Theories linking morphological awareness to reading development

Carlisle (Reference Carlisle and Feldman1995, Reference Carlisle2000) was among the first to define morphological awareness as “conscious awareness of the morphemic structure of words and [the] ability to reflect on and manipulate that structure.” Carlisle’s (Reference Carlisle2000) study with English-speaking learners at Grades One, Three, and Five showed that morphological awareness was significantly associated with reading comprehension across the grade levels. One main reason perhaps is that morphologically complex words are prevalent, particularly in written texts (Nagy & Anderson, Reference Nagy and Anderson1984).

Over the past three decades, the interest in the role of morphology in reading, both in L1 and L2, has grown (L1 theories: Carlisle, Reference Carlisle and Feldman1995, Reference Carlisle2000; Kirby and Bowers, Reference Kirby, Bowers, Cain, Compton and Parrila2017; Levesque et al., Reference Levesque, Breadmore and Deacon2021; Perfetti, Reference Perfetti2007; Perfetti & Hart, Reference Perfetti, Hart, Vehoeven, Elbro and Reitsma2002; Ziegler and Goswami, Reference Ziegler and Goswami2005; L2 theories: Chung et al., Reference Chung, Chen and Geva2019; Koda, Reference Koda2005, Reference Koda, Koda and Zehler2008). According to the Lexical Quality Hypothesis (Perfetti, Reference Perfetti2007; Perfetti & Hart, Reference Perfetti, Hart, Vehoeven, Elbro and Reitsma2002), successful reading comprehension depends on high-quality lexical knowledge represented by orthography, phonology, morphosyntax, meaning, and the binding of these four features. Kirby and Bowers (Reference Kirby, Bowers, Cain, Compton and Parrila2017) later proposed that morphology is the binding agent of phonological, orthographic, and semantic features of words. For children with RD, morphological awareness may compensate for the weak phonological skills (as cited in Georgiou et al., Reference Georgiou, Vieira, Rothou, Kirby, Antoniuk, Martinez and Guo2023). According to Ziegler and Goswami’s (Reference Ziegler and Goswami2005) psycholinguistic grain-size theory, the availability of a specific processing unit during word reading depends on the orthographic consistency and availability of the spoken unit in oral language. To bypass their phonological difficulties during word reading, individuals with dyslexia may resort to larger and more consistent grain-size units such as morphemes.

Substantial empirical evidence shows that morphological awareness is a critical predictor of the three major aspects of learning to read (Verhoeven & Perfetti, Reference Verhoeven and Perfetti2022), namely, becoming linguistically aware, acquiring word identification skills, and learning to comprehend. In (2021), Levesque et al. attempted to illuminate precise mechanisms by which morphology impacts word reading, spelling, and reading comprehension, and they proposed that morphological decoding and morphological meaning analysis are two important pathways through which morphology predicts reading development. In the reading literature, another widely cited theory is the Simple View of Reading (SVR), which posits that reading comprehension is the product of decoding and listening comprehension (Gough & Tunmer, Reference Gough and Tunmer1986; Hoover & Gough, Reference Hoover and Gough1990). Duke and Cartwright (Reference Duke and Cartwright2021, p.530) pointed out that “Morphological awareness is not named in the SVR or many other models of reading, yet research has documented the contributions of morphological awareness to reading, including through word recognition and language comprehension.”

The review above is based on L1 reading literature. In L2 reading, researchers are more concerned about the mechanisms through which one can transfer L1 morphological awareness to facilitate L2 reading development. According to Koda’s (Reference Koda2005, Reference Koda, Koda and Zehler2008) transfer facilitation model, there are complex interactions among L1 morphological awareness, evolving L2 morphological awareness, L2 linguistic knowledge (e.g., vocabulary knowledge), and L2 reading skills that develop subsequently. Furthermore, compared to other L1 metalinguistic resources (e.g., phonological awareness), the transfer facilitation of morphological awareness in L2 reading is constrained by L1-L2 distance. This is confirmed by Ke and Xiao’s (Reference Ke and Xiao2015) review of studies on the cross-linguistic transfer of morphological awareness between morphosyllabic Chinese and morphophonemic English, two typologically distinct languages. They found that there were small-to-moderate correlations between L1 and L2 morphological awareness in Chinese-English bilingual reading development, and that L2 exposure and task demands had notable impacts on cross-linguistic transfer. These findings highlight the complex nature of morphological awareness transfer, underscoring the importance of considering both linguistic and contextual factors when examining bilingual reading development.

Morphological awareness as a marker of Chinese and English reading difficulties

Chinese orthography is morphosyllabic, with characters representing morphemes at the syllable level (DeFrancis, Reference DeFrancis1984). In contrast, English orthography is morphophonemic, as English spelling represents word sounds while maintaining consistency in morpheme representation, even when pronunciation changes (Abbott et al., Reference Abbott, Fayol, Zorman, Casalis, Nagy and Berninger2016). Despite these typological differences, morphological awareness is a marker of reading difficulties in learners of both Chinese (e.g., Cheah et al., Reference Cheah, McBride, Meng, Lee and Huo2025; Ho et al., Reference Ho, Chan, Tsang and Lee2002; Pan et al., Reference Pan, Meng, Lee, Ng and McBride2024; Peng, Reference Peng, Wang, Tao and Sun2017) and English (e.g., Siegel, Reference Siegel2008).

In Chinese, Ho et al. (Reference Ho, Chan, Tsang and Lee2002) were among the first to propose that Chinese dyslexic children can have impairments in one or more of the four cognitive and linguistic domains, including rapid naming, orthographic processing, visual processing, and phonological processing. Building upon and expanding Ho et al.’s (Reference Ho, Chan, Tsang and Lee2002) empirical work in Hong Kong, Peng et al. (Reference Peng, Wang, Tao and Sun2017) conducted a meta-analysis of 81 studies published between 1964 and 2015 to investigate the deficit profiles of Chinese children with RD, and they identified two additional domains: working memory (cognitive) and morphological awareness (linguistic). Interestingly, Peng et al. (Reference Peng, Wang, Tao and Sun2017) did not find any significant moderating effect of location (Hong Kong versus mainland), which was echoed by Cheah et al. (Reference Cheah, McBride, Meng, Lee and Huo2025) in the cognitive and linguistic deficit profiles of Chinese readers across different Chinese diaspora societies (including Beijing, Hong Kong, Taipei). Cheah et al. (Reference Cheah, McBride, Meng, Lee and Huo2025) assessed 91 children with dyslexia from Hong Kong, Beijing, and Taipei using the same set of Chinese literacy tests to identify dyslexia subtypes. Cluster analysis based on four cognitive-linguistic skills revealed four subtypes: orthographic deficit, phonological deficit, morphological deficit, and global deficit, each with distinct cognitive profiles. The distribution of these subtypes was consistent across the three regions, indicating that the nature of Chinese dyslexia is stable despite differences in literacy experiences. Cheah et al. (Reference Cheah, McBride, Meng, Lee and Huo2025) also pointed out that different definitions and diagnostic criteria have been adopted in existing literature and highlighted the importance of assessing morphological awareness, along with orthographic processing, rapid automatized naming, and phonological awareness across different Chinese-speaking societies. In English, Siegel (Reference Siegel2008) compared the relation of morphological awareness to reading and spelling skills of children with dyslexia, children who are typical readers, and children who are English language learners. She found that students with dyslexia had significantly lower scores than normally achieving readers on the morphological awareness tasks, and that there were no significant differences between English language learners and the English L1 students.

Lately, some researchers have begun to examine the potential co-occurrence of cognitive and linguistic profiles in two languages among Chinese-English bilingual learners with or without RD (e.g., with RD: Chung & Ho, Reference Chung and Ho2010; Gao et al., Reference Gao, Zheng, Liu, Nichols, Zhang, Shang, Ding, Weng and Liu2019; Zhou et al., Reference Zhou, McBride-Chang, Law, Li, Cheung, Wong and Shu2014; without RD: Zang et al., Reference Zang, Wong and Lau2024). For example, Zhou et al. (Reference Zhou, McBride-Chang, Law, Li, Cheung, Wong and Shu2014) tracked the developmental pattern of Chinese and English reading skills in Hong Kong children with and without dyslexia from six to eight years of age. They observed that children with dyslexia manifested clear difficulties in vocabulary knowledge and morphological awareness in both Chinese and English reading. In the mainland China context, Gao et al. (Reference Gao, Zheng, Liu, Nichols, Zhang, Shang, Ding, Weng and Liu2019) investigated reading acquisition in a large representative sample of 1,824 Chinese-English bilingual children in Grades Four and Five from both urban and rural schools in Beijing and identified a 36% co-occurrence in reading difficulties between L1 Chinese and L2 English. In other words, L1 Chinese reading difficulty significantly increases the risk of L2 English reading difficulty. Gao et al. (Reference Gao, Zheng, Liu, Nichols, Zhang, Shang, Ding, Weng and Liu2019) recommended that, in addition to the assessment of L1 reading skill, the assessment of L2 reading skills is critical in early L2 readers, as a poor reader in one language may not necessarily be a poor reader in another language. More recently, Zang (Reference Zang, Wong and Lau2024) conducted a study comparing the predictors of word reading between Chinese-English bilingual children and Chinese-dominant monolingual children in mainland China. They found that, in terms of Chinese word reading, both morphological and orthographic awareness emerged as significant predictors in the monolingual group, whereas in the bilingual group, only morphological awareness was significant, and the contribution of phonological awareness was insignificant in either group. The findings suggest that, for bilingual readers, morphological awareness is more influential than phonological awareness in the development of word reading ability.

Previous meta-analyses

To date, there has been little meta-analytic evidence examining the morphological deficits in bilingual learners with RD. However, there are pertinent meta-analyses on the effectiveness of literacy intervention in remediating reading difficulties in Chinese (Ruan et al., Reference Ruan, Maurer and McBride2024) or English (Goodwin & Ahn, Reference Goodwin and Ahn2010), morphological, oral language, and phonological deficits in monolingual readers of different languages and writing systems (Georgiou et al., Reference Georgiou, Vieira, Rothou, Kirby, Antoniuk, Martinez and Guo2023; Melby-Lervåg & Lervåg, Reference Melby-Lervåg and Lervåg2012), and the impacts of morphological awareness on bilingual reading development among normally developing children (Ke et al., Reference Ke, Miller, Zhang and Koda2021, Reference Ke, Zhang and Koda2023) and deaf and hard-of-hearing (DHH) students (Zhang D. et al., Reference Zhang, Ke, Anglin-Jaffe and Yang2023).

Ruan et al. (Reference Ruan, Maurer and McBride2024) compared the efficacy of different reading interventions on literacy outcomes for Chinese children. In particular, they found that morphological training had a significant and medium effect on improving Chinese literacy skills, which was comparable between younger and older children. Similarly, for English reading, Goodwin and Ahn (Reference Goodwin and Ahn2010) reported a significant and medium impact of morphological instruction on various reading-related outcomes for children with reading difficulties, English language learners, and English L1 children. Notably, Goodwin and Ahn (Reference Goodwin and Ahn2010) found that the effect sizes were larger for children with reading difficulties and English language learners.

From a cross-language and cross-writing system perspective, Georgiou et al. (Reference Georgiou, Vieira, Rothou, Kirby, Antoniuk, Martinez and Guo2023) reviewed 40 studies covering child readers of seven languages (Arabic, Chinese, English, Finnish, French, Greek, and Hebrew). Their meta-analytic findings indicated that children with dyslexia experience significant difficulties in morphological awareness compared to their age-matched controls, and that the group differences in morphological awareness increase with age. Ke et al. (Reference Ke, Miller, Zhang and Koda2021) also identified a significant moderating effect of age when examining morphological awareness as a cross-language shareable resource in child bilingual readers without RD. Ke et al. (Reference Ke, Miller, Zhang and Koda2021) conducted a systematic review and meta-analysis of 34 studies and investigated the strength of associations between L1 and L2 morphological awareness and L2 reading development in child L2 learners. The correlation effect sizes between L2 morphological awareness and L2 reading comprehension are larger as age increases; yet the correlation effect sizes remain unchanged between L1 morphological awareness and L2 reading outcomes. In another meta-analysis, Ke et al. (Reference Ke, Zhang and Koda2023) compared the correlation effect sizes between morphological awareness and reading outcomes in L1 and L2 and observed that the effect sizes are consistently larger in L2. Ke et al. (Reference Ke, Zhang and Koda2023) conjectured that bilingual readers might rely more heavily on morphological awareness when reading in their L2 than in their L1.

In addition, Zhang D. et al. (Reference Zhang, Ke, Anglin-Jaffe and Yang2023) reported the first meta-analysis examining the correlations between morphological awareness and reading-related abilities among deaf and hard-of-hearing (DHH) students, revealing strong associations with word reading, vocabulary knowledge, and reading comprehension across multiple languages and educational contexts. Their findings underscore the critical role of morphological awareness as a foundational skill for literacy development in DHH learners, paralleling its significance in other populations with reading challenges, such as children with dyslexia. By highlighting the importance of morphology-based instruction for diverse learner groups, including those with sensory or learning disabilities, Zhang D. et al. (Reference Zhang, Ke, Anglin-Jaffe and Yang2023)’s study not only fills a notable gap in the literature on DHH students but also reinforces the broader relevance of metalinguistic skills within disability studies.

Viewed collectively, the diagnostic value of morphological awareness assessment has been confirmed by previous empirical and meta-analytic studies for monolingual learners with reading difficulties in Chinese only, English only, or across different languages and writing systems; however, to date, there is little meta-analytic evidence evaluating morphological awareness deficits in bilingual children with reading difficulties from a cross-language perspective, as well as the moderating factors that may influence the magnitude of these deficits. Accordingly, the present research focused on four potential moderators—RD severity, age, diglossia context, and methodological design—which were theoretically and empirically motivated to account for potential sources of heterogeneity in effect sizes across studies. First, RD severity (i.e., poor readers vs. diagnosed dyslexia) was included because previous research has demonstrated that the magnitude of cognitive and linguistic deficits may differ depending on the strictness of diagnostic criteria and the severity of reading difficulties (Cheah et al., Reference Cheah, McBride, Meng, Lee and Huo2025). Second, age was examined as a moderator given robust evidence that the role of morphological awareness in reading development increases with age and grade level, and that deficits may become more pronounced over time (Georgiou et al., Reference Georgiou, Vieira, Rothou, Kirby, Antoniuk, Martinez and Guo2023). Third, the diglossia context (Cantonese-speaking vs. Mandarin-speaking societies) was considered to capture the influence of linguistic environment and instructional practices, as prior studies have suggested that dialectal and orthographic differences may impact the development and manifestation of reading-related skills (McQuade et al., Reference McQuade, Pan, Ho, Ong, Ng, Meng and McBride2024; Peng et al., Reference Peng, Wang, Tao and Sun2017). Finally, methodological design (types of control or comparison groups) was included to address the possibility that study design choices—such as the selection of control groups—could systematically affect observed effect sizes (Goodwin & Ahn, Reference Goodwin and Ahn2010; Georgiou et al., Reference Georgiou, Vieira, Rothou, Kirby, Antoniuk, Martinez and Guo2023). By systematically examining these moderators, the meta-analysis aimed to provide a more nuanced understanding of the variability in morphological awareness deficits among Chinese-English bilingual children with reading difficulties. While our primary focus is on RD/dyslexia, the methodological framework and cross-linguistic perspective adopted in this study are relevant to broader disability studies, including studies of the DHH population, as they emphasize the importance of recognizing diverse learner profiles and the intersectionality of language and disability. Our results advocate for more inclusive diagnostic criteria and intervention strategies that account for the unique experiences of bilingual children with disabilities, thereby informing both research and practice in disability studies and related fields.

Literature search

The literature search proceeded in three major stages (as shown in Figure 1). First, an initial search was completed in May 2024. This included: (1) a Boolean search of databases (Web of Science and PsycINFO) with a combination of four keywords (i.e., reading difficulties OR dyslexia AND Chinese AND English) that appeared in the abstracts; and (2) a manual search was conducted by checking the references in existing meta-analyses and reviews (i.e., Ke et al., Reference Ke, Miller, Zhang and Koda2021, Reference Ke, Zhang and Koda2023; Goodwin & Ahn, Reference Goodwin and Ahn2010; Georgiou et al., Reference Georgiou, Vieira, Rothou, Kirby, Antoniuk, Martinez and Guo2023; McBride et al., Reference McBride, Wang and Cheang2018; Ruan et al., Reference Ruan, Maurer and McBride2024). These steps yielded 12 primary studies and 25 independent samples. Second, to update the database, another manual search was conducted via Google Scholar in October 2024. As a result, two studies (Chung et al., Reference Chung, Li, Lam, Fung and Liu2022; Tong et al., Reference Tong, Deng and Tong2022) with four independent samples were added. Third, an expanded search was conducted in October 2025 to ensure comprehensive coverage. This included: (1) updating the systematic database search across all five databases (Web of Science, PsycINFO, Scopus, ERIC, and LLBA); (2) manual-searching six key journals that specialize in reading research (i.e., Annals of Dyslexia, Dyslexia, Journal of Research in Reading, Reading Research Quarterly, Reading & Writing, and Scientific Studies of Reading), following the search strategy of a relevant prior meta-analysis (Melby-Lervåg & Lervåg, Reference Melby-Lervåg and Lervåg2012); (3) conducting a forward and backward citation search on Google Scholar for all included studies; and (4) setting the study publication time between 2000 and 2025. We identified one potential study (i.e., McQuade et al., Reference McQuade, Pan, Ho, Ong, Ng, Meng and McBride2024), but decided that the dataset might overlap with Pan et al. (Reference Pan, Meng, Lee, Ng and McBride2024)’s. Therefore, no additional studies meeting the inclusion criteria were identified during the final expanded search. The final dataset thus consists of 14 primary studies (comprising 29 independent samples).

Figure 1.

Literature search procedures.

Inclusion and exclusion criteria

Given the scope of this search, the selected studies must (1) include children with reading difficulties, (2) have a Chinese morphological awareness measure and another measure of either phonological awareness and word reading, (3) include reading-related measures in both Chinese and English, (4) written in English, and (5) report Hedge’s g or provide sufficient statistical information to calculate the effect size. Review and repetitive reports were excluded. To assess study quality, we reported instrument reliability as a key indicator, following the approach of Ke et al. (Reference Ke, Miller, Zhang and Koda2021) in their meta-analysis. As shown in Appendix 1, the mean and median reliability coefficients were all above .80, indicating that the reading measurements were reliable.

Coding and data analyses procedures

To answer RQ1, descriptive information of each independent sample was entered into an Excel file (see Table 1). Coding was first entered by the second author. The first author then recoded all cells. To answer RQ2 and RQ3, coding of the effect sizes was conducted by the two coders in a similar way. The intercoder reliability (Cronbach’s alpha) for effect size was .90. For categorical variable coding, Cohen’s kappa was .89. The first author resolved the inconsistencies by triple-checking. For studies that did not report Hedge’s g, effect size conversion was conducted via the Campbell Collaboration calculator (Littell & White, 2018). Regarding the size of g, the rule of thumb is that small, medium, and large effect sizes are at or above .30, .50, and .80, respectively.

Table 1.

Definitions, diagnostic assessment, and diagnostic criteria adopted in the primary studies

^a One: less severe RD; Two: more severe RD with diagnosed dyslexia.

^b a review of the subtests suggested that no MA test was reported in Chung and Ho (Reference Chung and Ho2010). N.A. = not applicable.

The coding of the moderator focused on four factors: (1) RD severity (two levels: poor readers vs. dyslexia); (2) Diglossia context (two levels: Cantonese-speaking vs.

Mandarin-speaking); (3) methodological design (two levels: Chinese RD vs. age-matched /reading control; Chinese RD vs. age-matched /English RD/bilingual RD); and (4) age (in years). Comprehensive Meta-analysis (CMA) Version 4 was employed to conduct subsequent analyses. The complete CMA files with coding are available at: https://osf.io/s3wz8/overview?view_only=4cf91fb65bec4fe48da0cd4e225db505.

First, publication bias was checked via three steps: (1) both published journal articles and an unpublished doctoral dissertation (i.e., Shen, Reference Shen2023) were included in the meta-analysis; (2) Funnel plots for random-effects models were used to determine the presence of retrieval bias. No asymmetric funnels were identified (see Figures 2–7). (3) The results of Duval and Tweedie’s (Reference Duval and Tweedie2000) trim and fill were reported in Table 2 below.

Figure 2.

Funnel plot for Chinese morphological awareness.

Figure 3.

Funnel Plot for Chinese Phonological Awareness.

Figure 4.

Funnel plot for Chinese word reading.

Figure 5.

Funnel plot for English morphological awareness.

Figure 6.

Funnel plot for English phonological awareness.

Figure 7.

Funnel plot for English word reading.

Table 2.

Descriptive statistics for the Mean Differences Between the RD and Control Groups

Note. MA = morphological awareness; PA = phonological awareness; N.A. = not applicable.

Second, potential outliers were examined for the six target outcome variables (i.e., morphological awareness, phonological awareness, and word reading in Chinese and English, respectively) and removed if the standard residuals were bigger than 2.5 in either positive or negative direction (see also Ke et al., Reference Ke, Miller, Zhang and Koda2021). Therefore, three independent samples from Chung et al. (Reference Chung, Li, Lam, Fung and Liu2022) were removed for the outcome of Chinese morphological awareness, and Chung and Ho (Reference Chung and Ho2010), as well as Kalindi et al. (Reference Kalindi, McBride, Tong, Wong, Chung and Lee2015), were removed for the outcome of English word reading.

Finally, in order to investigate whether there was variability among the correlations across primary studies, a heterogeneity test (Q test; Hedges & Olkin, Reference Hedges and Olkin1985) was conducted. A significant value on the heterogeneity test would provide evidence for variability among the correlations, and I² indicates the proportion of the heterogeneity that is real rather than due to chance. Given that three moderator variables in this study are categorical, a heterogeneity test (Q test) was performed based on random-effects modeling, both within and across the subgroups created by the moderator analyses. Additional meta-regressions were run with age (in years) as a continuous variable for outcomes with at least 10 independent samples (after Ke et al., Reference Ke, Zhang and Koda2023).