one dress, two dress: dialectal influence on spelling of english words among kindergarten children...

12

Click here to load reader

Upload: r-malatesha

Post on 25-Dec-2016

213 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: One dress, two dress: Dialectal influence on spelling of English words among kindergarten children in Singapore

Available online at www.sciencedirect.com

System 40 (2012) 214e225www.elsevier.com/locate/system

One dress, two dress: Dialectal influence on spelling of Englishwords among kindergarten children in Singapore

L. Quentin Dixon a,*, Jing Zhao b, R. Malatesha Joshi a

a Texas A&M University, 352 Harrington Tower, 4232 TAMU, College Station, TX 77843-4232, USAb Sun Yat-sen University, 135 Xingangxi Road, Guangzhou 510275, China

Received 1 March 2011; revised 11 October 2011; accepted 14 February 2012

Abstract

The present study examined the influence of Singapore Colloquial English (SCE) on Standard English word spelling througha plural formation task of four words (man, tooth, dress and child) among 168 Singaporean bilingual children with Chinesebackground. It was found that “dropping the plural” was the most prevalent type of error. Poor readers made more random errors,and good readers made more overgeneralization errors and formed more correct plural forms. Similar error patterns held forspelling achievement groups, except that good spellers also made more dropping plural errors than average spellers and poorspellers. The results of this study point to the importance of considering oral language influences on children’s development ofmorphological knowledge. The presence/absence of various morphological features in the spoken language may predict the relativedifficulty of acquiring certain morphological rules for the standard written variety.� 2012 Elsevier Ltd. All rights reserved.

Keywords: Bilingual; Dialect; Morphological awareness; Plural marking; Spelling

1. Introduction

Given the globalization and localization of English language, understanding the relationship between exposure tocertain English dialects and literacy acquisition among learners of English is essential. With the rapid expansion ofpeople learning English from many different first language (L1) backgrounds, many localized varieties of Englishhave developed (Kachru, 1997; Yano, 2001). According to Crystal (1997) and Hu (2007), the popularity of studyingEnglish around the world has increased enormously in recent decades. The purpose of this study was to examine theinfluence of Singapore Colloquial English (SCE) on the formation of English plurals as shown through morpho-logically based spelling patterns among Singaporean kindergartners.

2. Literature review

In English, a single phoneme can be represented by many different graphemes. This non-symmetry of orthographyor the degree of choice for alternative spellings is the source of many spelling errors (Bryant et al., 2006; Joshi et al.,

* Corresponding author. Tel.: þ1 281 210 8688; fax: þ1 979 845 9663.

E-mail address: [email protected] (L.Q. Dixon).

0346-251X/$ - see front matter � 2012 Elsevier Ltd. All rights reserved.

doi:10.1016/j.system.2012.02.003

Page 2: One dress, two dress: Dialectal influence on spelling of English words among kindergarten children in Singapore

215L.Q. Dixon et al. / System 40 (2012) 214e225

2008). This is true with both first language (L1) and second language (L2) English spellers. Although English hasa great degree of choice of alternative spellings, such as “right” or “rite” for homonyms, a majority of English wordsare regulated by spelling rules (Kessler and Treiman, 2003). For instance, Hayes et al. (2006) pointed out that Englishspellers across four different age groups, from elementary school children to adults, are sensitive to and are able toutilize the information of vowel context to spell the consonants. For example, the grapheme tch is determined by thevowel letter preceding it like in the words catch or notch. Among these spelling rules, many are based on morphemes,which are the smallest linguistic units that convey meaning.

It has been shown that children of very early grades can be aware of meaning relations in words when learning tospell (Carlisle, 1995). For example, in a study by Treiman et al. (1994), kindergarten to second grade monolingualEnglish-speaking children made fewer mistakes in the flap consonants in words like dirty than in city, because dirt isa morphemic unit. Morphological awareness is defined as the ability to recognize, distinguish, segment and constructmeaning subunits (Carlisle, 1995). Studies by McBride-Chang and her colleagues have shown that morphologicalawareness is important in reading and spelling development among monolinguals across a range of orthographies(McBride-Chang et al., 2008; Tong et al., 2009) and that morphological awareness transfers from one language toanother (Deacon et al., 2007; Wang et al., 2009).

First language (L1) spelling development, especially morphologically based spelling, is influenced by spokenlanguage (Sulzby, 1996). For example, Treiman (1993) found that English-speaking children in first grade frequentlyrepresented the past tense -ed ending with t (jumt for jumped). Similarly, French children learning to write in their L1have difficulty with spelling plurals, because plurals are not marked in oral French (Totereau et al., 1997). In Frenchimmersion programs in Canada, researchers found that L1 English-speaking L2 French learners in early elementarygrades made more errors on unarticulated French plural forms, such as -s in chiens and -aux in gateaux thanarticulated English plural forms (Cormier and Kelson, 2000). Although the focus of the Cormier and Kelson (2000)study was the influence of phonology in bilingual children’s use of plural morphemes, it also alluded to the possibilitythat L2 learners may find grammatical units that deviate from the morphemic features of their L1 more difficult tomaster.

For second language learners, it may be even more difficult to learn to mark plurals when their L1 does not usea morpheme to mark plurals. Jia (2003) reported a multiple case study in which she followed ten L1 Chinese-speakingchildren for five years and collected both elicited and naturally occurring speech containing plural marking in English.The results showed that the acquisition pattern for the English plural morpheme by nativeMandarin Chinese-speakingchildren was similar to English monolingual children in general. However, only seven out of the ten achieved native-like proficiency in forming English plurals after five years immersed in an English-speaking environment, whereas thesame kind of proficiency only requires three years for monolingual English speakers to achieve. Moreover, Jia (2003)found that plural omissions were more common than overgeneralization errors in this sample. Jiang (2004), froma psycholinguistic perspective, found that Chinese ESL students at university level were very insensitive to thesubject-verb disagreement in a sentence context, which suggests they have difficulty integrating plural or numbermorphemes into their L2 competence.

Spelling performance has also been shown to be highly influenced by dialect or variety of English. A studyconducted by Desberg et al. (1980) indicated that scores on the Social Dialect Feature Inventory (SDFI), a measurebased on the number of dialect features occurring in a child’s utterance in response to structured questions,explained a large amount of variance in children’s reading and spelling scores at three grade levels (2nd, 4th and 6thgrades). For children in the second grade, SDFI accounted for 21.4% of the variance in the spelling portion of WideRange Achievement Test (WRAT), whereas Peabody Picture Vocabulary Test (PPVT) scores only accounted for6.2% of the variance in WRAT-Spelling. At all three grade levels, SDFI explained more variance in spelling scoresthan in reading scores. Treiman (2004) more recently confirmed that speakers of African American VernacularEnglish (AAVE) demonstrate systematic differences from Standard English speakers in spelling production. AfricanAmerican university students in this study made more spelling errors that were associated with the confusion of finalconstant /t/ and /d/ than White students. This finding conforms to a salient characteristic in AAVE which is thedevoicing of the final consonant. African American students are more likely to spell the word rigid as rigit thanwhite students.

The phonology of the dialect or variety of English has also been shown to be salient to children’s spellingdevelopment. Treiman et al. (1997) studied how a specific difference in British English and American Englishregarding rhotic or non-rhotic has influenced spelling development of children aged 6e7½ years old. American

Page 3: One dress, two dress: Dialectal influence on spelling of English words among kindergarten children in Singapore

216 L.Q. Dixon et al. / System 40 (2012) 214e225

children, with a rhotic feature (vocalic /r/) in their variety of English, tend to spell, for example, hurt as hrt; whereasBritish children with a non-rhotic feature in British vowel pronunciation tend to spell it as hut. This type of errordecreases as the children’s age increases. However, overgeneralization errors influenced by different variations ofEnglish persist across age groups; for example, British children spelled bath as barth because the pronunciation of the/a:/ sound in bath is similar to the /a:/ sound in card in British English.

The influence of phonology is not restricted to spelling in children. Treiman and Barry (2000) conducted twostudies with American and British university students and showed that phonology continued to be an important factorin spelling development among adults. British students were more likely to spell polka as polker because ka and kerare alternative spellings in their variety, which is non-rhotic. They are also more likely to omit the final r in words likeleper or filter than American students even when their total score of spelling is higher overall.

Analysis of naturally observed composition data also showed an influence of dialect on written discourse. Ross(1971) studied written essays by elementary students who were also speakers of AAVE. Although the study’sfocus was on syntactic features of AAVE reflected in writing, omission of grammatical morphemes, such as plurals -sand past tense -ed, were common errors. This error pattern suggests a direct relationship between dialect and spelling,because in naturally occurring spoken AAVE, plurals are omitted if the determiner in the noun phrase alreadyindicates plurality. For example, in the phrase two apples, the determiner two marks the plurality, which makes the -sin apples redundant information. Many other languages, such as Chinese and French, share with AAVE this feature ofunmarked plurality in spoken form.

Previous research suggests that children in early school years have already acquired some knowledge of regularinflectional morphemes such as past tense and plural inflections (Berko, 1958). The development of noun pluralsfollows a U-shaped path, according to Marcus (1995), and is very similar to the development of past tense inflectionsin English-speaking children, based on the findings of Marcus et al. (1992). Marcus (1995) analyzed transcription ofspontaneous speech of ten children aged from 15 to 62 months from the CHILDES database (MacWhinney and Snow,1985). He found that children start to correctly represent a small number of plural forms, followed by over-regularization errors, and then finally correctly represent both regular and irregular inflections. The average rate ofovergeneralization in this sample was 8.5%. This rate was calculated based on the ratio of the number of irregularplural nouns that were overgeneralized to the number of correct irregular noun plurals plus the number of over-generalized noun plurals. Both mans and mens were counted as overgeneralization errors in this study.

In spoken form, native speakers of English reach 80e90% proficiency in producing correct plural forms by aboutthree years old (Brown, 1973). In spelling tasks, the rate of correct plural formation is much lower; Cormier andKelson (2000) reported that children in grade 1 in French immersion programs, with literacy instruction in bothFrench and English, spelled 16e45% of the regular English plurals correctly. In grade 2, the correct rate increased to31e60%, and the proportion of dropping plural errors decreased to 8e11%. Overgeneralization of the English -s andthe French -aux followed a different path than dropping plurals. This type of error tended to increase at first, thendecrease across grade levels 1e3. Students in grade 2 were found to make the most overgeneralization errors.

3. Research questions

Following this line of research, which takes children’s dialect or variety of English into account when examiningtheir spelling, we investigated whether English spelling is also influenced by SCE, a dialect of English which does notmark the plural. We hypothesized that the absence of inflectional morphemes in SCE would be reflected in spellingerrors made by Singaporean children of Chinese background. Because the children were exposed to both SCE andChinese at home or at kindergarten (or both), it was expected that spelling errors would reflect the lack of pluralmarking exhibited in both SCE and Chinese and would not differ among Chinese children of different home languagebackgrounds.

Little research has examined plural development of children exposed orally to non-standard dialects froma contrastive point of view and even fewer make the linkage between oral input with written output. Therefore, weasked the following questions:

1) What are the error types in spelling of Singaporean kindergartners from a Chinese background?2) Are there different spelling patterns for readers at different reading levels?3) Do these differences hold for spellers at different levels of spelling?

Page 4: One dress, two dress: Dialectal influence on spelling of English words among kindergarten children in Singapore

217L.Q. Dixon et al. / System 40 (2012) 214e225

3.1. Singapore colloquial English

Singapore Colloquial English (SCE) is considered a variety or dialect of English. Dialect is defined as “a variety ofa single language spoken by people in a particular region or in a particular social group” (Desberg et al., 1980: 70),which implies that if two spoken languages are mutually understandable, they are considered two different dialects;otherwise, they are two different languages (Desberg et al., 1980), although there are some exceptions, usually basedon national or ethnic identity (Sebba, 2007). SCE and English are similar enough that they are mutually intelligible inmost instances.

The Singapore population consists of three major ethnic groups: 77% Chinese, 14% Malay, and 8% Indian(Deterding, 2007). Although the different ethnic groups brought many different languages to Singapore, English iswidely spoken and understood in Singapore today. Among the four official languages (Chinese, Malay, Tamil andEnglish) in the pluralistic and meritocratic Singapore society, English is considered as the one directly linked withcareer promotion and economic advancement (Pakir, 1991). From a sociolinguistic perspective, “English-knowingbilingualism” (Kachru, 1983: 40e42) in Singapore involves the relative status of Standard English and the indige-nized form, SCE. The Singapore government actively promotes the use of Standard English, denigrating SCE(Wee, 2009). English users in Singapore with various proficiency levels differ in their level of mobility across theformality continuum with the Standard English at the upper end and SCE at the lower end (Pakir, 1991). Well-educated Singaporeans with a higher English proficiency level are more flexible in their movement on thecontinuum than those who are less proficient, according to Pakir (1991). Pakir notes that these elite Singaporeans canavoid using SCE and use only the standard form to signal their high status; however, they can also use SCE, a lowvariety, to establish rapport and express their sense of belonging to the Singaporean community. SCE, however, is thecommon variety of English used among the non-elite in Singapore, and many children are exposed primarily to SCE inthe community (Rubdy, 2007).

3.2. Morphology of SCE

SCE, due to the influence of diverse vernaculars, includes unique grammatical and morphological aspectscompared to International Standard English (ISE) (Pakir, 1991). The uniqueness of SCE morphology is mainlydemonstrated by inflectional suffixes (Deterding, 2007). SCE nouns are usually not inflected (Gupta, 1989, 1994).Noun plural -s (pronounced in Standard English as /s/ after unvoiced non-sibilants, /z/ after voiced non-sibilants and/iz/ after sibilants) is almost always non-articulated in SCE. SCE is heavily influenced by Mandarin and other Chinesedialects, which do not employ inflectional morphemes in oral or written form. Plurality, tense and comparative/superlative are marked by other lexical and syntactic devices in Chinese instead of intra-word inflectional affixes.Plural marking in Chinese is achieved by numerals or quantifiers modifying the noun, for example, yı ge pıng gǔo(‘one apple’) and lǐang ge pıng gǔo (‘two apple’); plural marking in Chinese does not necessitate any change inpronunciation or written form of the base noun. The unique feature of plural marking in SCE is also influenced byMalay (Tan, 2005) where plural marking is not obligatory, especially when the number information is clear from thesyntactical cues.

4. Method

4.1. Participants

A random sample of 28 (out of 308 total) PAP1 Community Foundation (PCF) kindergarten centers stratified bytheir geographic area in Singapore were invited to participate in the study. PCF kindergarten centers charge very lowfees, and government subsidies are available for families who cannot afford the fees. Thus, PCF kindergartens servea large swath of non-elite Singaporean children. Within each center a random sample stratified by ethnicity of theKindergarten 2 (K2) children in the center were invited to participate through letters to their parents. Approximately

1 PCF (PAP Community Foundation) is the largest provider of Kindergarten education in Singapore. It is a non-governmental community

organization run by Singapore’s ruling political party e People’s Action Party (PAP).

Page 5: One dress, two dress: Dialectal influence on spelling of English words among kindergarten children in Singapore

218 L.Q. Dixon et al. / System 40 (2012) 214e225

10% of the Chinese pupils in each center participated. A total of 168 six-year-old Chinese children (age: M¼ 72.7months; SD¼ 3.6) participated with their parents’ consent. Seventy-nine (47%) were boys. According to a homelanguage survey administered to the parents of the participants, 117 mothers (70.1%) and 141 fathers (87%) were bornin Singapore. Thirty-eight (22.6%) primary caretakers of these children spoke only English at home; forty-six (27.4%)spoke both English and Chinese (including Mandarin and Chinese dialects), and 84 (50%) spoke only Chinese to theirchild. The teachers in the kindergartens primarily used SCE in their instruction. The children were in the second halfof their K2 year at the time of data collection; all had 1½e2½ years of literacy instruction in English and Chinese. InSingapore, at the time of data collection, PCF kindergartens did not teach morphological units as part of Englishspelling.

4.2. Instruments

The Dictation and LettereWord Identification subtests of the Woodcock Language Proficiency Battery-Revised(Woodcock, 1991) were administered. The Dictation subtest is an assessment of spelling, word usage and punc-tuation delivered as a traditional dictation-style test. A total of 40 items are present in the test, among which six areusage items, two are contraction items and three are capitalization and punctuation items. One capitalization andpunctuation item on the Dictation subtest, “Dayton, Ohio” was not administered and was not counted toward thechild’s ceiling, because there is no Singaporean analogy to the spelling conventions of United States (U.S.) citiesand states. Among the six usage items, four required students to write an inflected form from the base form (e.g.,“Print the word that means more than one tooth.”). Responses made toward four items (men, teeth, dresses,children), all of which asked the children to create the plural, were used in this analysis. The LettereWordIdentification subtest is a measure of reading skills used in the current study. The test begins with matching rebusesto pictures, moves to letter naming and then to simple word identification. The split-half (oddeeven) reliability(with SpearmaneBrown correction) for our sample on Dictation and LettereWord Identification were 0.86 and0.97, respectively.

5. Results

5.1. What are the error types in spelling of Singaporean kindergartners from a Chinese background?

To understand participants’ overall spelling skill, we calculated their average standard score on the WoodcockDictation subtest, M¼ 111.34, SD¼ 11.32 and on the LettereWord Identification subtest, M¼ 112.88, SD¼ 17.03(see Table 1). Compared to the U.S. norming group, participants in the present study scored over 2/3 of a standarddeviation better on spelling and writing conventions. In other words, overall, this is a group of strong English spellersfor their age. We then coded the children’s spelling attempts into four categories for each of the four target words:0dno response or random error, 1ddropping plural, 2dovergeneralization, and 3dcorrect (see Table 2 for examplesof each category).

Table 1

Means and standard deviations based on standard scores on spelling (dictation) and reading (lettereword ID) tasks by spelling and reading groups.

Dictation Lettereword identification

M (SD) M (SD)

Spelling group

Poor speller (n¼ 45) 99.11 (6.85) 100.24 (10.67)

Average speller (n¼ 74) 110.55 (6.08) 110.26 (14.06)

Good speller (n¼ 49) 123.76 (6.94) 128.45 (14.00)

Reading group

Poor reader (n¼ 49) 102.65 (9.31) 95.59 (8.77)

Average reader (n¼ 71) 110.90 (8.66) 110.39 (7.21)

Good reader (n¼ 48) 120.85 (9.14) 134.21 (9.41)

Total (N¼ 168) 111.34 (11.32) 112.88 (17.03)

Page 6: One dress, two dress: Dialectal influence on spelling of English words among kindergarten children in Singapore

Table 2

Examples of spelling error coding categories.

Code Category men teeth dresses children

0 No response/random two/boy twos/theen dresser/skirt comb/dus

1 Dropping plural man tooth dress child

2 Overgeneralization mans/mens tooths/toothes dresss childs

3 Correct men teeth dresses children

219L.Q. Dixon et al. / System 40 (2012) 214e225

Aside from no response or random errors, dropping the plural was the most prevalent type of error, consisting of26.8%, 18.5%, 20.2% and 10.7% of all spelling attempts for each of the target words respectively (men, teeth, dresses,and children). This result indicates that the participants had the ability to correctly represent the base words but lackedthe knowledge to successfully provide morphemes that reflect plurality. Overgeneralization errors did not comprisea significant portion of the spelling attempts in this sample, especially for the word dress/dresses. The proportion ofovergeneralization error was 8.6%, and the proportion for each word, men, teeth, dresses, and children, was 9.5%,12.5%, 5.4% and 7.1%, respectively. An overgeneralization error of dresses would form dresss; the low rate ofcommitting this error for this particular word suggests that bilingual children who are learning English attend to thelegal letter combinations, even without explicitly being taught the orthographic rules: sss is simply not a permissibleletter string in English. Fig. 1 shows the percentage of each category by words.

The results did not differ as a function of home language groups (e.g., English Only, English plus Other, and OtherUsually). Aside from no response or random errors, all three home language groups produced more spelling attemptswith dropping plural errors than overgeneralization errors and correct responses for men and dresses, and differencesin error types among the groups were not statistically significant.

5.2. Are there different spelling patterns for readers at different reading levels?

The participants were divided into three reading groups based on their performance on the Woodcock JohnsonLettereword Identification subtest. They were classified as poor readers if they scored below the 30th percentile, asaverage readers if they scored between the 30th percentile and the 70th percentile and as good readers if they scoredabove the 70th percentile. Similar criteria of division were used in previous studies (e.g., D’Angiulli et al., 2001;Lesaux et al., 2007) to identify students who are at-risk for reading difficulties. This division results in 49 poorreaders, 71 average readers and 48 good readers. Fig. 2aed is a group of stacked bar charts which shows the

Fig. 1. Percentage of spelling categories by focal words. Note: Percentages do not add up to 100% due to dropping the category of ‘no response/

random error’.

Page 7: One dress, two dress: Dialectal influence on spelling of English words among kindergarten children in Singapore

Fig. 2. Percentage of errors made by the three reading groups (Poor, Average, and Good) for a. Men, b. Teeth, c. Dresses and d. Children.

220 L.Q. Dixon et al. / System 40 (2012) 214e225

percentages of different types of errors made by reader groups for each individual word. It is not surprising to see thatpoor readers are more likely to not respond or make a random error, whereas good readers tend to answer the itemcorrectly. For the word men, within poor readers, 87.8% fell into “no response/random error” category and only 2%responded correctly. Within average readers, 52.1% fell into “no response/random error” category and none respondedcorrectly. Within good readers, however, 29.2% fell into “no response/random error” category, while 25.0% respondedcorrectly.

Another finding is that good readers typically made more overgeneralization errors than the other reading groups.For example, for the word teeth, a large percent (29.2%) of good readers spelled the plural form of tooth as tooths ortoothes, followed by 8.5% of average readers; no poor readers made this kind of error. Poor readers were more likely tomake random errors or did not respond to the test. Overgeneralization errors by good readers for words men, dresses,and children, were 18.8%, 10.4%, and 12.5% respectively, which were the largest among the three reading groups. Fortheword dresses, good readers tended to drop the plural more than overgeneralizing it, which indicates their awarenessof the rules governing English letter sequences.

To validate the findings presented above, we conducted a latent class analysis (LCA) using error codes for each ofthe four words as predictors of the latent class membership. The relationships are described in a set of logisticregression equations for ordered categorical latent class indicators and a set of multinomial logistic regressions forunordered categorical latent class indicators. Using the Mplus software to perform the calculations (Muthen andMuthen, 2007), we applied LCA to a 168 * 4 data matrix. The rows corresponded to the 168 participants, whilethe four columns corresponded to responses of each subject to the four target items.

We then fit the data to a three-class solution. The number of classesdtwo, three or fourdis arbitrary, but the resultsincluding the most salient indicatordentropydhelped to confirm that a three-class solution was favorable. Individualparticipants were then assigned class membership based on the likelihood of their particular response profile. In otherwords, all individual differences of the errors made across the four words are characterized by class membership.These differences can be modeled as a function of predictors of class membership. The results showed a high

Page 8: One dress, two dress: Dialectal influence on spelling of English words among kindergarten children in Singapore

221L.Q. Dixon et al. / System 40 (2012) 214e225

classification quality index, entropy¼ 0.982. The average latent class probabilities for most likely latent classmembership range from 0.984 to 1.00. The analysis yielded the following classes:

Class 1¼mostly random error/no response (n¼ 112)Class 2¼mostly dropping plural (n¼ 27)Class 3¼mostly overgeneralization or correct (n¼ 29)

A Pearson chi-square test was computed to determine if the latent classes differ as a function of the three readinggroups. The result indicated a significant association between the latent class membership and whether the childbelongs to poor, average or good reading group, Pearson c2(4)¼ 55.25, p< 0.01, Cramer’s V¼ 0.41, p< 0.001,indicating a medium effect size. As expected, poor readers made significantly more “random/no response errors” andgood readers made the least amount of this type of error. Inversely, a majority of Class 3 members (75.9%), which hasan error profile of mostly overgeneralization or correct, were good readers. None of the poor readers fell into Class 3and a significantly smaller portion of poor readers were in Class 2. The standardized residues were�2.1 and�2.9 forpoor readers to be in Class 2 and Class 3, respectively (see Table 3).

5.3. Do these differences hold for spellers at different levels of spelling?

To determine whether the error patterns found for different reading groups still hold for spellers at different levelsof spelling, we divided the spellers into poor spellers (<30th percentile), average spellers (30th percentilee70thpercentile) and good spellers (>70th percentile) according to their original Dictation scores. Among the 48 goodreaders, 35 were good spellers (72.9%), 11 were average spellers (31.4%) and 2 were poor spellers (5.7%). Among the71 average readers, 44 were average spellers (62.0%), 13 were good spellers (18.3%) and 14 were poor spellers(19.7%). Among the 49 poor readers, 29 were poor spellers (59.2%), 19 were average spellers (38.8%) and only onewas a good speller (2.0 %). The Pearson correlation coefficient between reading and spelling scores was 0.74(p< 0.01). The division of the spelling groups resulted in 45 poor spellers, 74 average spellers and 49 good spellers.Means and standard deviations of standard scores for these three spelling groups are shown in Table 2.

A chi-square cross-tabulation test was performed and the result showed a significant association between the latentclass membership based on error profile and whether the child belonged to the poor, average or good spelling group(Pearson c2(4)¼ 69.685, p< 0.01, Cramer’s V¼ 0.46, p< 0.001). The chi-square test result for the spelling groupswas similar to that of the reading groups (see Table 4), and the effect was even stronger. All poor spellers, like poorreaders, fell into Class 1. More good spellers fell into Class 3 of overgeneralization or correct plural marking than theother two spelling groups. Interestingly, a significant number of good spellers also fell into Class 2 which has a patternof making errors that mostly consisted in dropping the plural.

A close examination of the error patterns for each group by each individual word confirmed this finding. Onehundred percent of poor spellers made “no response/random” errors for the words children and dresses. Largepercentages of the good spellers were able to spell the plural forms of the target words correctly (24.5%, 14.3%, 22.4%and 16.3% for men, teeth, dresses, and children, respectively) or at least to capture the basic form of the word and

Table 3

LCA membership predicted by error profile as a function of reading groups.

Reading group Class 1 Class 2 Class 3

Poor reader Count 47 2 0

% within class 42.00% 7.40% 0.00%

Std. residual 2.5 �2.1 �2.9

Average reader Count 50 14 7

% within class 44.60% 51.90% 24.10%

Std. residual 0.4 0.8 �1.5

Good reader Count 15 11 22

% within class 13.40% 40.70% 75.90%

Std. residual �3 1.2 4.8

Note: Result is statistically significant if standard residual is greater than þ2.0 or smaller than <�2.0.

Page 9: One dress, two dress: Dialectal influence on spelling of English words among kindergarten children in Singapore

Table 4

LCA membership predicted by error profile as a function of spelling groups.

Spelling group Class 1 Class 2 Class 3

Poor speller Count 45 0 0

% within class 40.20% 0.00% 0.00%

Std. residual 2.7 �2.7 �2.8

Average speller Count 55 13 6

% within class 49.10% 48.10% 20.70%

Std. residual 0.8 0.3 �1.9

Good speller Count 12 14 23

% within class 10.70% 51.90% 79.30%

Std. residual �3.6 2.2 5.0

Note: Result is statistically significant if standard residual is greater than þ2.0 or smaller than <�2.0.

Fig. 3. Percentage of errors made by the three spelling groups (Poor, Average, and Good) for a. Men, b. Teeth, c. Dresses and d. Children.

222 L.Q. Dixon et al. / System 40 (2012) 214e225

make dropping plural errors (28.6%, 24.5%, 40.8% and 20.4% for men, teeth, dresses, and children, respectively)(see Fig. 3aed). Among good spellers, 40.8% made “dropping plural” errors on the word dress and this pattern wasconsistent with the pattern found for good readers.

Good spellers are the group of spellers who were most likely to make overgeneralization errors compared toaverage and poor spellers. The percentage of good spellers who made overgeneralization errors were 20.4%, 24.5%,40.8% and 20.4% for men, teeth, dresses, and children, respectively. Average spellers made fewer overgeneralizationerrors than good spellers; the percentages of overgeneralization for average spellers were 8.1%, 6.8%, 4.1% and 4.1%for the four words. Hardly any poor spellers made overgeneralization errors.

Page 10: One dress, two dress: Dialectal influence on spelling of English words among kindergarten children in Singapore

223L.Q. Dixon et al. / System 40 (2012) 214e225

6. Discussion and conclusion

The influence of SCE, which lacks inflectional morphemes, was evident in the spelling of Singaporean childrenwith Chinese background. Many participants correctly represented the base words but failed to provide morphemesthat reflect plurality. Dropping the plural was the most common type of error observed in this sample. This findingechoes Ross (1971) study with AAVE-speaking students, who also tend to omit the plural forms in spellingproductions reflecting the AAVE dialect in which plurality is often unmarked.

The current study complements from a morphological perspective the findings of Yeong and Rickard Liow (2010),which was also conducted in the Singaporean context. Yeong and Rickard Liow investigated the relationship betweenlinguistic experience and phonemic representations in spelling in two groups: Mandarin-dominant and English-dominant bilingual children aged five to six years old. Mandarin-dominant bilingual children performed signifi-cantly more poorly on representing the word-final phonemes such as /v/ or /b/ that only appear in English but not inChinese than their English-dominant counterparts. Our study shows that Chinese-English bilingual children’slinguistic experience with SCE also influences their morphological representations in English spelling, with nosignificant differences according to home language background. Additional research is needed to explore howextensive the dropping of the plural is on a wider variety of English words, and in Singapore’s other ethnic groups.

The overgeneralization (8.6%) of irregular noun plurals in spelling production from our studywas comparable to theovergeneralization rate of English-speaking children of similar age in oral production found in Marcus (1995). Thelower rate of committing overgeneralization errors for the word dress/dresses suggests that children who are learningEnglish as a second language attend to permissible letter combinations even without explicitly being taught theorthographic rules. Recent studies suggest that knowledge of graphotactic conventions for permissible letter sequencesis acquired in very early grades for native English-speaking children (Cassar and Treiman, 1997; Treiman, 1993).

Caravolas et al. (2005) found that children’s sensitivity to permissible letter strings is influenced by a variety offactors including frequency of printed words. The Chinese children in our sample were exposed to English print for1½e2½ years in kindergarten, which probably helped them understand permissible letter sequences. LearningEnglish through print was also found to lead to more accuracy in spelling tasks (Joshi et al., 2006), which may be whychildren in our sample had better overall spelling skills than the U.S. norms. The good readers made fewer (or none) ofthese errors, indicating that those who are reading and are therefore exposed to more print are inductively recognizingthose permissible letter strings rules. English as a second language instruction is highly context-sensitive: instructionshould be altered as context changes, which requires language teachers to always be aware of the specific varieties ofEnglish to which students are exposed. The major finding of this study was that variations of English play an importantrole in bilingual children’s morphologically determined spelling.

In the context of Singapore, where SCE is widely used, teachers need to pay attention to the systematic differencesbetween SCE and Standard English. It is recommended that explicit instruction be implemented to address thesedifferences. A search of the literature indicated that no programs for teaching English morphological rules tokindergarten-aged children have been described or tested (Reed, 2008), although morphological programs forkindergartners in Norwegian (Lyster, 2002) and French (Casalis and Cole, 2009) met with some success. Specificinstruction on certain morphological rules, such as the use of -s or -es to mark plurality, should help Singaporeanchildren learn Standard English spelling. This same recommendation can be broadened to include speakers of othervarieties of English that also do not mark plurality, such as AAVE. Because other studies have shown that dialectdifferences can influence spelling even among university students (Treiman, 2004; Treiman and Barry, 2000), it isimportant to address this issue to help students acquire the standard written form of the language as expected forclerical, secretarial and professional occupations. One study indicated that Singaporean students in grades 3e6dropped their plurals in written work, even though they reported not using SCE for writing (Rubdy, 2007). Thisdisparity between stated practice and actual use suggests that many students may not be aware that plural forms arerequired in Standard English. Future research is needed to determine whether Singaporean university students alsodrop plurals in their writing, and whether targeted interventions at kindergarten or later could remedy this situation.

In addition, these results suggest that second language learners of English from languages that do not mark pluralityin speech (such as Chinese or French) may similarly benefit from explicit instruction in English morphological rules formarking plural in spelling. Generally, explicit instruction in morphology and morphological rules for spelling canenhance any learner’s English spelling skills. A recent meta-analysis exploring the effects of morphological instruction

Page 11: One dress, two dress: Dialectal influence on spelling of English words among kindergarten children in Singapore

224 L.Q. Dixon et al. / System 40 (2012) 214e225

on literacy skills found that the average effect size (Cohen’s d) across 21 studies on spelling skills was 0.49 favoring theexperimental groups receiving explicit morphological instruction over the control groups (Bowers et al., 2010).

In conclusion, more research needs to be conducted to determine the extent of the influence of SCE on differentaspects of children’s English spelling, and to disentangle the possible effects of other languages (Chinese, Malay,Tamil) also spoken in Singapore. A longitudinal study following Singaporean children’s spelling development wouldalso help illuminate which aspects of SCE persist in formal writing and which disappear as children experience moreexposure to print and to Standard English through school. The current study begins this process by indicating theinfluence of SCE on young Singaporeans’ spelling of English plurals.

References

Berko, J., 1958. The child’s learning of English morphology. Word 14, 150e177.

Bowers, P., Kirby, J., Deacon, H., 2010. The effects of morphological instruction on literacy skills. Review of Educational Research 80, 144e179.

Brown, R., 1973. A First Language: The Early Stages. Harvard University Press, Cambridge, MA.

Bryant, P., Deacon, H., Nunes, T., 2006. Morphology and spelling: what have morphemes to do with spelling? In: Joshi, R.M., Aaron, P.G. (Eds.),

Handbook of Orthography and Literacy. Erlbaum, Mahwah, NJ, pp. 601e616.Caravolas, M., Kessler, B., Hulme, C., Snowling, M., 2005. Effects of orthographic consistency, frequency, and letter knowledge on children’s

vowel spelling development. Journal of Experimental Child Psychology 92, 307e321.

Carlisle, J., 1995. Morphological awareness and early reading achievement. In: Feldman, L.B. (Ed.), Morphological Aspects of Language Pro-

cessing. Erlbaum, Mahwah, NJ, pp. 189e209.

Casalis, S., Cole, P., 2009. On the relationship between morphological and phonological awareness: effects of training in kindergarten and in first-

grade reading. First Language 29, 113e142.

Cassar, M., Treiman, R., 1997. The beginnings of orthographic knowledge: children’s knowledge of double letters in words. Journal of

Educational Psychology 89, 631e644.

Cormier, P., Kelson, S., 2000. The roles of phonological and syntactic awareness in the use of plural morphemes among children in French

immersion. Scientific Studies of Reading 4, 267e293.

Crystal, D., 1997. English as a Global Language. Cambridge University Press, Cambridge, England.

D’Angiulli, A., Siegel, L.S., Serra, E., 2001. The development of reading in English and Italian in bilingual children. Applied Psycholinguistics

22, 479e507.

Deacon, H., Wade-Woolley, L., Kirby, J., 2007. Crossover: the role of morphological awareness in French immersion children’s reading.

Developmental Psychology 43, 732e746.Desberg, P., Elliott, D., Marsh, G., 1980. American Black English and spelling. In: Frith, U. (Ed.), Cognitive Processes in Spelling. Academic

Press, London, pp. 69e82.

Deterding, D., 2007. Singapore English. Edinburgh University Press, Edinburgh.

Gupta, A.F., 1989. Singapore colloquial English and standard English. Singapore Journal of Education 10 (2), 33e39.

Gupta, A.F., 1994. The Step-Tongue: Children’s English in Singapore. Multilingual Matters, Clevedon, UK.

Hayes, H., Treiman, R., Kessler, B., 2006. Children use vowels to help them spell consonants. Journal of Experimental Child Psychology 94,

27e42.Hu, G., 2007. The juggernaut of Chinese-English bilingual education. In: Feng, A.W. (Ed.), Bilingual Education in China: Practices, Policies and

Concepts. Multilingual Matters, Clevedon, UK, pp. 94e126.

Jia, G., 2003. The acquisition of the English plural morpheme by native Mandarin Chinese-speaking children. Journal of Speech, Language, and

Hearing Research 46, 1297e1311.Jiang, N., 2004. Morphological insensitivity in second language processing. Applied Psycholinguistics 25, 603e634.

Joshi, R.M., Hoien, I., Feng, X., Chengappa, R., Boulware-Gooden, R., 2006. Learning to spell by ear and by eye: a cross-linguistic comparison.

In: Joshi, R.M., Aaron, P.G. (Eds.), Handbook of Orthography and Literacy. Erlbaum, Mahwah, NJ, pp. 569e577.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 16 (6), 42e43.

Kachru, B., 1997. World Englishes 2000: resources for research and teaching. World Englishes 14, 209e251.

Kachru, B., 1983. The Other Tongue: English Across Cultures. Pergamon Press, Oxford.

Kessler, B., Treiman, R., 2003. Is English spelling chaotic? Misconceptions concerning its irregularity. Reading Psychology 24, 267e289.

Lesaux, N.K., Rupp, A.A., Siegel, L.S., 2007. Growth in reading skills of children from diverse linguistic backgrounds: findings from a 5-year

longitudinal study. Journal of Educational Psychology 99, 821e834.

Lyster, S.-A.H., 2002. The effects of morphological versus phonological awareness training in kindergarten on reading development. Reading and

Writing: An Interdisciplinary Journal 15, 261e294.

MacWhinney, B., Snow, C.E., 1985. The child language data exchange system. Journal of Child Language 12, 271e295.

Marcus, G., 1995. Children’s overregularization of English plurals: a quantitative analysis. Journal of Child Language 22, 447e460.

Marcus, G., Pinker, S., Ullman, M., Hollander, M., Rosen, J., Xu, F., 1992. Overregularization in language acquisition. Monographs of the Society

for Research in Child Development 57 (4), 1e182.

McBride-Chang, C., Tardif, T., Cho, J., Shu, H., Fletcher, P., Stokes, S., Wong, A., Leung, K., 2008. What’s in a word? Morphological awareness

and vocabulary knowledge in three languages. Applied Psycholinguistics 29, 437e462.

Page 12: One dress, two dress: Dialectal influence on spelling of English words among kindergarten children in Singapore

225L.Q. Dixon et al. / System 40 (2012) 214e225

Muthen, L., Muthen, B., 2007. Mplus User’s Guide, fifth ed. Muthen & Muthen, Los Angeles, CA.

Pakir, A., 1991. The status of English and the question of ‘standard’ in Singapore: a sociolinguistic perspective. In: Tickoo, M.L. (Ed.), Language

and Standards: Issues, Attitudes, Case Studies. SEAMEO Regional Language Centre, Singapore, pp. 109e130.Reed, D.K., 2008. A synthesis of morphology interventions and effects on reading outcomes for students in grades K-12. Learning Disabilities

Research & Practice 23, 36e49.

Ross, S., 1971. On the syntax of written Black English. TESOL Quarterly 5, 115e122.

Rubdy, R., 2007. Singlish in the school: an impediment or a resource? Journal of Multilingual & Multicultural Development 28, 308e324.Sebba, M., 2007. Spelling and Society: The Culture and Politics of Orthography Around the World. Cambridge University Press, Cambridge, UK.

Sulzby, E., 1996. Roles of oral and written language as children approach conventional literacy. In: Pontecorvo, C., Orsolini, M., Burge, B.,

Resnick, R. (Eds.), Early Text Construction in Children. Erlbaum, Hillsdale, NJ, pp. 25e46.Tan, C., 2005. English or Singlish? The syntactic influences of Chinese and Malay on the learning of English in Singapore. Journal of Language

and Learning 3, 156e179.

Tong, X., McBride-Chang, C., Shu, H., Wong, A., 2009. Morphological awareness, orthographic knowledge, and spelling errors: keys to

understanding early Chinese literacy acquisition. Scientific Studies of Reading 13, 426e452.Totereau, C., Thevenin, M., Fayol, M., 1997. The development of the understanding of number morphology in written French. In: Perfetti, C.A.,

Rieben, L., Fayol, M. (Eds.), Learning to Spell: Research, Theory, and Practice across Languages. Erlbaum, Mahwah, NJ, pp. 97e114.

Treiman, R., 1993. Beginning to Spell: A Study of First-grade Children. Oxford University Press, New York.

Treiman, R., 2004. Spelling and dialect: comparisons between speakers of African American Vernacular English and White speakers. Psycho-

nomic Bulletin & Review 11, 338e342.

Treiman, R., Barry, C., 2000. Dialect and authography: some differences between American and British spellers. Journal of Experimental

Psychology: Learning, Memory, and Cognition 26, 1423e1430.Treiman, R., Cassar, M., Zukowski, A., 1994. What types of linguistic information do children use in spelling? The case of flaps. Child

Development 65, 1318e1337.

Treiman, R., Goswami, U., Tincoff, R., Leevers, H., 1997. Effects of dialect on American and British children’s spelling. Child Development 68,

229e245.Wang, M., Ko, I., Choi, J., 2009. The importance of morphological awareness in Korean-English biliteracy acquisition. Contemporary Educational

Psychology 34, 132e142.

Wee, D., 2009. Singapore language enhancer: identity included. Language and Intercultural Communication 9, 15e23.

Woodcock, R., 1991. Woodcock Language Proficiency Battery-Revised. Riverside Publishing, Itasca, IL.

Yano, Y., 2001. World Englishes in 2000 and beyond. World Englishes 20, 119e132.

Yeong, S., Rickard Liow, S., 2010. Phonemic representation and early spelling errors in bilingual children. Scientific Studies of Reading 14,

387e406.