Printer Friendly

Influence of L1 orthography on spelling English words by bilingual children: a natural experiment comparing syllabic, phonological, and Morphosyllabic first languages.

Abstract. This study examined the influence of first language (L1) orthography on bilingual children's spelling performance in their second language (L2), English. The subtests of spelling and letter-word identification from the Woodcock Proficiency Battery were administered to a sample of 285 six-year-olds in Singapore. All children received literacy instruction in English through the "look-say" method. Analyses of covariance showed a statistically significant effect of L1 on conventional spelling but not on phonological spelling, controlling for reading proficiency. The Chinese (morphosyllabic) group not only scored higher than the Malay (alphabetic) and Tamil (syllabic) groups overall, but also made more real-word substitution and transposition errors. The results are discussed in terms of the influence of L1 orthographic depth on spelling, and how learning disability specialists can utilize the information while evaluating the spelling/literacy skills of English L2 learners.


Children having difficulty learning to spell may be dyslexic or have an underlying cognitive or learning disability. Students who are learning English as a second language (L2), however, may demonstrate atypical English spellings that are due to the influence of exposure to literacy in their first language (L1). Given concerns regarding the persistent over- and underclassification of English L2 learners as qualifying for special education services (Rueda & Windmueller, 2006), it is essential for diagnosticians and teachers to recognize patterns of spelling errors that may be due to L1 influences as opposed to those that may indicate a need for special education services.

Children may be diagnosed as having reading disorders due to difficulties in either decoding ("sounding out") or encoding (spelling) words. Compared to receptive decoding skills, productive spelling performance in dictation is a better indicator of the quality of children's word representations, which, according to Perfetti and Hart (2002), are dependent on the accuracy of the representations of constituent letters in words, phonologically, orthographically, and semantically. Before children are able to use knowledge of orthographic systems to represent words analytically, their word representations usually contain inaccurate or incomplete phonological and orthographic representations of certain letter constituents (Wang, Koda, & Perfetti, 2003).

Typically developing monolingual children display a pattern of errors that reflects their movement through phases or stages of literacy development (Ehri, 1997; Moats, 1995). Bilingual, or multilingual, children's lexical representations follow a similar pattern with the addition that both, or all, languages, consecutively or simultaneously acquired, impact the representation in the other (August & Shanahan, 2006). Therefore, for children speaking more than one language who are learning to read, spell, and write in English, the orthography in L1 must be taken into consideration when examining their reading and spelling acquisition in English.

One prominent theory that seeks to explain literacy processes in different orthographies is the Orthographic Depth Hypothesis (ODH) (Katz & Frost, 1992). According to the ODH, reading processes differ according to the degree of "depth" of the orthography. Starting at the very "shallow" end, where orthography has a one-to-one phoneme-grapheme correspondence, orthographies are placed on a continuum depending on grapheme and phoneme mapping consistency. Thus, readers and spellers of a shallow orthography, such as Finnish, Spanish, or Malay, depend more on the phonological operations, whereas readers and spellers of an extremely deep orthography, such as Chinese, may depend more on visual script or pictorial operations in processing words (Katz & Frost, 1992). Syllabic scripts such as Tamil and Japanese Kana belong somewhere in the middle of the orthographic depth continuum, using both phonological and visual operations in processing words. Studying reading and spelling performance in bilinguals with L1s displaying orthographic, phonological, and morphological differences from their second language (L2) can help distinguish spelling differences that are due to L1 literacy influences from those due to learning disabilities.

Rickard Liow and Lau (2006) investigated the spelling performance of 80 children (average age = 6 years) with three language backgrounds (English L1 and Mandarin L2, Mandarin L1 and English L2, Malay L1 and English L2) in Singapore. Mandarin Chinese, with its morphosyllabic orthography, utilizes little phonological information in reading and writing. The orthography of English is deep, whereas Malay has a shallow orthography, with a nearly one-to-one phoneme-grapheme relationship (Rickard Liow, 1999). Rickard Liow and Lau found all three groups used some phonological awareness (PA) for spelling, but the children with Malay literacy background performed better on PA measures.

Joshi, Hoien, Feng, Chengappa, and Boulware-Gooden (2006) studied the English spelling performance of students in grades 3-8 from the United States, Norway, India, and China. American third graders performed better than Norwegian fifth graders, Indian seventh graders, and Chinese fifth graders on conventional spelling (i.e., spelling scored as correct or incorrect). There was no difference among Norwegian, Indian, and Chinese students who were exposed to formal English instruction for three years. However, in their fourth year of English instruction, Norwegian and Indian students performed better than Chinese students on conventional spelling. As for phonetic spelling, the Chinese group scored lower than the comparable groups from Norway and India. When the Chinese students did not know the correct spelling, they tended to either skip the word or use a known real word as a substitute, whereas Norwegian and Indian students tended to make phonetic approximations of the target word. Joshi et al.'s findings suggest that Chinese students relied more on visual processing in spelling English words. The findings may also reveal the limitations of using a primarily visual approach to English spelling, as each word must be learned more or less independently, rather than being built from known phonemegrapheme correspondences.

In addition, Wang and Geva (2003) compared spelling performance of monolingual English-speaking children and Chinese ESL children in grade 2. Their study showed that there was no difference in real-word spelling between the two groups, but monolingual English-speaking children performed better on pseudoword spelling than Chinese children, lending further support to the argument that Chinese children rely more on visual processes in spelling English and may not be utilizing phonological information when attempting to spell English pseudowords.

Does a bilingual's L1 impact the developmental stages of bilingual reading and spelling? Yin, Anderson, and Zhu (2007) addressed this question in an investigation of 118 Chinese children learning English as a second language in grades 2, 4, and 6 in mainland China. They found that Chinese children followed the stages of prealphabetic, partial alphabetic, and full alphabetic like those of English monolingual children. They also found that Chinese children more readily made onset-vowel analogies than vowel-coda analogies, whereas monolingual children acquiring written English tended to make more vowel-coda analogies.

Based on a sample of children from Singapore, Rickard Liow (1999) differed from Yin and colleagues (2007), arguing for the need to consider the oral language foundation of bilingual children. Due to limited exposure to Standard English and a lack of explicit phonics-based instruction, the Chinese children seemed to adopt visual strategies for an extended period. Yin et al.'s (2007) mainland China sample was exposed to instruction in the shallow alphabetic script called Pinyin to help them learn the pronunciations of newly introduced Chinese characters. Exposure to Pinyin may have helped the children shorten the "extended" logographic stage seen in Singaporean children who had not received Pinyin instruction.

Thus, children who are literate in a shallow orthography may rely on phonological processes, perhaps to the neglect of visual processes. These children may require literacy instruction that, in addition to teaching the English phoneme-grapheme relationships, emphasizes that some English words are not easily decodable and, therefore, must be memorized. On the other hand, children who are literate in a very deep orthography such as Chinese may need a longer period of explicit instruction in the phoneme-grapheme correspondences in English and encouragement to use these correspondences to attempt to spell (or read) unfamiliar words.

Language Policy in Singapore: A Primer

Singapore, as a multilingual state, provides a natural setting for studying the influence of L1 on mastering literacy skills in L2. Singapore's population is comprised of three major ethnic groups: 77% Chinese, 14% Malays, and 8% Indians (Singapore Department of Statistics, 2002). After gaining independence from Britain in 1959, Singapore became an officially multilingual state, selecting four official languages: English, Mandarin Chinese, Malay, and Tamil. Government-provided schooling requires all students, beginning at age 6, to learn their general subject-area content (e.g., science, history) through the medium of English, but children must also study their ethnic group's language (Mandarin Chinese, Malay, or Tamil) as a single subject. (1) To prepare their children for their formal education, virtually all parents send their children to at least two years of kindergarten, which introduces children to basic literacy skills in both English and their group's language (Mandarin Chinese, Malay, or Tamil).

Orthography and Literacy Instruction in English and L1s

English. English uses an alphabetic writing system in which letters or groups of letters represent phonemes. English phoneme-grapheme correspondences are complex and not very transparent (Treiman, 1993); nonetheless, English spelling relies on activation of phonological processes. In Singapore at the time of data collection, kindergartens utilized a whole-word approach to teaching English literacy. Thus, systematic phoneme-grapheme connections were not taught nor encouraged in English spelling.

Malay. Malay (also known as Bahasa Melayu or Bahasa Malaysia) uses a shallow or transparent orthography with the Roman alphabet; thus, it is processed by the assembled phonology, similar to processing the English word cat by converting each letter into the corresponding sounds /k/-/ss/-/t/. Although Malay and English are visually very similar, the two languages differ in orthographic depth. Given the consistent phoneme-grapheme mapping in Malay script, Malay literacy is taught using explicit instruction in phonemegrapheme correspondence. This type of instruction familiarizes Malay children with phonological segmenting and reassembling, which they potentially can apply to English spelling.

Tamil. Tamil, part of the Dravidian language family, is referred to as a syllabic or alpha-syllabic language (Aaron & Joshi, 2006). It consists of 10 primary vowels, 8 secondary vowels, 2 diphthongs, and 16 consonants in Tamil (Arulmani, 2003). Each consonant has an inherent vowel/a/, and when the consonant phoneme has to be changed, secondary vowels or diacritical marks are added before, after, below, or above the letter. Each consonant, therefore, represents a syllable. The mapping of phonemes to graphemes is indistinct; for example, voiced and unvoiced stops are often written in the same way but are pronounced depending on the context (Aaron & Joshi, 2006). The Tamil language also does not differentiate between aspirated and unaspirated consonants. Tamil is taught by introducing sound-symbol correspondences, but the differences between spoken Tamil and written Tamil can be striking and are potential sources of spelling problems for children learning Tamil.

Mandarin Chinese. Mandarin Chinese is morphosyllabic in nature (Perfetti & Zhang, 1995). The Chinese writing system is often regarded as iconic (Luk & Bialystok, 2005); for example, the Chinese character [??] [guo2], country, has two semantic components, the outside one, [??], symbolizing a country's border and the inside one, [??], which means jade on its own and refers to the authoritative figure of a country. In such characters, there is no phonetic component. However, 80% of Chinese characters contain both a semantic and a phonetic component that aid Chinese readers in character reading (Hanley, 2005). For example, the character [??] [cheng2], city, contains the semantic component [??], earth, soil, and the phonetic component, [??] [cheng2], which, on its own, means succeed, finish, or become.

At the time of data collection, kindergarteners in Singapore were not taught Pinyin or any other phonetic representation of Chinese when introducing Chinese literacy. For children with Chinese L1, a "whole-word" reading approach was used in both Chinese and English. Thus, according to the ODH, the Chinese group was likely to rely more on visual processes for learning to read and write in English than the Malay and Tamil groups. Compared to Tamil and Mandarin Chinese, Malay is orthographically much closer to English, and procedures involved in phonological decoding and activation in Malay are more similar to those required in English.

Following the ODH, it was hypothesized in the present study that children with Malay as L1 would use their knowledge of phoneme-grapheme correspondences from Malay and, therefore, make more phonologically plausible errors in their English spellings than the Chinese and Tamil groups, whereas the Chinese group would make fewer of those errors in spelling English words. Additionally, we hypothesized that the Chinese group would make more real-word substitution and transposition errors in their spellings than the Malay and Tamil groups due to their reliance on visual-based reading and writing strategies, as hypothesized by the ODH.

The specific research questions underlying this study were as follows:

1. Does conventional English spelling performance differ by L1 language group, controlling for English reading scores?

2. Are there any differences by L1 language group in English spelling performance when phonologically plausible spellings are counted as correct, controlling for English reading scores?

3. What is the nature of errors among five focal words (he, six, green, house, and table) among different L1 groups?



A random sample of 28 (out of 308) PCF2 kindergarten centers stratified by their geographic area in Singapore were invited to participate in the study. Within each center, a random sample stratified by ethnicity of the Kindergarten 2 (K2) children was invited to participate through letters to their parents. Approximately 10% of the Chinese pupils in each center participated. Malay and Indian pupils were intentionally oversampled to ensure the statistical feasibility of comparisons across groups; therefore, about half of the Malay and Indian pupils from each center participated. Children from ethnic groups other than Chinese, Malay, and Indian, and children identified as having visual, hearing, or learning difficulties, were excluded from the sample. Teachers and center directors indicated that all the sampled students were achieving normally in their L1 and English.

A total of 297 children were recruited with their parents' consent to participate in a larger study. For the current study, we planned to examine children's spelling performance by L1 language group (or L1 group). Thus, 12 children whose ethnicity was Indian but did not speak Tamil were removed from the original sample, leaving 285 children in the final sample: 168 Mandarin Chinese (58.9%), 72 Malay (25.3%), and 45 Tamil (15.8%). All of the children were bilingual; less than 2% of the sample were exposed to three languages, and less than 1% were exposed to four languages at home. About 4% spoke two languages (one of which was English) at home and were learning Mandarin Chinese as a third language at school. Fifty-four percent of parents reported that L1 was the primary home language, with another 31% reporting use of both L1 and English at home. Only 14% reported English (3) as the main home language, but these children had also been learning their ethnic group's L1 for at least 1 1/2 years. Thus, assignment of children to L1 group was based on the language studied in kindergarten in addition to English. All children had received the same English literacy instruction for 1 1/2 years in kindergarten classes. In addition, children in the same L1 group had received the same L1 literacy instruction for 1 1/2 years. Table 1 shows the percentages of home languages used in three L1 groups.

There were 141 boys and 144 girls, constituting 49.5% and 50.5% of the sample, respectively. Nearly all (91.9%) of the children were born in Singapore, 2.8%, in India, 2.1% in Malaysia, 1.4% in Indonesia, and the remaining 1.8% of the sample were born in China, Australia, Sri Lanka, or countries that were not specified (two children). As for parents' origin, 72.3% of the mothers and 83.2% of the fathers were born in Singapore.

Because children must be 5 years old to start K2 in January, the age of the current sample, tested in July 2003, ranged between 66 to 79 months with a mean of 72.29 months (SD = 3.47 months). Years of education completed were similar for mothers (M = 10.69, SD = 3.41) and fathers (M = 11.43, SD = 3.90). In Singapore, 10 years of education marks the end of secondary school. Family income ranged from less than $1,000 Singapore dollars (SGD) per month, or approximately $7,500 U.S. dollars (USD) per year, to over SGD $10,000 per month (USD $75,000 per year), with the median category being SGD $3,000-$3,999 per month, which is considered middle income in Singapore.


Used as the outcome in this study, the Dictation subtest of the Woodcock Language Proficiency Battery-Revised (WLPB-R; Woodcock, 1991) is an assessment of spelling, word usage, and punctuation delivered as a traditional dictation-style spelling test. The test consists of six items on children's fine-motor ability with regard to writing (e.g., draw a straight line) and five items on letters (e.g., print a capital Y) at the beginning; it then proceeds with items of word spelling and word usage (e.g., print the word that means more than one tooth) and punctuation. Of the 38 items completed by at least some of the participants, eight were usage items, two were contraction items, and five were capitalization and punctuation items. No standardized measures were available that had been normed on the Singapore population; thus, raw scores based on the number of correct words according to the scoring rubric of the WLPB-R (e.g., one point for one correct word) were used in the analysis.

The WLPB-R Letter-Word Identification subtest, used as a covariate in these analyses, is an assessment designed to measure early reading skills; it begins with matching rebuses to pictures, moves to letter naming and then to simple word identification. Children with higher reading scores are likely to recognize more words, and having a larger pool of familiar words from reading helps these children visually recognize and, therefore, accurately spell more words. To control for this factor, we used English reading performance as a covariate in our analyses of overall spelling performance. The split-half (odd-even) reliability for our sample on Letter-Word Identification and Dictation were .97 and .86, respectively (with Spearman-Brown correction).


Bilingual Singaporean research assistants followed standard administration procedures when conducting a series of assessments for the larger study, including the Dictation and Letter-Word Identification tests, with each testing session lasting about 45 minutes. For the Dictation test, the examiner said, for example, "Number 16. Spell the word green. The grass was green. Green." The child was then supposed to write on the answer sheet "g-r-e-e-n." If the child did not respond, the examiner encouraged a response; if there was still no response, the examiner continued by presenting the next item until the child wrote six items in a row incorrectly. Dictation raw scores were ceiling-adjusted.

Spelling scoring systems. Dictation raw scores were calculated according to the WLPB-R scoring criteria as described in the instrument section. Standardized scores for the Dictation subtest converted from raw scores showed that, overall, the Singapore sample in the current study scored higher than the U.S. normative sample of the same age by 2/3 standard deviations. Other studies confirm that English L2 learners generally achieve at similar levels as children with L1 English on word-level spelling and reading (August & Shanahan, 2006). Conventional spelling scores (CSS) used in the current study were recalculated from the dictation raw score with 15 usage, contraction, punctuation, and capitalization items removed. The maximum possible score for CSS is 23; however, this may be misleading, because many children reached ceiling well before being administered enough items to reach a CSS score of 23.

Treiman (1993) argued that "any study of children's spelling that is confined to the level of whole words is incomplete" (p. 66). Therefore, we analyzed phonologically plausible spellings as follows: (a) omitting one of the letters in the vowel doublet such as gren for green; (b) omitting vowels but putting the consonants in the right order such as sx for six, tbl for table, and pochs for purchase; (c) substituting vowel sounds with plausible letters, such as grin for green and hi for he; and (d) doubling the consonant for example tabble for table. Spelling scores calculated based on this coding, when phonologically plausible spelling is treated the same as conventionally correct spelling, is called conventional plus phonological spelling scores (CPSS).

The classification of spelling errors. For error coding, we selected five focal words--he, six, green, house, and table--based on the number of responses. For other items, such as comb, purchase, or garage, there was not a sufficient number of responses in our sample for statistical analysis; therefore, these items were excluded from the error coding. Of all responses to the focal words, 43.6% were conventionally correct spelling, 22.8% no response, and 33.6% spelling errors.

After removing conventionally correct spellings and no responses, spelling errors were categorized according to a comprehensive coding scheme adapted from Treiman (1993) to capture possible orthographic influences of the L1. As shown in Appendix A, the errors were first coded into three categories, according to their phonological properties: (a) first phoneme correctly represented only, (b) vowel omission/illegal substitution with consonants in the correct order, and (c) major consonant omission/illegal substitution.

Not all errors were captured in these three categories; hence, we developed a second coding scheme according to the orthographic properties of the errors. All of the errors were captured using this coding scheme and were categorized as (a) real-word substitution (e.g., he spelled as hi, green spelled as grass, house spelled as horse); (b) transposition (e.g., table spelled as tabel, six spelled as xis, or house spelled as houes); and (c) other. Further examples of this coding are provided in Appendix B.

To carry the analysis further, we calculated the proportion of phonemes represented in errors categorized as "other" in the second coding scheme. For example, the error "Grnn" has g, r, n represented, which means three of the four phonemes in green were plausibly represented; therefore, the proportion of phonemes represented is 75%. With this calculation, we were able to compare proportion of phonemes represented across L1 groups for each focal word (he, six, green, house, and table).


The results will be discussed in terms of the research questions underlying the study.

1. Does conventional English spelling performance differ by L1 language group, controlling for English reading scores?

An analysis of covariance (ANCOVA) showed a statistically significant effect of L1 language group on CSS, after controlling for the influence of reading scores, F(2, 281) = 11.07, p < .05, partial [[eta].sup.2] = 7.30%. Least Significant Difference (LSD) post-hoc contrasts revealed that the children with Chinese L1 scored statistically significantly higher than children with Malay and Tamil L1s at the same level of word recognition skill, but there was no statistically significant difference between the Malay and Tamil L1 groups (see Table 2 for group means and standard deviations).

2. Are there any differences by L1 language group in English spelling performance when phonologically plausible spellings are counted as correct, controlling for English reading scores?

Phonologically plausible responses reflect children's knowledge of the various phoneme-grapheme correspondences of English. Despite differences in conventional spelling favoring children with Chinese L1, no statistical significance among the L1 group means was found in the analysis of covariance for CPSS, controlling for the influence of reading scores, F(2, 281) = 3.00, p .05, partial [[eta].sup.2] = 2.09% (see Table 2 for M and SD). Thus, children with Chinese L1, who had more conventionally correct spellings, did not maintain a spelling advantage once phonologically plausible spellings were included, compared to children with Malay and Tamil L1s at the same level of word recognition skill.

3. What is the nature of errors among five focal words (he, six, green, house, and table) among different L1 groups?

There are a total of 1,425 spelling instances for the five focal words (285 participants x 5 focal words = 1425 spelling instances). Spelling attempts include conventionally correct spelling, errors, and no response. The conventionally correct spellings summed over all five focal words were distributed among the Chinese, Tamil, and Malay groups by 49.4%, 40.9%, and 31.7%, respectively. If children's spelling did not differ by L1 group, it was expected that these percentages would reflect the percentages in our sample (58.9% Chinese, 25.3% Malay, and 15.8% Tamil). Instead, the Malay and Tamil groups, who are exposed to other alphabetic or syllabic orthographies, performed better on the five focal words than their percentage of the sample would indicate.

For the Malay group, aside from "no response," "first phoneme represented only" was the most common error type. For the Tamil group, the highest percentage of errors was in the category "major consonant omission and illegal substitution," aside from "no response." However, children with Malay L1 committed 45.5% of the total vowel omission and substitution errors but only 17.9% of the consonant omission errors (see Figure 1). These results suggest that children who are exposed to Malay orthography have a better grasp of letter-sound correspondence, being more likely to spell the consonants right but have some uncertainty about how to represent the vowels. The Chinese children made fewer "first phoneme represented only" errors than the other two groups despite its larger group size.

A chi-square test was computed to analyze the spelling errors coded as "transposition," "real-word substitution," and "other." As shown in Table 3, there was a statistically significant difference of error types across the three L1 groups, Pearson [chi square] (4) = 10.687, p < 0.05. The Chinese group committed statistically significantly more real-word substitution errors than the other two groups. The Chinese group also made more errors in transposition; however, the result was not statistically significant.

The results are in line with our hypothesis that the Chinese group would not only make more real-word substitution errors but also more transposition errors due to their lack of phonological sensitivity and their experience with print in Chinese, a language that requires more visual memory in spelling than the other two languages. No statistically significant difference was found between the Tamil and Malay groups. We also checked whether there were differences in the proportion of phonemes represented in errors categorized as "other" across L1 groups. One-way ANOVAs comparing the mean proportion of phonemes represented across groups found no statistically significant differences for all focal words combined or for each focal word separately.


We began this article by describing the importance of recognizing the differences in spelling errors based on L1 from spelling errors that may signal a need for special education services. Results from the present study can help teachers and diagnosticians understand the importance of recognizing L1 influences on English spelling. In our study, children with Chinese L1 performed better overall than children with Malay and Tamil L1s on the Woodcock spelling measure. Our finding of a high rate of real-word substitution errors by the Chinese group supports the ODH as well as Ho and Bryant's (1999) assertion that experience with morphosyllabic Chinese characters leads to enhanced visual processing ability. In addition, the test items in our study were mostly high-frequency words, which might have given an advantage to the Chinese group because their visual-orthographic processing could be better utilized, despite weak phonological representation skills.

Using conventional spelling measures, children with Chinese L1 may perform well, especially on spelling high-frequency words. However, these children's proficiency with sight-word spelling may mask difficulties in utilizing phonological information to spell or decode unfamiliar words, which in turn may limit their ability to acquire the vast reading vocabulary necessary to become highly proficient English readers and writers.

However, difficulties in phonological awareness and in mapping phoneme-grapheme correspondences should not lead to an immediate diagnosis of dyslexia in children literate in Chinese. Rather, children with a Chinese L1 background can be provided with explicit training in phonemic awareness and the common English spelling patterns corresponding to the phonemes to determine whether additional instruction results in a better grasp of English phoneme-grapheme correspondences (Joshi, Treiman, Carreker, & Moats, 2008). Only if problems persist after such instruction should a diagnosis of dyslexia or another cognitive or learning disability be made. Further, more exposure to oral and written English could help children with Chinese L1 to understand the phonetic base of English spelling and recognize common spelling patterns (Rickard Liow & Lau, 2006).

Conversely, readers of a shallow orthography, such as Malay, may develop better PA and thus have an advantage in learning the phonetic base of English spelling (Landerl, 2006). Although evidence from the current study on this point was indirect, Malay children were able to spell as many words "correctly" as Chinese children were when phonologically plausible spellings were counted as correct alongside conventionally correct spellings. Thus, it may be advantageous for such children to learn to read first (or simultaneously) in the shallow orthography before tackling the complexity of English spelling.

These findings from Malay may be applicable to children of other L1s with shallow orthographies, such as Spanish or Korean. Spelling errors from children literate in a shallow L1 may reflect more precisely the phonemes they are hearing, which may be expressed with the simplest phoneme-grapheme correspondence they know (such as /z/ always being expressed with the letter z even in words such as is or dresses). If the shallow L1 orthography also uses the Roman alphabet, children literate in L1 will likely apply their L1 phoneme-grapheme correspondences until they have learned the specifics and complexities of English spelling.

Our study included only one grade level. Future studies should include different grade levels and cognitive measures, and examine the influence of motivation, home background, and other languages with different orthographic depth. In addition, within a particular orthography, measures of L1 proficiency and literacy levels should be administered to further determine the effect of L1 literacy level on spelling English words.

Implications for Practice

The results from the present study have two important implications for learning disability (LD) specialists. First, the influence of L1 orthography must be taken into consideration when assessing literacy skills, especially spelling, in L2. Second, spelling errors, which are a good indicator of literacy skills (Shankweiler, Lundquist, Dreyer, & Dickinson, 1996), must be analyzed qualitatively, examining the phonetic, morphological, and orthographic nature of the errors instead of scoring words as correct or incorrect.

Regarding the first point, diagnosticians and teachers should determine whether a child has experienced instruction and/or home exposure to a different orthography, and consider the relative depth of that orthography when assessing children for disabilities related to their spelling or reading performance. Children familiar with a shallow orthography are likely to apply their PA and their awareness of the alphabetic principle to the new orthography and may need less PA training. For these children, instruction that teaches transfer to the English phonemic system, with specific attention to sounds not present in the L1, could be helpful. However, students with a background in a deep orthography may rely more on visual processes or whole-word memorization to learn to spell. It is important to first provide these children with explicit phonemegrapheme correspondence instruction in English to see if their spelling improves, rather than rushing to diagnose a cognitive or learning disability.

Concerning the second point, a more detailed examination of children's spelling errors provides much more information than a simple correct/incorrect scoring system. If only a correct/incorrect scoring system had been employed, the Malay children in our study could have been over-identified as having spelling difficulties, whereas the Chinese children could have been underidentified. With a more in-depth scoring system, like that used in this study, a specific spelling difficulty that is related to the L1 orthography may be pinpointed. Therefore, intervention for spelling difficulties can be more focused and more effective. With the increase of English L2 learners in the United States, as well as misdiagnosis of LD among this population, it is imperative that the depth of L1 orthography be taken into consideration and that spelling errors be scored taking linguistic properties of English into consideration.
A. Phonological Coding Scheme Focal Words

Code   Type of Error            Example

1      1st phoneme correctly    He: h
       represented only         Six: S
                                Green: g
                                House: h
                                Table: t

2      Vowel omission/illegal   Six: sx,
       substitution with        Green: grnne, grnn,
       consonants               House: hoes, heos,
       in the correct order     heonse, hnis, hooss,
                                hns, horse, hos, haiws,
                                hoaes, haes, hus, horss,
                                Table: tible, tblea

3      Major consonants         He: Ki, e, Se
       omission/illegal         Six: X
       substitution             Green: geey, geer, garee,
                                gieen, gree, geem, greer
                                House: hou, houer, houe
                                Table: tadal, tadol, teber

B. Orthographic Coding Scheme Focal Words

Code   Error Type               Example

1      Real-word substitution   He: here, his, has
                                Six: sin, sex, he
                                Green: grass, teen
                                House: has, hose, three,
                                home, seventeen, how, he
                                Table: taller, tap, tea, tales,

2      Transposition            He: Eh
                                Six: Xis, Sxi
                                Table: Tebal, Tabel

3      Other                    He: iS, th, ch,
                                Six: isi,
                                Green: ult
                                House: r, n, inboy,
                                Table: s, Tang


Aaron, P. G., & Joshi, R. M. (2006). Learning to spell from print and learning to spell from speech: A study of spelling of children who speak Tamil, a Dravidian language. In R. M. Joshi & P. G. Aaron (Eds.), Handbook of orthography and literacy (pp. 551568). Mahwah, NJ: Lawrence Erlbaum Associates.

Arulmani, S. N. (2003). Reading difficulties in Indian languages. In N. Goulandris (Ed.), Dyslexia in different languages (pp. 235-253). London: Whurr.

August, D., & Shanahan, T. (Eds.). (2006). Developing literacy in second-language learners: Report of the National Literacy Panel on Language-Minority Children and Youth. Mahwah, NJ: Lawrence Erlbaum Associates.

Ehri, L. C. (1997). Learning to read and learning to spell are one and the same, almost. In C. A. Perfetti, L. Rieben, & M. Fayol (Eds.), Learning to spell: Research, theory and practice across languages (pp. 237-269). Mahwah, NJ: Lawrence Erlbaum Associates.

Hanley, J. R. (2005). Learning to read in Chinese. In C. Hulme & M. J. Snowling (Eds.), The science of reading: A handbook (pp. 272-289). Oxford, UK: Blackwell Publishing.

Ho, C. S.-H., & Bryant, P. (1999). Different visual skills are important in learning to read English and Chinese. Educational and Child Psychology, 16, 4-14.

Joshi, R. M., Hoien, T., Feng, X., Chengappa, R., & Boulware Gooden, R. (2006). Learning to spell by ear and by eye: A crosslinguistic comparison. In R. M. Joshi & P. G. Aaron (Eds.), Handbook of orthography and literacy (pp. 569-577). Mahwah, NJ: Lawrence Erlbaum Associates.

Joshi, R. M., Treiman, R., Carreker, S., & Moats, L. (2008). How words cast their spell: Spelling is an integral part of learning the language, not a matter of memorization. American Educator, 32(4), 6-16, 42-43.

Katz, L., & Frost, R. (1992). The reading process is different for different orthographies: The orthographic depth hypothesis. In R. Foster & L. Katz (Eds.), Orthography, phonology, morphology, and meaning (pp. 67-84). Amsterdam: Elsevier.

Landerl, K. (2006). Reading acquisition in different orthographies: Evidence from direct comparisons. In R. M. Joshi & P. G. Aaron (Eds.), Handbook of orthography and literacy (pp. 513-530). Mahwah, NJ: Lawrence Erlbaum Associates.

Luk, G., & Bialystok, E. (2005). How iconic are Chinese characters? Bilingualism: Language and Cognition, 8, 79-83.

Moats, L. C. (1995). Spelling: Development, disability, and instruction. Baltimore: York Press.

Perfetti, C. A., & Hart, L. (2002). The lexical quality hypothesis. In L. Vehoeven., C. Elbro, & P. Reitsma (Eds.), Precursors of functional literacy (pp. 189-213). Amsterdam: John Benjamins.

Perfetti, C. A., & Zhang, S. (1995). Very early phonological activation in Chinese reading. Journal of Experimental Psychology. Learning, Memory and Cognition, 21, 24-33.

Rickard Liow, S. J. (1999). Reading skill development in bilingual Singaporean children. In M. Harris & G. Hatano (Eds.), A cross-linguistic perspective on learning to read (pp. 196-213). Cambridge, UK: Cambridge University Press.

Rickard Liow, S. J., & Lau, L. H.-S. (2006). The development of bilingual children's early spelling in English. Journal of Educational Psychology, 98, 868-878.

Rueda, R., & Windmueller, M. P. (2006). English language learners, LD, and overrepresentation: A multiple-level analysis. Journal of Learning Disabilities, 39, 99-107.

Shankweiler, D., Lundquist, E., Dreyer, L. G., & Dickinson, D. D. (1996). Reading and spelling difficulties in high school students: Causes and consequences. Reading and Writing: An Interdisciplinary Journal, 8, 267-294.

Singapore Department of Statistics. (2002). Key indicators of the resident population. Retrieved from keystats/c2000/indicators.pdf

Treiman, R. (1993). Beginning to spell: A study of first-grade children. New York: Oxford University Press.

Wang, M., & Geva, E. (2003). Spelling performance of Chinese children using English as a second language: Lexical and visual-orthographic processes. Applied Psycholinguistics, 24, 1-25.

Wang, M., Koda, K., & Perfetti, C. A. (2003). Alphabetic and nonalphabetic L1 effects in English word identification: a comparison of Korean and Chinese English L2 learners. Cognition, 87, 129-149.

Woodcock, R. W. (1991). Woodcock Language Proficiency Battery-Revised. Itasca, IL: Riverside Publishing.

Yin, L., Anderson, R. C., & Zhu, J. (2007). Stages in Chinese children's reading of English words. Journal off Educational Psychology, 99, 852-866.


(1) Even students whose dominant home language is English must study Mandarin Chinese, Malay, or Tamil, according to their ethnic background. Non-Tamil L1 Indian students may study one of a few other languages (e.g., Gujarati, Hindi) to fulfill this requirement, but the government does not provide trained teachers or classes in government schools for these low-incidence languages.

(2) pCF (PAP Community Foundation) is the largest provider of kindergarten education in Singapore. It is a nongovernmental community organization run by Singapore's ruling political party, People's Action Party (PAP).

(3) All but two of these children are Chinese, which is about 20% of the Chinese subsample. Because some of these children might technically have English as their first language, we removed them from the sample and reconducted our analyses. The results were substantially the same, so in this article we report the results for the entire sample. We believe these children had sufficient exposure to Mandarin Chinese literacy instruction so that they too would be more likely to use visual processes in spelling English words than the children exposed to Malay or Tamil L1 literacy instruction.

Please send correspondence about this article to: L. Quentin Dixon, 352 Harrington Tower, TAMU 4232, College Station, TX 77843; e-mail:

L. QUENTIN DIXON, Ed.D., Texas A&M University, College Station, Texas.

JING ZHAO, Ph.D. candidate, Texas A&M University, College Station, Texas.

R. MALATESHA JOSHI, Ph.D., Texas A&M University, College Station, Texas.
Table 1
Percentage of Home Language Use by L1 Group (N = 285)

Group                                Percentage

Chinese (n = 168)
  Mandarin Chinese mostly               50%
  Mandarin Chinese and English          27%
  English mostly                        23%

Malay (n = 72)
  Malay mostly                          64%
  Malay and English                     35%
  English mostly                         1%

Tamil (n = 45)
  Tamil mostly                          60%
  Tamil and English                     38%
  English mostly                         2%

Overall (N = 285)
  Ethnic L1 mostly                      55%
  Ethnic L1 and English                 31%
  English mostly                        14%

Table 2
Means and Standard Deviations of the Conventional Spelling
Scores (CSS) and Conventional Plus Phonological Spelling Scores
(CPSS) by LI Group

Outcome                    CSS             CPSS

L1 Group                M      SD        M      SD

Chinese (n = 168)     13.26   2.52     13.39   2.61
Malay (n = 72)        11.85   2.18     12.53   2.69
Tamil (n = 45)        12.22   2.60     12.58   2.82

Table 3
Count and Standard Residual of Error Types Across L1 Groups

                                           L1 Group

                                Chinese   Malay   Tamil   Total

Transposition   Count             23       12       7      42
                Std. Residual    0.61     -0.25   -0.62

Real-Word       Count             37       11       7      55
Substitution    Std. Residual   2.03 *    -1.43   -1.35

Other           Count             171      124     87      382
                Std. Residual    -0.97    0.63    0.72

Total           Count             231      147     101     479
                of Total         48.23    30.69   21.09

* p < 0.05.

Figure 1. Proportion of three errors made by the L1 groups.

                             Chinese   Malay    Tamil
1st phoneme only             40.70.%   52.20%   36.40%
C omission and illegal sub   39.60%    17.90%   45.50%
V omission and illegal sub   19.80%    29.90%   18.20%

Note: Table made from bar graph.
COPYRIGHT 2010 Council for Learning Disabilities
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2010 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:first language
Author:Dixon, L. Quentin; Zhao, Jing; Joshi, R. Malatesha
Publication:Learning Disability Quarterly
Article Type:Report
Geographic Code:9SING
Date:Jun 22, 2010
Previous Article:African American English and spelling: how do second graders spell dialect-sensitive features of words?
Next Article:Critical issues in response-to-intervention, comprehensive evaluation, and specific learning disabilities identification and intervention: an expert...

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters