Abstract
This study investigated the rate of school-aged Chinese–English language learners at risk for reading difficulties in either Chinese or English only, or both, among second and fifth graders in Hong Kong. In addition, we examined the metalinguistic skills that distinguished those who were poor in reading Chinese from those who were poor in reading English. The prevalence of poor English readers among children identified to be poor in Chinese word recognition across the five participating schools was approximately 42% at Grade 2 and 57% at Grade 5. Across grades, children who were poor readers of both languages tended to have difficulties in phonological and morphological awareness. Poor readers of English only were found to manifest significantly poorer phonological awareness, compared to those who were poor readers of Chinese only; their average tone awareness score was also lower relative to normally developing controls. Apart from indicating possible dissociations between Chinese first language (L1) word reading and English second language (L2) word reading, these findings suggested that the degree to which different metalinguistic skills are important for reading in different writing systems may depend on the linguistic features of the particular writing system.
Keywords
Learning to read two languages represents a growing reality for children; almost two thirds of the world’s children grow up in bilingual environments (Crystal, 1997; Wolf, 2007). Understanding the cognitive and linguistic processes involved in dual-language reading development has become a central topic in the study of biliteracy acquisition and development (e.g., Cook & Bassetti, 2005; Koda, 2005). Over the past three decades, research on biliteracy acquisition has specified that dual-language word reading development is a process that involves language-universal and language-specific components (Bialystok & Luk, 2007; Leong, Hau, Cheng, & Tan, 2005; McBride-Chang & Kail, 2002; Wang, Perfetti, & Liu, 2005), and this process is characterized by individual differences in understanding differences and similarities of linguistic structure between first language (L1) and second language (L2; reflected by metalinguistic awareness) and in the development of the processing capacity of these languages to be learned (e.g., Bialystok & Hakuta, 1994; Durgunoglu, 2002a; Durgunoglu & Hancin, 1992; Goswami, 2008). However, there remain questions concerning the nature of the relations between L1 word reading and L2 word reading, as well as the effect of L1 linguistic characteristics on the development of L2 reading.
These questions become particularly evident and complex for Hong Kong Chinese children who are learning to read two distinct writing systems, that is, Chinese and English. Chinese is a monosyllabic writing system (e.g., DeFrancis, 1989; Leong & Joshi, 1997), which means the basic writing unit is the square-shaped character that maps onto one spoken syllable and is a meaningful unit (morpheme). In contrast, English is a morphophonemic writing system, which means the basic unit of writing is the letter that entails phonemic and morphological properties (Bolger, Perfetti, & Schneider, 2005). Given these two complexities, is it possible for children to have difficulties in learning to read one writing system and not another when the writing systems are as different as English and Chinese? This is an important but underexplored “dissociation” question concerning whether two distinct but interrelated word reading skills, that is, Chinese L1 word reading and English L2 word reading, are learned independently. We addressed the potential “dissociation” issue in the present study by identifying “a double deficit group” (i.e., children who are poor in both Chinese L1 and English L2 word reading, referred to as “poor in both” hereafter), “a non-deficit group” (defined as children who are adequate both in Chinese L1 and English L2 word reading, referred to as “controls” hereafter), and two “single-deficit” groups (children who are poor in either Chinese L1 word reading or English L2 word reading only, referred to as “poor in Chinese/English only”) to look at the overlap and dissociation in reading difficulties between Chinese and English. We were also interested in examining the key metalinguistic skills—referring to the individual’s ability to “identify, analyse, and manipulate language forms” (Koda, 2005, p. 311)—that distinguished those who were poor in reading Chinese only from those who were poor in reading English only.
The investigation of possible dissociations between Chinese L1 word reading and English L2 word reading was inspired by theories of second-language reading models (Cummins, 1981, 1984; Dulay, Burt, & Krashen, 1982; Koda, 2005). The first type of second-language reading models can be generally characterized as reading universal accounts. An important aspect of the universal account is the assumption of the developmental similarity between L1 and L2 reading processes (e.g., Dulay et al., 1982; Frost, 2011; Goswami, 2008). The earliest identity hypothesis (L1 = L2) claimed that the cognitive processes of reading L1 and L2 are universal and that the process of children learning L2 is similar to that of learning L1 (Dulay et al., 1982). Recently, Frost (2011) advocated a reading universal approach by proposing that the universal dimensions of word properties—phonology, semantics (morphology) and orthography—can be used to adequately explain all commonly observed reading phenomena in all languages. One of the most noticeable limitations of the universal account is that little attention has been paid to the possible interaction between L1 and L2 reading, as well as possible cross-language differences in acquiring certain reading-related skills, such as phonological awareness.
In contrast, the interactive account of second-language reading models has recognized the interplay between L1 and L2 reading and has assumed some cross-language transfer of certain reading-related skills (Cummins, 1981, 1984; Koda, 2005). One of the most important interactive models is the linguistic interdependence hypothesis (Cummins, 1981, 1984), which assumes that L1 and L2 reading competencies are interrelated and certain linguistic or cognitive skills that are well-established for acquiring L1 reading can transfer to facilitate the acquisition of L2 reading. However, the precondition of L1 reading skill transfer is that L1 reading reaches a certain proficiency level (referred to as a “threshold”). Koda (2005) has expanded the linguistic interdependence hypothesis to the transfer facilitation model and further specified that the transfer of L1 metalinguistic skills to L2 reading arises from the interaction between the child’s L1 metalinguistic competency and the amount of L2 print input to which the child is exposed. Specifically, the visual presentation of L2 leads to the activation of particular L1 metalinguistic skills, and these activated L1 metalinguistic skills lay the foundation for initial L2 reading. Moreover, the transferred L1 metalinguistic skills are modularized by L2 print experience and are turned into an “optimal utility” in learning L2 reading.
Although the universal and interactive accounts of second-language reading models hold different assumptions about the nature of the relations between L1 and L2 reading, both have recognized the commonality between L1 and L2 reading. This provides initial theoretical support for the possible overlap between L1 and L2 word reading in which children who are poor in L1 reading are more likely to have problems in L2 reading (i.e., poor in both). However, these two accounts lead to different predictions about the existence of “single-deficit” readers in dual-language readers. According to the universal account, there should be no single-deficit readers (poor in Chinese or English only) because there should be no difference in the cognitive process between L1 and L2 reading. However, on the basis of the interactive account, there could be a group of children who are poor in either Chinese or English because the dual-language reading development route is shaped by both L1 metalinguistic skills as well as the visual input experience of L2. Furthermore, the interactive account supports the transfer of L1 metalinguistic skills to L2 word reading, whereas the universal account does not. These hypotheses are not only theoretically important but also practically important for educators, teachers, policy makers, and parents who are interested in L2 education among children. For example, in Hong Kong, most Chinese children are educated bilingually, and they are expected to develop literate competences in both Chinese and English. A frequently asked question by teachers, parents, and educators is whether Chinese children who have difficulties in Chinese word reading will also have problems in English.
Indeed, this potential dissociation in Hong Kong Chinese children has been examined in one recent study by McBride-Chang, Liu, Wong, Wong, and Shu (2011). In their study, there were 8 (around 5%) children with word reading scores below the 25th percentile both in Chinese and English and 16 (around 10%) with either English or Chinese word reading scores below the 25th percentile among 154 Hong Kong Chinese children 8 to 9 years old. Moreover, these three groups were compared with controls, whose Chinese and English word reading were both above the 25th percentile, on phonological awareness, morphological awareness, vocabulary, and rapid number naming. Results revealed that children who were poor in both Chinese and English word reading tended to have multiple weaknesses in Chinese phonological awareness, morphological awareness, rapid number naming, and English vocabulary skills. Those who were poor in Chinese word reading performed significantly worse only on Chinese phonological awareness and Chinese vocabulary as compared to the controls, whereas those who were poor in English scored lower than controls only on Chinese phonological awareness and English vocabulary. Children with poor Chinese reading performed worse only on Chinese morphological awareness and Chinese vocabulary as compared to children who had poor English reading only. An important finding of this study is that all three groups (poor both, poor Chinese only, and poor English only) consistently demonstrated weakness in Chinese phonological awareness.
Findings from the study of McBride-Chang et al. (2011) affirm that, by looking into the potential diversities of L1 and L2 reading profiles, differences across groups can be identified—a discovery that could potentially be important to both practice and theory. Moreover, the fact that children who have reading problems regardless of the specific writing system are also weak in phonological awareness is consistent with the interactive account of second-language reading model, which states that certain L1 metalinguistic skills can transfer to L2 reading. However, given that this was a single study investigating the dissociation between Chinese and English word reading, and it focused on a relatively young age group (age M = 8 years old), one may ask whether the dissociation of L1 reading and L2 reading would occur in children in other studies as well.
There are some possible reasons that this might or might not be true across different studies. For example, it remains possible that the discrepancy gap between L1 and L2 reading can be filled in with reading instruction and print experience. Also, despite the fact that, in the previous study, all three groups of poor readers appeared to have difficulties with Chinese phonological awareness tasks tapping into the manipulation of syllables and phonemes, the Chinese phonological awareness task used in that study was specific to segmental phonological awareness (syllable and phoneme). Whether such weaknesses might expand to the suprasegmental feature of Chinese phonology, that is, lexical tone awareness, remains unclear. Given the complexity and prominence of lexical tone in Cantonese as described below, this was an important issue to consider and pursue in the present study.
In Cantonese, spoken words can be recognized and distinguished by both segmental and suprasegmental levels of information. The segmental information refers to vowels and consonants that are composed of speech sounds in languages. For example, the Cantonese syllable /fu/ is made up of the fricative initial consonant /f/ and the midback rounded vowel /u/. The suprasegmental feature concerns vocal effects that span “more than one sound segment in an utterance such as a pitch, stress or juncture pattern” (Crystal, 2003, p. 446), and these manifest predominantly as lexical tones in Cantonese. Lexical tones are pitch patterns that are used to convey differences in meaning of words with identical phonetic segments in Cantonese. There are six contrastive tones (see Note 1): high level (55, T1), high rising (25, T2), mid level (33, T3), low falling (21, T4), low rising (23, T5), and low level (22, T6; Bauer & Benedict, 1997). Figure 1 shows the vocal traces of six Cantonese tones of /fu/ produced by a female Cantonese speaker. As illustrated in Figure 1, each tone is characterized by pitch height and contours. Pitch height refers to the average F0, whereas pitch contour refers to the change of overall slope of the F0 over time (e.g., Wong, Perrachione, Gunasekera, & Chandrasekaran, 2009). Of these six tones, there are three level tones, including high level (55, T1), midlevel (33, T3), and low level (22, T6), and three contour ones, high rising (25, T2), low falling (21, T4), and low rising (23, T5). Cantonese tones are lexically distinctive. As shown in Figure 2, an identical phonetic segment pattern /fu/ can represent six completely different words with meanings of skin /fu1/膚, tiger /fu2/虎, trousers /fu3/褲, symbol /fu4/符, woman /fu5/婦, and father /fu6/父 when speaking in six different tones, respectively.
F0 traces of six Cantonese tones of the syllable /fu/ produced by a native female Cantonese speakers. T1 = high level, T2 = high rising, T3 = midlevel, T4 = low falling, T5 = low rising, T6 = low level.
The hierarchy of levels of phonology in Chinese.
Despite the fact that both segmental and tonal distinctions can serve to distinguish meanings between words, there is substantial empirical evidence showing that they are perceptually distinctive and that there is a developmental perceptual bias of phones over tones (e.g., Burnham et al., 2011; Cutler & Chen, 1997; Ye & Connine, 1999). Our interest in the present study was in how phonological awareness at the segmental level and lexical tone awareness at the suprasegmental level would be related to dual-language reading development in Cantonese children. Specifically, on the basis of findings by McBride-Chang et al. (2011) that all three types of poor readers (poor in both, poor in Chinese only, poor in English only) demonstrated weakness on L1 Chinese phonological awareness at the segmental level, we wanted to examine whether such a weakness would also be exhibited in processing other phonologically relevant information, that is, lexical tones.
Lexical tone awareness refers to individuals’ sensitivity to pitch patterns (tonemes) of spoken syllables and their ability to distinguish different pitch patterns (Burnham et al., 2011). There is evidence showing that individual differences in lexical tone awareness predict individual differences in word reading abilities (Burnham et al., 2011; McBride-Chang et al., 2008). For example, McBride-Chang et al. (2008) reported that lexical tone awareness bootstraps Chinese word reading development in that lexical tone awareness uniquely contributed to Chinese word reading in 4- to 5-year-old Hong Kong Cantonese-speaking children. Studies on Chinese word reading difficulties have reported similar findings, that is, that difficulty in perceiving lexical tone is an early linguistic marker of reading difficulties in Chinese children (e.g., Cheung et al., 2009; Chung et al., 2008).
Despite considerable evidence showing the link between lexical tone awareness and word reading ability, the nature of the relations between lexical tone awareness and reading abilities remains controversial. For example, a recent study by Burnham et al. (2011) demonstrated that both lexical tone awareness and phonological awareness were predicted by reading abilities for Thai children learning the Thai alphabetic writing system, where tones are represented orthographically, whereas no such association was found for Cantonese-speaking children learning the Chinese morphosyllabic writing system, in which tone was not orthographically represented. It is worth noting, however, that in that study, lexical tone awareness was measured with Thai tones, so that this represented nonnative tone awareness for Cantonese children. Wang et al. (2005) reported that lexical tone sensitivity in Chinese was related to reading of English pseudowords. However, this finding was not replicated in a study by McBride-Chang et al. (2008), in which tone sensitivity was predictive of Chinese word reading, but not English word reading in Cantonese-speaking children, perhaps because of constraints in their English language learning experiences. Given these conflicting findings, in the present study we continued to focus on this theme of explicitly how individual differences in L1 lexical tone awareness relate to reading difficulties in both L1 Chinese word reading and L2 English word reading.
The purpose of the present study was to analyze the nature of the relations between L1 word reading and L2 word reading among Hong Kong children in the intermediate and advanced stages of learning to read Chinese and English, two distinct writing systems, and to examine possible L1 metalinguistic transfer, with a particular focus on lexical tone awareness. Two hypotheses were examined. First, we tested for a potential dissociation between the difficulties in Chinese L1 word reading and English L2 word reading. With a cross-sectional design, we considered the extent to which there might be an appearance of developmental difference in the relations between first-language reading and second-language reading across second and fifth grade. We expected that Chinese word reading and English word reading are interrelated but that they could also develop separately.
There were at least three reasons for choosing children in second and fifth grade. The first one comes from the theoretical concerns in distinguishing “reading problems” from “general reading/learning difficulties” in studying second-language reading (Durgunoglu, 2002a; Geva, 2000). Relative to L1 reading, L2 reading acquisition is much more complex and may require more time, instruction, and effort. Moreover, the initial development of L2 reading is based on L1 metalinguistic skills and L2 print input (Cummins, 1981; Koda, 2005). Also, there exists cross-linguistic variability in the development of L1and L2 metalinguistic skills. For example, there is empirical evidence showing that phonemic awareness develops faster and earlier for young alphabetic language readers but not for nonalphabetic language learners, such as Chinese learners (Goswami, 2008). Considering this, it may not be theoretically ideal to study “dissociation” between L1 and L2 in beginning readers because they are faced with difficulties in building up L1 metalinguistic sophistication and lack a certain amount of L2 print input.
Also, if children were identified to have lower L2 reading scores, it may be difficult to distinguish whether this might be a result of lower linguistic proficiency in L1 or insufficient L2 input or a developmental discrepancy (Durgunoglu, 2002a). Therefore, we selected children from second and fifth grade who had developed L1 metalinguistic sophistication and also had relatively solid L1 and L2 print input. This rationale comes from previous studies of cross-language transfer of metalinguistic skills in Chinese children (e.g., Keung & Ho, 2009; McBride-Chang et al., 2011; Wang et al., 2005). Most previous studies of cross-language associations between Chinese L1 and English L2 have selected participants who were second graders or older because these children were supposed to reach a certain level of reading proficiency in both L1 and L2. We therefore selected second graders as target participants. At the same time, we also selected fifth graders to examine the differences of L1 and L2 reading across ages.
In addition, there is converging evidence showing that by age 10, Cantonese children can reliably differentiate six tones, and their identification accuracy is comparable to that of adults (Ciocca & Lui, 2003), although children have mastered tone production by age 2 (Dodd & So, 1994). We therefore included second graders who are developing tone perception and fifth graders who have stable tone perception skills. Also, testing the older children enabled us to control for a potential confound, namely, a developmental difference in acquisition of the six Cantonese tones.
Our second hypothesis focused on the potential for the transfer of metalinguistic skills, with a particular focus on lexical tone awareness. We expected that children who were poor in reading English only would have lower phonological awareness than those who were poor in reading Chinese only. This hypothesis was posited because the English word reading process relies primarily on phonology–orthography mapping, with the mastery of the alphabetic principle and various “grain size” (i.e., syllable, rime, and phoneme; Ziegler & Goswami, 2005) units being prominent for reading development and reading difficulty, whereas Chinese word reading difficulties have been described as fitting a multiple deficits model, of which phonological awareness is not among the strongest predictors, at least in Hong Kong Chinese children (Ho, Chan, Lee, Tsang, & Luan, 2004). Unlike previous work that examined only these grain-sized phonological units (McBride-Chang et al., 2011), in the present study we sought to determine whether such potential differences across groups might extend to lexical tone awareness as well.
Method
Participants
Participants were 172 second graders (79 girls; age M = 8.09 years, SD = 5 months) and 165 fifth graders (78 girls; age M = 11.07 years, SD = 5 months) from five elementary schools across the New Territories and Kowloon areas in Hong Kong. These students demonstrated a wide range of ability levels within Hong Kong; all were typically developing students. All children were from families with medium house income levels according to the 2006 Hong Kong census (Hong Kong Census and Statistics Department, 2006).
A classification criterion of the lowest 25% on the respective word reading test (e.g., Francis et al., 2005; McBride-Chang et al., 2011) was employed to select children who were designated as poor readers in Chinese, English, or both, for the purposes of this study, in second and fifth grade. We identified 25 children who were poor in Chinese word reading but normal in English word reading (poor in Chinese; PC), 25 children who were poor in English word reading but normal in Chinese word reading (poor in English; PE), and 18 children who were poor in reading in both Chinese and English (poor in both Chinese and English; PB) in Grade 2. As for Grade 5, there were 18 poor readers in the PC, 18 poor readers in the PE, and 24 poor readers in the PB conditions. In addition, we selected, respectively, 25 and 24 students as average readers from among children who scored in the top 50% in both Chinese and English word reading tests (n = 53 in the second grade, n = 58 in the fifth grade) as controls (abbreviated as C) for Grade 2 and Grade 5, respectively. The mean ages in years of the PC, PE, PB, and C groups were 8.04 (SD = 0.3), 8.11 (SD = 0.5), 8.05 (SD = 0.7), and 8.10 (SD = 0.5) in Grade 2, and 11.05 (SD = 0.3), 11.21 (SD = 0.4), 11.02 (SD = 0.3), and 11.08 (SD = 0.2) in Grade 5, respectively.
According to parents’ reports, the selected children in these four groups were Hong Kong–born native Cantonese speakers who had been raised in Hong Kong Cantonese-speaking families. They did not have any hearing impairments. They had minimum exposure to Mandarin or other dialects, and they did not attend Mandarin classes after school.
Measures
Syllable and phoneme segmentation
Segmental phonological awareness was assessed with the combination of syllable deletion and phoneme-onset deletion. The syllable deletion subtest asks children to remove the specific syllable (first, second, or third) from three-syllable spoken words, such as “/huŋ4 luk6 tɐŋ1/ (red-green-light) without /luk6/ (green) would be /huŋ4–tɐŋ1/ (red light).” There are 29 test items including 15 real words and 14 nonwords in this subtest. Testing discontinued after six consecutive failed items. The phoneme-onset deletion subtest, which followed the syllable deletion items, requires children to take away the initial sound from spoken syllables, for example, saying /huk1/ without saying the initial sound would be /uk1/). It consists of 22 items, with 10 real words and 12 nonwords. All items are arranged in order of increasing difficulty level. Testing halted after four consecutive errors. The obtained Cronbach’s alphas for the phonological awareness task, combining both subtests, were .95 and .96 for second and fifth grades, respectively.
Tone awareness
An odd-one-out tone discrimination task was used to assess children’s ability to perceive and distinguish different lexical tone contrasts. The odd-one-out format was chosen because of concerns of discrimination validity (no ceiling effect) and testing tacit knowledge of lexical tones, as well as zero involvement of speech production (Burnham et al., 2011). There were three blocks (same onsets different rimes, different onsets same rimes, different onsets different rimes) of eight test trials/tone contrasts. Each trial consisted of four spoken Cantonese words, along with four pictures corresponding to these four words. Children were aurally presented each trial and were then asked to pick the odd one out by pointing to its corresponding picture. Each trial tested a specified tone contrast, and each block, therefore, consisted of eight tone contrasts. These eight tone contrasts were high level (55, T1)–midlevel (33, T3), high level (55, T1)–low level (22, T6), midlevel (33, T3)–low level (22, T6), high rising (25, T2)–low rising (23, T5), high level (55, T1)–high rising (25, T2), low rising (23, T5)–low level (22, T6), low falling (21, T4)–low rising (23, T5), and low falling (21, T4)–low level (22, T6), which were selected on the basis of Ciocca and Lui (2003). The first four tone contrasts, that is, T1 (55)–T3 (33; level vs. level), T1 (55)–T6 (22; level vs. level), T3 (33)–T6 (22; level vs. level), and T2 (25)–T5 (23; rising vs. rising), have the same contours but different pitch heights, whereas the latter four tone contrasts, that is, T1 (55)–T2 (25; level vs. rising), T5 (23)–T6 (22; rising vs. level), T4 (21)–T5 (23; falling vs. rising), and T4 (21)–T6 (22; falling vs. level), have different contours but the same pitch heights. Each block represents a specific manipulation of the similarity of spoken words at the segmental level. The first block consisted of same onsets different rimes, in which four syllables shared the same onsets but different rimes, such as /si1/, /sɐm1/, /siŋ1/, /sɐw2/. The second block consisted of different onsets same rimes pairs, such that the four spoken words had the same rimes but different onsets, such as /lεy4/, /khεy4/, /mεy4/, and /phεy5/. The third block comprised different onsets different rimes, such as /jœŋ4/, /thɔw4/, /khɐm4/, and /luk6/. Across all trials of these three blocks, the familiarity of onsets and rimes used in stimuli was approximately matched. The presentation order of the eight tone contrasts was counterbalanced across three blocks. All items are listed in the appendix.
Prior to testing, children were presented with six Cantonese tones in examples by experimenters, and they were also presented with an additional practice trial to enable them to familiarize themselves with the testing procedure. During the practical session, children’s responses were followed with oral feedback; testing started following correct response of practice trials. The obtained Cronbach’s alphas of the phonological awareness task were .81 and .77 for second and fifth grades, respectively.
Morphological awareness
A composite morphological awareness task including morphological construction and homophone production was used (e.g., Chung et al., 2008). In the morphological construction section of the task, a three-sentence scenario describing one object or concept was aurally presented to children, and they were then asked to generate a newly described object or concept based on the description. For example, “There is one type of traffic light having red and green colors, we call it a red-green-light. What would we call a type of traffic light having blue and green colors instead?” The correct answer is “blue-green-light.” There were 27 such items, ranked according to increasing difficulty level. Testing stopped following four consecutive failed items. The homophone production subtest consisted of 14 items. In this task, a two-syllable word (e.g., 月亮/jyt6 lœŋ6/) having the target monosyllable (i.e., 月/jyt6/) was aurally presented to the children. They were then asked to construct new words that included the same morpheme representing the target monosyllable (e.g., 月/jyt6/), such as 月光 (moonlight) or 月饼 (moon cake), within 10 s to ensure they knew the target tonal syllable. Following this, children were asked to produce as many words as possible within 10 s that contained a homophone of the given morpheme 月/jyt6/, for example, 洞穴/tuŋ6 jyt6/ or 阅读 /jyt6 tuk6/. The obtained Cronbach’s alphas for this morphological awareness task were .95 and .72 for second and fifth grades, respectively.
Chinese word reading
A 211-item word reading test was formed from a list of 27 single-character and 34 two-character words (McBride-Chang & Ho, 2000) and a list of 150 two-character words adopted from the Hong Kong Test of Specific Learning Difficulties in Reading and Writing (Ho, Chan, Tsang, & Lee, 2000). These words were arranged in order of increasing difficulty. This task requires children to read aloud each word, and testing is discontinued following 15 consecutive failed items. The obtained Cronbach’s alphas of this task were .97 and .96 for second and fifth grades, respectively.
English word reading
Children were asked to read aloud a word list consisting of 60 English words. The list contained both orthographically simple and complex words. These words were selected from Hong Kong primary school English textbooks and displayed in order of increasing difficulty (Tong & McBride-Chang, 2010). Testing stopped following 15 consecutive errors. The obtained Cronbach’s alphas for this task were .96 and .95 for second and fifth grades, respectively.
Procedure
After parents gave consent for their children’s participation, which was obtained via the schools, we conducted testing in a quiet room in the school by trained psychology-major undergraduate students. All tasks were administered individually to the child in two approximately 50- to 60-min sessions. This included other tasks that were part of a larger research project that were not included in this study. The tasks were presented to children in a pseudo-random order, such that the same two tasks did not occur consecutively across children.
Results
We first examined the data for the presence of univariate and multivariate outliers. The scatterplots of all variables as well as bivariate plots across samples demonstrated that the normality and linearity were statistically acceptable. Using the criterion of median plus or minus two interquartile ranges, there were no outliers across samples (Tabachnick & Fidell, 2007). Thus, all analyses were carried out on raw scores.
Is There Any “Dissociation” or “Co-occurrence” of First-Language Chinese Reading Difficulties and Second-Language English Reading Difficulties Among Hong Kong Chinese Children as English Language Learners (ELL)?
Table 1 shows the prevalence rates of Grade 2 and Grade 5 Hong Kong Chinese children’s Chinese word reading difficulties (PC), their English word reading difficulties (PC), and their average reading (C) or reading difficulties with both writing systems (PB). The chi-square analysis suggested that the probability of poor Chinese readers also being poor English readers was significantly above the baseline level in both second grade and fifth grade, χ2(1, N = 172) = 8.69, p < .01, and χ2(1, N = 165) = 29.81, p < .001, for second and fifth grade, respectively.
The Distribution of the Different Types of Chinese–English Readers in Second and Fifth Grades Hong Kong Chinese Children.
To confirm the reading profiles of the four groups of readers, a multivariate analysis of variance (MANOVA) was conducted by entering Chinese word reading and English word reading as dependent variables (see the upper part of Table 2). There was a significant overall group effect for Grade 2, Λ = .07, F(6, 176) = 85.90, p < .001, ηp2 = .75, and Grade 5, Λ = .35, F(6, 156) = 40.77, p < .001, ηp2 = .61, respectively. The univariate F test revealed that the four groups differed significantly in both Chinese word reading, F(3, 89) = 70.65, p < .001, ηp2 = .71, and English word reading, F(3, 93) = 120.14, p < .001, ηp2 = .80, in Grade 2. Similarly, in Grade 5, the four groups also differed significantly in both Chinese word reading, F(3, 80) = 30.62, p < .001, ηp2 = .53, and English word reading, F(3, 80) = 75.44, p < .001, ηp2 = .74. A post hoc test with a Bonferroni adjustment suggested that those who demonstrated reading difficulties in both Chinese and English (PB) and those with reading difficulties only in Chinese (PC) scored significantly lower on Chinese word reading than those selected as controls (C) and those manifesting reading difficulties only in English (PE; ps < .001). Those with reading difficulties only in English (PE) or in both Chinese and English (PB) performed significantly worse on English word reading tasks than did those in the control group and those who were poor only in Chinese (PC) across both grade levels (ps < .001).
Means, Standard Deviations, and Comparisons of Children Who Are Poor Readers in Both Chinese and English (PB), Poor Chinese Readers (PC), Poor English Readers (PE), and Controls Who Were Average in Chinese and English Across Second and Fifth Grade.
Note. There is no maximum score for the production task.
Bonferoni correction was used.
p < .05. **p < .01.
Do Different Types of Poor Readers Have Specific Weaknesses on Certain L1 Metalinguistic Skills?
Table 2 shows the raw scores for each group on each of the three Chinese metalinguistic tasks, segmental phonological awareness, tone awareness, and morphological awareness, and the comparisons of the four groups on these tasks. To examine whether certain Chinese metalinguistic skills (segmental phonological awareness, tone awareness, and morphological awareness) that might underlie these separate but related difficulties, we conducted MANOVAs to examine group differences on metalinguistic skills, separately for second and fifth grade. Because we conceptualized tasks of syllable and phoneme deletion as segmental phonological awareness and lexical tone awareness as suprasegmental level awareness, and given that these tasks are correlated (McBride-Chang et al., 2008), we treated segmental phonological awareness and tone awareness initially as a single phonological processing construct to increase statistical power for these analyses. The z composite score was first created for phonological processing in second and fifth grade by averaging z scores of syllable, phoneme-onset deletion, and tone discrimination. In the MANOVA analysis, the composite score of phonological awareness and the z score of morphological awareness were entered simultaneously as dependent variables and the group was the between-subjects factor.
At Grade 2, the overall effect of group was significant, Λ = .60, F(6, 176) = 8.46, p < .001, ηp2 = .22, with significant group differences on phonological processing, F(3, 89) = 15.62, p < .001, ηp2 = .35, and morphological awareness, F(3, 89) = 7.06, p < .001, ηp2 = .19. The post hoc comparisons with Bonferroni adjustment suggested that both the PB group and the PE group scored significantly lower than the control group (ps < .001), and the PC group on phonological processing (ps < .01). For the morphological awareness measure, the PB group scored significantly lower than the control (ps < .05) and PC groups (p < .01). There were no other statistically significant differences.
At Grade 5, there was a significant overall group effect, Λ = .68, F(6, 158) = 5.73, p < .001, ηp2 = .18, with a significant group difference on both phonological processing, F(3, 80) = 8.02, p < .001, ηp2 = .23, and morphological awareness, F(3, 80) = 5.10, p < .01, ηp2 = .16. The post hoc comparisons with Bonferroni adjustment suggested that the PB group and the PE group scored significantly lower than the control group (ps < .05) on phonological processing. The PC group performed approximately equally as compared to the control group (p = 1.0), and they performed better than the PE group (p < .01) on phonological processing. For morphological awareness, the PB group scored significantly lower than the control group only (p < .01). No other differences were found between groups (ps > .05).
A Closer Examination of Chinese Phonological Processing: Segmental Phonological Awareness and Tone Awareness
To determine what aspects of phonological processing might more strongly distinguish among groups, another set of MANOVAs was conducted by entering the phonological awareness (segmental phonology) and tone awareness (suprasegmental phonology) as dependent variables and group as a between-subjects factor, separately for second grade and fifth grade.
Table 2 shows the comparisons of four groups on phonological awareness and tone awareness (see the lower part). At Grade 2, there was a significant overall effect of group, Λ = .63, F(6, 176) = 7.63, p < .001, ηp2 = .21, with significant group differences on phonological awareness, F(3, 89) = 14.61, p < .001, ηp2 = .33, and tone awareness, F(3, 89) = 5.22, p < .01, ηp2 = .15. The post hoc comparisons with Bonferroni adjustment on phonological awareness further revealed that the PB group and the PE group scored significantly lower than the control group (ps < .001). The PB group performed less well than the PC group (p < .01). However, they performed approximately equally as compared to the PE group (p = 1.0). Moreover, the PE group performed less well than the PC group (p < .01). For tone awareness, the PB group and the PE group performed worse than the control group (ps < .05). There was a trend that the poor English group performed less well than the PC group (p = .08) as well. No such trend was found between the PB group and the PE group.
At Grade 5, the overall effect of group was significant, Λ = .71, F(6, 158) = 4.85, p < .001, ηp2 = .16, with significant group differences on phonological awareness, F(3, 80) = 8.39, p < .001, ηp2 = .24, and tone awareness, F(3, 80) = 3.28, p < .05, ηp2 = .11. The post hoc comparison with Bonferroni adjustment on syllable and phoneme-onset deletion further revealed that the PB group and the PE group performed less well than the control group (ps < .01). The PB group performed less well than the PC group (p < .05). Similarly, the PE group scored significantly lower than the PC group (p < .05). The post hoc comparison with Bonferroni adjustment on tone awareness suggested that the PE group performed worse than the control group only (p < .05). No other differences were found among groups (ps > .05).
Discussion
The goals of this study were twofold: First, we were interested in determining the prevalence of Grade 2 and Grade 5 Hong Kong Chinese children’s risk for reading difficulties in either Chinese or English only as well as in both writing systems. Second, we wanted to examine the Chinese metalinguistic skills-morphological awareness and phonological awareness that might distinguish poor readers in Chinese only, in English only, or in both English and Chinese. In particular, we focused on whether the phonological awareness weakness found in previous studies (McBride-Chang et al., 2011) is specific to segmental phonological awareness that is common to both Chinese and English language or whether it could be expanded to suprasegmental lexical tone awareness that seems specific to the Chinese writing system. We found a co-occurrence and dissociation of word reading difficulties in morphosyllabic Chinese and the alphabetic English writing systems. Specifically, the chance of being a poor reader in English given that a child was poor in word recognition in Chinese was approximately 42% (18 out of 43) at Grade 2 and 57% (24 out of 42) at Grade 5. The likelihood of being a poor reader in English given that the child was adequate in Chinese was 19% (25 out of 129) at Grade 2 and 15% (18 out of 123) at Grade 5. The same likelihoods applied to being a poor reader in Chinese given children being adequate in English. Also, weakness in Chinese segmental phonological awareness and morphological awareness were shown for children with word reading difficulties in both Chinese and English writing systems, whereas those with difficulty in English only showed weaknesses in both segmental phonological awareness and tone awareness. In addition, those children with reading difficulties in English only perform poorly on tone awareness than those children with reading difficulty in Chinese reading only at Grade 2 but not Grade 5.
The comorbidity and dissociation observed between Chinese L1 word reading and English L2 word reading is consistent with the findings of McBride-Chang et al. (2011), showing that approximately 33% of young children ages 8 to 9 years who are at risk for Chinese word reading difficulty are also at risk for English word reading difficulty. However, this co-occurrence rate was found to be much larger in the present study than in the previous one. A probable reason is that, in the present study, children were selected from five elementary schools, whereas the previous study included children from many schools given that it was a statistically representative sample of all of Hong Kong. Including only a few schools in the present study may have increased the overlap in word recognition difficulties for Chinese and English because of possible similarities in children’s learning skills or shared teaching methods to which the children were exposed.
There are at least three main plausible explanations for the co-occurrence between Chinese L1 word reading difficulty and English L2 word reading difficulties. The first explanation centers on L2 reading models (Cummins, 1984; Dulay et al., 1982; Frost, 2011; Goswami, 2008; Koda, 2005). According to a reading universal account (Frost, 2011; Goswami, 2008), word reading is a fundamental mapping process (grapheme to sound or meaning) across writing systems such that certain cognitive demands necessary for operating this process by the human cognitive system are universal across different writing systems (Frost, 2011; Share, 2008). Moreover, the interactive account (Cummins, 1984; Koda, 2005) also recognizes that there exist common prerequisite skills for reading acquisition in both L1 and L2, such as linking language forms (sound or meaning) with the print unit (orthography; Perfetti, 2003, Perfetti, Cao, & Booth, 2013). Therefore, we observed a strong association between L1 Chinese word reading difficulties and L2 English word reading difficulties.
The second explanation we considered comes from empirical evidence on dual-language reading development in different biliterate readers (e.g., Deacon, Wade-Woolley, & Kirby, 2007; Keung & Ho, 2009; Pasquarella, Chen, Lam, Luo, & Ramirez, 2011). There is ample evidence showing that L1 word reading and L2 word reading are mutually correlated and reinforce each other in the process of children’s biliteracy development, and the correlations between L1 and L2 word reading skills range from moderate to high (e.g., Deacon et al., 2007; Pasquarella et al., 2011). In our study, the correlation between Chinese word reading and English word reading was r = .62 (p < .001). This is consistent with research examining cross-language transfer of Chinese and English phonological awareness, with a correlation between Chinese word reading and English word reading of r = .58 (Keung & Ho, 2009).
In addition, the shared learning setting and sociolinguistic contexts for the acquisition of L1 Chinese word wording and L2 English word reading are also possible factors that contribute to the overlapping between reading difficulties in morphosyllabic Chinese and morphophonemic English. Unlike the curricula in Mainland China, Taiwan, and other Chinese societies, the school curriculum in Hong Kong is generally characterized as bilingual, promoting Chinese–English biliteracy acquisition. In such a bilingual program, Chinese and English are both taught in a holistic rather than an analytical way, and the language curriculum emphasizes understanding and reasoning. No explicit or systematic instruction is provided for children on English phonetics or phonemes to Chinese children. Rather, children learn English words by rote memorization, which is similar to the way they learn to read Chinese characters (Leong, Cheng, & Tan, 2005). Given that the teaching approach affects the route of reading development (McBride-Chang et al., 2013), it is highly likely that Hong Kong Chinese children use some similar strategies in learning to read Chinese and English words. Furthermore, there is also research showing that Hong Kong Chinese children tend to used well-established Chinese learning strategies to learn English (Keung & Ho, 2009). Therefore, children who are failing in learning to read Chinese characters are more likely to have difficulty in reading English words. It might be both theoretically and practically interesting and useful to examine to what extent such an overlap originates from the source of cognitive demands or educational contexts in future research. In addition, it may well be the case that Pinyin is important as a mediator between Chinese and English learning. For example, McBride-Chang et al. (2013) demonstrated that the co-occurrence of difficulties in reading Chinese and English was higher in Beijing as compared to Hong Kong children and suggested that one major reason for this might be Pinyin learning. Traditionally, Hong Kong Chinese children have not had strong Pinyin training, if any. They do learn Mandarin in school, but Pinyin has not been consistently taught during Mandarin lessons, though this is now changing. Therefore, it might be interesting to consider the effects of Pinyin training (e.g., after-school tutoring or in-school consistent lessons) for the association between reading status (poor vs. normal reader) across reading of Chinese and English. This is a potentially important focus for future research.
Apart from the co-occurrence, the existence of groups of poor Chinese only and poor English only readers further extends the study by McBride-Chang et al. (2011) by showing that some disassociation between L1 Chinese word reading and L2 English word reading is evident across both second and fifth graders. Also, the existence of disassociation between L1 Chinese word reading and L2 English word reading provides evidence against the L1 = L2 identity hypothesis and suggests that there exists a great individual difference at the levels of reading proficiency in L1 and L2 for children learning two writing systems, and that there exist differences in certain cognitive or linguistic underpinnings necessary for reading in Chinese and English. This provides evidence that there exist some aspects that are specific to reading Chinese (Perfetti et al., 2013). This further reinforces the idea that developing a universal model of reading is theoretically appealing, but may not be practically possible (McBride-Chang, Chen, et al., 2012). Also, the existence of a poor Chinese reading only group may indicate that L1 difficulties do not necessarily lead to L2 reading difficulties, as assumed by interactive modes of L2 reading (Cummins, 1981, 1984; Koda, 2005).
An important finding of the present study is that phonological awareness (segmental phonology) indexed by syllable and phoneme segmentation distinguished those who had reading difficulties in English from those who had difficulties in reading Chinese. Children with word reading difficulty in English performed worse than children with word reading difficulties in Chinese. No difference was found between control and children with reading difficulties in Chinese. This indicates that segmental phonological awareness is an important predictor of English word acquisition for Hong Kong Chinese children as English language learners, and that the mere presence of phonological awareness weakness by itself may not be the crucial source of Chinese word reading difficulties. This finding is both theoretically and clinically significant and noteworthy. Theoretically, it suggests that differences in language features might have impacts on the skills involved in word reading, here referring specifically to segmental phonological awareness, and that dual-language word reading development is a process that involves some language-specific skills apart from universal cognitive and linguistic skills. As Bialystok and Hakuta (1998) stated, dual-language learners may develop a “common representation” that embodies language general information and “separate representations” that characterize language-specific information. This becomes uniquely evident for Hong Kong Chinese children learning to read morphosyllabic Chinese and alphabetic English. Chinese is a morphosyllabic language with no grapheme–phoneme correspondences. The most phonological salient units in Chinese are syllables that directly map onto characters or morphemes (McBride-Chang et al., 2008). This possibly explains why it remains challenging for school-aged children to develop fine-grained sensitivity to phonological units that are smaller than syllables, such as phonemes (McBride-Chang et al., 2005).
Also, there is a distinctive difference between Chinese and English in terms of the relation between visual-orthographic unit and sound. As Keung and Ho (2009) stated, Chinese is “a part-whole derived phonology” because the phonetic radical, one component of a Chinese complex character, provides a clue to the sound of the whole character. For example, the character 燈 /tɐŋ1/ (lamp) contains the phonetic radical 登 on the right, indicating the sound of /tɐŋ1/. In contrast, English is an alphabetic language for which Roman letters are the basic building blocks of the word. English word reading is a grapheme–phoneme mapping process that places considerable demands on children’s acquisition of the alphabetic principle and smaller phonological units, that is, phonemes (Ziegler & Goswami, 2005). The English alphabet is processed via assembled phonology, which involves segmenting words into letters and mapping them onto phonemes, such as the segmentation of “cat” into /k/-/æ/-/t/ (e.g., Hamada & Koda, 2008). However, such demands are less obvious in reading Chinese words (Ho et al., 2004; McBride-Chang et al., 2005).
Given that reading is a process of mapping print units onto language forms (Perfetti, 2003), we postulate that this contrastive difference between Chinese and English in terms of the phonology and orthographic features leads to different demands on phonological awareness in reading single words. These findings corroborate and extend the psycholinguistic grain size theory (Ziegler & Goswami, 2005) developed on the basis of alphabetic writing systems, by suggesting that the importance of phonological awareness in reading highly depends on phonological features of the language and changes with the experience of language. Clinically, these findings provide early evidence that phonological sensitivity is a critical factor that can be used to distinguish children with specific reading difficulties in Chinese from those with specific reading difficulties in English. This finding might facilitate the development of an early diagnostic test for Chinese–English learners at risk for reading difficulties in English and suggest possible intervention or remediation programs.
One of the most striking findings of this study is that the transfer of tone awareness to English word reading was evident: Children with reading difficulties in English only also manifested weaknesses in tone awareness across Grade 2 and 5, and tone awareness distinguished children with reading difficulty in English from children with reading difficulty in Chinese at Grade 2. This finding corroborates and extends existing evidence of the cross-linguistic transfer of Chinese lexical tone processing to English pseudoword reading found in normally developing Chinese–English bilingual children (Wang et al., 2005). In a study by Wang et al. (2005), Chinese immigrant children who were learning to read Mandarin Chinese and English were asked to carry out a tone matching task, for example, deciding whether the target Mandarin tone carried by the spoken syllable /maŋ2/ matched one of the two candidates, that is, /hwei2/ or /ʂu1/. They found that children’s tone processing skill was significantly correlated with both English real word (r = .31) and pseudoword reading (r = .34), and it remained significant in explaining English pseudoword reading even after controlling for English phonemic processing. In our study, we found a similar correlation between Cantonese tone awareness and English real word reading (r = . 44 and r = . 34 for Grade 2 and Grade 5, respectively) in Grade 2 and Grade 5 Hong Kong Cantonese children. These moderate correlations conjoining the tone weakness manifested by poor English only extends our understanding to suggest that phonological skills transfer between Chinese and English may not be restricted to the syllable (e.g., McBride-Chang, et al., 2008), rime (e.g., Gottardo, Yan, Siegel, & Wade-Woolley, 2001), and phoneme (e.g., Keung & Ho, 2009) only, but include also the suprasegmental levels of lexical tones. Furthermore, these findings provide evidence that phonological transfer can occur in language with different structures (Durgunoglu, 2002a; Durgunoglu & Hancin, 1992). More generally, the transfer of lexical tone, the speech prosodic feature that is absent from English words, broadens the idea of “transfer” as defined in second-language reading models (e.g., Cummins, 1984; Koda, 2005), in which the transfer could occur in the suprasegmental level of phonological processing in two writing systems with distinct suprasegmental features and the L1 learning experience might influence the acquisition of L2 reading in a much more pervasive way.
One possible explanation for this finding is that the source of the contribution of lexical tone to English word reading is at the level of acoustic-auditory processing of speech sounds. This explanation rests largely on the characteristics of lexical tones and findings from research on perception of linguistic pitch and nonlinguistic pitch (e.g., Bent, Bradlow, & Wright, 2006; Wang et al., 2005). As shown in Figure 1, Cantonese lexical tones are characterized as perceptually distinctive pitch patterns and serve to distinguish meanings between words. Lexical tones are not orthographically represented or marked in Chinese orthography, and it is therefore not surprising that reading experience does not influence the distinction of Cantonese lexical tones (Burnham et al., 2011). However, research on second-language speech perception has clearly shown that perception of lexical tones is shaped by language experience (e.g., So & Best, 2010). For example, So and Best (2010) found that Cantonese speakers showed perceptual difficulties in distinguishing the contrast of nonnative Mandarin Tone 1 versus Tone 4 because Cantonese speakers tended to assimilate nonnative Mandarin Tone 1 and Tone 4 with the native Cantonese Tone 1. Moreover, the effect of tonal language experiences is not restricted to the linguistic pitch, and it can even be extended to nonspeech pitch contour identification (Bent et al., 2006). Therefore, it is highly likely that Cantonese children’s tonal language experience influences the encoding of English speech sounds at a very general or low level, such as acoustic-auditory processing.
An alternative explanation for this finding is that lexical tone skill that has been established in Cantonese facilitates the acquisition and development of lexical stress sensitivity in English and further promotes subsequent English word reading. Our reasoning behind this is that despite the fact that Chinese lexical tones and English lexical stress are two prosodically distinct features, these features both rely on the changes of fundamental frequency (F0), the amplitude and duration of speech that extend beyond segmental units (consonants and vowels; Cutler & Chen, 1997; L. Lee & Nusbaum, 1993). For example, in Cantonese, an identical segmental pattern such as /fu/ can represent six completely different meanings of words by varying pitch pattern (as shown in Figure 1). On the other hand, in English, changes in pitch, intensity, and duration at the between-syllable level, referred to as lexical stress, can result in a change of meaning and word class, such as in the word PREsent versus preSENT. Apart from the structure and functional similarity in linguistic perspective, there is also empirical evidence showing that native Cantonese adults use fundamental frequency as a primary prosodic cue in stress perception, in the same way they use it for tone perception (Chan, 2008). Also, there is evidence that sensitivity to lexical stress accounts for substantial variance in English word reading (Holliman, Wood, & Sheehy, 2010), and English language children at risk for reading difficulties demonstrated specific difficulty in lexical stress (Wood & Terrell, 1998). Our reasoning is also consistent with a transfer facilitation model (Koda, 2005), which suggests that L1 metalinguistic skills provide an initial foundation for the development of second-language skills that is essential to second-language reading. Therefore, it seems plausible that Cantonese lexical tone skills promote the development of English stress sensitivity and subsequent English word reading. However, we must state that this is an assumption, and future research should test this by including parallel English lexical stress measures. Also, although we speculated that either low-level acoustic-auditory processing or high-level lexical function processing can lead to the transfer of lexical tone to English reading, it is also possible that lexical tone transfer takes place as a result of the interaction between two levels of processing. This is an important topic relating to the mechanisms of lexical tone transfer and merits further investigation.
The present study was novel in examining the co-occurrence and dissociation between L1 word reading and L2 word reading among dual-language learners, and this provided a somewhat comprehensive understanding of different types of readers learning to read two writing systems (Durgunoglu, 2002a). Also, it was novel to examine phonological transfer at the suprasegmental level, and this may provide a new angle for the diagnosis and intervention of Chinese children with read difficulties. However, the sample size of the present study is considered to be small relative to the samples in studies of biliteracy acquisition. Also, there were limited numbers of measures used in the present study. Given that dual-language reading development involves the interaction of two sets of languages skills (e.g., Koda, 2005), future research should consider having parallel measures of L1 and L2 metalinguistic skills. In particular, both Chinese lexical tones and English lexical stress should be included. Also, both Chinese and English reading tasks were single word reading and were decontextualized, and future research should consider including sentence and short passage reading tasks. In this way, the phonetic and phonological realization of stress, including duration, loudness, and pitch, can be better assessed as compared with linguistic parameters of suprasegmental tones. In addition, the present study tested Hong Kong Chinese children who are English language learners. A possible extension of this study is to involve Mandarin-speaking children. Mandarin-speaking children in China learn Pinyin as an adjunct to learning to read characters in the early school years (see Note 2). Also, Mandarin has four tones rather than six. There is empirical evidence showing that Mandarin-speaking children performed better than Cantonese-speaking children in lexical tones and phonological sensitivity tasks (Leong, Cheng, et al., 2005). It would be theoretically interesting to test whether the weakness emerging in lexical tones and segmental phonological awareness in Cantonese children with reading difficulties in learning English would also be found in Mandarin-speaking children. In addition, given that there is research showing the facilitative effect of music training on lexical tone perception (Wong, Skoe, & Russo, Dees, & Kraus, 2007), a better control of children’s music learning experience is worth considering in future research on lexical tone transfer.
Despite these limitations, the present study offers a new direction in reading research across the majority of the world’s population of children who are bilingual and often biliterate. Two findings are of particular importance in relation to Chinese–English bilinguals from this work. First, children who have reading difficulties in both Chinese and English showed some tendency to manifest multiple deficits in phonological awareness and morphological awareness. Second, phonological sensitivity across both segmental and suprasegmental levels could distinguish children who were poor readers in English from those who had difficulties in Chinese reading only. Together, these findings support the idea that a variety of reading-related skills are important for word recognition, but some skills, such as phonological awareness, may be more important for alphabetic writing systems such as English, when compared to a nonalphabetic or morphosyllabic system such as Chinese, and the influence of phonology on L2 reading can be extended to the suprasegmental level, including lexical tones in Chinese.
Footnotes
Appendix
Acknowledgements
We thank the children and their parents who participated in our study, and we thank the many administrators and teachers at the children’s schools for their cooperation, as well as the research assistants who helped with data collection.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported, in part, by Hong Kong Government Research Grants Council Grant 448907 to Catherine McBride-Chang, and Direct Grant 2020910 from the Chinese University of Hong Kong to Catherine McBride-Chang and Xiuli Tong.