Korean–English Biliteracy Acquisition: Cross-Language Phonological and Orthographic Transfer

Min Wang, Yoonjung Park, and Kyoung Rang Lee University of Maryland

Cross-language phonological and orthographic relationship in the biliteracy acquisition of children learning to read Korean and English was investigated in this study. Forty-five Korean–English bilingual children were tested in first-language (L1; Korean) and 2nd-language (L2; English) reading skills focusing on 2 reading processes—phonological and orthographic processing. The authors found that phonological skills in L1 and L2 were strongly correlated, and Korean phonological skills explained a unique amount of variance in English pseudoword reading beyond English phonological and orthographic skills. However, there was limited orthographic skill transfer between the 2 systems. Results are discussed within the framework of universal phonological processes in learning to read. The authors conclude that bilingual reading acquisition may be a joint function of general phonological processes and orthographic-specific skills.

Keywords: Korean, English, biliteracy, cross-language transfer

Recent research on biliteracy development has yielded consis- and orthography transfer in two alphabetic systems that share a tent results for a strong facilitation between first-language (L1) mapping principle but not a visual form. In this study, we focused reading skills and second-language (L2) reading skills. However, on two basic reading components: phonological processing and these studies have only focused on children learning to read orthographic processing. These two variables have been shown in English whose L1 is another Roman alphabetic system, such as the literature to be critical for learning to read in all languages, Spanish, French, or Italian (e.g., Comeau, Cormier, Grandmaison, whether alphabetic or nonalphabetic. We were interested in ad- & Lacroix, 1999; D’Angiulli, Siegel, & Serra, 2001; Durgunoglu, dressing the issue of how cognitive processes of phonology and Nagy, & Hancin-Bhatt, 1993). The present study seeks to inves- orthography contribute differentially to L1 and L2. tigate such biliteracy development for children learning to read English whose L1 is a non-Roman alphabetic system, that is, The Roles of Phonological and Orthographic Processing Korean. Korean and English share a fundamental alphabetic prin- in Learning to Read Alphabetic Languages ciple; graphemes in both languages correspond to phonemes. How- ever, Korean Hangul has a unique visual and spatial configuration For alphabetic languages, a large volume of research has re- compared with other orthographic systems. Korean has a nonlinear vealed that early phonological processing skills in general predict spatial layout, just like Chinese. Korean–English biliteracy acqui- later reading success (e.g., Goswami & Bryant, 1990; Lundberg, sition provides an excellent opportunity to study cross-language Olofsson, & Wall, 1980; Mann & Liberman, 1984; Rack, Snowl- ing, & Olson, 1992; Shankweiler, 1999; Torgesen, 1999). Phono- logical processing skills generally refer to the ability to analyze and manipulate sound units of spoken language. According to Min Wang and Yoonjung Park, Department of Human Development, Treiman and her colleagues’ linguistic structure hypothesis (e.g., University of Maryland; Kyoung Rang Lee, Department of Curriculum and Bruck, Treiman, & Caravolas, 1995; Treiman, 1995; Treiman, Instruction, University of Maryland. The research reported here was supported by a National Academy of Mullennix, Bijeljac-Babic, & Richmond-Welty, 1995), the syllable Education/Spencer Fellowship and a Summer Research Award from the is at the top of the hierarchical structure—it is the largest and most General Research Board of the University of Maryland awarded to Min accessible unit. The phoneme is at the bottom of the hierarchical Wang. Preparation of this article was supported by a National Institutes of structure—it is the smallest unit and is a later developing one for Health/National Institute of Child Health and Human Development Grant children. Between syllables and phonemes lie intermediate onset 1R01HD048438-01 awarded to Min Wang. We thank all the teachers and and rime units. Studies demonstrate the importance of processing parents in the Korean Language School in Ellicott City, Maryland for their both large and small phonological units for reading skills. For warm support and the children for their active participation. We also thank example, processing of relatively large phonological units in tasks Kirsten VanMeenen, Aleck Chen, and Stella Chen for their assistance with such as rhyme and alliteration has been shown to be important for collecting and scoring the data. Thanks also go to Charles A. Perfetti, promoting young children’s learning to read (e.g., Bradley & Catherine McBride-Chang, Che Kan Leong, Jennifer Cromley, and Donald J. Bolger for their valuable comments on an earlier version of this article. Bryant, 1983; Bryant, MacLean, Bradley, & Crossland, 1990; Correspondence concerning this article should be addressed to Min Goswami & Bryant, 1990; see also Bryant, 2002, for a review). It Wang, Department of Human Development, 3304P Benjamin Building, has been suggested in other studies that children’s skills at pro- University of Maryland, College Park, MD 20742. E-mail: cessing the smallest phonological units—phonemes—are powerful [email protected] predictors of individual differences in learning to read and that

148 KOREAN–ENGLISH BILITERACY ACQUISITION 149 training children in phonemic-level skills can benefit their later Korean Language and Hangul Orthography reading progress (e.g., Byrne & Fielding-Barnsley, 1995; Hulme et al., 2002; Lundberg, Frost, & Petersen, 1988; Muter, Hulme, The Korean language has 14 basic consonants (e.g., , , , ,/k, Snowling, & Taylor, 1998). n, t, l /) and 10 basic vowels (e.g., , , , /a, ja, o, jo /). Some Orthographic knowledge has also been shown to facilitate chil- Korean phonemes are the same as English phonemes (e.g., /m/ and dren’s reading acquisition. Orthographic knowledge generally re- /n/). However, Korean does not have the two English “th” sounds, fers to “children’s understanding of the conventions used in the as in this and think; there is also no /f/ or /v/ sound. Just like writing system of their language” (Treiman & Cassar, 1997). One Chinese and Japanese speakers, Koreans have difficulty distin- important orthographic processing skill is children’s ability to guishing /l/ and /r/ sounds in English. The syllable is an important detect acceptable and unacceptable letter sequences and their re- unit in Korean, but its syllable structure is simpler compared with lation to letter positions in words (Cassar & Treiman, 1997; English. Korean syllables can be in the form of CV (e.g., /ga/), Treiman, 1993). For example, Cassar and Treiman (1997) found CVC (e.g., /muk/), and CVCC (e.g., /dak/) structures. Please that by late kindergarten, children have acquired some knowledge note that the first consonant in a final consonant cluster in Korean of the acceptable form and position of consonant doublets. These CVCC words is usually not pronounced. There are no initial young children preferred spellings with final doublets (e.g., baff)to consonant clusters in Korean syllables; final consonant clusters are those with beginning doublets (e.g., bbaf). limited. This simpler syllable structure may pose less of a These young children also preferred spellings with acceptable phonemic-level phonological processing challenge for learning to doublets (e.g., yill) to those with unacceptable ones (e.g., yihh). read Korean in comparison to learning to read English. Some researchers have argued that phonological and orthographic The nature of the Korean writing system, Hangul, is alphabetic; knowledge mutually facilitate each other and that grapheme- it maps letters onto phonemes just as English, Russian, and Italian phoneme knowledge provides young readers with a powerful tool do. One important feature of Hangul is that it is nonlinear: The to bind the spelling patterns of individual and multiple letters with composition of letters is shaped into a squarelike syllable block. their pronunciations in words (e.g., Ehri, 1991, 1998). Recent The symbols are arranged from left to right and from top to empirical research suggests that this orthographic knowledge may bottom. Each Hangul syllable is built of 2–4 symbols that in contribute significantly to word recognition skill in children over various combinations represent each of the 24 phonemes. Its and above phonological factors (e.g., Cunningham, Perry, & overall shape makes Korean appear more similar to Chinese than Stanovich, 2001; Cunningham & Stanovich, 1990). to its fellow alphabetic orthographies. Each Hangul letter contains It has been well documented that the transparency of grapheme- between one and eight strokes, compared with English letters that phoneme mappings, or orthographic depth (Frost, 1994; Katz & contain between one and four strokes. The simplest syllable block Frost, 1992), affects children’s progress in learning to read (e.g., contains two strokes (e.g. /i/), and the most complex block has 13 Cossu, Shankweiler, Liberman, Tola, & Katz, 1988; Durgunoglu strokes (e.g., /bbal/). Taylor and Taylor (1995) pointed out that & Oney, 1999; Frith, Wimmer, & Landerl, 1998; Geva & Siegel, the uneven visual complexity of Hangul syllable blocks should 2000; Goswami, Gombert, & Barrera, 1998; Shimron, 1999). help recognition because the more varied the visual shapes of These differences have been observed in various languages— graphs, the more easily they are distinguished from one another. English versus German, Spanish, and Italian. For example, readers There are some letter position constraints for Hangul syllables of a shallow orthography (e.g., Italian) demonstrate an advantage (Simpson & Kang, 2004). A horizontal vowel is always placed in phonological awareness over readers of English, a less trans- under an initial consonant (e.g., , /so/, “cow”). A vertical vowel parent orthography (Cossu et al., 1988). In Italian, there are direct is always placed to the immediate right of the initial consonant letter-sound correspondences, whereas in English, the letter-sound (e.g., /na/, I). A written syllable always starts with a consonant correspondences are not one-to-one. For example, the letter a in (or a null consonant that does not have a pronunciation). There are English words can be pronounced in many different ways, as in also constraints on final consonant clusters and simple vowel about, fall, cake, says, and so forth. Furthermore, Goswami and combinations; certain combinations of two consonants will lead to colleagues found that children who learn to read English benefit illegal final clusters. Likewise, certain combinations of two simple more from processing large orthographic units, such as rimes, than vowels will lead to illegal Hangul syllables. children who learn to read Spanish, a highly transparent orthog- Because the Hangul syllable blocks are separated, there is a raphy. Frith and colleagues reported a similar result when com- clear syllable boundary for a Hangul word (e.g., , /ha sim paring English with German learners. Neuroimaging studies of ni ka/, hello). As a result, Korean Hangul is sometimes more adult reading suggest that not only can reading strategies be accurately called an alphabetic syllabary. It is noteworthy that the affected by the orthography but also that these strategy differences syllabary aspect of Hangul is different from Japanese Kana, which become manifest in the brain structures that are activated during is a true syllabary system. The Japanese Kana syllables cannot be word reading (Paulesu et al., 2000). further segmented into consonants and vowels; therefore, memo- In the present study, we examined phonological and ortho- rization of the existent Kana syllables is necessary for beginning graphic skills in a group of Korean–English bilingual children and learners. Hangul learners, however, do not need to memorize the analyzed their contribution to basic reading skills in both L1 and syllable blocks; instead, they need to learn the alphabetic symbols L2. We did this by designing parallel experiments that tap phono- and then the simple and reliable rules to package them. Hence, logical and orthographic skills in the two languages. In the fol- Hangul is a shallow orthography. Finally, although Chinese char- lowing section, we provide brief descriptions of the Korean lan- acters are used in conjunction with Hangul in South Korea, their guage and orthography in order to facilitate understanding of the use is relatively rare because they are not taught in primary Korean tasks. schools. 150 WANG, PARK, AND LEE

Phonology is important in reading Korean Hangul, just like processing, word reading, and spelling in a group of Spanish– other alphabetic languages. Research has shown that reading English bilingual children, many of whom had English reading Hangul words involves prelexical phonological information pro- difficulties. More recently, this strong facilitation from L1 to L2 cessing and that prelexical phonology is activated rapidly and phonology was shown in research in Canada on English-speaking automatically even in semantic processing of Korean words (e.g., children learning to read French (e.g., Comeau et al., 1999), Italian Kim, 1999; Yoon, Bolger, & Perfetti, 2002). Cho and Chen (1999) (e.g., D’Angiulli et al., 2001), and Hebrew (Geva & Siegel, 2000). found that Korean readers demonstrated strong homophone inter- For example, D’ Angiulli and colleagues found strong correlations ference effects for pseudohomophones in a category judgment in skills on various reading and spelling tasks between English and task. That is, phonological similarity of the stimuli (e.g., sute for Italian. The authors argued that exposure to a shallow orthography, the category “an article of clothing”) caused greater confusion for such as Italian, which involves relatively direct letter-sound cor- Hangul readers when making meaning judgments compared with respondences, may promote phonological skills in learning to read control items (e.g., blane for the same category). Although the English. However, it is not certain whether a similar pattern of subjects also made more errors on visually similar trials than on cross-language relationship exists across two different ortho- controls, the effect size for homophone interference was much graphic systems when they share similar mapping principles but stronger than that for visual similarity. In another study suggesting differ in visual forms. the importance of phonology in reading Hangul, Kim and Davis The aforementioned studies mainly focused on cross-language (2004) found that Korean poor readers performed poorly on Ko- transfer at the phonological level. These studies have successfully rean phonological awareness measures when compared with good demonstrated a robust and universal cross-language phonological readers. In contrast to studies on phonological processing, little transfer phenomenon for various alphabetic systems. However, research has been conducted regarding the role of orthographic little is known about any possible transfer at the orthographic level, processing skills in learning to read Korean Hangul. In this study, in particular, for those orthographic systems that differ in visual we examined the contribution of both phonological and ortho- forms. Because of the alphabetic nature of the Korean Hangul graphic skills to learning to read Korean and their relationship to system, we hypothesize that there will be a close relationship learning to read English among a group of bilingual children. The between Korean and English phonological processing. We predict unique feature of the Korean Hangul orthographic system forms an that onset, rhyme, and phonemic processing will be highly corre- excellent comparison with other Roman alphabetic systems in lated in the two languages and that phonological processing skills studying bilingual literacy acquisition. will predict word-reading skills in both languages. We also hypothesize that phonological skills in Korean L1 will contribute a Bilingual and Biliteracy Acquisition Within Alphabetic significant amount of unique variance to English word identifica- Languages tion even after English phonological and orthographic skills are taken into consideration. With respect to orthographic processing Cummins (1979, 1986, 1991) proposed the linguistic interde- skills in the two languages, there may be two possible results. One pendence hypothesis. This hypothesis suggests that once the child is that orthographic skills may not contribute a unique amount of develops reading skills in L1, he or she is able to transfer those variance to reading in each of the two languages over and above skills to L2. Recent research on biliteracy acquisition across two phonological skills, given the fundamental phonological skills alphabetic languages and orthographic systems, such as Spanish– required to read alphabetic languages. The other possibility is that English and French–English, has yielded two major findings. First, orthographic skills are also a critical factor in acquiring alphabetic there is clearly a close phonological relationship between two languages and that orthographic skills will explain a unique alphabetic languages. Phonological skills in one language are amount of variance in reading both Korean Hangul and English highly correlated with phonological skills in the other language. after taking into account phonological skills in the two languages. Second, phonological skills in one language contribute to word- However, because of the nature of the nonlinear orthography of the reading skills in the other language. For example, first-grade Korean Hangul system, in contrast to the linear orthographic Spanish-speaking children who were enrolled in a transitional system of English, orthographic skills in the two systems may not bilingual education program on both Spanish and English reading be significantly related to each other. We also predict limited skills were tested in a study by Durgunoglu et al. (1993). Their orthographic transfer in reading across these two languages. For results demonstrated that children who could perform well on within-language phonological skills, we hypothesize that the per- Spanish phonological awareness tasks were more likely to be able formance difference between onset rhyme and phonemic-level to read English words and pseudowords than were children who tasks will be greater for learning English compared with learning performed poorly on these tasks. The phonological awareness Korean. This is because of the simpler syllabic structure of the tasks included different linguistic units (onset-rime and phoneme) Korean language compared with English. in Spanish words. Moreover, phonological awareness was a significant predictor of performance on word recognition tests both Method within and across languages. Cisero and Royer (1995) used longitudinal data to demonstrate this cross-language phonological Participants transfer in Spanish–English bilingual children. Their results Forty-five Korean-speaking children were recruited from one Korean showed that phoneme detection in Spanish at Time 1 contributed language school in the Washington, DC area. Most of the parents were to phoneme detection in English at Time 2. Ganschow and Sparks first-generation Korean immigrants and spoke Korean at home. There were (1995) further showed that explicit, direct training in Spanish 24 boys and 21 girls. They simultaneously attended first-grade (n ϭ 16, phonology resulted in significant gains in English phonological mean age ϭ 6.71 years), second-grade (n ϭ 14, mean age ϭ 7.84), and KOREAN–ENGLISH BILITERACY ACQUISITION 151 third-grade (n ϭ 15, mean age ϭ 8.74) English classes in public schools has been found to be the most difficult for young children (e.g., Stanovich and Level 1 (n ϭ 22, mean age ϭ 7.05), Level 2 (n ϭ 12, mean ageϭ 8.34), et al., 1984). Nonwords were used in this task to control for the effect of and Level 3 (n ϭ 11, mean age ϭ 8.46) Korean classes at the Korean lexicality on children’s performance. Children were encouraged to play a school. word game with the experimenter. A female native English speaker re- The Korean language school provided a unique opportunity to study corded the audio stimuli, which were presented via a CD player. The child Korean-English biliteracy acquisition. This school enrolled preschool to heard a word first and was asked to repeat the word. Then the child was Grade 9 children. The school was open on Friday evenings, with 2 hr asked to remove a sound in the word. For example, the child was instructed devoted to Korean literacy instruction. There were also other music and as follows: “Say mab. Now say it again, but don’t say /b/.” Twenty items sports activities, which were taught in Korean. The children attended were included in this task, similar to those used by Wade-Woolley (1999) public American schools in the area on weekdays. All of the children were and Gottardo, Yan, Siegel, and Wade-Woolley (2001). The position of the fluent English speakers; there were no reports from their English school phoneme to be deleted from the word was varied in order to test the teachers concerning English proficiency problems. The teachers in the difficulty level associated with phoneme positions. There were two items Korean school were experienced Korean language teachers. each for beginning and ending consonants. There were four items each for the first phoneme of a beginning consonant cluster, the second phoneme of Demographic and Family Literacy Survey a beginning consonant cluster, the first phoneme of a final consonant cluster, and the second phoneme of a final consonant cluster. Children’s Parents of the children were asked to fill out a short questionnaire with responses were recorded via a digital voice recorder. Children’s responses basic demographic information in addition to family language and literacy were scored as correct if the target phoneme was deleted accurately. experiences. Thirty-three children were born in the United States, and 12 Cronbach’s alpha internal consistency reliability was .77 for this task. were born in South Korea. All but 5 children learned Korean as their first Orthographic choice task. The child was presented with a pair of language. Forty-one of the children currently spoke both Korean and nonwords on a card one at a time. The child was instructed to point to the English at home; 4 spoke only Korean. About half the parents (20) spoke one that looked more like a real word. The stimuli set included 28 items, both Korean and English at home, and half spoke Korean only (23) at similar to those used by Treiman (1993) and Siegel, Share, and Geva home. The majority of these parents had a high school diploma or under- (1995). Pseudowords were used. The task tapped into children’s sensitivity graduate degree. The majority of the families engaged in Korean literacy to various orthographic patterns in English. For example, there is a legal activities at home during the week, such as reading Korean books and position of certain double consonants (e.g., ff does not occur at the practicing Korean Hangul writing. beginning of a word). Five examples were given and accuracy data were recorded. Cronbach’s alpha internal consistency reliability for this task was English Tasks .77. Real-word naming. The child was shown one word at a time on a card Experimental phonological tasks. Experimental tasks were designed and was instructed to say the word aloud as best as he or she could. The for testing English phonological skills. These included onset detection, materials contained 35 words from the Word Recognition subtest of the rhyme detection, and a phoneme manipulation task (i.e., phoneme Wide Range Achievement Test–Revised (WRAT-R; Jastak & Jastak, deletion). 1984). Five examples were given. Children’s naming responses were Onset-rhyme detection. This detection task was designed to tap into recorded via a digital voice recorder. Children’s responses were scored as children’s ability to manipulate and differentiate between phonological correct or incorrect, and only fully accurate pronunciations were accepted units in spoken English words. One-syllable nonwords were used in this as correct. task, ranging from 3 to 5 letters in length. Nonwords were used to exclude Pseudoword naming. The child was shown two items at a time on a any potential confound resulting from influence of lexical access of the card and was instructed to sound out the letter string aloud as best as he or items; 15 items were included for each condition. Materials were similar to she could. Forty items from the Word Attack subtest of the Woodcock those used by Bradley and Bryant (1983); Stanovich, Cunningham, and Reading Mastery Test—Revised (Woodcock, 1987) were administered. Cramer (1984); and Gottardo (2002). In the onset-detection condition, Five examples were given. Children’s naming responses were recorded via onsets for all of the items were single consonants. In the rhyme-detection a digital voice recorder. Children’s responses were scored as correct or condition, 4 of 15 items had consonant clusters as onsets. Native English incorrect, and only fully accurate pronunciations were accepted as correct. speakers recorded the materials via a digital voice recorder; the digital audio files were then saved as audio files on a CD. Korean Tasks Children were tested individually in a quiet room equipped with a CD player on a desk. The child was asked to listen carefully through a pair of Onset-rhyme detection task. Parallel to the English onset-rhyme detec- earphones connected to the CD player. The child heard three spoken words, tion task, two sets of 15 Korean pseudosyllables were created for the for example, bap, bep, and gonk, with a time interval of 5 s between the Korean onset and rhyme detection tasks, respectively. For example, in the three words, and an interval of 15 s between each item. This time interval rhyme detection task, the child was asked to choose one of the three words was sufficient to allow children to respond. In order to reduce the memory that did not have the same rhyme as the other two words, /bot/, /sap/, demands, the child was shown three number signs on the desk; the first /sot/. The procedure was the same as that for the English tasks. Cron- word corresponded to the “1” sign on the desk, the second word corre- bach’s alpha internal consistency reliability was .78 for both measures. sponded to the “2” sign, and the third word corresponded to the “3” sign. Phoneme deletion task. Parallel to the English phoneme deletion task, The child’s task was to choose which of these three words (a) did not start a set of 20 Korean pseudowords was created to assess children’s skill in with the same sound as the other two words in the onset-detection condi- manipulating phonemes in different word positions. However, because tion or (b) did not end with the same sound as the other words in the Korean syllables do not contain beginning consonant clusters, there were rhyme-detection condition. The child was instructed to point to one of the only items for deleting single beginning consonants. Also, because one of numbers on the desk to represent the one word that was different from the the two final consonants in Korean is normally silent, only single final other words. Three practice items were given for each condition. Accuracy consonants were used for the Korean phoneme deletion task. Furthermore, data were recorded. Using Cronbach’s alpha, the internal consistency because the final consonants in Korean are all voiceless, it is difficult to reliability of the onset and rhyme tests was .72 and .80, respectively. single out the final consonant in a word and pronounce it. We decided that Phoneme deletion task. Of the commonly used phonological process- instead of pronouncing the final sound in the word, we would label it as the ing tasks (e.g., blending, segmenting, rhyming, oddity), phoneme deletion final sound. For example, the child was asked to say /sub/ and then say 152 WANG, PARK, AND LEE it again without pronouncing the final sound / / (/b/). The correct answer a missing piece in it and asked to choose one from five options to complete would be (/su/). There were 10 such items, each targeting the beginning the pattern. The items require the child to detect the shapes and directions or final sound of a Korean syllable. The internal consistency reliability in an abstract diagram to decide what option fits the pattern. The child was (Cronbach’s alpha) for this task was .85. Because of the simpler syllable allowed either to point or to say aloud the number of his or her answer. structure of Korean, children were expected to show less difficulty with this task in relation to onset and rhyme tasks compared with English. Procedure Orthographic choice task. There were 28 items in total, targeting four types of constraints in the Hangul orthographic system. First, Korean Children were tested in two sessions. One session was devoted to Hangul is a nonlinear orthographic system, and there are some constraints English tasks, which were administered by a trained graduate student, a on placing vowels in a square-shape form. As described earlier, a horizon- native English speaker. The other session was devoted to Korean tasks, tal vowel always should be placed under the initial consonant, whereas a which were administered by a trained graduate assistant, a native Korean vertical vowel always should be placed to the immediate right of the initial speaker. The order of the two sessions was randomized among the children. consonant. Six items were designed to measure children’s sensitivity to the The order of the tasks within the sessions was also randomized for each legality of the vowel position, for example, (illegal) versus (legal), child. Each session took about 30 min, and children were given a break in and (illegal) versus (legal). Second, six items were presented to see the middle of each testing session. Children were given school-related gifts whether children could identify the legality of the combination of simple for participation at the end of each testing session. vowels, for example, (illegal) versus (legal). Third, three items were designed to test children’s knowledge about the required initial consonants Results in Hangul syllables, for example, (illegal) versus (legal). Finally, 13 items were designed to measure children’s ability to detect the legality of Means and standard deviations for the English and Korean tasks final consonant clusters, for example, (illegal) versus (legal). There by grade are shown in Table 1. Overall, English language and were four practice items at the beginning of the testing. The internal reading skills were significantly improved from Grade 1 to Grade consistency reliability for this task was .66 (Cronbach’s alpha). 3, for all of the tasks (F Ͼ 5, all ps Ͻ .05). For the Korean tasks, Real-word and pseudoword reading. There were 36 items each for the real-word and pseudoword reading tasks. Each reading task consisted of 12 children’s performance was improved significantly for the ortho- ϭ Ͻ one-syllable, 12 two-syllable, and 12 three-syllable words. We selected the graphic task and phoneme deletion task, F(2, 42) 9.37, 3.30, p real-word items from the Korean language school curriculum. There were .001 and p Ͻ .05, respectively. Children’s performance on the three levels of familiarity in the real-word naming task: high, medium, and other Korean tasks still showed improvement across age; however, low. Level of familiarity was rated by the language school teachers. Three the changes did not reach significance. To increase the number of teachers together discussed and reached agreement for each item’s level of participants, we pooled data from the three grades together and familiarity. All of the items in the pseudoword reading task were pro- focused on the cross-language transfer in the subsequent correla- nounceable Hangul nonwords by replacing phonemes in real words that tion and regression analyses. were not used in the real-word naming task. Of the 12 two-syllable and 12 three-syllable nonwords, half of the items in each subset contained 1 pronounceable non-Hangul syllable, and the remaining half contained 2 Correlations Among All the Variables non-Hangul syllables. There were five practice items for each naming task. Correlations among all of the Korean and English tasks, after Internal consistency reliability (Cronbach’s alpha) was .96 and .97 for the real-word and pseudoword tests, respectively. partialing out children’s age, are shown in Table 2. Several im- Nonverbal ability test. The Pattern Completion subtest of the Matrix portant observations were obtained: English rhyme detection, En- Analogy Test (MAT-expanded form; Naglieri, 1985) was used to test glish phoneme deletion, and English orthographic skills were children’s nonverbal ability. This measure includes 14 abstract designs of strongly correlated with English real-word and pseudoword read- the standard progressive matrix type. The child was shown a picture with ing (all rs Ͼ.4, p Ͻ .01). English onset detection, rhyme detection,

Table 1 Means and Standard Deviations of Percentages of Accuracies for All Chinese and English Tasks by Grade

Grade 1 Grade 2 Grade 3 Full sample

MSDMSDMSDMSD

English tasks Matrix Analogy Test 53.85 17.16 75.27 16.91 87.50 7.54 71.80 20.06 Onset detection 64.17 18.99 74.29 20.70 83.11 14.66 73.63 19.59 Rhyme detection 67.08 24.03 82.38 10.33 92.44 12.31 80.30 19.80 Phoneme deletion 27.33 19.07 41.43 13.93 53.21 20.90 40.35 20.80 Orthographic choice 71.65 14.19 87.24 6.38 91.90 5.31 83.25 13.01 Real-word naming 47.97 16.28 78.39 8.97 78.39 16.50 67.72 20.52 Pseudoword naming 35.54 16.09 64.49 11.02 76.76 15.11 58.29 22.67 Korean tasks Onset detection 56.67 24.23 66.11 17.16 73.33 25.65 63.26 23.51 Rhyme detection 63.03 20.95 72.78 11.18 79.39 20.18 69.63 22.27 Phoneme deletion 48.64 28.92 70.00 18.95 71.82 27.39 60.00 30.54 Orthographic choice 70.94 12.34 81.85 6.36 85.39 7.22 77.38 11.70 Real-word naming 62.88 31.25 76.39 18.97 79.29 34.75 70.49 29.80 Pseudoword naming 54.67 33.27 72.22 27.09 78.28 36.02 65.12 33.44 KOREAN–ENGLISH BILITERACY ACQUISITION 153

Table 2 Correlations Among English and Korean Variables Controlling for Age

Variable 1 2345678910111213

English tasks 1. Matrix Analogy Test — 2. Onset detection .11 — 3. Rhyme detection .28 .60** — 4. Phoneme deletion .10 .41* .46** — 5. Orthographic choice .21 .34* .36* .43** — 6. Real-word naming .35* .27 .48** .43** .71** — 7. Pseudoword naming .35* .35* .49** .61** .69** .80** —

Korean tasks

8. Onset detection .36* .60** .47** .41* .60** .52** .63** — 9. Rhyme detection .32 .63** .74** .60** .32 .50** .59** .52** — 10. Phoneme deletion .11 .34* .30 .43* .28 .41* .61** .31 .55** — 11. Orthographic choice .21 .21 .38* .32 .28 .28 .39* .19 .39* .18 — 12. Real-word naming .13 .33 .36* .29 .46** .35* .44** .19 .32 .37* .64** — 13. Pseudoword naming .05 .38* .40* .29 .41* .32 .43* .18 .25 .40* .54** .90** —

* p Ͻ .05. ** p Ͻ .01.

and phoneme deletion skills were closely correlated with Korean skill was also entered when English-reading measures were the onset detection, rhyme detection, and phoneme deletion skills (r ϭ dependent variables; however, it was not entered in predicting .60, .74, and .43, respectively; all ps Ͻ .05). Korean onset detec- Korean-reading skills because of the nonsignificant correlations tion, rhyme detection, and phoneme deletion were highly corre- shown between these variables. lated with English real-word and pseudoword reading (all rs Ͼ .41, For predicting English word reading, the order of entry was age, ps Ͻ .05). Korean phoneme deletion and orthographic skills were MAT, English phonological processing, English orthographic pro- strongly correlated with Korean real-word and pseudoword read- cessing, Korean phonological processing, and Korean ortho- ing (all rs Ͼ .37, p Ͻ .05). Real-word and pseudoword reading graphic processing. Korean tasks were entered after the English were significantly correlated with each other in both orthographies tasks to examine the unique variance explained by Korean lan- (r ϭ .80 and .90, respectively; both ps Ͻ .01). Korean orthographic guage and orthographic tasks over and above the English tasks. and English orthographic processing skill were not significantly Orthographic tasks were entered after phonological tasks in order correlated (r ϭ .28, p Ͼ .05). We also found that the English onset to explore the unique variance explained by orthographic tasks and rhyme detection tasks tended to be more highly correlated with after considering phonological tasks. In order to reduce the number Korean Hangul reading than did Korean onset and rhyme detection of predictors in the regressions, we created English and Korean tasks. For example, the correlation between English onset detec- phonological composite scores using principal-component scores tion and Hangul pseudoword reading was .38, whereas the corre- for the English and Korean phonological measures. For the English lation between Korean onset detection and Hangul pseudoword phonological composite, the first principal component had an reading was .18. eigenvalue of 2.15 and explained 71.77% of the variance in the three phonological variables. Loadings on the first principal com- Cross-Language Phonology and Orthography Prediction ponent were large (onset: .86; rime: .86; phoneme: .82). Principal- In this section, we focused our analyses on cross-language and component scores were then calculated from these loadings. These orthography prediction from Korean to English reading and from principal-component scores were then used in the regression anal- English to Korean reading. We were interested in whether phono- yses. For the Korean phonological composite, the first principal logical and orthographic skills could be transferred from one component had an eigenvalue of 2.35 and explained 78.17% of the language to the other. Using a set of hierarchical regression anal- variance in the three phonological variables. Loadings on the first yses, we first examined the variables that predict English word principal component were large (onset: .85; rhyme: .93; phoneme: reading using both English and Korean tasks. Our goal was to .88). Principal-component scores were then calculated from these determine whether Korean tasks explained a significant amount of loadings. For predicting Korean word reading, the order of entry variance in English word reading after English tasks were taken was age, Korean phonological processing, Korean orthographic into consideration. Second, we examined variables that predict processing, English phonological processing, and English ortho- Korean word reading using both English and Korean tasks. Similar graphic processing. to our analyses of English word reading, we aimed to examine the Results of the hierarchical regressions on English reading are unique variance in Korean word reading accounted for by English shown in Table 3, and those for Korean reading are in Table 4. Age language and orthographic skills. In both sets of analyses, age was clearly contributed a significant amount of variance to learning to always entered first to control for its effect when considering read both English and Korean when entered first. The size of this cross-language and orthography prediction. Nonverbal (MAT) effect was larger for learning to read English than for reading 154 WANG, PARK, AND LEE

Table 3 respectively. Therefore, the present sample size of 45 was appro- Hierarchical Regression Analyses Predicting English Word priate for predicting English reading skills but relatively small for Reading Using English and Korean Tasks predicting Korean reading skills.

Mult. Mult. Variable R R2 ⌬R2 ⌬F Within-Language Analyses on Phonological Processing As we laid out in the introduction, there are different levels of Predicting English real-word reading Step 1 phonological processing skills. Phonemic-level processing is more Age .62 .38 .38 19.86*** fine-grained than onset- and rime-level processing. In order to get Step 2 a glimpse of bilingual children’s performance at the different MAT .68 .47 .08 4.89* levels of phonological processing, we compared their performance Step 3 English phonological processing .75 .56 .10 6.69* on the onset and rhyme tasks with the phoneme deletion task in Step 4 each language. We found that children were significantly poorer at English orthographic processing .85 .73 .16 17.54*** English phoneme deletion than at English onset and rhyme detec- Step 5 tion, t(44) ϭ 11.01, 13.97, respectively, both ps Ͻ .001. This Korean phonological processing .87 .75 .02 2.62 difference was much smaller for Korean, t(44) ϭ 0.84, p Ͼ .1, for Step 6 Korean orthographic processing .87 .75 .00 0.01 the difference between Korean onset detection and phoneme deletion and, t(44) ϭ 3.21, p Ͻ .01, for the difference between Predicting English pseudoword reading Korean rhyme detection and phoneme deletion. This was because of the simpler syllabic structure in Korean. Recall that the pho- Step 1 Age .71 .50 .50 32.02*** neme deletion task included five types of phonemes to be deleted. Step 2 We further analyzed children’s accuracy on the different phoneme MAT .76 .58 .08 5.57* positions. The pooled Grade 1–Grade 3 means and standard devi- Step 3 ations for the different levels of phonological processing are as English phonological processing .84 .71 .13 13.50** Step 4 follows, each with a maximum total score of four: (a) the begin- English orthographic processing .90 .81 .10 14.37** ning and ending sound deletion combined (M ϭ 2.28, SD ϭ 1.28); Step 5 (b) the first sound deletion in the beginning consonant cluster Korean phonological processing .93 .87 .06 12.94** (M ϭ 1.37, SD ϭ 1.31); (c) the second sound deletion in the Step 6 beginning consonant cluster (M ϭ 1.84, SD ϭ 1.36); (d) the first Korean orthographic processing .93 .87 .00 0.81

Note. Mult. ϭ Multiple; MAT ϭ Matrix Analogy Test. * p Ͻ .05. ** p Ͻ .01. *** p Ͻ .001. Table 4 Hierarchical Regression Analyses Predicting Korean Word Korean Hangul. Phonological skills in Korean predicted a signif- Reading Using English and Korean Tasks icant amount of unique variance for Korean word reading. This Mult. Mult. was true for both real words and pseudowords. Phonological skills Variable R R2 ⌬R2 ⌬F in English also predicted a significant amount of unique variance for English word reading, including real words and pseudowords. Predicting Korean real-word reading Orthographic processing skill in both English and Hangul orthog- Step 1 Age .36 .13 .13 5.71* raphy contributed a significant amount of unique variance to Step 2 reading both real and pseudowords in the same language. Korean phonological processing .60 .36 .23 13.46** More important, the Korean phonological skills further contrib- Step 3 uted a significant amount of unique variance to English Korean orthographic processing .73 .53 .18 14.08** Step 4 pseudoword reading even after English phonological and ortho- English phonological processing .73 .53 .00 0.01 graphic skills were taken into account. This cross-language pho- Step 5 nological transfer did not occur for reading real English words. English orthographic processing .73 .54 .00 0.20 There was no additional cross-language contribution for Korean Predicting Korean pseudoword reading word reading from English tasks once the Korean phonological and orthographic skills were taken into account. Step 1 To examine whether we had adequate participants for the afore- Age .44 .20 .20 9.72** mentioned regression analyses, we calculated appropriate sample Step 2 sizes, taking into account the effect sizes. Green (1991) suggested Korean phonological processing .60 .36 .16 9.77** Step 3 that in order to achieve an adequate case-to-independent variable Korean orthographic processing .68 .47 .10 7.06* ratio, the sample size for each predictor should be equal to or Step 4 greater than (8/F 2) ϩ (I Ϫ 1), where F 2 ϭ R2/(1 Ϫ R2), R2 is the English phonological processing .70 .49 .02 1.41 total multiple R2 in each regression, and I is the number of Step 5 English orthographic processing .70 .49 .00 0.17 predictors. The needed sample size for predicting English real- word reading, English pseudoword reading, Korean real-word Note. Mult. ϭ Multiple. reading, and Korean pseudoword reading was 46, 37, 54, and 61, * p Ͻ .05. ** p Ͻ .01. KOREAN–ENGLISH BILITERACY ACQUISITION 155 sound deletion in the ending consonant cluster (M ϭ 1.02, SD ϭ Chinese–English, in which Chinese is a nonalphabetic writing 0.94); and (e) the second sound deletion in the ending consonant system, suggest that there are common phonological processes cluster (M ϭ 1.56, SD ϭ 1.16). Paired sample t tests revealed that even in reading two different writing systems (Gottardo et al., children performed best on deleting word initial and final sounds 2001; Wang, Perfetti, & Liu, 2005). For example, Wang and compared with all other positions (all ps Ͻ .05). Children had the colleagues found that the Chinese contrastive phonological prop- most difficulty with the first sound of the ending consonant cluster, erty, tone, explained a modest but significant amount of unique compared with all other positions (all ps Ͻ .05), except the first variance in English pseudoword reading even after taking English sound of the beginning cluster. phonological tasks into consideration. Taken together, we can Korean phoneme deletion items were further grouped into two make a strong claim that bilingual reading builds on general categories: first consonant and final consonant deletion. The mean phonological processes of the two spoken languages. accuracies (and standard deviation) were 6.04 (2.91) and 5.96 It is also worth noting that this cross-language phonological (3.89), each with a maximum total score of 10. There was no transfer was weak in the direction of English to Korean. It appears significant difference between students’ performance at deleting that the influence from L2 to L1 was not as strong as that from L1 these two consonant types. to L2. We believe this is because learners can draw upon a well-established L1 phonological system in learning to read L2. Discussion However, weak L2 oral language proficiency limits such reliance for learners. Some bilingual researchers have also found this Our results add to those from other studies on bilingual and asymmetric link between L1 and L2 meaning representation. For biliteracy acquisition to support cross-language phonological example, Kroll and Stewart (1994) found that there is shared transfer across various alphabetic languages. Korean Hangul and conceptual understanding of L1 and L2 vocabularies; however, the English orthographic systems share a fundamental alphabetic prin- connection from L1 to the shared concept is much stronger than ciple. Learning to read Hangul and English both involve mapping that from L2 to a shared concept. graphemes to phonemes. Our study successfully demonstrated that We found limited orthographic transfer in learning to read two for two alphabetic languages, phonological skills are highly cor- different orthographies. Korean and English orthographic skills related with each other, and this finding suggests that better pho- were not significantly correlated with each other. Korean ortho- nological skills in one language will lead to better phonological graphic skill did not predict English word reading over and above skills in another language. Korean onset detection, rhyme detec- English phonological and orthographic skills. We suggest that this tion, and phoneme deletion are highly correlated with English reflects the differences in visual forms and possibly the ortho- onset detection, rhyme detection, and phoneme deletion. Korean graphic transparency between the two languages. English is a phonological skills, including onset detection, rhyme detection, linear orthographic system, whereas Korean is nonlinear. The and phoneme deletion, were also significantly correlated with visual-spatial configuration of Korean Hangul is in contrast with English real-word and pseudoword reading. English. Wang et al. (2005) found among a group of Chinese– More important, Korean phonological skills explained a unique English bilingual children that Chinese orthographic skills did not amount of variance in English pseudoword reading even after correlate significantly with English orthographic skills and did not English phonological and orthographic skills were taken into ac- predict English word reading. Korean and Chinese characters share count. This result suggests that L1 phonological skills facilitate L2 a nonlinear and squarelike structure. Recent neuroimaging work on pseudoword reading. This is not surprising because reading En- Chinese–English bilingual adults (e.g., Liu & Perfetti, 2003; Tan et glish pseudowords requires true alphabetic knowledge, in other al., 2001, 2003) has shown that reading Chinese resulted in more words, strong letter-sound correspondence skills, and phonological activation in some brain areas that are responsible for coordinating processing skills are essential for processing these nonwords. Real and integrating visual-spatial analyses of logographic Chinese words, however, can be read either via letter-phoneme mapping, an characters compared with reading English. We expect that reading assembly route, or a whole-word access lexical route, as was Korean Hangul words would involve similar activation in those described in a well-known dual-route reading model (e.g., Colt- brain areas relating to visual-spatial analyses of nonlinear Hangul heart, 1978; Coltheart, Curtis, Atkins, & Haller, 1993; Coltheart, words. Another explanation for this nonsignificant orthographic Rastle, Perry, Langdon, & Ziegler, 2001). Therefore, real-word transfer is that the Hangul system is a shallow orthography, in reading may not be as sensitive to the phonological processing which single letters directly map onto individual phonemes, and skills as is pseudoword reading. Our finding that L1 phonological one letter stands for only one sound. However, English is a deep skills contribute to L2 word reading complements and reinforces orthography; the relation between letters and sounds is opaque, and previous findings for bilingual groups of children such as Spanish– in many cases one letter can be pronounced as more than one English (e.g., Durgunoglu et al., 1993), French–English (e.g., sound. Therefore, not much help would be expected from a shal- Comeau et al., 1999), Italian–English (e.g., D’Angiulli et al., low orthographic experience for learning a deep orthography. 2001), and Portuguese–English children (e.g., DaFontoura & Sie- Both phonological and orthographic processing skills are im- gel, 1995) in which both languages are of a Roman alphabetic portant for learning to read English and Korean Hangul systems. nature. In the present study, we demonstrated cross-language pho- These findings are consistent with the literature on roles of phonological transfer across two different alphabetic orthographies nology and orthography in learning to read alphabetic languages that differ in visual forms as well as orthographic transparency. (e.g., Cunningham et al., 2001; Ehri, 1991, 1998; Goswami & Phonological processing is an important general process in learn- Bryant, 1990; Hulme et al., 2002; Lundberg et al., 1980). In the ing to read different alphabetic orthographies. Two recent studies present study, we found that orthographic skills in both Korean and on bilingual reading across different writing systems such as English not only explained unique variance in reading both real 156 WANG, PARK, AND LEE and pseudowords in the two orthographies but also predicted far Korean Hangul system requires children to be sensitive to both more variance than did phonological skills for English real-word syllabic and subsyllabic linguistic units. Their results are consis- reading. This finding indicates that orthographic learning plays an tent with our study on the prediction of subsyllabic processing important role for this group of bilingual children. The strong skills in reading Hangul. However, their results also point to the prediction from English orthographic skill to English real-word importance of syllable-level skills for successful Hangul reading. reading is not surprising. This echoes the claim made by some Cho and McBride-Chang argued that the high salience of syllables researchers that basic reading skill is highly related to building in Korean language as well as in Korean script relative to other orthographic knowledge. That is, one of the critical factors in word languages, such as English, is responsible for such a strong pre- identification skill is the development of strong orthographic rep- diction of Korean syllable-level skills for learning to read Hangul. resentations of lexical items (e.g., Perfetti, 1992; Share, 1995). A Future research in studying Korean–English biliteracy acquisition recent extended version of the orthographic depth hypothesis sug- could include both syllable-level and subsyllabic-level tasks so gests that reading in a shallow orthography, such as Italian and that a comparison can be made directly as to the contribution of German, may rely primarily on a single phonological processes, syllable-level skills in learning to read Hangul versus English in a whereas a deep orthography, such as English or Dutch, entails both group of bilingual children. One more limitation of our study is phonological and visual-orthographic processes (e.g., Seymour, that a measure of oral language proficiency in either Korean or Aro, & Erskine, 2003; Share, 2004). Our results for learning to English was not included. Previous research suggested that English read English from a group of bilingual children supported the language proficiency could influence word-reading performance in aforementioned hypothesis about deep orthography. Nevertheless, our result on learning to read the Korean Hangul system chal- English as a second language (e.g., Geva, Yaghoub-Zadeh, & lenged the hypothesis on the shallow orthography part. Share Schuster, 2000). Thus, care should be taken in future research to (2004) suggested that children who learn to read Hebrew, a shal- control for this factor. Various measures tapping the oral language low alphabetic orthography, were not sensitive to word-specific proficiency, such as vocabulary, grammatical knowledge, and orthographic detail until Grade 3 despite their high levels of discourse-related knowledge, can be used in cross-language trans- decoding accuracies. Korean Hangul is a perfectly shallow orthog- fer studies. raphy; however, Hangul orthographic processing skill contributed Finally, findings from our study have important educational a significant unique amount of variance to Hangul word reading implications. Classroom teachers need to pay attention to bilingual over and above phonological processing skill for a group of Grade children’s strong L1 language skills and be aware that these strong 1–Grade 3 children. We speculate that the unique visual form of first L1 language skills can be transferred to L2 learning. Further- the Hangul system may be responsible for the difference in the role more, teachers can make use of the children’s language skills in of orthographic skill between reading the Korean Hangul alpha- their L1 as a “facilitator” or “springboard” to develop their L2 betic system and other Roman alphabetic systems. We further language and literacy skills (see Durgunoglu, 2002, for this sug- suggest that it is necessary to take into account the visual form of gestion, p. 192). the alphabetic system when considering the degree of orthographic In the case of bilingual children whose L1 and L2 are both transparency among languages in the world. alphabetic languages, phonological processing skills transfer from The poorer performance in processing English phonemic infor- L1 to L2 could be fully used in the classroom. Durgunoglu (2002) mation compared with English onset and rhyme information is in also proposed a way to use cross-language transfer as a screening line with the findings in reading literature. The ability to attend to tool for L2 children’s reading disabilities. There are two possible phonemes develops later than the ability to attend to larger pho- sources of difficulties associated with L2 reading difficulties: One nological units such as onset-rhyme divisions of the syllable (e.g., is children’s low levels of linguistic proficiency in L2, and the Goswami, 1993; Goswami & Bryant, 1990; Perfetti, Beck, Bell, & other is more general cognitive and learning difficulties. For chil- Hughes, 1987). The English phoneme deletion task is an especially dren who have good L1 language and literacy skills, we should challenging task. It requires that children isolate a given sound in expect transfer to their L2 literacy. Presumably, if children have the word, hold the resulting sound in memory, and then blend strong L1 language and literacy skills but still have not acquired sounds together; therefore, phoneme deletion is sometimes called strong L2 literacy skills, then it is possible that they need more a compound phonemic awareness task (e.g., Yopp, 1988). The time and exposure to L2 for improving their L2 language profi- children had the most difficulty with the first consonant in the final ciency. Their reading difficulty is not because of a true cognitive/ consonant cluster; they were also poor in processing the first consonant in the beginning consonant cluster. This is not surpris- learning disability. This way we can distinguish a true L2 reading ing because consonant clusters impose a higher demand when disability and a delayed L2 learning because of limited L2 lan- children have to manipulate phonemes in words (e.g., Bruck & guage proficiency. Treiman, 1990; Treiman, 1985; Treiman & Weatherston, 1992). To conclude, the findings from our study suggest that bilingual According to Treiman and her colleagues, children tend to treat the reading acquisition may be a joint function of general phonological consonant cluster as a whole unit; therefore, segmentation of the processes and orthographic-specific skills. Phonological skills in two sounds appears to be a daunting task even for children in the L1 contributed to phonological skills and reading in L2. However, primary school grades. in the case of two orthographic systems that share the alphabetic A recent study on Korean children learning to read Hangul at principles but differ in visual forms, there is limited facilitation of kindergarten and Grade 2 found that both syllabic- and orthographic skills from one to the other. Both phonological and subsyllabic-level skills uniquely predicted Hangul word recogni- orthographic skills are important predictors for reading in two tion (Cho & McBride-Chang, 2005). This finding suggests that the different alphabetic orthographies. KOREAN–ENGLISH BILITERACY ACQUISITION 157

References memory skills of bilingual Portuguese-English Canadian children. Read- ing & Writing: An Interdisciplinary Journal, 7, 139–153. Bradley, L., & Bryant, P. E. (1983, February). Categorizing sounds and D’Angiulli, A., Siegel, L., & Serra, E. (2001). The development of reading learning to read: A causal connection. Nature, 301, 419–421. in English and Italian in bilingual children. Applied Psycholinguistics, Bruck, M., & Treiman, R. (1990). Phonological awareness and spelling in 22, 479–507. normal and dyslexics: The case of initial consonant clusters. Journal of Durgunoglu, A. Y. (2002). Cross-linguistic transfer in literacy develop- Experimental Child Psychology, 50, 156–178. ment and implication for language learners. Annals of Dyslexia, 52, Bruck, M., Treiman, R., & Caravolas, M. (1995). Role of the syllable in the 189–204. processing of spoken English: Evidence from a nonword comparison Durgunoglu, A. Y., Nagy, W. E., & Hancin-Bhatt, B. J. (1993). Cross- task. Journal of Experimental Psychology: Human Perception and Per- language transfer of phonological awareness. Journal of Educational formance, 21, 469–479. Psychology, 85, 453–465. Bryant, P. (2002). It doesn’t matter whether onset and rime predicts reading Durgunoglu, A. Y., & Oney, B. (1999). A cross-linguistic comparison of better than phoneme awareness does or vice versa. Journal of Experi- phonological awareness and word recognition. Reading and Writing: An mental Child Psychology, 82, 41–46. Interdisciplinary Journal, 11, 281–299. Bryant, P. E., MacLean, M., Bradley, L. L., & Crossland, J. (1990). Rhyme Ehri, L. C. (1991). Learning to read and spell words. In L. Rieben & C. A. and alliteration, phoneme detection, and learning to read. Developmental Perfetti (Eds.), Learning to read: Basic research and its implications Psychology, 26, 429–438. (pp. 57–73). Mahwah, NJ: Erlbaum. Byrne, B., & Fielding-Barnsley, R. (1995). Evaluation of a program to Ehri, L. C. (1998). Research on learning to read and spell: A personal- teach phonemic awareness to young children: A 2- and 3-year follow-up historical perspective. Scientific Studies of Reading, 2, 97–114. and a new preschool trial. Journal of Educational Psychology, 87, Frith, U., Wimmer, H., & Landerl, K. (1998). Differences in phonological 488–503. recoding in German- and English-speaking children. Scientific Studies of Cassar, M., & Treiman, R. (1997). The beginnings of orthographic knowl- Reading, 2, 31–54. edge: Children’s knowledge of double letters in words. Journal of Frost, R. (1994). Prelexical and postlexical strategies in reading: Evidence Educational Psychology, 89, 631–644. from a deep and shallow orthography. Journal of Experimental Psychol- Cho, J. R., & Chen, H. C. (1999). Orthographic and phonological activa- ogy: Learning, Memory, and Cognition, 20, 116–129. tion in the semantic processing of Korean Hanja and Hangul. Language Ganschow, L., & Sparks, R. (1995). Effects of direct instruction in Spanish and Cognitive Processes, 14(5–6), 481–502. phonology on the native-language skills and foreign-language aptitude Cho, J. R., & McBride-Chang, C. (2005). Correlates of Korean Hangul of at-risk foreign-language learners. Journal of Learning Disabilities, acquisition among kindergartners and second graders. Scientific Studies 28, 107–120. in Reading, 9, 3–16. Geva, E., & Siegel, L. (2000). Orthographic and cognitive factors in the Cisero, C. A., & Royer, J. M. (1995). The development of cross-language concurrent development of basic reading skills in two languages. Read- transfer of phonological awareness. Contemporary Educational Psy- ing and Writing: An Interdisciplinary Journal, 12(1–2), 1–30. chology, 20, 275–303. Geva, E., Yaghoub-Zadeh, Z., & Schuster, B. (2000). Understanding Coltheart, M. (1978). Lexical access in simple reading tasks. In G. Under- individual differences in word recognition skills of ESL children. Annals wood (Ed.), Strategies of information processing (pp. 151–216). New York: Academic Press. of Dyslexia, 50, 123–154. Coltheart, M., Curtis, B., Atkins, P., & Haller, M. (1993). Models of Goswami, U. (1993). Toward an interactive analogy model of reading reading aloud: Dual-route and parallel-distributed-processing ap- development: Decoding vowel graphemes in beginning reading. Journal proaches. Psychological Review, 100, 589–608. of Experimental Child Psychology, 56, 443–475. Coltheart, M., Rastle, K., Perry, C., Langdon, R., & Ziegler, J. (2001). Goswami, U., & Bryant, P. (1990). Phonological skills and learning to DRC: A dual route cascaded model of visual word recognition and read. Hove, England: Erlbaum. reading aloud. Psychological Review, 108, 204–256. Goswami, U., Gombert, J., & Barrera, L. (1998). Children’s orthographic Comeau, L., Cormier, P., Grandmaison, E., & Lacroix, D. (1999). A representations and linguistic transparency: Nonsense word reading in longitudinal study of phonological processing skills in children learning English, French, and Spanish. Applied Psycholinguistics, 19, 19–52. to read in a second language. Journal of Educational Psychology, 91, Gottardo, A. (2002). The relationship between language and reading skills 29–43. in bilingual Spanish-English speakers. Topics in Language Disorders, Cossu, G., Shankweiler, D., Liberman, I. Y., Tola, G., & Katz, L. (1988). 22, 46–70. Awareness of phonological segments and reading ability in Italian Gottardo, A., Yan, B., Siegel, L., & Wade-Woolley, L. (2001). Factors children. Applied Psycholinguistics, 9, 1–16. related to English reading performance in children with Chinese as a first Cummins, J. (1979). Linguistic interdependence and the educational de- language: More evidence of cross-language transfer of phonological velopment of bilingual children. Review of Educational Research, 49, processing. Journal of Educational Psychology, 93, 530–542. 222–251. Green, S. B. (1991). How many subjects does it take to do a regression Cummins, J. (1986). Empowering minority students: A framework for analysis? Multivariate Behavioral Research, 26, 499–510. intervention. Harvard Educational Review, 56, 18–36. Hulme, C., Hatcher, P. J., Nation, K., Brown, A., Adams, J., & Stuart, G. Cummins, J. (1991). Interdependence of first- and second-language profi- (2002). Phoneme awareness is a better predictor of early reading skill ciency in bilingual children. In E. Bialystok (Ed.), Language processing than onset-rime awareness. Journal of Experimental Child Psychology, in bilingual children (pp. 70–89). New York: Cambridge University 82, 2–28. Press. Jastak, J. R., & Jastak, S. R. (1984). Wide Range Achievement Test- Cunningham, A., Perry, K., & Stanovich, K. E. (2001). Converging evi- Revised. Wilmington, DE: Guidance Associates. dence for the concept of orthographic processing. Reading & Writing: Katz, L., & Frost, R. (1992). The reading process is different for different An Interdisciplinary Journal, 14(5–6), 549–568. orthographies: The orthographic depth hypothesis. In R. Frost & L. Katz Cunningham, A., & Stanovich, K. E. (1990). Assessing print exposure and (Eds.), Orthography, phonology, morphology and meaning (pp. 67–84). orthographic processing skill in children: A quick measure of reading Amsterdam: Elsevier Science. experience. Journal of Educational Psychology, 82, 733–740. Kim, J. (1999). Investigating phonological processing in visual word DaFontoura, H. A., & Siegel, L. S. (1995). Reading, syntactic and working recognition: The use of Korean Hangul (alphabetic) and Hanja (logo- 158 WANG, PARK, AND LEE

graphic) scripts. Dissertation Abstracts International: Section B: The Stanovich, K. E., Cunningham, A. E., & Cramer, B. B. (1984). Assessing Physical Sciences and Engineering, 59(11-B), 6093. phonological awareness in kindergarten children: Issues of task compa- Kim, J., & Davis, C. (2004). Characteristics of poor readers of Korean rability. Journal of Experimental Child Psychology, 38, 175–190. Hangul: Auditory, visual, and phonological processing. Reading & Writ- Tan, L. H., Liu, H.-L., Perfetti, C. A., Spinks, J. A., Fox, P. T., & Cao, J.-H. ing: An Interdisciplinary Journal, 17, 153–185. (2001). The neural system underlying Chinese logographic reading. Kroll, J., & Stewart, E. (1994). Category interference in translation and NeuroImage, 13, 836–846. picture naming: Evidence for asymmetric connections between bilingual Tan, L. H., Spinks, J. A., Feng, C. M., Siok, W. T., Perfetti, C. A., Xiong, memory representations. Journal of Memory and Language, 33, 149– J., et al. (2003). Neural systems of second language reading are shaped 174. by native language. Human Brain Mapping, 18, 158–166. Liu, Y., & Perfetti, C. A. (2003). The time course of brain activity in Taylor, I., & Taylor, M. M. (1995). Writing and literacy in Chinese, reading English and Chinese: An ERP study of Chinese bilinguals. Korean, and Japanese. Philadelphia: John Benjamins. Human Brain Mapping, 18, 167–175. Torgesen, R. K. (1999). Phonological-based reading disabilities: Toward a Lundberg, I., Frost, J., & Petersen, O.-P. (1988). Effects of an extensive coherent theory of one kind of learning disability. In R. J. Sternberg & program for stimulating phonological awareness in preschool children. L. Spear-Swerling (Eds.), Perspectives on learning disabilities (pp. Reading Research Quarterly, 23, 263–284. 106–111). New York: Westview Press/HarperCollins. Lundberg, I., Olofsson, A., & Wall, S. (1980). Reading and spelling skills Treiman, R. (1985). Onsets and rimes as units of spoken syllables: Evi- in the first school years predicted from phonemic awareness skills in dence from children. Journal of Experimental Child Psychology, 39, kindergarten. Scandinavian Journal of Psychology, 21, 159–173. 161–181. Mann, V. A., & Liberman, I. Y. (1984). Phonological awareness and verbal Treiman, R. (1993). Beginning to spell. New York: Oxford University short-term memory. Journal of Learning Disabilities, 17, 592–599. Press. Muter, V., Hulme, C., Snowling, M., & Taylor, S. (1998). Segmentation, Treiman, R. (1995). Errors in short-term memory for speech: A develop- not rhyming, predicts early progress in learning to read: Erratum. Jour- mental study. Journal of Experimental Psychology: Learning, Memory, nal of Experimental Child Psychology, 71, 3–27. and Cognition, 21, 1197–1208. Naglieri, J. (1985). Matrix Analogies Test: Expanded form. New York: Treiman, R., & Cassar, M. (1997). Spelling acquisition in English. In C. A. Harcourt Brace Jovanovich Inc. Perfetti & L. Rieben (Eds.), Learning to spell: Research, theory, and Paulesu, E., McCrory, E., Fazio, F., Menoncello, L., Brunswick, N., Cappa, practice across languages (pp. 61–80). Mahwah, NJ: Erlbaum. S. F., et al. (2000). A cultural effect on brain function. Nature Neuro- Treiman, R., Mullennix, J., Bijeljac-Babic, R., & Richmond-Welty, E. D. science, 3, 91–96. (1995). The special role of rimes in the description, use, and acquisition Perfetti, C. A. (1992). The representation problem in reading acquisition. In of English orthography. Journal of Experimental Psychology: General, P. B. Gough, L. C. Ehri, & R. Treiman (Eds.), Reading acquisition (pp. 124, 107–136. 145–174). Hillsdale, NJ: Erlbaum. Treiman, R., & Weatherston, S. (1992). Effects of linguistic structure on Perfetti, C. A., Beck, I., Bell, L. C., & Hughes, C. (1987). Phonemic children’s ability to isolate initial consonants. Journal of Educational knowledge and learning to read are reciprocal: A longitudinal study of Psychology, 84, 174–181. first grade children. Merrill-Palmer Quarterly, 33, 283–319. Rack, J., Snowling, M., & Olson, R. (1992). The nonword reading deficit Wade-Woolley, L. (1999). First language influences on second language on developmental dyslexia. Reading Research Quarterly, 27, 29–53. word reading: All roads lead to Rome. Language Learning, 49, 447– Seymour, P. H. K., Aro, M., & Erskine, J. M. (2003). Foundation literacy 471. acquisition in European orthographies. British Journal of Psychology, Wang, M., Perfetti, C. A., & Liu, Y. (2005). Chinese-English biliteracy 94, 143–174. acquisition: Cross-language and writing system transfer. Cognition, 97, Shankweiler, D. (1999). Words to meaning. Scientific Studies of Reading, 67–88. 3, 113–127. Woodcock, R. C. (1987). Woodcock Reading Mastery Test. Circle Pines, Share, D. (1995). Phonological recoding and self-teaching: Sine qua non of MN: American Guidance Service. reading acquisition. Cognition, 55, 151–218. Yoon, H. K., Bolger, D. J., & Perfetti, C. A. (2002). Sublexical processes Share, D. (2004). Orthographic learning at a glance: On the time course and in learning to read: A difference between Korean and English. In L. developmental onset of self-teaching. Journal of Experimental Child Vehoeven, C. Elbro, & P. Reisman (Eds.), Precursors of functional Psychology, 87, 267–298. literacy (pp. 139–163). Amsterdam and Philadelphia: John Benjamins. Shimron, J. (1999). The role of vowel signs in Hebrew: Beyond word Yopp, H. K. (1988). The validity and reliability of phonemic awareness recognition. Reading & Writing: An Interdisciplinary Journal, 11, 301– tests. Reading Research Quarterly, 23, 159–177. 319. Siegel, L. S., Share, D., & Geva, E. (1995). Evidence for superior orthographic skills in dyslexics. Psychological Science, 6, 250–254. Received September 23, 2004 Simpson, G. B., & Kang, H. (2004). Syllable processing in alphabetic Revision received July 11, 2005 Korean. Reading & Writing: An Interdisciplinary Journal, 17, 137–151. Accepted July 13, 2005 Ⅲ