Combined Language and Code Emergent Literacy Intervention for At-Risk Preschool Children: A Systematic Meta-Analytic Review

Literacy acquisition in early schooling significantly impacts academic and life outcomes. Global illiteracy rates, with 23% of 15-year-olds in 2018 having low literacy, highlight the need for effective interventions. Theoretical frameworks like the Simple View of Reading and the Direct and Indirect Effect Model of Reading (DIER) emphasize the interaction of decoding (code skills) and linguistic comprehension (language skills) in reading. Studies show strong predictive relationships between preschool oral language and code skills and later reading comprehension, with specific subskills including vocabulary, grammar, phonological awareness, and letter knowledge. A significant portion of children, especially those from socially disadvantaged backgrounds, enter school with low emergent literacy levels, often presenting with both language and code difficulties. This underscores the importance of preventive reading interventions in the preschool years, suggesting that combined language and code interventions could be a solution, considering the complex interaction between these skill sets.

Combined language and code emergent literacy interventions are considered a logical approach to prevent future reading failure, supported by theories such as the Lexical Restructuring Model, which links vocabulary and phonemic awareness. Research indicates that vocabulary knowledge contributes to phonological awareness. While the concept of 'synergistic intervention effects' – where simultaneously targeting oral language and code subskills provides reciprocal benefits – has been explored, findings are mixed. Some studies found that combined interventions, despite reduced dosage for individual skills, yielded similar outcomes to domain-specific interventions, suggesting potential synergistic benefits. This meta-analytic review aimed to address three key questions: characterize combined interventions for at-risk preschoolers, assess their effectiveness, and determine if synergistic relationships exist between code and language outcomes.

The review followed PRISMA guidelines and was registered with PROSPERO. A comprehensive search across 10 databases was conducted for studies published before March 2023, focusing on combined language and code interventions for preschool children (4–6 years) at risk of literacy difficulties or diagnosed with speech/language disorders. Inclusion criteria required randomized controlled trials reporting quantitative outcomes for at least one language and one code subskill. Studies involving neuro-developmental disorders or diagnosed disabilities were excluded. Data extraction included participant and intervention characteristics, and quantitative outcome measures for vocabulary (distal and proximal), alphabet knowledge, print awareness, and phonological awareness. Interventions were coded as 'distributed' or 'explicit' and based on emphasis ('code,' 'language,' or 'even'). Study quality was assessed using the PEDro scale. Meta-analyses used multi-level random effects models to calculate effect sizes and explored heterogeneity, subgroup differences, and publication bias using Egger's test and funnel plots. A meta-regression examined the relationship between composite language and code effects to investigate synergistic relationships, alongside subgroup analyses by intervention emphasis.

The systematic search identified 5692 unique studies, with 29 meeting the eligibility criteria, reporting on 43 interventions and involving 9333 children. Most studies (25/29) were conducted in the United States. Methodological quality was deemed 'fair' for 18 studies and 'good' for 11, according to PEDro scores. The majority of participants were identified as at-risk for literacy failure due to socioeconomic status (SES) or low academic scores, with three studies focusing on children with diagnosed speech/language difficulties. The participant cohort was 55% male, comprising 45% African American, 30% Hispanic, and 20% White children. Child age averaged 4-6 years. Half of the studies reported on maternal education levels.

Most interventions (39/43) were delivered in preschool settings, primarily by preschool educators (32/43) or trained interventionists. Non-computer-based interventions often provided supplementary curricula with educator training and coaching (most common support), along with program resources and parent involvement. Small group instruction was the dominant delivery method, often combined with individual or whole-class activities. While a majority (63%) of interventions were explicit with defined session plans and times, there was high variability in dosage reporting, hindering precise cumulative intervention time calculations for all studies. Cumulative intervention time ranged from 8 to 85 hours. Vocabulary (39/43) was the most frequently targeted language subskill, followed by narrative and grammar, with many interventions also addressing general oral language and listening comprehension. All major code subskills were well-represented, including phonological awareness (42/43), alphabet knowledge, and print awareness. Interventions, on average, targeted five code subskills simultaneously. Shared book reading was a common component (35/43). Intervention emphasis was largely split: 44% equally emphasized code and language, 42% emphasized code, and 14% emphasized language.

Combined language and code interventions showed a very small positive effect on language outcomes (g = 0.11) and a small positive effect on code outcomes (g = 0.23). At the subskill level, code intervention effects were comparable to domain-specific meta-analytic findings for phonological awareness and alphabet knowledge. However, distal vocabulary effects (g = 0.08) were significantly lower than previous meta-analyses. This small language effect may be due to the prevalence of shared book reading, which has a small, non-specific impact on language outcomes, and the negative effect of 'distributed' (less specific) language interventions. Explicit interventions yielded higher language effects (g = 0.12) compared to distributed interventions (g = -0.01). Language gains were also found to depend on targeted subskill, dosage, and meaningful application. Proximal vocabulary effects (g = 0.53) were higher than distal effects (g = 0.08), and proximal expressive vocabulary effects (g = 0.80) outstripped receptive (g = 0.40). Significant between-study heterogeneity remained across some outcomes, influenced by publication bias, intervention type, and outcome measures. The varied study populations (including both typically developing at-risk children and those with diagnosed language disorders) and inconsistent reporting of implementation fidelity likely contributed to this heterogeneity.

Two methods were used to explore synergistic relationships: a meta-regression and subgroup meta-analyses by intervention emphasis. The meta-regression revealed a positive, statistically significant relationship (p=0.032) between composite language and code effect sizes, indicating that as language effects increased, code effects also increased. Subgroup analyses by intervention emphasis showed that interventions with an 'even' emphasis on both code and language targets produced the largest language effects (g = 0.18) and comparable code effects (g = 0.22) to code-emphasized interventions (g = 0.29). Language-emphasized interventions showed negative effects on both language and code outcomes, though this finding is cautious due to a small number of studies in this subgroup. These results suggest a potential synergistic relationship, where code effects are maintained while language effects are enhanced when both are targeted equally. Despite this, the overall small meta-analytic outcomes suggest that other intervention characteristics, such as dosage and target selection, also heavily influence the results.

A key limitation is the reliance on reported intervention details for emphasis coding, which varied widely among studies. The review concludes that combined language and code emergent literacy interventions are effective for at-risk preschoolers but the effects are small, particularly for language. Future research should aim to enhance oral language outcomes by targeting a broader range of explicit language subskills (e.g., vocabulary, grammar, narrative) and incorporating discourse-level activities beyond shared book reading. More consistent and detailed reporting of dosage and implementation fidelity in intervention studies is crucial for future meta-analyses and clinical translation. The observed positive relationship between code and language outcomes and the benefits of 'even' emphasis interventions provide some evidence for a synergistic relationship, suggesting that future research should focus on optimizing both language and code outcomes simultaneously.