew and English tasks, 17% of the variance was explained, F(1, 73) = 17.31,
p < .001.
Copyright © 2013 UKLA
EFL GOOD AND POOR SPELLERS
17
Discussion
The present study examined differences between good and poor EFL spellers amongst L1
Hebrew-speaking children in the fifth, eighth and 10th grades, focusing on the development of novel phonemes and orthographic pattern knowledge, as well as the connections
between L1 underlying phonological, orthographic and spelling ability, and EFL spelling.
Differences in accuracy for spelling amongst good and poor spellers
First, we investigated the development of accurate EFL spelling across three grades for
good and poor spellers. Accuracy scores for good spellers in the fifth grade on both realword and pseudoword spellings showed basic grapheme–phoneme correspondence
knowledge together with some orthographic pattern knowledge. This is in contrast with
poor spellers in the fifth grade whose scores were remarkably low, reflecting minimal
orthographic pattern knowledge and suggesting difficulties even at the level of basic
grapheme–phoneme correspondence knowledge. These findings for EFL poor spellers
are in contrast with findings regarding poor English L1 spellers in the second grade who
were found to represent all phonemes in dictated words (Arndt & Foorman, 2010). Similar
to Moats’ (1996) findings amongst dyslexic English L1 spellers, poor spellers in the current study demonstrated greater heterogeneity in their EFL spelling ability than good
spellers. Further, despite the fact that the accuracy levels of the poor spellers continued
to improve significantly, the gap between good and poor spellers remained significant
across grades. Similar findings were reported by Sparks et al., (1998) who found that although language abilities of weak FL students improved over time, the gap in spelling between good and poor language learners did not close. By the eighth grade, the good
spellers were already spelling real words with over 90% accuracy. However, this does
not necessarily reflect their independent encoding abilities as can be seen in the fact that
their spellings for novel phonemes as well as novel orthographic patterns remained lower
in comparison. In other words, good spellers seem to have good orthographic memory for
highly frequent words. This could be a reflection of the type of instruction they have
received, which places emphasis on whole-word memorisation.
The current study also tested differences between good and poor EFL spellers on underlying English orthographic measures. The correlation between these orthographic tasks
indicates that they both tap into a basic level of orthographic awareness (r = .43, p < 0.01).
However, the orthographic choice task is more word specific, in that children must
recognise the correct spelling for a real word, whereas for the orthographic sensitivity task,
children are required to identify legal orthographic strings embedded within pseudowords.
This may explain the overall lower scores for the orthographic sensitivity task, as opposed
to the orthographic choice task where scores were almost at ceiling. Scores for orthographic tasks amongst poor spellers from the eighth grade up reflect similar results in
EL1 research, where it was found that poor spellers perform as well as good spellers on
orthographic recognition tasks (Cassar et al., 2005; Lennox & Siegel, 1996; Stanovich,
Siegel & Gottardo, 1997). However, in the case of the EFL learners in the present study,
this orthographic recognition ability did not seem to help the poor spellers spell more
accurately, as can be seen in the real-word and pseudoword spelling scores. In other words,
despite the fact that the poor spellers scored commensurately to the good spellers on
measures of orthographic knowledge, this knowledge did not enable them to spell at a
Copyright © 2013 UKLA
18
RUSSAK and KAHN-HORWITZ
similar level of accuracy to their good spelling peers. It appears that although the poor
spellers are able to recognise a misspelled word from a choice of a pair of correct and
incorrectly spelled words, they are still not able to apply this knowledge in the production
of accurate spellings.
In sum, the gap between good and poor spellers is already evident at early stages of FL
acquisition and remains significant across grades. Whereas good spellers show improvement in spelling accuracy for real words up to the eighth grade, poor spellers continue to
develop their spelling accuracy on all English spelling measures across grades.
The development of knowledge of novel phonological and orthographic patterns in EFL
Another focus of this study was a more fine-grained investigation of the development of
knowledge of novel phonemes and novel orthographic patterns amongst good and poor
EFL spellers. A developmental trend was found for consonantal clusters, initial h and /ð/,
whereby significant differences between good and poor spellers were found in the fifth grade
only. Although the scores for poor fifth-grade spellers on these three target items were not the
lowest of their scores of target items in general, their scores here were low enough to differentiate them from the good spellers. These results indicate that accurate spelling for these
three targets discriminates very clearly between good and poor spellers in the initial stages
of letter–sound knowledge acquisition. Information of this nature may be useful in designing
assessment measures to identify struggling Hebrew L1 EFL learners at early stages of EFL
literacy acquisition. Further, for the good spellers, there was no significant development in accuracy of spelling these target items across grades. For non-novel targets, the good spellers
reached ceiling by the eighth grade. The fact that good spellers did not show improvement
in accuracy scores for novel targets could indicate that across the years, these learners had
not acquired sufficient phonological, orthographic and lexical knowledge. This is particularly
evident where even in the tenth grade, accuracy scores for the silent ‘e’ were not above 82%
for good spellers. These continued difficulties might be due to the complex nature of the
English orthography, coupled with a lack of explicit and direct EFL instruction in Israel.
The striking development in sight word spelling accuracy observed from the fifth to eighth
grades might be further evidence of the strong emphasis placed on whole-word learning
strategies, as opposed to attention to smaller linguistic units in EFL instruction in Israel. In
addition, the fact that EFL instruction takes place on an average of 3–4 hours per week, as
opposed to the total immersion experience of English L1 literacy instruction, may illustrate
that this is simply not enough exposure to facilitate the construction of accurate phonological
and orthographic word-level representations.
Difficulties with the graphemic representation of consonant clusters were only found
amongst the poor spellers in the fifth grade. In English L1 spelling research, there have
been mixed findings. Whereas Bruck and Treiman (1990) found that children with dyslexia
had more difficulties than normally developing children with the spelling of initial
consonant clusters, Cassar et al. (2005) found that both first-grade typically developing
and older dyslexic learners exhibited difficulties in spelling consonant clusters. Difficulties
with the representation of consonant clusters in spelling were also found amongst Dutch
poor spellers (Van Bon & Uit De Haag, 1997). The current study found that the spelling
of consonant clusters was particularly difficult for the poor spellers in the fifth grade. In
general, we did not find consonant clusters to be a problematic orthographic pattern,
possibly because it is the linguistic concept of two consecutive consonant sounds being
Copyright © 2013 UKLA
EFL GOOD AND POOR SPELLERS
19
represented by two separate graphemes that is initially challenging, and once this concept
has been mastered, it is easily transferred across language-learning experiences. This
trend was also found in a study by Kahn-Horwitz et al. (2011), where fifth-grade
Russian–Hebrew-speaking children who could read in Russian and Hebrew, as well as
Russian–Hebrew-speaking and Hebrew-speaking children who could only read in Hebrew,
did not demonstrate difficulties in spelling consonant clusters.
Initial h was included as a target non-novel phoneme because Hebrew speakers often
omit its pronunciation when it appears in a word initial position in English. Therefore,
we were interested to see whether there would be difficulties with the representation of this
phoneme in EFL spelling. Representation of initial h only differed between good and poor
spellers in the fifth grade. All other spellers scored close to ceiling on the spelling of this
phoneme. This indicates that although native Hebrew speakers may omit this phoneme
in word initial position in English oral language production, they do represent it in English
spelling.
It is noteworthy that the scores for the non-novel /e/ sound were not higher than scores
for some of the novel phonemes. In other words, the fact that this phoneme is familiar to
the native Hebrew-speaking pupils does not guarantee that they will represent it accurately
in the EFL context. Despite the fact that this is considered a familiar phoneme from the L1
inventory, there are two confounding issues to consider. Firstly, the idea of equal
orthographic representation of vowels and consonants is not to be taken for granted for
these EFL learners as their experience in Hebrew is that the vowels appear fully
represented in initial grades and then are gradually dispensed with. Secondly, because of
phonemic proximity with other short vowel sounds in English, the choice of which grapheme
to use may present a challenge.
Correct spelling of the /Λ/ sound proved to be the most challenging of all novel
phonemes examined in the present study. Even the good spellers did not score much higher
than 60% accuracy across grades. In the case of EFL instruction in Israel, inaccurate and
incorrect pronunciation may explain these difficulties. Many EFL teachers pronounce the
/æ/ as /Λ/. Further, Hebrew names written in English that contain the /Λ/ are graphemically
represented as ‘a’, for example, Tel Aviv. This graphemic misrepresentation could further
hinder the already poor phonological representation of the /Λ/, leading to incorrect
spelling.
Higher scores for the /ð/ as opposed to the /θ/ sound could possibly attest to the salience
of the voicing feature (Jaeger, 1992; Maassen, Groenen, Crul, Assman-Hulsmans &
Gabreëls, 2001) so that, despite the fact that these targets are novel both phonologically
and orthographically, the voiced target proved to be easier to spell.
The impact of underlying Hebrew L1 abilities and English underlying abilities
on EFL spelling
Finally, the present study investigated the connection between Hebrew L1 underlying
linguistic and literacy abilities and EFL spelling across grades. Hebrew spelling was found
to be a strong predictor of EFL real-word and pseudoword spellings in both fifth and eighth
grades. This may point to common underlying linguistic components for spelling in these
two completely different orthographies. This finding provides support for the Linguistic
Coding Differences Hypothesis. Further support comes from a longitudinal study that
showed that Hebrew L1 spelling combined with Hebrew word attack tested at the end of the
Copyright © 2013 UKLA
20
RUSSAK and KAHN-HORWITZ
fourth grade explained EFL spelling at the end of the ninth grade (Kahn-Horwitz et al., 2012).
In addition to the finding that Hebrew spelling predicted EFL spelling, findings of the present
study highlight the significant contribution of L1 phonological components to EFL spelling
across grades. Here, Hebrew phonological choice was also found to predict EFL real-word
spelling in the fifth and 10th grades, thereby supporting earlier cross-linguistic studies showing phonological predictors of L2 literacy skills, specifically reading (Comeau, Cormier,
Grandmaison & Lacroix, 1999; Kahn-Horwitz et al., 2005). The current study expands the
scope of the cross-linguistic influence of phonological knowledge measured in L1 to include
EFL spelling.
The absence of Hebrew orthographic knowledge as a predictor of EFL real-word and
pseudoword spellings across grades is noteworthy, in that it indicates a strongly languagespecific component to spelling development in EFL. In other words, English orthographic
measures supersede the Hebrew orthographic measure in predicting EFL spelling. This was
evident by the finding that for real-word spelling, English orthographic word choice contributed significant variance over and above Hebrew spelling and phonological choice. Thus, we
see language-general as well as language-specific contributors to accurate EFL spelling.
Hebrew phonological choice contributed language-general knowledge to EFL spelling,
whereas English orthographic choice contributed language-specific knowledge.
Conclusions and educational implications
In conclusion, it appears that exposure and practice with the English orthography does not
necessarily lead to more accurate EFL spelling amongst all L1 Hebrew-speaking pupils. It
is possible that the amount of exposure and practice that the good spellers get with the
English orthography allows them to reach an orthographic accuracy threshold by the eighth
grade and that for them to continue to improve their accuracy levels in spelling, they would
need direct and explicit instruction regarding novel orthographic patterns and conventions.
On the other hand, the continued improvement in spelling amongst the poor spellers might
be due to the extra years of exposure to the English orthography. However, the gap in
spelling accuracy levels remains significant between the good and poor spellers across
grades. It is possible that explicit spelling instruction would benefit the poor speller group
as well. Carefully designed intervention studies focusing on explicit spelling instruction
could shed light on this question.
A second conclusion is that poor EFL spellers show a disadvantage early on in their FL
learning experience. In all probability, these linguistic differences would be evident in the
L1 learner performance profiles as well (Kahn-Horwitz, Shimron & Sparks, 2006). For
assessment purposes, it would be advantageous to include the specific novel and non-novel
phonemes and orthographic patterns that were found to distinguish between good and poor
spellers in the fifth grade, namely initial h, consonant clusters and /ð/. Further, as these
specific phoneme–grapheme patterns, in addition to the others tested, proved to be
challenging, it is highly recommended to focus on these target items through the use of
a language teaching programme that is both direct and explicit.
Finally, this study highlights the complex nature of EFL spelling, which develops in an
asynchronous manner. Accurate phoneme–grapheme correspondence knowledge occurs at
different rates and reaches different levels of accuracy at different grades, and in relation to
different ability levels, despite the uniform nature of the EFL language instruction provided.
Copyright © 2013 UKLA
EFL GOOD AND POOR SPELLERS
21
Acknowledgements
This research was funded by a grant from the Oranim College of Education Graduate
Department to the second author. Special thanks are due to Professor Linda Siegel for
her assistance in the initial stages of this research.
References
Arndt, E.J. & Foorman, B.R. (2010). Second graders as spellers: What types of errors are they making? Assessment for
Effective Intervention, 36(1), 57–67. doi:10.1177/1534508410380135
Benuck, M. & Peverly, S.T. (2004). The effect of orthographic depth on reliance upon semantic context for oral
reading in English and Hebrew. Journal of Research in Reading, 27, 281–299.
Bernstein, S.E. (2008). Phonology, decoding, and lexical compensation in vowel spelling errors made by children
with dyslexia. Reading and Writing, 22, 307–331. doi:10.1007/s11145-008-9116-z
Bourassa, D. & Treiman, R. (2003). Spelling in dyslexic children: Analyses from the Treiman–Bourassa Early
Spelling Test. Scientific Studies of Reading, 7, 309–333.
Brown Corpus (2009). Available from: http://www.lextutor.ca/freq/lists_download/
Bruck, M. (1990). Word-recognition skills of adults with childhood diagnoses of dyslexia. Developmental
Psychology, 26(3), 439–454.
Bruck, M. & Treiman, R. (1990). Phonological awareness and spelling in normal children and dyslexics: The case
of initial consonant clusters. Journal of Experimental Child Psychology, 50, 156–178.
Caravolas, M., Hulme, C. & Snowling, M.J. (2001). The foundations of spelling ability: Evidence from a 3-year
longitudinal study. Journal of Memory and Language, 45(4), 751–774.
Cassar, M. & Treiman, R. (2004). Developmental variations in spelling: Comparing typical and poor spellers. In
C.A. Stone, E.R. Silliman, B. Ehren & K. Apel (Eds.), Handbook of language and literacy: Development and
disorders. pp. 627–643. New York: Guilford.
Cassar, M., Treiman, R., Moats, L., Pollo, T.C. & Kessler, B. (2005). How do the spellings of children with dyslexia
compare with those of typical children? Reading and Writing, 18, 27–49. doi: 10.1007/s11145-004-2345-x
Central Bureau of Statistics (2011). January. Retrieved from http://www.cbs.gov.il/locals.htm
Comeau, L., Cormier, P., Grandmaison, E. & Lacroix, D. (1999). A longitudinal study of phonological processing
skills in children learning to read in a second language. Journal of Educational Psychology, 91, 29–43.
Conrad, N.J. (2008). From reading to spelling and spelling to reading: Transfer goes both ways. Journal of Educational Psychology, 100, 869–878.
Cummins, J. (1979). Linguistic interdependence and the educational development of bilingual children. Review of
Educational Research, 49, 222–251.
Cunningham, A.E., Perry, K.E. & Stanovich, K.E. (2001). Converging evidence for the concept of orthographic
processing. Reading and Writing, 14, 549–568.
Davis, C. & Bryant, P. (2006). Causal connections in the acquisition of an orthographic rule: A test of Uta Frith’s
developmental hypothesis. Journal of Child Psychology and Psychiatry, 47(8), 849–856. doi:10.1111/j.14697610.2006.01597.x
Ehri, L.C. (1991). The development of reading and spelling in children: An overview. In M. Snowling & M.
Thomson (Eds.), Dyslexia: Integrating theory and practice. (pp. 63–79). London: Whurr Publishers.
Ehri, L.C. (2000). Learning to read and learning to spell: Two sides of a coin. Topics in Language Disorders,
20(3), 19–49.
Elbro, C. (1998). When reading is ‘readn’ or somthn. Distinctness of phonological representations of lexical items
in normal and disabled readers. Scandinavian Journal of Psychology, 39, 149–153.
Ellis, N.C. & Cataldo, S. (1990). The role of spelling in learning to read. Language and Education, 4, 1–28.
Fender, M. (2008). Spelling knowledge and reading development: Insights from Arab ESL learners. Reading in a
Foreign Language, 20(1), 19–42.
Figueredo, L. (2006). Using the known to unchart the unknown: A review of first-language influence on the
development of English-as-a-second-language spelling skill. Reading and Writing, 19, 873–905.
doi:10.1007/s11145-006-9014-1
Fischer, F., Shankweiler, D. & Liberman, I.Y. (1985). Spelling proficiency and sensitivity to word structure. Journal
of Memory and Language, 24, 423–441.
Copyright © 2013 UKLA
22
RUSSAK and KAHN-HORWITZ
Foorman, B.R. & Petscher, Y. (2010). Development of spelling and differential relations to text reading in Grades
3–12. Assessment for Effective Intervention, 36(7), 7–20. doi:10.1177/1534508410379844
Frost, R. (2005). Orthographic systems and skilled word recognition processes in reading. In M.S. Snowling &
C. Hulme (Eds.), The science of reading: A handbook. (pp. 272–295). Boston, MA: Blackwell.
Geva, E. & Siegel, L. (2000). Orthographic and cognitive factors in the concurrent development of basic reading
skills in two languages. Reading and Writing, 12, 1–30.
Goswami, U. (1986). Children’s use of analogy in learning to read: A developmental study. Journal of Experimental Child Psychology, 42, 73–83.
Goswami, U. (1988). Orthographic analogies and reading development. Quarterly Journal of Experimental Psychology, 40A, 239–268.
Jaeger, J.J. (1992). Phonetic features in young children’s slips of the tongue. Language and Speech, 35(1–2),
189–205. doi:10.1177/002383099203500215
James, C.P., Scholfield, P., Garrett, P. & Griffiths, Y. (1993). Welsh bilinguals’ English spelling: An error
analysis. Journal of Multilingual and Multicultural Development, 14, 287–306.
Kahn-Horwitz, J., Schwartz, M. & Share, D. (2011). Acquiring the complex English orthography:
A triliteracy advantage? Journal of Research in Reading, 34(1), 136–156. doi:10.1111/j.14679817.2010.01485.x
Kahn-Horwitz, J., Shimron, J. & Sparks, R.L. (2005). Predicting foreign language reading achievement in elementary school students. Reading and Writing, 18, 527–558. doi:10.1007/s11145-005-3179-x
Kahn-Horwitz, J., Shimron, J. & Sparks, R.L. (2006). Weak and strong novice readers of English as a foreign
language: Effects of first language and socioeconomic status. Annals of Dyslexia, 56(1), 161–186.
Kahn-Horwitz, J., Sparks, R.L. & Goldstein, Z. (2012). English as a foreign language spelling development: A
longitudinal study. Applied Psycholinguistics 33(2), 343–363. doi:10.1017/S0142716411000397
Kemp, N. (2009). The acquisition of spelling patterns: Early, late, or never? In C. Wood & V. Connelly (Eds.),
Contemporary perspectives on reading and spelling. (pp. 76–91). London: Routledge.
Lennox, C. & Siegel, L.S. (1993). Visual and phonological spelling errors in subtypes of children with learning
disabilities. Applied Psycholinguistics, 14, 473–488.
Lennox, C. & Siegel, L.S. (1996). The development of phonological rules and visual strategies in average and
poor spellers. Journal of Experimental Child Psychology, 62, 60–83.
Leppanen, U., Niemi, P., Aunola, K. & Nurmi, J.E. (2006). Development of reading and spelling Finnish from
preschool to Grade 1 and Grade 2. Scientific Studies of Reading 10, 3–30.
Loney, N. (2005). Insights. Raanana, Israel: Eric Cohen Books.
Maassen, B., Groenen, P., Crul, T., Assman-Hulsmans, C. & Gabreëls, F. (2001). Identification and discrimination of
voicing and place-of-articulation in developmental dyslexia. Clinical Linguistics & Phonetics, 15(4), 319–339.
Masterson, J.J. & Apel, K. (2010). Linking characteristics discovered in spelling to intervention: Goals and
methods. Learning Disabilities Quarterly, 33(3), 185–198.
Moats, L.C. (1996). Phonological spelling errors in the writing of dyslexic adolescents. Reading and Writing, 8, 105–119.
Navon, D. & Shimron, J. (1984). Reading Hebrew: How necessary is graphemic representation of vowels? In L.
Henderson (Ed.), Orthographies and reading. (pp. 91–102). Hillsdale, NJ: Erlbaum.
Ravid, D. (2001). Learning to spell in Hebrew: Phonological and morphological factors. Reading and Writing, 14,
459–485.
Ravid, D. (2011). Spelling morphology: The psycholinguistics of Hebrew spelling. New York: Springer.
Raynolds, L.B. & Uhry, J.K. (2010). The invented spellings of non-Spanish phonemes by Spanish–English
bilingual and English monolingual kindergarteners. Reading and Writing, 23(5), 495–513. doi:10.1007/
s11145-009-9169-7
Ritchey, K.D. (2008). The building blocks of writing: Learning to write letters and spell words. Reading and
Writing, 21, 27–47.
Russak, S. (2007). A cross-linguistic study of reading disability in native Hebrew speaking adults learning
English as a foreign language. PhD Thesis. Bar Ilan University, Ramat Gan.
Russak, S. & Saiegh-Haddad, E. (2011). Phonological awareness in Hebrew (L1) and English (L2) in normal and
disabled readers. Reading and Writing, 24(4), 427–442. doi:10.1007/s11145-010-9235-1
Saiegh-Haddad, E. (2003). Linguistic distance and initial reading acquisition: The case of Arabic diglossia.
Applied Psycholinguistics, 24, 431–451. doi:10.1017/S0142716403000225
Sawyer, D.J., Wade, S. & Kim, J.K. (1999). Spelling errors as a window on variations in phonological deficits
among students with dyslexia. Annals of Dyslexia, 49, 137–159.
Schlagel, B. (2001). Traditional, developmental, and structured language approaches to spelling: Review and
recommendations. Annals of Dyslexia, 51, 147–178.
Copyright © 2013 UKLA
EFL GOOD AND POOR SPELLERS
23
Shankweiler, D. & Fowler, A.E. (2004). Questions people ask about the role of phonological processes in learning
to read. Reading and Writing, 17, 483–515.
Share, D. & Levin, I. (1999). Learning to read and write in Hebrew. In M. Harris & G. Hatano (Eds.), Learning to
read and write: A cross-linguistic perspective. (pp. 89–111). Cambridge: Cambridge University Press.
Shatil, E. & Nevo, B. (2007). The ELUL Battery for Assessment of Learning Disabilities and Difficulties. Haifa,
Israel: Haifa University.
Shimron, J. (2006). Reading Hebrew: The language and the psychology of reading it. Hillsdale, NJ: Erlbaum.
Siegel, L.S., Share, D. & Geva, E. (1995). Evidence for superior orthographic skills in dyslexics. Psychological
Science, 6, 250–254.
Sparks, R.L. (1995). Examining the linguistic coding differences hypothesis to explain individual differences in
foreign language learning. Annals of Dyslexia, 45, 187–214.
Sparks, R.L. & Ganschow, L. (1993a). Searching for the cognitive locus of foreign language learning difficulties:
Linking native and foreign language learning. Modern Language Journal, 77, 289–302.
Sparks, R.L. & Ganschow, L. (1993b). The impact of native language learning problems on FL learning: Case
study illustrations of the linguistic coding deficit hypothesis. Modern Language Journal, 77, 58–74.
Sparks, R.L., Patton, J., Ganschow, L. & Humbach, N. (2009). Long term cross linguistic transfer of skills from
L1 to L2. Language Learning, 59, 203–243.
Sparks, R., Artzer, M., Patton, J., Ganschow, L., Miller, K., Hordubay, D. & Walsh, G. (1998). Benefits of multisensory language instruction for at-risk learners: A comparison study of high school Spanish students. Annals
of Dyslexia, 48, 239–270.
Spear-Swerling, L. & Sternberg, R. (1996). Off track when poor readers become learning disabled. Boulder, CO:
Westview Press.
Stanovich, K. & Siegel, L. (1994). The phenotypic performance profile of reading-disabled children: A
regression-based test of phonological core variable-difference model. Journal of Educational Psychology, 86,
24–53.
Stanovich, K.E., Siegel, L.S. & Gottardo, A. (1997). Converging evidence for phonological and surface subtypes
of reading disability. Journal of Educational Psychology, 89, 114–127.
State of Israel Ministry of Education English Inspectorate (2009). Guidelines for the teaching of English at the
pre-foundation level. Available from: http://cms.education.gov.il/EducationCMS/Units/Mazkirut_Pedagogit/
English/Publications
Sun-Alperin, M.K. & Wang, M. (2005). Cross-language transfer of phonological and orthographic processing
skills in Spanish-speaking children learning to read and spell in English. Paper presented at the Annual Meeting
of the Society for the Scientific Study of Reading, Toronto, Canada, July.
Treiman, R. & Bourassa, D.C. (2000). The development of spelling skill. Topics in Language Disorders,
20, 1–18.
Treiman, R. & Cassar, M. (1997). Spelling acquisition in English. In C.A. Perfetti, L. Rieben & M. Fayol (Eds.),
Learning to spell: Research theory, and practice across languages. (pp. 61–80). Mahwah, NJ: Erlbaum.
Uhry, J.K. & Raynolds, L.B. (2011). Developmental spelling at three points in the kindergarten school year for L1
English- and L1 Spanish-speaking urban children. Paper presented at the Annual Meeting of the Society for the
Scientific Study of Reading, St. Pete’s, Florida, USA, July.
Van Bon, W.H.J. & Uit De Haag, I.J.C.A.F. (1997). Difficulties with consonants with the spelling and segmentation of CCVCC pseudo words: Differences among Dutch first graders. Reading and Writing, 9, 363–386.
Varnhagen, C.K., McCallum, M. & Burstow, M. (1997). Is children’s spelling naturally stage-like? Reading and
Writing, 9, 451–481.
Venezky, R.L. (1999). The American way of spelling: The structure and origins of American English orthography.
New York: Guilford Press.
Wang, M. & Geva, E. (2003). Spelling acquisition of novel English phonemes in Chinese children. Reading and
Writing, 16(4), 325–348.
Wilkinson, G.S. (1993). The Wide Range Achievement Test 3. Wilmington, DE: Wide Range, Inc.
Wimmer, H. & Landerl, K. (1997). How learning to spell German differs from learning to spell English. In C.A.
Perfetti, L. Rieben & M. Fayol (Eds.), Learning to spell: Research theory, and practice across languages.
(pp. 81–96). Mahwah, NJ: Erlbaum.
Copyright © 2013 UKLA
24
RUSSAK and KAHN-HORWITZ
Appendix
Words from experimental word spelling task: name, went, three, because, found, hand,
south, this, help, had, with, just, our, want, about, that, left, but, people, house, does, time,
question, them, home, like, best, school.
Words from experimental word spelling task: hast /hæst/, fent /fent/, thrun /θrΛn/, beft /
beft/, thack /ðæk/, nith /nIθ/, thip /ðIp/, hab /hæb/, nug /nΛg/, cust /kΛst/, oup / p/, tound
/t nd/, agout /g t/, louth /l θ/, theg /ðeg/, fime /faIm/, rike /raIk/, helt /helt/, mome /
mom/, hame /hem/, yest /yest/, houk /h k/
Susie Russak is a lecturer at Beit Berl Academic College. Her research interests include individual
differences in language learning and their impact on the acquisition of English as a foreign language,
the effects of novel phonological and orthographic features on the acquisition of additional languages, coping mechanisms of students in higher education with learning disabilities and success
stories of adults with learning disabilities.
Janina Kahn-Horwitz is a lecturer at Oranim College of Education. Her research interests include
individual differences in language acquisition and their impact on English as a foreign language
(EL2) acquisition, EL2 reading and spelling development, development of EL2 assessment measures,
EL2 teacher orthographic knowledge and intervention studies in EL2.
Received 17 January 2012; revised version received 21 March 2013.
Address for correspondence: Janina Kahn-Horwitz, Oranim Academic College of
Education, Kiryat Tivon, 36006, Israel. E-mail: [email protected]
Copyright © 2013 UKLA