Abstract
In this study, we examine the impact of the COVID-19 pandemic on children’s language development and experiences, particularly focusing on a socioeconomically disadvantaged population of Spanish–English emergent bilinguals. Using propensity score matching, we compared first- and third-grade children in the pandemic (n = 73) and pre-pandemic (n = 73) cohorts across multiple dimensions of language skills and experiences. Results revealed no significant differences between cohorts in English and Spanish language proficiency measures, English reading abilities, language experiences in school, relative English experience at home, or reading frequency, after correction for multiple comparisons. However, pandemic-cohort children demonstrated enhanced benefits from increased English and Spanish language experience in English language and literacy skills than their peers in the pre-pandemic cohort. The observed differential effects of English and Spanish experience between cohorts suggest potential compensatory mechanisms in language learning during disrupted formal education.
Keywords
The COVID-19 pandemic caused interruptions in learning for all K–12 students in the United States due to widespread school closures (Kuhfeld et al., 2020). This unprecedented situation led to a rapid change in how teaching and learning occurred, such that all instruction and learning moved from face-to-face to remote (Abuosbeh et al., 2024). However, millions of emergent bilinguals (EBs), who speak a language other than English at home, and their families were disproportionately affected by such a significant loss of exposure to learning opportunities (Sayer & Braun, 2020). Often coming from families with constraints in material and economic resources, such as limited access to books, technology, and reliable internet, thus facing additional barriers to academic success (Gibney & Henry, 2020), EBs are especially vulnerable to slower language and literacy gains due to the disruptions and insufficient support from schools (Crosson & Silverman, 2022; Umansky, 2020). Moreover, during the pandemic, these students may have experienced substantial changes in their language-learning environments, including reduced exposure to English in academic settings and potential shifts in home language patterns (Sheng et al., 2021).
Research has established that language experience, including exposure to and use of both languages, plays a critical role in bilingual language development (Hammer et al., 2020; Shen & Choi, 2024). However, the pandemic’s impact on these patterns of language experience and their relation to language and literacy outcomes remain understudied among EBs. The forced shift to remote learning may have altered not only the quantity but also the quality of language exposure, potentially affecting different aspects of language development differently. The current study addresses these knowledge gaps by examining how this unprecedented change due to COVID-19 affected the language and reading skills, as well as the language experiences, of Spanish–English EBs. Leveraging the pandemic as a natural experiment, we compared matched samples of Spanish–English EBs who were tested before and during the pandemic.
Language and Literacy Development for Spanish–English EBs
The development of language and literacy skills in Spanish–English EBs involves multiple interconnected components that develop across both languages at varying degrees of proficiency. Core language abilities, including vocabulary, phonological awareness (PA), morphological awareness (MA), and syntactic awareness (SA), form the foundation for literacy development and academic success (Shen & Coker, 2023; Sohail et al., 2022). These skills are typically assessed in prior research through standardized tasks: receptive and expressive vocabulary is commonly measured through recognition or picture-naming tasks (e.g., Peabody Picture Vocabulary Test [PPVT]; Dunn, 2019); PA is measured through phoneme segmentation or elision tasks (e.g., Comprehensive Test of Phonological Processing [CTOPP]; Wagner et al., 2013); MA is measured through morphological production and judgment tasks (e.g., Cho et al., 2011); and SA is measured through sentence repetition or grammaticality judgment tasks (e.g., Peña et al., 2010). Such tasks are similar in nature to those employed in the present study, where children’s performance was measured with standardized and widely used assessments of language and literacy development. However, no prior research has applied these standardized assessments in both English and Spanish to examine pandemic-related changes specifically among Spanish–English EBs in dual language immersion settings, making this study uniquely positioned to explore whether, and how, the pandemic influenced bilingual children’s language skills in ways that cannot be inferred from monolingual-focused or pre-pandemic studies.
These language skills often develop interdependently across languages, with research showing facilitative relations (i.e., cross-language transfer) between Spanish and English development (Goodrich et al., 2023; Proctor et al., 2017). For instance, PA skills have demonstrated strong cross-linguistic transfer, supporting reading acquisition in both languages (Durgunoğlu et al., 1993; Melby-Lervåg & Lervåg, 2011). Similarly, MA development can benefit from children’s exposure to both languages’ word formation rules, particularly given the prevalence of Spanish–English cognates (Ramírez et al., 2013). Importantly, due to EBs’ emerging English oral proficiency, they often rely on their Spanish heritage language skills to support English reading (Shen & Goodrich, 2024). Thus, during the early years of formal schooling (kindergarten through third grade), Spanish–English EBs’ oral and literacy development in Spanish plays a critical role in scaffolding their English literacy growth.
Reading development in Spanish–English EBs builds on these foundational language skills while following unique trajectories compared to monolingual peers. English decoding skills rely heavily on PA and letter-sound knowledge, while reading comprehension draws upon both decoding and broader language proficiency, such as vocabulary and SA (Hoover & Gough, 1990; Proctor et al., 2012). English monolingual children learn to associate print with speech sounds in their native language with established oral vocabulary knowledge supporting such mapping through decoding and comprehension of words read (Long, 2001). However, for EBs who are not yet proficient in English, schools frequently represent their first exposure to formal English reading instruction (Bedore & Peña, 2008), and they are tasked with simultaneously learning to understand spoken English and decode written English words. Consequently, it is not surprising to observe initial disparities in reading abilities at school entry, along with differences in reading growth rates between native English speakers and EBs (Johnson, 2023; Soland & Sandilos, 2021). Research has also demonstrated that Spanish language skills can support English reading development, suggesting the importance of maintaining and developing both languages (Huang et al., 2022; Proctor et al., 2017). However, this development is highly dependent on consistent exposure and opportunities for practice in both languages (Cummins, 1979). As such, young EBs may be disproportionately affected by school closures during the pandemic, particularly due to the lack of formal instruction in their home language.
Bilingual Language and Reading Experiences
Language experience, encompassing both exposure to and active use of each language, plays a fundamental role in bilingual development (Place & Hoff, 2016; Shen & Choi, 2024; Sussman et al., 2023). Research has consistently demonstrated that the quantity and quality of language input are associated with bilingual children’s vocabulary growth and overall language proficiency (Paradis & Jia, 2017; Place & Hoff, 2011). For Spanish–English EBs in U.S. schools, language experience typically follows distinct patterns: English exposure predominantly occurs in academic settings through teacher and peer interactions, while Spanish exposure often centers on home and community environments.
Importantly, the type of educational model that EBs are enrolled in greatly shapes their language experiences. In some states and districts, bilingual children have access to dual language immersion (DLI) programs, which aim to develop proficiency in both English and Spanish through structured instruction in both languages. In contrast, other states have enacted restrictive language education policies that limit or prohibit dual language programming, resulting in English-only instructional models (Gándara & Hopkins, 2010). For students in Spanish–English DLI schools, however, such distribution of language experience across contexts may not be as distinct, since both English and Spanish are the languages of instruction at school. However, with school closures and increased time spent at home with families (thus less time spent at school learning English), children’s language experience at home likely plays a more prominent role, compared to that before the pandemic (G. Li & Lin, 2023). This shift in language experience may have increased reliance on Spanish for some children while simultaneously reducing opportunities for structured English input and cross-linguistic transfer in academic contexts.
Reading frequency and habits represent a crucial component of language experience, particularly for language and reading development (Guo et al., 2021; Schiefele et al., 2012). Studies have shown that access to and engagement with literacy materials in both languages support not only reading skills but also vocabulary development, reading comprehension abilities, verbal fluency, and general knowledge development (Schiefele et al., 2012; Shen & Shea, 2025; Silinskas et al., 2020). Research has also shown that variations in home literacy environments and school resources can lead to significant differences in children’s reading habits and, consequently, their language and literacy development (Jabbar et al., 2021; Yi et al., 2019). With school closure leading to fewer resources children could leverage, it is possible that Spanish–English EBs may engage less frequently in reading, given that they disproportionately live in families with socioeconomic disadvantage.
Impacts of the COVID-19 Pandemic
Language and literacy development requires consistent practice and instruction (Allington & McGill-Franzen, 2018), with schools playing a critical role by offering communication opportunities and explicit reading instruction. Prolonged disruptions, like those caused by COVID-19, interrupt these opportunities, leading to significant changes in children’s language and reading development trajectories (Engzell et al., 2021; Nevo, 2024). Studies conducted around the world have suggested learning losses in language and literacy skills. For instance, Engzell et al. (2021) revealed a learning loss equivalent to one-fifth of a school year among elementary school children in the Netherlands. A meta-analysis of 42 studies from 15 countries found that students across elementary and secondary grades experienced a learning loss of about 35% of a school year (Betthäuser et al., 2023). Such learning losses were particularly pronounced for children from low-socioeconomic status (SES) backgrounds (Betthäuser et al., 2023; Engzell et al., 2021; Shaul et al., 2024) and for younger children in the early elementary grades (Hammerstein et al., 2021; König & Frey, 2022). Children from lower SES backgrounds may have faced limited access to educational and human resources, along with unreliable technology and internet connectivity, during remote learning (UNESCO, 2021). Younger children often require substantial instructional support and scaffolding, as their skills for independent and self-regulated online learning are still developing. However, limited teacher experience with digital materials for young learners, the need for significant parental involvement, which some families find challenging, and children’s lack of interest or ability to engage meaningfully in online learning can hinder the success of distance education at this age (Nevo, 2024; Relyea et al., 2023).
In typical school settings, EBs benefit from rich peer interactions and scaffolded oral support to develop both social and academic language skills. However, the shift to remote learning disrupted key strategies for making input comprehensible, such as verbal and nonverbal cues, and replaced oral interaction with reading and writing-focused tasks (H. Li et al., 2024). This shift particularly impacted emergent readers in the early grades, as they lost access to differentiated, small-group literacy instruction and the oral scaffolding essential for building foundational reading skills (Sayer & Braun, 2020). Moreover, Spanish–English EBs are disproportionately from low-SES backgrounds with limited home resources to support children’s home-based learning (Gibney & Henry, 2020). Further, children in DLI programs face additional challenges, as these programs may lack infrastructure for online instruction and materials in both languages (Sugarman & Lazarin, 2020). Teachers in dual language programs also reported limited time devoted to bilingual instruction (especially in the target language other than English) to maintain the DLI model (Renn et al., 2025), as well as children’s limited engagement in social interaction in both languages (Marcus, 2021).
Indeed, some studies have also suggested disproportionate influences of the pandemic on bilingual children. For example, in their study of second-grade children in Germany, Förster et al. (2023) revealed a widening of the gap in reading achievement between students with and without a migration background over time. Likewise, Relyea et al. (2023) studied U.S. elementary school children in Grades 3 through 5 and revealed a gap in reading gains between EBs and their English-fluent peers during 2020–2021, with EBs significantly lagging behind their peers. An analysis with nationally representative data from 5 million U.S. students in Grades 3 through 8 also showed that the negative effects of the COVID-19 pandemic were more pronounced for Hispanic children (Kuhfeld et al., 2023).
In contrast, other research has suggested no adverse effects of the pandemic on children’s language and literacy development (Förster et al., 2023; Gore et al., 2021; X. Sun et al., 2023). For instance, Förster et al. (2023) compared Grade 2 students from two pandemic cohorts with children of four pre-pandemic cohorts on a reading comprehension measure in Germany. Interestingly, children in the first pandemic cohort outperformed their peers in the pre-pandemic cohorts, while the second cohort demonstrated reading comprehension skills comparable to the pre-pandemic groups. The researchers suggested that parents spent more time reading with their children during lockdowns, which helped offset the pandemic’s negative effects on reading development. In another study, Abuosbeh et al. (2024) worked with English-French bilinguals in Grades 1 and 2 in Canada who were enrolled in French immersion programs and found that the pandemic had a moderately negative impact on French vocabulary but no significant effect on French PA or word reading, or English skills, highlighting the need to enhance L2 vocabulary input in virtual learning settings. The authors concluded that this might be due to students typically coming from high-SES families with substantial literacy support from their parents. X. Sun et al. (2023) compared reading skill growth in monolingual English, Chinese–English, and Spanish–English students at elementary school ages. All participants came from relatively high-SES homes and attended English-only programs. Bilingual students showed greater growth in English vocabulary than monolinguals and maintained age-appropriate progress in English PA and word reading, likely due to their independent reading practices, rich English exposure in the community, and high-SES backgrounds. Sheng et al. (2021) matched a group of children in the pandemic cohort to a pre-pandemic cohort of Mandarin–English bilingual children (ages 4–8) and found that the two groups performed comparably in English sentence comprehension and production; however, the pandemic group outperformed the pre-pandemic group in Mandarin skills, likely due to the richer Mandarin language environment observed in the pandemic group.
As such, extant literature on the impacts of the pandemic on bilingual children largely focuses on language and literacy development; less is known about the pandemic impacts on their language experience (cf. Sheng et al., 2021) and reading frequency in their two languages. Moreover, although language experience (Sheng et al., 2021) and reading frequency (X. Sun et al., 2023) have shown importance in bilinguals’ language and literacy acquisition during the pandemic, whether they may function differently before and during the pandemic is unknown. In addition, even less is known about the impact of the pandemic among children in DLI programs. Spanish–English EBs in DLI programs may have experienced the pandemic in ways that fundamentally differ from monolinguals or bilinguals in different instructional settings due to their participation in dual language models, the distribution of English and Spanish across home and school contexts, and their reliance on varied sources of linguistic input. The present study fills this gap by simultaneously assessing bilingual children’s experiences in English and Spanish, their reading behaviors in both languages, and their performance on widely used standardized measures. Examining this could provide valuable insights into how shifts in language experience and reading habits during the pandemic may have influenced bilingual children’s development differently compared to pre-pandemic periods, shedding light on the role of environmental and contextual factors in shaping bilingual language acquisition.
The Current Study
Leveraging the COVID-19 pandemic as a natural experiment, we aim to contribute to the literature by investigating how educational disruptions affected bilingual language skills and English reading across multiple dimensions. These include vocabulary, PA, MA, and SA in English and Spanish, English decoding, and English reading comprehension. Additionally, we examined whether the role of language experiences in language and literacy development shifted as a result of the pandemic. To do so, we first matched a pandemic cohort with a pre-pandemic cohort of children on child and school characteristics. We then compared whether differences in language and literacy skills and language experiences exist across the two groups. Three research questions guided our study: (1) Did the pandemic significantly alter bilingual language skills across vocabulary, PA, MA, and SA in English and Spanish, as well as English decoding and reading comprehension? (2) Were there differences in language experiences and reading frequency between pandemic and pre-pandemic cohorts? (3) Did language experience and reading frequency play differential roles in skill development between pandemic and pre-pandemic cohorts?
Materials & Methods
Participants
We leveraged a secondary dataset from a larger multiyear study aiming to examine Spanish–English bilingual children’s language and literacy development (Huang et al., 2022). The original analytical sample for the current study included Spanish–English emergent bilinguals (N = 194) from both the pandemic (n = 82) and the pre-pandemic (n = 112) cohorts. All children were recruited from Spanish–English DLI programs from five elementary schools in a Southwestern city in the United States, where there was a high concentration of students of Mexican heritage. The DLI program is designed to support students in becoming proficient in two languages, developing literacy skills in both, achieving strong academic outcomes, and building intercultural competence. The program follows a balanced 50/50 model, where instructional time is evenly distributed between English and Spanish. Students receive literacy instruction separately in each language through dedicated periods for Spanish language arts and English language arts. All teachers were state-certified bilingual education teachers and delivered instruction independently in both languages. Thus, unlike models where students rotate between a partner English teacher and a partner target-language teacher, each classroom had one teacher who provided both the English and target-language portions of instruction. In terms of content instruction, the district assigns specific subjects to each language: mathematics is taught in Spanish, while science and social studies are delivered in English. Prior to the pandemic, schools operated under regular in-person instruction. During the pandemic, schools were closed and transitioned to fully remote instruction. To support students and families in this shift, several schools secured funding to provide iPads and/or internet services to families in need. Data were collected in fall 2019 for the pre-pandemic cohort and in fall 2020 for the pandemic cohort.
All of the participants were classified as English language learners by their schools. At the time of data collection, the classification process began with a home language survey completed by parents or guardians during school registration. If the survey indicated that a language other than English was spoken at home, the student was referred for an English language proficiency assessment. Students who scored below the established proficiency benchmarks were formally identified as English learners and became eligible for language assistance programs.
Table 1 presents the demographic characteristics of children in the pandemic cohort and pre-pandemic cohort pools separately. Children in the pandemic cohort were on average 7.81 years old. Over half were female (53.67%). About 52.46% were in Grade 1 and 47.56% were in Grade 3, as we relied on a secondary dataset drawn from a larger longitudinal project, which was designed follow children at two early elementary grade levels to capture language and reading relationships across different reading development stages. The majority of them (65.85%) were in schools with 80% of children qualified for free or reduced lunch. About 75.61% of them were in schools with a zip code median income of $30,000–$40,000. Children in the pre-pandemic cohort pool were on average 7.59 years old. Over half of them were female (52.68%). About 48.21% were in Grade 1 and 51.79% were in Grade 3. The majority of them (74.11%) were in schools with 80% of children qualified for free or reduced lunch, with a school zip code median income of $30,000–$40,000.
Demographic Information Before and After Propensity Score Matching
Measures
English Reading Measures
The Letter Word Identification (LWI) and Word Attack (WA) subtests in the Woodcock-Johnson IV Tests of Achievement (WJ-IV ACH; Schrank et al., 2014a) were used to evaluate English decoding skills, while the Passage Comprehension (PC) subtest was used to measure English reading comprehension skills. According to the publisher, the reliability for the three subtests (LWI, WA, PC) ranges from .94 to .98, .92 to .96, and .87 to .93, respectively. Given the high correlation between LWI and WA, we used the average scores as the English Decoding variable.
Language Measures in Spanish and English
We evaluated children’s bilingual language skills in both Spanish and English in the following domains: phonological awareness (PA), morphological awareness (MA), vocabulary, and syntactic awareness (SA). We used the Comprehensive Test of Phonological Processing–Second Edition (CTOPP-2; Wagner et al., 2013) and the Test of Phonological Processing in Spanish (TOPPS; Francis et al., 2001), respectively, to measure PA in English and Spanish. The publisher-reported reliability indices are .88–.94 for CTOPP-2 and .83 for TOPPS. To measure MA, we used the early lexical morphology measure (ELMM) in English and its Spanish parallel form (ELMM-S; see Marks et al., 2022). Both the English and Spanish MA tasks evaluate children’s knowledge of compound morphology and derivational morphology. The reliability indices for ELMM and ELMM-S are .93 and .95. We also used the English and Spanish vocabulary subtests in the Woodcock-Johnson IV Tests of Oral Language (WJ-IV OL; Schrank et al., 2014b) to evaluate expressive vocabulary. The reliability indices for both subtests range from .77 to .94. Finally, to assess SA in both English and Spanish, we used the Morphosyntax subtest in Bilingual English Spanish Assessment - Middle Extension (BESA-ME; Peña et al., 2010). The reliability estimate is .88 for both English and Spanish.
Teacher Survey
The Teacher Survey included six questions about child participants’ use of Spanish and English in school. Three of the questions are about Spanish experience, whereas the other three parallel questions are about English experience. Specifically, teachers reported the frequency of participants’ use in each language with the teacher, with their peers, and with other adults in school on a Likert response scale. We treated responses of “Don’t know” as missing data. A composite score averaged across the three English experience questions was created for “Teacher-reported English experience,” and a parallel score was also created for “Teacher-reported Spanish experience” across the three ratings of participants’ Spanish experience in school.
Child Survey
A child survey was administered with each child participant to gather information about their demographic backgrounds and language experience. There were six questions about language experience that required the child participants to respond using a Likert scale. For example, participants responded to questions such as “How frequently do you use English at breakfast in the morning?” We created a “child-reported relative English experience” composite averaging child participants’ responses across the six language experience questions. The child survey also included five questions about their frequency of reading various genres (i.e., magazines, fiction, nonfiction, comic books, emails, and webpages) in English and five parallel questions in Spanish. For example, participants were asked “How often do you read magazines in English/Spanish outside of school?” and responded on a Likert-type scale. Two composite variables, “English reading frequency” and “Spanish reading frequency,” were created by averaging participants’ answers across the reading frequency questions.
The research team developed the Teacher Survey and Child Survey by drawing on both prior published studies in the field as well as a previous study we conducted with a similar population (Huang et al., 2021). The complete surveys can be found in Appendix A and Appendix B of Huang and Shen (2025).
Procedure
All assessment procedures followed standardized protocols across cohorts; however, the mode of administration differed due to pandemic-related restrictions. For the pre-pandemic cohort, all assessments were conducted in person at the participating schools, with trained research assistants meeting individually with children in a quiet space. For the pandemic cohort, we shifted to fully remote administration via Zoom, using digital versions of the tasks and screen-sharing. All research assistants followed the same scripted instructions and scoring protocols, and children were tested individually in a quiet space at home. Their caregivers assisted only with technology and logistics and did not remain present to support item responses once the testing session began. Across both cohorts, children completed two sessions approximately four or more days apart, with the order of English and Spanish sessions counterbalanced. Sessions were administered by trained research assistants who were native or heritage speakers of the target language and followed the same standardized scripts across modes. To maintain data quality, all raw entries were verified by a quality-check team, and 20% of assessments were double-scored for accuracy.
Analytical Approach
To examine the impacts of the COVID-19 pandemic on Spanish–English EB children’s language development and experiences, we first used propensity score matching to match a cohort of pandemic children to a pre-pandemic cohort. The propensity score matching was conducted in R with the R package matchit (Ho et al., 2007). The propensity scores were estimated using a logistic regression model that included child characteristics (grade and sex), as well as school characteristics including percentage of children qualified for free or reduced lunch and school zip code median income as predictors of treatment assignment. One-to-one and nearest neighbor matching and matching without replacement were used to match one child in the pandemic cohort to one child in the pre-pandemic cohort. We implemented nearest neighbor matching with a 1:1 ratio without replacement, incorporating a caliper of .1 standard deviations to restrict matches to units with sufficiently similar propensity scores and improve covariate balance quality. After matching, independent sample t-tests were used to compare children’s language development and experiences between children in the pandemic and pre-pandemic cohorts.
To explore differential roles of language experience between children in the pandemic and pre-pandemic cohorts, linear regression models were used with each of the language development and experience variables as the dependent variable. The independent variables included (1) pandemic cohort (pandemic cohort = 1, pre-pandemic cohort = 0); (2) language experience variables: teacher-reported English and Spanish experience, child-reported relative English language experience, and English and Spanish reading frequencies; and (3) the interactions between pandemic cohort and each of the language experience variables. All language experience variables were grand mean-centered. Simple slope analyses were conducted following significant interactions. Missing data were handled using listwise deletion. This approach was appropriate given the low level of missing data in the matched sample (see Table 3).
Results
Propensity Score Matching
Our original sample included 82 children and 112 children in the pandemic and pre-pandemic cohorts, respectively. Before matching, the pandemic cohort had a mean propensity score of .47 and the pre-pandemic cohort had a mean propensity score of .39. Figure 1 and Table 2 show the covariate balance before and after matching. The matching procedure successfully balanced the pandemic and pre-pandemic groups across all covariates. Results revealed substantial improvements in standardized mean differences, with the propensity score difference reducing from .47 to .003, and SES variables achieving perfect balance (standardized mean differences of .00). All covariates met the conventional threshold of |.10| for adequate balance. The final matched sample retained 73 children in the pandemic cohort (89% of the original sample in the pandemic cohort) and 73 matched children in the pre-pandemic cohort, providing sufficient statistical power while maintaining high covariate balance. Following successful matching, we conducted the subsequent analyses on the matched dataset to estimate the impacts of the pandemic on language skills and experiences.
Covariate balance before and after matching.
Summary of Covariate Balance Before and After Propensity Score Matching
Table 1 presents the demographic information for pandemic and pre-pandemic cohorts after matching. Children in the pandemic and pre-pandemic cohorts were on average 95.11 months/7.93 years old (range 59.60–125 months) and 89.82 months/7.49 years old (range 72–116 months), respectively. Over half of them were female (pandemic cohort: 53.42%; pre-pandemic cohort: 57.53%). About half were in Grade 1 (pandemic cohort: 46.58%; pre-pandemic cohort: 50.68%), with the rest in Grade 3. For both groups of matched children in the pandemic and pre-pandemic cohorts, the majority (73.97%) were in schools with 80% of children qualified for free or reduced lunch, with a school zip code median income of $30,000 to $40,000.
Impacts of COVID-19 Pandemic on Language Development and Experiences
Language skills and language experience differences between children in the pandemic and pre-pandemic cohorts and results from t-tests are presented in Table 3. Results revealed three significant differences; that is, children in the pandemic cohort scored higher in Spanish SA (t[135] = 2.01, p = .046, Cohen’s d = .34) but had fewer Spanish language experiences at school (t[128] = 2.13, p = .035, Cohen’s d = .38) and engaged in Spanish language reading less frequently (t[137] = 2.01, p = .024, Cohen’s d = .39) compared to their peers in the pre-pandemic cohort. However, the differences were not significant after correction for multiple comparisons (see Table 3 for both Benjamini-Hochberg false discovery rate (FDR)-adjusted and Bonferroni-adjusted p values).
Comparisons in English Reading and Bilingual Language Skills Between Matched Pandemic and Pre-Pandemic Cohorts
Note. PA = phonological awareness, MA = morphological awareness, SA = syntactic awareness.
Differential Roles of Language Experiences for Children in the Pandemic and Pre-Pandemic Cohorts
Prior to interpreting the regression results, we conducted comprehensive diagnostic testing to verify that the data met the assumptions required for linear regression analysis. The residuals versus fitted values plots showed random scatter around zero with no discernible patterns, confirming linearity, while the Breusch-Pagan tests yielded nonsignificant results for all (ps > .05) but one model, indicating that the homoscedasticity assumption was satisfied for all but one model with English vocabulary as the outcome. The Durbin-Watson tests produced values between 1.44–2.16, demonstrating acceptable independence of residuals for the matched dataset. Normality of residuals was confirmed through visual inspection of the Q-Q plots, which showed residuals closely following the theoretical normal distribution line, and the Shapiro-Wilk tests (ps > .05), which were nonsignificant, with three exceptions (English MA, Spanish PA, and Spanish MA models). Multicollinearity assessment using variance inflation factors (VIF) revealed that all VIF values were below 5 (maximum value = 4.17), indicating no multicollinearity issue. Cook’s distance analysis showed no influential outliers, with all values (maximum value = .15) well below the conventional threshold of 1.
For the model predicting English vocabulary, the Breusch-Pagan test revealed evidence of heteroscedasticity (p = .032). To address this violation, we employed robust standard errors (HC3 estimator) for all statistical inferences, which provide valid hypothesis tests and confidence intervals in the presence of unequal variances. For the models predicting English MA and Spanish PA, the Shapiro-Wilk tests indicated nonnormal residuals (p < .05). To address this violation, we applied a square root transformation to the outcome variables, which resulted in normally distributed residuals (English MA: p = .369; Spanish PA: p = .100) while maintaining other assumption requirements. The Spanish MA variable showed a bimodal distribution with clustering around 0–0.40 and 0.80–1, suggesting two distinct performance groups within the sample. Assumption testing of the untransformed model revealed a significant violation of normality (Shapiro-Wilk test, p = .011). To address the normality violation, several transformation approaches were systematically evaluated. The final model used the square root–transformed Spanish MA, resulting in substantial improvement in assumption compliance: normality (p = .029), with alternative transformations including logit, Box-Cox, and inverse transformations tested but yielding worse normality violations.
Results from the moderation models are presented in Table 4 for English and Spanish language outcomes. Results showed that both teacher-reported Spanish experience (B = .08, 95% confidence interval [CI] = .01, .14, p = .015) and child-reported relative English experience (B = 0.08, 95% CI = .01, 0.14, p = .017) significantly moderated the differences in English reading comprehension between children in the pandemic and pre-pandemic cohorts (see Figure 2). Significant moderators when predicting English PA were teacher-reported English (B = .12, 95% CI = .01, .22, p = .027) and Spanish experience (B = .12, 95% CI = .01, .23, p = .031), child-reported relative English experience (B = .12, 95% CI = .01, .24, p = .035), and Spanish reading frequency (B = .18, p = .039; see Figure 3). In addition, Spanish reading was a significant moderator when predicting English MA (B = .18, 95% CI = .05, .32, p = .009; see Figure 4). Teacher-reported English (B = .10, 95% CI = .02, .18, p = .017) and Spanish experience (B = .13, 95% CI = .05, .22, p = .003) were significant moderators when predicting English SA (see Figure 5). Finally, child-reported relative English experience (B = .06, 95% CI = .01, .11, p = .025) moderated the differences in Spanish vocabulary between children in the pandemic and pre-pandemic cohorts (see Figure 6).
Regressions Predicting English and Spanish Outcomes
Note. PA = phonological awareness, MA = morphological awareness, SA = syntactic awareness.
Interaction plot of English reading comprehension.
Interaction plot of English phonological awareness.
Interaction plot of English morphological awareness.
Interaction plot of English syntactic awareness.
Interaction plot of Spanish vocabulary.
Simple slopes analyses were used to probe the interactions. First, teacher-reported Spanish experience predicted English reading comprehension for children in the pandemic cohort only (B = 0.07, p = .005; pre-pandemic cohort: B = −.01, p = .733). However, child-reported relative English experience did not significantly predict English reading comprehension for either cohort (pandemic cohort: B = .04, p = .083; pre-pandemic cohort: B = −.04, p = .092). Second, in terms of English PA, teacher-reported English experience positively predicted it among children in the pandemic cohort only (B = .15, p = .001; pre-pandemic cohort: B = .03, p = .338). Similarly, teacher-reported Spanish experience positively predicted English PA among children in the pandemic cohort only (B = 0.11, p = .009; pre-pandemic cohort: B = −.01, p = .845). However, child-reported relative English experience negatively predicted English PA among children in the pre-pandemic cohort (B = −.11, p = .020) but not those in the pandemic cohort (B = .02, p = .618). Likewise, Spanish reading frequency negatively predicted English PA for children in the pre-pandemic cohort only (B = −.18, p = .001; pandemic cohort: B = −.002, p = .980). Third, when predicting English MA, Spanish reading frequency was a significant predictor for children in the pre-pandemic group only (B = −.13, p = .003; pandemic cohort: B = .05, p = .321). Fourth, in terms of English SA, teacher-reported English experience predicted it among both cohorts (pandemic cohort: B = .16, p < .001; pre-pandemic cohort: B = .06, p = .017); teacher-reported Spanish experience was a significant predictor for children in the pandemic cohort only (B = .12, p = .001; pre-pandemic cohort: B = −.01, p = .604). Finally, child-reported relative English experience negatively predicted Spanish vocabulary among children in the pre-pandemic cohort only (B = −.08, p < .001; pandemic cohort: B = −.02, p = .218).
Discussion
The COVID-19 pandemic caused significant disruptions to schooling and family routines. This study investigated the effects of these pandemic-related disruptions on the language and reading development of Spanish–English EB children in the United States. The EB children in the study predominantly came from economically disadvantaged backgrounds, spoke Spanish at home, and were enrolled in Spanish–English DLI programs at the time of participation. To ensure comparability, propensity score matching was employed to create two cohorts: a pre-pandemic cohort and a pandemic cohort. Children in these cohorts were matched on key demographic variables, such as grade, sex, and the median income of their school zip code. Our study revealed nuanced insights into Spanish–English EBs’ language development and experience during the pandemic. Contrary to potential expectations of significant learning disruptions, we found remarkable resilience in core language skills. No statistically significant differences emerged between pandemic and pre-pandemic cohorts across vocabulary, PA, MA, or SA in both English and Spanish, and English decoding and reading comprehension abilities, after multiple comparisons correction.
Two critical findings distinguished the cohorts: (1) children in the pandemic cohort demonstrated higher Spanish SA, but fewer opportunities to engage in Spanish experience at school and Spanish reading at home, compared to children in the pre-pandemic cohort (although the differences were no longer significant after correction for multiple comparisons), and (2) pandemic-era children exhibited unique patterns of language skill development. Specifically, they showed enhanced benefits from increased English and Spanish language experience in developing English reading and language skills. These findings suggest that educational disruptions may have activated compensatory mechanisms that enhanced the cognitive benefits of both within-language and cross-linguistic experience in bilingual development, highlighting children’s capacity to leverage their full bilingual repertoire as a unified resource for academic language learning despite challenging circumstances.
The observed patterns underscore the complex, dynamic nature of bilingual development, particularly during extraordinary periods like the COVID-19 pandemic. The results challenge simplistic narratives of learning loss and instead point to more nuanced, context-dependent processes of language skill maintenance and development.
Declines in Spanish Language Experience and Reading Frequency
Our findings revealed declines in Spanish–English EB children’s Spanish language experience and reading frequency, although the differences were not significant after multiple comparisons correction. These declines in Spanish language engagement are particularly notable given that these students were in Spanish–English DLI programs, where both languages were emphasized in content instruction (Shen et al., 2022). The convergent evidence from both teacher observations and child self-reports suggests a systematic reduction in Spanish language exposure and literacy practices during the pandemic period. Our finding is largely consistent with Mak et al. (2025), who found that among preschool Spanish–English bilinguals, their parents reported a decrease in the proportion of time children spoke in Spanish from pre-pandemic to during the pandemic.
These declines likely stem from multiple interconnected factors that fundamentally altered language engagement opportunities for EBs. First, home literacy environments may have transformed significantly during the pandemic (G. Li & Lin, 2023), often compensating for reduced school-based input. With schools closed, children spent more time at home, and parents reported taking a more active role in providing literacy activities to support children’s development in both English and the heritage language (G. Li & Lin, 2023; G. Li et al., 2023). Many families also relied more heavily on digital resources to access materials in both English and heritage languages (H. Sun et al., 2023). Collectively, these shifts suggest that bilingual children’s exposure to language and literacy opportunities became increasingly dependent on the resources and practices available in their home environments during the pandemic. This could potentially reduce access to Spanish reading materials and structured literacy activities.
School-related factors may also contribute to the reduced Spanish language experience and reading frequency. Although DLI programs entail content instruction in both languages, remote learning models may have inadvertently changed this instructional model, affecting both the quantity and quality of Spanish language exposure that teachers observed in their students. This shift aligns with research showing that even in well-implemented bilingual programs, maintaining Spanish literacy engagement requires intentional support and structured opportunities (Lindholm-Leary, 2016; Palmer et al., 2014). The digital divide disproportionately affected socioeconomically disadvantaged families (Cherewka, 2020), potentially further limiting access to Spanish-language digital reading resources that could have compensated for reduced in-person instruction. However, it is worth mentioning that between pandemic and pre-pandemic cohorts, children’s English language experience, either in school or out of school, did not show significant differences, along with English reading frequency. This suggests that remote learning or disruptions in schools disproportionately affected Spanish–English EBs’ Spanish language engagement across multiple contexts, while English language experiences remained relatively stable.
Spanish–English EBs’ Resilience in Language and Literacy Development
Our findings reveal a nuanced and adaptive process of language skill maintenance among Spanish–English EBs. The consistent performance across core language dimensions (vocabulary, PA, MA, and SA in both English and Spanish) suggests the robust and potentially adaptive nature of bilingual language development during educational disruptions. Our results contribute to clarifying the mixed findings in the existing pandemic literature. Some studies report significant learning losses (e.g., Relyea et al., 2023), while others find limited impact or even compensatory gains among bilingual children (Abuosbeh et al., 2024; Mak et al., 2025; X. Sun et al., 2023). For example, X. Sun et al. (2023) observed age-appropriate progress in English PA and vocabulary among elementary school Spanish–English bilinguals. Moreover, they found that compared to monolinguals, Spanish–English bilinguals demonstrated a significantly higher growth rate in English vocabulary. However, their sample of Spanish–English bilinguals is from middle and high SES backgrounds, which may not be representative of the EB population in the United States. Likewise, among French-English immersion first and second graders in Canada, Abuosbeh et al. (2024) found that the pandemic had a limited impact on English language skills, as well as on French PA, although they showed a negative effect on French vocabulary. Mak et al. (2025) even found improved English and Spanish vocabulary knowledge for preschool Spanish–English bilinguals during the pandemic. By focusing specifically on children in DLI programs, our study suggests that programmatic supports may have buffered against sharp declines, helping explain why some bilingual populations fared better than expected. In fact, specific features of the DLI program may also help explain the observed resilience among Spanish–English EBs. DLI programs provide structured bilingual instruction, ensuring consistent exposure to both English and Spanish, which likely supported cross-linguistic transfer and allowed children to leverage skills from one language to support development in the other during remote learning. Our multimethod approach combining teacher reports, child self-reports, and standardized assessments also strengthens these conclusions. For example, while teacher reports captured school-based patterns of exposure, child self-reports provided insight into out-of-school language use, a dimension often overlooked in large-scale pandemic research.
We also showed that Spanish–English EBs’ English literacy acquisition in decoding and reading comprehension was largely unaffected by COVID-19 school closures, underscoring both children’s adaptability and the ways in which instructional emphases during remote learning shaped outcomes. Our findings corroborated those of (Kuhfeld et al. (2023), who reported no significant changes in reading achievement among third to eighth graders in the United States in fall 2020. Similarly, among French-English immersion students, Abuosbeh et al. (2024) did not observe significant differences in word reading in French and English between a pandemic cohort and a pre-pandemic cohort in Grades 1 and 2. X. Sun et al. (2023) also found that Spanish–English EBs still read within the typical developmental age range in decoding and reading comprehension despite the pandemic.
Such resilience in English literacy skills may be attributed to several factors. First, many schools prioritized reading instruction, especially in code-related skills (e.g., decoding) during remote learning, with digital tools often emphasizing reading activities (H. Li et al., 2024). Indeed, in Crosson and Silverman’s (2022) study of K–2 literacy instructional practices during COVID-19, they found the smallest declines in instructions promoting code-focused skills, but the largest declines in language-focused (e.g., vocabulary) practices. As such, remote learning may have potentially provide consistent opportunities for code-related reading practices despite the changed learning environment, which helped maintain children’s English decoding and reading comprehension skills. This pattern may also help explain why children experienced declines in measures tied more directly to Spanish oral exposure and reading frequency. In fact, teachers also reported challenges fostering language-focused interaction in DLI programs (Renn et al., 2025). Thus, our findings point to the importance of considering the interaction between program features (e.g., bilingual instruction), family literacy practices, and instructional priorities when evaluating the impact of the pandemic on student learning. This is consistent with the complex dynamic systems theory, which conceptualizes bilingual development as nonlinear and sensitive to different environmental factors (Larsen-Freeman & Cameron, 2008). More broadly, our findings suggest that bilingualism itself does not pose a heightened risk during periods of disruption and may even provide a protective resource when paired with consistent school structures and enriched home environments.
Differential Benefits of Language Experience
Children in the pandemic cohort demonstrated enhanced benefits from language exposure that differed from pre-pandemic patterns, suggesting sophisticated compensatory mechanisms in language acquisition during educational disruptions. Teacher-reported Spanish language experience emerged as a particularly powerful predictor for pandemic-era children, positively predicting English reading comprehension, PA, and SA, whereas these relations were nonsignificant in the pre-pandemic cohort. Similarly, teacher-reported English experience showed enhanced predictive power for English PA and SA specifically among pandemic-era children. Our findings suggest that during periods of educational disruption, both heritage and societal language experiences provided unique scaffolds for English language and literacy development, and in particular, the cross-linguistic transfer mechanism became amplified during remote learning conditions. This aligns with prior work showing that cross-language transfer is a central mechanism in bilingual development (Melby-Lervåg & Lervåg, 2011; Shen & Goodrich, 2024) but extends it by highlighting that such transfer may become particularly salient when school-based exposure is disrupted and children rely more heavily on language experiences at home. In other words, children appeared to flexibly draw on both languages to support English reading and language development during a time when typical instructional supports were less available.
At the same time, not all forms of language experience operated similarly across cohorts. Child-reported relative English experience negatively predicted English PA and Spanish vocabulary for the pre-pandemic cohort only. This finding is not expected and seems counterintuitive. It may suggest that for the pre-pandemic group, more self-reported English experience may have displaced opportunities for rich Spanish input at home, resulting in lower Spanish vocabulary. For English PA, which is a code-focused skill that often requires explicit instruction, general English experience alone may not be sufficient to promote its development.
In contrast, the absence of these negative associations in the pandemic cohort suggests that bilingual children and their families may have altered their language practices in ways that supported both English and Spanish development. Importantly, the resilience of the pandemic cohort highlights altered language environments in the face of disruption for bilingual families. During the pandemic, many families increased their use of heritage languages at home while also supporting English learning (Sheng et al., 2021) through digital platforms and school-provided resources (Mak et al., 2025). This shift may have mitigated the trade-offs observed in the pre-pandemic group, creating conditions where relative English use no longer came at the expense of Spanish vocabulary or English PA.
Limitations
While the findings of EBs’ language and literacy skill maintenance are encouraging, they do not eliminate the possibility of a delayed negative impact on EBs’ language and reading development. Data for this study were collected in Fall 2020, which is the first semester following the major disruptions in Spring 2020. It is plausible that the negative effects of the pandemic had not yet manifested at the time of data collection for the pandemic cohort (e.g., Betthäuser et al., 2023; Engzell et al., 2021; G. Li & Lin, 2023; Shaul et al., 2024). Longitudinal data tracking EBs’ language and literacy skills beyond the pandemic year is essential to understanding the long-term impact of the pandemic on their development. We also noted the limited variables included in our propensity score-matching model. While we included demographic and school SES variables, our analysis was constrained by the variables available in the secondary dataset. Future studies may wish to include more potential confounders, such as family SES variables (e.g., parental education, household income) and home literacy environment, to reduce bias.
Another limitation of the study is related to sampling, specifically a potential self-selection bias in the pandemic cohort. Participants in this cohort were those who still actively participated in school during the pandemic, had reliable internet access at home, and had parents who were reachable and willing to participate. This suggests that the sampled pandemic cohort, despite being economically disadvantaged, maintained access to schooling and instruction. Conversely, their peers who dropped out of school, were unreachable, or declined to participate may have experienced greater negative impacts during the pandemic.
The study is also limited in the measurement of EBs’ language experience. We relied on teacher reports to assess EBs’ language experience at school and on children’s self-reports to capture their language and literacy experiences outside of school. However, the validity of self-reported data, particularly from young children, may be questionable as it depends on their ability to accurately recall and articulate their experiences. Additionally, we lacked data on home literacy activities from parents who may have provided more reliable and detailed information than the children themselves. Prior research indicates that home literacy activities, such as library visits, access to reading materials, and reading habits, strongly predict EBs’ language and literacy development (Duursma et al., 2007; Yeomans-Maldonado & Mesa, 2021). Future studies incorporating this construct into EBs’ language experience could enhance our understanding of the relationship between language experience and skill development.
Further, the generalizability of our findings is limited to Spanish–English bilingual children enrolled in DLI programs, which provide structured bilingual instruction. Bilingual children not enrolled in DLI settings, who may have less consistent exposure to both languages or fewer instructional supports, could experience different outcomes under similar circumstances. Thus, caution is warranted when extending these results to the broader population of bilingual children, particularly those in non-DLI or less resourced educational contexts.
Finally, assessments were administered in different modes across cohorts: in-person for the pre-pandemic cohorts and remotely via videoconferencing for the pandemic cohort. Except for the morphological awareness task, all of our measures were standardized assessments, and some of them, such as the Woodcock-Johnson IV Tests of Achievement (WJ-IV), have already established remote or tele-assessment protocols. Prior studies have also consistently shown that remote online assessments yield comparable performance as in-person administration when procedures are carefully standardized (e.g., standardized instructions, trained examiners) (Burchell et al., 2022; Castilla-Earls et al., 2022; Manning et al., 2020; Pratt et al., 2022). Nevertheless, mode equivalence is not guaranteed, especially for tasks not originally standardized for remote delivery or for domains requiring tightly controlled environments (e.g., timed tasks that are sensitive to timing or attentional control). In our study, we did not have any timed tasks, and we implemented safeguards such as standardized scripts, trained research assistants, consistent scoring procedures, and monitoring of session quality. However, we cannot rule out that remote administration may have introduced additional sources of variability, such as subtle environmental distractions or differences in the quality of the technology. Consequently, differences observed between cohorts may in part reflect administration mode rather than solely underlying developmental or cohort effects.
Conclusions
Our study reveals the complex ways in which Spanish–English EBs adapted to educational disruptions during the COVID-19 pandemic. While these students demonstrated remarkable resilience and adaptability in maintaining core language abilities and reading skills, the pandemic appeared to specifically impact Spanish language experience and Spanish reading frequency, suggesting vulnerability in heritage language maintenance even within DLI programs. The enhanced benefits of English and Spanish language experience observed in the pandemic cohort point to potential compensatory mechanisms in bilingual development during crisis periods. These findings carry important implications for educational policy and practice, highlighting the need for targeted support in maintaining balanced biliteracy development during educational disruptions. As schools continue to address pandemic-related challenges, particular attention should be paid to preserving Spanish language experiences and Spanish reading opportunities and leveraging children’s adaptive language learning capabilities to support their continued bilingual development. Future research should examine the long-term implications of these altered patterns of language experience and explore effective strategies for supporting heritage language maintenance during periods of educational disruption.
Footnotes
Acknowledgements
The study was supported by the National Institutes of Health (NIH) under Award Number SC2HD100362. The content is solely the responsibility of the authors and does not necessarily reflect the official views of the NIH. We are grateful to Lisa Bedore and Ioulia Kovelman for their valuable input on measure selection and for generously sharing the ELMM and BESA-ME measures with us. We also thank members of the former Language Learning and Assessment Laboratory at the University of Texas at San Antonio (now housed at the Ohio State University) for their assistance with data collection and data analysis. Finally, we extend our sincere appreciation to the school district offices, principals, teachers, parents, and children for their support. All remaining errors are our own.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The study was supported by the National Institutes of Health (NIH) under Award Number SC2HD100362.
Authors
YE SHEN is an assistant professor in the Department of Counseling and Educational Studies in the School of Education at the Johns Hopkins University; email:
BECKY H. HUANG is a professor in the Department of Teaching and Learning and a faculty associate at the Crane Center for Early Childhood Research and Policy at the Ohio State University, Columbus, OH; email: