Tier 2 oral language and early reading interventions for preschool to grade 2 children: a restricted systematic review

ABSTRACT

This systematic review investigated small-group Tier 2 interventions to improve oral language or reading outcomes for children during preschool and early primary school years. Literature published from 2008 was searched and 152 papers selected for full-text review; 55 studies were included. Three strength of evidence assessment tools identified a shortlist of six interventions with relatively strong evidence: (a) Early Reading Intervention; (b) Lonigan and Philips (2016) Unnamed needs-aligned intervention; (c) PHAB+WIST (PHAST)/PHAB+RAVE-O; (d) Read Well-Aligned intervention; (e) Ryder and colleagues’ (2008) Unnamed Phonological Awareness and Phonics intervention; and (f) Story Friends. Investigation of intervention componentry found common characteristics included 3–5 students, 4–5 sessions per week, minimum 11-week duration, content covering a combination of skills, modelling and explicit instruction, and trained personnel. Shortlisted interventions provide a useful foundation to guide further interventions and inform educators and policymakers seeking to implement effective evidence-based interventions in the early years of schooling.

Introduction

Importance of oral language and literacy skills in the preschool and primary school years

Relationship between oral language and literacy

The ability to use and understand oral language to communicate effectively is central to children’s learning and academic success (Snow, 2016). In the early years of school, oral language competence underpins the ability to learn to read and the emergence of literacy skills more broadly (Castles, Rastle, & Nation, 2018; Dickinson, Golinkoff, & Hirsh-Pasek, 2010). There is a distinction to be made between the terms reading and literacy. For the purposes of this paper, and in line with the Simple View of Reading (Gough & Tunmer, 1986), reading refers to the ability to decode, recognise, and understand the printed word, and the associated higher-level skills of sentence and text comprehension (Buckingham, Wheldall, & Beaman-Wheldall, 2013). Literacy encapsulates a broader set of skills, including reading, writing, and spelling and the ability to produce and engage with a variety of text types across the curriculum at all year levels (Buckingham et al., 2013; Snowling & Hulme, 2012). A reciprocal relationship exists between oral language and literacy, with improvements in one helping to advance skills in the other (Muter, Hulme, Snowling, & Stevenson, 2004; Snow, 2016).

Long-term and broader impact of oral language and literacy difficulties

Oral language competence is also strongly linked to a broad range of outcomes including social and emotional well-being, success in making and retaining friendships, prosocial problem solving, school engagement, and subsequent mental health (Snow, 2020). It is well recognised that if children do not learn to understand and use spoken language, and to read and write to communicate their ideas, there can be significant consequences for their self-esteem, social development, and educational and employment opportunities (Conti-Ramsden & Botting, 2008; Eadie et al., 2018; Law, Rush, Schoon, & Parsons, 2009; Lyon, 2003). Longitudinal research shows that students who struggle in the early years of school also tend to have difficulties in later grades (Maughan et al., 2009; St Clair, Pickles, Durkin, & Conti-Ramsden, 2011) and face long-term academic challenges with the risk of early disengagement from school (Snow, 2020). There is also evidence that some students, particularly those whose literacy difficulties are comorbid with behaviour disturbances, enter the so-called “school-to-prison” pipeline early in their school years. These are students who exit school prematurely following histories of suspension and exclusion and enter the youth justice system via community-based or custodial orders (Snow, 2019).

Prevalence of oral language and literacy difficulties

Difficulties with oral language and literacy are among the most common developmental problems in the preschool and primary school population (Snow, 2020). The estimated prevalence of developmental language disorder in young children (aged 4–5 years) is 7.6% (Norbury et al., 2016). Similar prevalence levels are reported for reading disorder in the population (Bishop & Snowling, 2004), whereas reading difficulty rates during primary school have been reported as approximately 10–30% of the school-aged population (Adolf & Hogan, 2019; Hempenstall, 2013; Snow, 2020).¹ However, it is important to note that language and reading difficulties are often undiagnosed or misdiagnosed in the school system (Adolf & Hogan, 2019; Hendricks, Adolf, Alonzo, Fox, & Hogan, 2019; Snow, 2020).

Social and economic cost of oral language and literacy difficulties

Difficulties with oral language and literacy have serious ramifications for both individual life trajectories and social and economic costs (Cronin, Reeve, McCabe, Viney, & Goodall, 2020; Heckman & Masterov, 2007; Le et al., 2020). The causal link between educational attainment and a range of health and social outcomes (Gakidou, Cowling, Lozano, & Murray, 2010; Rhodes, Cordie, & Wooten, 2019) suggests the educational and social impact of oral language and literacy difficulties must be viewed as a population-level cost to society (Lyon, 2003; Tomblin et al., 1997) incurred across the lifespan (Law et al., 2009). In a review of research on outcomes for students experiencing reading difficulties, Slavin, Lake, Davis, and Madden (2011) noted that reading failure incurs significant costs to education systems, in terms of special education, remediation, grade repetition and dropout. Language and literacy difficulties and low educational achievement in childhood are also associated with indirect costs in the longer term through reduced workforce participation and productivity losses in adulthood (Cronin et al., 2020).

According to a report from the Industry Skills Council Australia (2011) examining the challenge of low language, literacy, and numeracy skills, approximately 46% of Australian adults have difficulty with reading, and 13% are classified in the lowest literacy category. Such difficulties with literacy make secure engagement in the workplace impossible for some adults. Working to optimise students’ oral language and literacy development, and subsequent educational outcomes, should pay significant dividends for individual students and society more broadly.

Importance of early intervention in the preschool and primary school setting

Research suggests oral language and literacy difficulties emerge early in childhood and become increasingly difficult (and expensive) to treat over time (Hempenstall, 2013; Spira, Bracken, & Fischel, 2005). Prevention and early intervention are generally considered more efficacious and cost-effective than attempts to remediate entrenched difficulties (Hempenstall, 2013, 2016). Evidence from reviews and meta-analyses consistently demonstrates that oral language and literacy interventions targeting children in the early years of school are particularly effective (Ehri, Nunes, Stahl, & Willows, 2001; Ehri et al., 2001; Piasta & Wagner, 2010) and potentially cost-effective (Hollands et al., 2016). However, addressing well-established deficits in the upper grades is more challenging (Wanzek et al., 2013). It has been noted that it takes four times as many resources to address a literacy problem in Year 4 than it does in Year 1 (Pfeiffer et al., 2001); yet, in many cases, children progress through the primary school years carrying a widening reading deficit relative to more able peers and relative to the demands of the curriculum (Hempenstall, 2013, 2016). This has been referred to as the Matthew Effect (Stanovich, 1986) and reinforces the fact that “the importance of getting children off to a good start in reading cannot be overstated” (Slavin et al., 2011, p. 2). Crucially, research evidence suggests that exemplary teaching, and effective and timely intervention, can lead to high levels of oral language and literacy achievement for at-risk or vulnerable students (Buckingham et al., 2013). It is therefore imperative that practicable and effective early interventions supported by the strongest levels of evidence are identifiable and available to educators and policymakers.

Education policy context

Historically, education policy in the United States of America (USA) has been criticised for creating conditions in which students struggling with literacy have either not been identified or have not received formal support or appropriate intervention until officially diagnosed with specific learning disabilities, which does not usually occur until Grade 2 or 3 (Gersten et al., 2009). In Australia, it has been argued that an “unnecessary and avoidable” level of illiteracy in students is due to “poorly conceived government policies and university education faculties wedded to outdated and unproven teaching methods” (Buckingham et al., 2013, p. 28). In addition, longitudinal studies have demonstrated there can be substantial individual variability in oral language performance across different measures or time points. Some children experience early delays and catch up with their peers, whereas others will develop later language difficulties following an apparently positive start, both in the preschool years and after school-entry (McKean et al., 2017). This instability has major implications for the provision of educational support, and careful monitoring of at-risk students over time has been recommended (Eadie et al., 2014; McKean et al., 2017).

In the USA, the No Child Left Behind Act of 2002 and Individuals with Disabilities Education Act of 2004 have been instrumental in the wide-scale implementation of early oral language and literacy interventions and interest in Response to Intervention (RTI) approaches particularly (Gersten et al., 2009; Gersten, Newman-Gonchar, Haymond, & Dimino, 2017; Slavin et al., 2011). Following a shift in education reform to a more preventive approach to addressing learning difficulties, RTI (Fuchs & Fuchs, 2006) and the broader Multi-Tiered Systems of Support (MTSS; Eadie et al., 2018; Hall, 2018) frameworks have been introduced and are gaining momentum in research and school communities (Gilbert et al., 2013). RTI is a proactive, systematic improvement framework that is used to provide targeted support to struggling students, supporting academic growth and achievement (Fuchs & Fuchs, 2006). MTSS is more comprehensive and includes RTI but also supports many other areas such as social and emotional needs, and may include whole-of-school policies, professional development, and collaboration with parents (Eadie et al., 2018; Hall, 2018; Sugai, Horner, & Gresham, 2002). Within RTI frameworks, tiers of support increase in intensity from one level to the next and interventions are typically categorised into three levels of support (Fuchs, Fuchs, & Vaughn, 2008; Gersten et al., 2009). Tier 1 represents whole-class, high-quality, evidence-based classroom instruction that will result in most students benefiting from satisfactory rates of progress. Tier 2 supports involve additional instruction for small groups of students who do not make adequate progress with whole-class instruction alone, or who fail to meet benchmarks on screening measures. Tier 2 interventions are typically delivered to small groups of students in 20–40-min sessions, 3–5 times per week, for at least 5 weeks and are generally not offered for more than 25–30 weeks (Gersten et al., 2009; Wanzek et al., 2016). Tier 3 interventions are further intensified efforts to provide students with more intensive one-to-one support when they continue to struggle despite provision of Tier 1 and Tier 2 support after a reasonable amount of time. It is important to note that the interventions offered at Tiers 2 and 3 typically build directly on Tier 1 approaches, but with greater intensity, duration, and frequency (Fuchs & Fuchs, 2006).

The RTI framework presents a comprehensive, promising, and pragmatic approach to addressing difficulties with oral language and reading once children have commenced formal schooling. Indeed, it has been estimated that the incidence of reading difficulties can be dramatically reduced following effective Tier 1 classroom instruction, and further reduced following effective Tier 2 intervention, so that the number of students requiring more intensive and costly Tier 3 support can be minimised (Coyne et al., 2013; Gilbert et al., 2013; Lovett et al., 2017; Scanlon, Gelzheiser, Vellutino, Schatschneider, & Sweeney, 2008). According to some estimates, 80% of students respond satisfactorily to general instruction, and 15% are adequately served by small-group intervention, leaving 5% requiring more intensive Tier 3 support (Spencer, Petersen, & Adams, 2015). It has been argued that a tiered approach to intervention should decrease the number of students inappropriately referred to more intensive and costly individualised intervention (Tier 3), or special education (Fuchs, Compton, Fuchs, Bryant, & Davis, 2008). For many students, some small-group interventions can be as effective as more expensive one-to-one interventions (Elbaum, Vaughn, Tejero Hughes, & Watson Moody, 2000; Gilbert et al., 2013; Vaughn et al., 2003).

Overview of Tier 2 oral language and reading intervention

As briefly described above, Tier 2 intervention involves additional intensive student support that is compatible with, and builds upon, the core instruction provided at the universal level (Gersten et al., 2009). Tier 2 intervention should involve regular collection of student achievement data in order to (a) identify student needs regarding specific skill components, (b) choose appropriate interventions, and (c) monitor the impact of intervention on student progress so as to determine whether they are able to return to a lower level of support (i.e. Tier 1 classroom instruction only) or whether increased intensity of support is needed (i.e. Tier 3) (Gersten et al., 2009; Mellard, McKnight, & Jordan, 2010).

Tier 2 reading intervention content typically involves systematic and explicit instruction in the core skill components of reading instruction such as phonemic awareness (identifying, segmenting, and blending sounds orally), phonics and print knowledge (e.g. letter name and sound correspondence, including consonant clusters and digraphs), fluency (accuracy, speed, and expression), vocabulary development, and reading comprehension (Gersten et al., 2009; Hempenstall, 2016). Tier 2 intervention to support oral language tends to take a broad approach, typically focusing on areas such as vocabulary acquisition, attention and listening, sentence building and syntax rules (e.g. plurals, past tense, prepositions), and listening comprehension (e.g. following multi-step directions) and narrative (Sedgwick & Sothard, 2019).

Research on oral language and reading interventions in the preschool and primary school setting

There is growing research interest in Tier 2 oral language and reading interventions, and several recent systematic reviews and meta-analyses have included an examination of small-group interventions (e.g. Gersten, Haymond, Newman-Gonchar, Dimino, & Jayanthi, 2020; Gersten et al., 2017; Sedgwick & Sothard, 2019; Slavin et al., 2011; Wanzek et al., 2016). For the most part, a broad approach has been taken, incorporating interventions beyond the parameters of small-group Tier 2 intervention.

Early childhood context

In early childhood contexts, there is some evidence that small-group interventions can have significant impacts on early language and literacy skills of preschool children at risk of academic difficulties (e.g. Lonigan, Schatschneider, & Westberg, 2008; Marulis & Neuman, 2010; Shepley & Grisham-Brown, 2019). Notably, however, the focus is on interventions across the tiers, rather than on Tier 2 explicitly. For example, the National Early Literacy Panel (National Early Literacy Panel, 2008) reported on meta-analyses of intervention studies aimed at the development of early literacy skills. Findings indicated moderate to large effects on outcomes related to language and literacy (e.g. phonological awareness and early reading skills), although the review included interventions conducted at the individual level as well as small-group interventions (Lonigan et al., 2008). Marulis and Neuman (2010) conducted a meta-analysis examining the effects of vocabulary interventions on language development with children in preschool and the early years of primary school. Although results indicated a strong effect size overall (g = 0.88), the meta-analysis was not focused on a specified group size or support tier, thus included individualised and large-group interventions as well as those delivered to small groups. Also looking across the tiers, Shepley and Grisham-Brown (2019) examined the available literature on MTSS (or a specific tier within it). Interventions for preschool children (ages 3–5 years) targeting a range of developmental/content domains, including literacy and language, were reviewed. Mean effect sizes indicated that clinically significant findings were evident for language and literacy outcomes; however, when considering only the studies with the most rigorous research designs, the effects typically lessened or were less frequently detected (Shepley & Grisham-Brown, 2019).

Primary school context

In the primary school context, Sedgwick and Sothard (2019) and Slavin et al. (2011) both conducted reviews of oral language and reading interventions across the RTI tiers (including small-group interventions) and positive effects of small-group interventions were demonstrated. Specifically, Sedgwick and Sothard (2019) conducted a recent review of mainstream school-based oral language interventions for children with speech, language, and communication needs in the early years of school in the United Kingdom. Five studies comparing the effects of intensified small-group instruction with a usual-practice or other control group were identified, and they all reported positive effects (although information regarding comparison groups or results for each of the studies were not reported in detail in the review). Similarly, Slavin et al.’s (2011) systematic review of 97 studies evaluating a range of reading interventions for struggling readers identified 20 small-group studies and the overall mean effect size for these interventions was 0.3. However, Slavin et al. (2011) did not include a separate analysis of small-group interventions specifically targeting children in the early years of school in their review.

Tier 2 interventions

To date, there have been only three systematic reviews with an explicit focus on Tier 2 interventions (rather than across the tiers) for children with reading difficulties in the early years of school (Gersten et al., 2020, 2017; Wanzek et al., 2016). Wanzek and colleagues conducted a meta-analysis of 72 predominantly experimental and quasi-experimental studies that focused on children in Grades K-3 and were published between 1995 and 2013. Interestingly, however, the authors defined Tier 2 reading interventions in terms of session quantity (specifically 15–99 sessions), and thereby included interventions delivered in a one-on-one instructor-to-student format. Wanzek and colleagues reported positive effects of intervention for foundational reading skills measured with standardised (Hedges’ g = 0.49) and non-standardised (g = 0.61) tests, as well as standardised (g = 0.38) and non-standardised (g = 1.03) measures of language or reading comprehension. Effect sizes across all measures appeared higher for interventions delivered to student pairs/trios, compared to groups of 4–5 students. These findings show that, on average, Tier 2 interventions targeting struggling readers have demonstrated positive effects on foundational reading skills and on measures of language or reading comprehension.

A more recent review (Gersten et al., 2017) and meta-analysis (Gersten et al., 2020) of US-based reading interventions also provide evidence of positive effects. Gersten et al. (2017) focused on 20 interventions trialled in studies meeting What Works Clearinghouse standards published between 2002 and 2014, then conducted a meta-analysis of 33 studies including those published up to 2017 (Gersten et al., 2020). As in the Wanzek et al. (2016) meta-analysis, Gersten and colleagues’ reviews included interventions with a one-to-one format in their definition of Tier 2. Interestingly, however, they acknowledged that in usual RTI practice, schools almost always implement small-group (3–5 students) interventions at Tier 2. Of the 20 included Tier 2 interventions, the 2017 review identified only nine small-group interventions. Positive effects (defined in the review as the treatment group performing better than the comparison group by a statistically significant margin in at least one area of reading in at least one study) were reported for eight of these nine interventions. Similarly, of the 33 included studies in the Gersten et al. (2020) meta-analysis, 21 were of interventions conducted one-to-one. Significant positive effects on a range of reading outcomes were reported, with a mean effect size of .39 (Hedges’ g).

Rationale for this review

Although the findings from the Wanzek et al. (2016) meta-analysis and the Gersten et al. (2020); (2017) reviews are encouraging, further research is needed for a number of reasons. For pragmatic purposes, schools need to know (a) whether small-group Tier 2 oral language and reading interventions are generally effective, and (b) what components effective Tier 2 interventions have in common. Such components may be the active ingredients and indicative of programs that schools could reasonably expect to be effective. If schools are unable to implement programs with a strong evidence base, then they can at least identify programs that share the most common characteristics to those programs that do have a strong evidence base. Moreover, educators need to know which interventions are most likely to be effective when implemented within a particular educational context or with particular groups of students, and which are feasible to deliver with high levels of fidelity. Fidelity of implementation is typically difficult to evaluate because it is not consistently operationally defined and measured (Gersten et al., 2020). Further research is also needed to identify interventions trialled in recent years and not included in previous reviews.

Information about the features that characterise the most effective interventions with a robust evidence base may also prove instructive. However, to date, there has been no systematic review of the componentry associated with the strongest Tier 2 language or reading interventions for children in preschool and the early years of primary school. Although Wanzek et al. (2016) and Gersten et al. (2020) investigated several intervention features as potential moderators, the analyses were not specific to small-group interventions and were underpowered in many cases.

Aims

In this restricted systematic review (Pluddermann, Aaronson, Onakpoya, Heneghan, & Mahtani, 2018), we sought to identify effective small-group Tier 2 interventions that schools could implement to improve oral language or reading outcomes for children. The review focuses on interventions that have been tested with children in the preschool and early primary school years, and was guided by the following research questions:

1. Which small-group Tier 2 interventions for preschool to Grade 2 children improve oral language or reading outcomes?

2. Which of these interventions are supported by relatively strong levels of evidence?

3. What components do the most effective Tier 2 oral language and reading interventions have in common?

In this review, we have chosen to focus on literacy skills broadly defined and have also included studies with a narrower focus on reading, which in turn has implications for the acquisition of broader literacy skills.

Method

A detailed study protocol covering review questions, search strategy, selection criteria, coding procedures, risk of bias assessment, and data synthesis, was developed and prospectively registered with the international register of systematic reviews (PROSPERO; CRD42019126121). A restricted systematic review approach was utilised. Restricted systematic reviews use similar methods to full systematic reviews but make concessions to the depth and breadth of the process for the purpose of providing critical information and rigorous assessment relatively quickly. Methodological decisions were guided by existing recommendations for best practice in such reviews (Ganann, Ciliska, & Thomas, 2010; Kelly, Moher, & Clifford, 2016; Pluddermann et al., 2018). This review adhered to the guidelines in the PRISMA statement (Moher, Liberati, Tetzlaff, & Altman, 2009).

Search procedures

A comprehensive literature search was conducted across several academic databases (Education Resources Information Center (ERIC), Education Research Complete (ERC), Ovid MEDLINE, PsycINFO and SCOPUS), reputable program registers (Blueprints for Healthy Youth Development; Promising Practices Network on Children Families and Communities; and the California Evidence-Based Clearinghouse for Child Welfare), and language and literacy-specific databases (What Works Clearinghouse; Speechbite repository; Communication Trust What Works). The Cochrane and Campbell databases were also searched for relevant systematic reviews. Reference lists of included papers were also searched.

Database searches were developed in consultation with an experienced librarian and applied a PICO-based framework (Schardt, Adams, Owens, Keitz, & Fontelo, 2007). Each had the general form: (preschool to primary school age terms) AND (language or reading difficulty terms) AND (general intervention or Tier 2 specific terms) AND (study design terms) AND (language or literacy outcome terms). Search keywords and example syntax are listed in Table S1 (online only). Initial searches were restricted to publications from 1990 to the date of search (October 2018). A substantive body of potentially relevant literature was identified; therefore, in accordance with the registered study protocol (PROSPERO; CRD42019126121), the search range was further restricted to journal articles published from 2008 onward (i.e. to strengthen applicability of findings to the current educational context, while also balancing resource and time constraints with review scope).

Selection criteria

Studies were considered potentially relevant if the title or abstract indicated that the study included (a) children in the preschool to Grade 2 range (or ages 4–9 years as children usually receive formal reading instruction over this period, with some variation across different education settings and systems²), (b) targeted supplemental instruction in literacy or oral language skills, and (c) a treatment and comparison group. Studies were excluded if information explicitly contraindicated full-text inclusion criteria.

At full-text screening, studies were selected if they also met the following inclusion criteria: (a) children attended a mainstream education setting (i.e. not segregated special education settings); (b) children were identified as at-risk of or experiencing oral language or reading difficulties, on the basis of screening tests, formal diagnoses, or teacher or parent concern; (c) intervention specifically targeted students with emerging or established language or reading problems; (d) intervention was implemented in a small-group format of supplementary instruction (i.e. not universal classroom practices or one-to-one tutoring); (e) intervention occurred on-site at school during regular school terms and hours; and (f) intervention was implemented by regular school staff, community members, allied health professionals, or researchers.³ In addition, studies had to (g) compare a randomly or non-randomly assigned Tier 2 intervention group with a usual practice, alternative treatment, or waitlist control group, (h) on at least one oral language (e.g. expressive language, receptive language, vocabulary) or reading skill (e.g. phonics knowledge, word recognition, reading fluency, passage comprehension) and (i) analyse group mean difference scores following intervention. Finally, studies had to (j) be published from 2008 in an English language journal article.

Studies were excluded if they did not meet all of the above criteria or met any of the following exclusion criteria: intervention (a) exclusively targeted second-language learners, (b) was designed for children with severe and specific diagnoses not primarily defined by difficulty with oral language or literacy (e.g. autism, ADHD, Down syndrome), (c) included systematic provision of one-to-one support to a portion of students (d) was delivered in low-/middle-income countries or with education systems considered substantially different from North America, the United Kingdom, Australia, or New Zealand. Studies were also excluded if analysis was limited to (a) comparison with a historical control group only, (b) a proportion of “responders”, or (c) a redundant data set (i.e. analysis of the same outcomes, at the same timepoints, with an overlapping sample of students).⁴ See Table S2 (online only) for the rationale behind each of the selection criteria.

As per the registered study protocol (PROSPERO; CRD42019126121), at both the abstract and full-text screening phases, 10% of papers were randomly selected for double-coding by a second independent reviewer. Inter-rater agreement for inclusion/exclusion was high (90% abstracts, 100% full-text), and examination of discrepancies showed no instances where prevention of relevant papers progressing to full-text screening would occur erroneously. Coders discussed differences and reached 100% consensus.

Coding procedures

Studies that met inclusion criteria were coded by an experienced reviewer with the following data extracted: research design (e.g. randomised or quasi-experimental study; type of comparison group/s), participant information (e.g. mean age and grade range, selection criteria, school context, country, socio-economic status), intervention componentry (e.g. specific oral language and literacy skills targeted, general pedagogical strategies, group size, intensity and duration, type of provider), outcomes assessed (e.g. standardised and treatment-aligned measurements, time of assessment), results (e.g. statistical significance of group mean differences and effect size), intervention feasibility (e.g. cost, implementation fidelity, social validity), and overall relevance (e.g. percent of sample in preschool to Grade 2 years). A 10% sample of papers was randomly selected and double-coded by a second independent reviewer. Inter-rater agreement for data extraction was high (94%) and differences were resolved through discussion.

Study quality and intervention-level evidence assessment

As part of the process for identifying interventions with the strongest levels of evidence, risk of bias was evaluated for each study. Specifically, study quality was evaluated using an 11-item⁵ checklist adapted from recent meta-analyses in education research (e.g. Graham et al., 2018) to assess both education-specific risks and critical design features recognised across many scientific fields (see Table S3; online only). Appropriate education-specific items were identified in a recent meta-analysis of literacy interventions (Graham et al., 2018) and general items from established quality and bias checklists (e.g. NICE, 2012). Items covered the following quality criteria: experimental design, sample size and pre-test equivalence, attrition (overall and differential), measurement, condition descriptions, blinded assessment, and statistical analysis. Items were coded 1 if fully satisfied, 0.5 if partially satisfied, and 0 if inadequately addressed. A total score was calculated, and the following cut-points applied to categorise study quality as low (<50%), moderate (50–74%), and high (75+%), consistent with other systematic reviews (e.g. Bond, Wood, Humphrey, Symes, & Green, 2013; Sedgwick & Sothard, 2019).

An evidence rating system was also conducted for each intervention. For pragmatic purposes,⁶ three distinctive evidence ranking systems were used to shortlist interventions supported by the strongest evidence: (a) a customised strength of evidence evaluation, and two established frameworks; (b) the California Evidence-Based Clearinghouse (CEBC, 2016) Scientific Rating Scale; and (c) the Australian National Health and Medical Research Council (NHMRC) Body of Evidence matrix (Hiller et al., 2011).

The customised evidence rating system was adapted from established frameworks (e.g. CEBC Scientific Rating Scale; Institute of Education Sciences levels of evidence for practice guidelines; What Works Clearinghouse evidence standards; Promising Practices Network on Children Families and Communities evaluation criteria; Blueprints for Healthy Youth Development evidence standards) and considered several factors (i.e. study design, risk of bias, sample size, direction and magnitude of effect, convergence of evidence across studies – see Table 1). Interventions rated as having Very Strong evidence according to the customised evidence rating were shortlisted, as were interventions receiving a Supported or Well Supported ranking on the CEBC or meeting Level A or Level B criteria using the NHMRC assessment.⁷

Table 1. Customised evidence rating system.

Strength	Definition
Very Strong ++++	Converging evidence from multiple evaluations and the evidence from at least one evaluation was rated strong or promising (with others rated favourably).
Strong +++	RCT or CRCT with low risk of bias, acceptable sample size^a, statistically significant positive effects equivalent to Cohen’s d of 0.25^b magnitude or larger (relative to usual treatment or an alternative intervention) on more than 1 standardised measure.
Promising ++	RCT or CRCT with low-moderate risk of bias, acceptable sample size, and statistically significant positive effects (may be <0.25 Cohen’s d or equivalent) on at least 1 standardised measure.
Preliminary +	RCT/CRCT/QED with low–moderate risk of bias, and statistically significant positive effects (relative to usual treatment) on at least one outcome measure (may be treatment-aligned or researcher-designed), or trend on at least one standardised measure … Or RCT/CRCT/QED with low–moderate risk of bias demonstrates equivalent effect of intervention to an effective alternative treatment. Sample size may be small.
Weak ±	RCT/CRCT/QED with low–moderate risk of bias or effects of treatment fail to reach statistical significance on any measure.
Not adequately assessed	RCT/CRCT/QED with high risk of bias, or critical flaw.
Contraindicated	Study shows consistent pattern of negative effect.

^a n ≥50 students in each condition.

^bConsistent with What Works Clearing House criteria for substantively important effects.

RCT = randomised controlled trial; CRCT = cluster randomised controlled trial; QED = quasi-experimental design.

Data synthesis

As the primary purpose of the review was to identify effective school-based Tier 2 oral language and literacy interventions supported by the strongest levels of evidence, and their key components, a narrative synthesis of the findings is presented.

Results

Study selection

Figure 1 provides an overview of the search and selection process. After screening 2107 abstracts, 152 articles were selected for full-text review. Ninety-nine articles were excluded (listed in Table S4 [online only] with reasons), leaving 53 articles describing 55 included studies (two publications each described two separate studies). See Tables 2 and 3 for an overview of included studies.⁸ For further demographic information (e.g. age, gender, SES, language, race) see Table S5 (online only). From the 55 included studies, 47 separate interventions were identified.

Figure 1. Search and selection process.

Table 2. List of included studies.

Study	Design	N^a	Grade (age)^b	Schools (classes)	Intervention	Comparison(s)	Measures	Relevant results (effect sizes)^c	Study quality^e	Relative strength of positive evidence^f
Bailet, Repper, Piasta, and Murphy (2009)	CRCT	217	Pre-K	NR (38 classes)	Unnamed (targets: Phonological awareness & Phonics) Teacher-delivered	Usual Practice (waitlist, NOS)	20-item early literacy skills	ns	Mod	Preliminary
							Get it, Got it, Go (PA)	0.44
Boyle et al. (2009)	RCT	161	NR (6–11)	NR	Unnamed – language (targets: language comprehension, vocabulary, narrative) SLT or Assistant-delivered	Usual Practice & Same intervention (1 community-based SLT vs. 4 school-based) Schools: (group vs. 1:1) x (SLT vs. Assistant)	CELF: Expressive	Groups vs. UP: 0.37, p <.10 @PT ns @ f/up 4 school-based: ns	High	Preliminary
							Receptive	Groups vs. UP: ns @ both PT & 1 year f/up 4 school-based: ns
Buckingham, Wheldall, and Beaman (2012)	RCT	22	K-2	1 school	MiniLit (Australia) Teacher/TA-delivered	Usual Practice (Tier 1)	Burt Word Reading Test	η2= 0.36, p <.05	Mod	Preliminary
							South Australian Spelling Test	ns
							Non-Word Reading	η^2= 0.59, p <.05
							Wheldall Assessment of Reading	ns
Case et al. (2010)	RCT	30	1	3 schools (9 classes)	Unnamed (targets: PA, phonics, spelling, vocabulary, fluency, comprehension) RAs with teaching experience	Usual Practice (Tier 1)	CTOPP: Rapid Letter Naming, Elision WRMT: Word Attack, Word Identification Researcher-designed: Letter Sound Fluency; Decodable Word Fluency; Word Identification Fluency; Spelling	Trends only: Word Attack (d = 0.86, p = .07) Word Identification (d = 0.49, p <.10)	Mod	Preliminary
Case et al. (2014)	RCT	123	1	NR (15 classes)	Unnamed (targets: PA, phonics, spelling, vocab, fluency, comprehension) Graduate RA-delivered	Usual Practice (strong Tier 1)	TOWRE: Phonemic Decoding Efficiency WRMT: Word Identification, Word Attack Other: Word Spelling	No significant mean group differences @ post-test or 1 year f/up	High	Weak
Chambers et al. (2011)	CRCT	195 177	1 2	33 schools (classes NR)	Team Alphie (computer-based) Teacher-supervised	Alternative intervention (1:1 Success for All tutoring)	WRMT:	+ve (Grade 1s)	Mod	Promising
							Letter-Word Ident.	0.17, p = .05
							Word Attack	0.21, p = .04
							Passage Comprehension	0.15, p = .05
								Grade 2s: ns
Coyne et al. (2013)	RCT	103	K	9 schools (40 classes)	Early Reading Intervention Reading specialist & TA-delivered	Same intervention (tests flexible delivery componentry)	WRMT: Supplementary Letter Checklist, Word Attack, Word Identification, Passage Comprehension CTOPP: Sound Matching, Blending Words DIBELS: Phoneme Segmentation Fluency, Nonsense Word Fluency Others: Passage fluency (ORF); Test of Written Spelling	Sig. +ve @post-test: WRMT (0.54–0.76) & ORF (0.46) +ve @ 1 year follow-up: all WRMT (0.39–0.64), Spelling (ES = .69) & ORF (ES = .61). ES = Hedges’ g	High	Strong
Denton, Fletcher, Taylor, Barth, and Vaughn (2011)	RCT	192	1	9 schools (32 classes)	Read Well-Aligned Tutor-delivered	Same intervention (frequency & duration componentry)	WJIII: Letter-Word Identification, Word Attack, Spelling, Passage Comprehension TOWRE: Word Reading Efficiency, GRADE: Passage Comprehension	All ns	Mod	Preliminary
Denton et al. (2014)	RCT	218	1–2	9 schools (classes NR)	Fountas & Pinnell Guided Reading Teacher & Tutor-delivered	Both usual practice (UP) & alternative intervention (explicit instruction: EI)	WJII: Letter-Word Identification, Word Attack, Passage Comprehension TOWRE: Sight Word Reading, Phonemic Decoding Efficiency Texas Primary Reading Inventory: Progress Monitoring for Beginning Readers) Gates-MacGinitie: Reading Comprehension; Silent Reading Efficiency, Comprehension	Sig. +ve GR>UP Letter-Word Identification (d = 0.32). Sig. +ve EI>UP on all WJII: Letter-Word Identification (d = 0.33), Word Attack (d = 0.59) & Passage Comprehension (d = 0.46)	High	Promising
Denton et al. (2010)	RCT	507	1	31 schools (classes NR)	Responsive Reading Intervention Teacher-delivered	Usual practice (multi-tier NOS)	CTOPP: Blending, Segmenting TOWRE: Sight Word Efficiency, Phonemic Decoding WJIII: Letter-Word Identification, Word Attack, Spelling, Passage comprehension Comprehensive Monitoring of Early Reading Skills: Oral Reading	+ve sig on all measures except CTOPP segmenting (ESs: 0.12– 0.72)	High	Strong
Foorman, Herrera, and Dombek (2018)	CRCT	3447	K-2	55 schools (classes NR)	Composite interventions: Embedded (Curious About Words & Journeys Strategic Intervention) vs. Stand Alone (Sound Partners, Bridge of Vocabulary & Language in Motion) Teacher & TA-delivered	Alternative interventions (Tier 2)	Florida Reading Assessment: Phonological Awareness (K); Word Reading, Vocabulary, Following Directions, Sentence comprehension (K, 1, 2); Spelling (2) SESAT: Word Reading (K), Sentence Reading (K), SAT: Reading Comprehension (1,2) Grade 1 & 2 assessments ARE NOT follow-ups, they are separate cohorts.	Sig: Stand Alone > Embedded on Spelling (g = 0.18) Trend: SA>E for sentence comprehension All others ns for full sample	High	Promising
Gallagher and Chiat (2009)	RCT	24	Pre-K	NR	General approach: Nursery-based consultative SLT model (NB) Teacher with SLT-delivered	Both No Treatment (waitlist) & alternative (intensive clinical SLT)	Reynell Developmental Scales III: comprehension of grammar British Picture Vocabulary Scales Renfrew Tests: Action Picture Test (expressive language), Word Finding Test (expressive vocabulary)	NB>waitlist ns, except parent-report impact scores, p < .01, but ESs positive for all measures (ds 0.06–1.6) CLINICAL SLT>NB (p < .05): receptive grammar, receptive & expressive vocabulary & expressive information. ES: NR for clinical vs. school.	Mod	Weak
Gilbert et al. (2013)	RCT	212	1	11 schools (69 classes)	Unnamed (targets: PA, phonics, sight words, spelling, fluency) Trained graduates-delivered	Both usual practice (Tier 1) & alternative (Tier 3 for non-responders)	Word Reading (latent construct from multiple measures including: TOWRE Sight Word Reading Efficiency, Phonemic Decoding Efficiency; & WRMT Word Attack, Word Identification)	Tier 2 > Tier 1 (d = 0.19, p <.05 @ PT; ns @ 1 & 2 year f/up). Tier 2 = Tier 3	High	Promising
Goldstein et al. (2016)	CRCT	163	Pre-K	NR (32 classes)	Story Friends (automated embedded vocabulary & comprehension instruction) Educator-supervised	Control (same storybooks but no instruction)	Researcher-designed: Unit Vocabulary Test (treatment aligned), Comprehension (proximal) Standardised: CELF & PPVT	+ve sig effect on Tx aligned Vocabulary only (large Cohen f) All others ns	High	Preliminary
Haley, Hulme, Bowyer-Crane, Snowling, and Fricke (2017)	RCT	103	Pre-K	13 preschools	Nursery Language 4 Reading (L4R) TA-delivered	Waitlist (NOS)	CELF-Preschool 2: Sentence Structure, Vocabulary Renfrew: Action Picture Test (expressive) York Assessment of Reading Comprehension: listening comprehension, Core Letter subset, phonemic awareness (alliteration), Early Word Reading, Other: taught vocabulary test	Standardised measures ns +ve sig: naming taught words (d = 1.04, p < .001) & definitions (d = 0.66, p <.001)	Mod	Preliminary
Helf, Cooke, and Flowers (2009)	CRCT	54	1	3 schools (classes NR)	Early Reading Tutor TA-delivered	Same intervention (tests 1:1 vs. group delivery)	DIBELS: Phoneme Segmentation Fluency (PSF), Nonsense Word Fluency (NWF), Oral Reading Fluency (ORF)	All ns	Mod	Weak
Helf, Cooke, and Konrad (2014)	QED	303	K	14 schools (classes NR)	Early Reading Tutor Teacher & TA-delivered	Usual practice (Tier 2)	DIBELS: Phoneme Segmentation Fluency (PSF), Nonsense Word Fluency (NWF)	Sig time x group (p < .01): PSF η^2= 0.06; NWF η^2= 0.05	Low	Weak
Horne (2017)	RCT	38	NR (6–11 years)	2 schools (classes NR)	Comprehension Booster program (computer-based) Researcher with Teacher & TA-supervised	Usual practice (multi-tier)	Neale Analysis of Reading Ability: accuracy, fluency, comprehension	Sig group x time effects: accuracy (time d = 0.79 for Tx, ns for Co); comprehension (time d = 0.68 for Tx, ns for Co)	Low	Not adequately assessed
Hudson, Isakson, Richman, Lane, and Arriaza-Allen (2011)	RCT	58	2	7 schools (classes NR)	Unnamed (focus: repeated reading, PA, phonics) Teacher, TA, Volunteer, Uni-Student-delivered	Same intervention (componentry: reading cues for accuracy or accuracy & speed (A vs. AS)	KTEA II: decoding accuracy TOWRE: Phonemic Decoding Efficiency (PDE) DIBELS: reading fluency WJIII: comprehension Researcher designed: automaticity measure (RDAM)	Sig. effect: AS>A PDE (d = 0.88) RDMA (d = 0.80) A>AS for comprehension (d = −0.70)	High	Promising
Hund-Reid and Schneider (2013)	RCT	37	K	9 schools (10 classes)	Road to Code TA-delivered	Usual practice (NOS)	DIBELS:	+ve sig. effect @ post-test & 1 month respectively	Mod	Preliminary
							Initial Sound Fluency	d = 2.17, 1.33
							Phonemic Segmentation Fluency	d = 1.39, 1.25
							Letter Naming Fluency	d = 0.72 (ns), 1.06
							Nonsense Word Fluency	d = 1.47, 1.50
Kamps et al. (2008)	QED	83	K-2	8 schools (classes NR)	Direct Instruction programs: Reading Mastery, Early Interventions in Reading & Read Well) Teacher & TA-delivered	Alternative interventions (Tier 2 Guided Reading, Programmed Reading & Open Court interventions)	DIBELS: Nonsense Word Fluency, Oral Reading Fluency, WRMT: Word Identification (WI), Word Attack, Passage Comprehension (comp)	DI>GR sig: NWF (d = 0.66). WI & WA (Grades 1 & 2). Comp (Grade 2)	Low	Not adequately assessed
Kelley, Goldstein, Spencer, and Sherman (2015)	RCT	18	Pre-K	3 preschools (3 classes)	Story Friends Researcher-supervised	Usual practice (NOS)	PPVT CELF Researcher-designed: vocabulary (intervention-aligned) comprehension (inferential & literal)	+ve sig. RD-only: vocabulary (d = 1.37 to 2.62) inferential comprehension. (d = 1.10) All others ns.	Mod	Preliminary
Kerins, Trotter, and Schoenbrodt (2010)	RCT	23	1	1 school (classes NR)	Unnamed (comprised adaptation of Lindamood Phoneme Sequencing & Orton-Gillingham programs) SLT & Special Ed Teacher-delivered	Usual practice (Tier 1 only)	TOWRE Other: Individual Language Assessment (county-based phonological awareness measure)	All ns	Low	Not adequately assessed
Lee and Pring (2016)	CRCT	180	K-2	18 schools (10 classes)	Talk Boost TA-delivered	Usual practice (waitlist)	Renfrew: Renfrew Action Picture Test RAPT Grammar Test	Sig. group x time effects (both p = .001, ES: NR^g)	Mod	Promising
Little et al. (2012)	RCT	90	K	8 schools (classes NR)	Early Reading Intervention Reading specialist & Teacher-delivered	Usual practice (very strong Tier 2)^h	DIBELS: Letter Naming Fluency, Phoneme Segmentation Fluency, Nonsense Word Fluency WRMT: Word Identification, Word Attack, Supplementary Letter Checklist CTOPP: Sound Matching (SM), Rapid Object Naming, Blending Words Test of Written Spelling: spelling Mac Gets Well reading test	Trends only: SM:0.59 (p = .06) Spelling: 0.50 (p = .08) Some sig ideographic differences	Mod	Preliminary
Lonigan and Phillips (2016) Experiment 1	RCT	93	Pre-K	12 preschools (classes NR)	Unnamed (focus: needs-aligned intervention: code-based, language or both)	Usual practice (strong Tier 1)	CELF-Preschool TERA-3 Other: Preschool Comprehensive Test of Phonological & Print Processing (researcher designed), Code & Language Intervention Post-test (curriculum-aligned assessment)	Trend: CELF Expressive Language; language-intervention> UP (p = .09, d = 0.27). CLIP Letter Names Test; code-focused> UP (p = .07, d = 0.25)	High	Preliminary
Lonigan and Phillips (2016) Experiment 2	RCT	184	Pre-K	19 preschools (classes NR)	Unnamed (focus: needs-aligned intervention: code-based, language or both) Implementer NR	Usual practice (strong Tier 1)	CELF-Preschool TERA-3 Other: Preschool Comprehensive Test of Phonological & Print Processing (researcher designed), Code & Language Intervention Post-test (curriculum-aligned assessment)	+ve sig. RD-only: language-Tx > UP & code-Tx>UP (ds = 0.21−0.5, ps <.05). Note: analyses only included outcomes targeted.	Mod	Preliminary
Lovett et al. (2017)	QED	219	1–3	NR	Triple Focus Reading Program (PHAB + WIST + RAVE-O) Special Ed Teachers-delivered	Usual practice (NOS) Componentry: Grade at intervention	WRMT: Word Identification, Word Attack, Passage Comprehension WRAT: Reading SRI: Reading Comprehension TOWRE: Sight Word Efficiency, Phonemic Decoding Efficiency GORT: Reading Comprehension PIAT: Spelling Others: phonics & vocabulary.	Sig +ve Tx effects on all 14 reading measures (mean d = 0.80 standardised, 1.69 for others).	Mod	Preliminary (QED)
Madden and Slavin (2017) Experiment 1	CRCT	427	1–3	14 schools (classes NR)	Tutoring with Alphie (computer program) TA-supervised	Usual practice (limited multi-tier)	Success for All periodic reading assessment combined with teacher observation & classwork progress.	Sig +ve Tx effect: d = 0.46	Low	Not adequately assessed
Madden and Slavin (2017) Experiment 2	CRCT	478	1–3	13 schools (classes NR)	Tutoring with Alphie (computer program) TA-supervised	Usual practice (limited multi-tier)	Success for All periodic reading assessment combined with teacher observation & classwork progress.	Sig +ve Tx effect: d = 0.40	Low	Not adequately assessed
Morris et al. (2012)	CRCT	279	2–3	NR	PHAB+RAVE-O, PHAB+WIST (PHAST)^i Vs. Class skills + Math (PT only) PHAB+ Class skills (PT & FU) Teacher-delivered	Alternative interventions: (language & non-language) (Tier 2)	WRMT: Word Identification, Word Attack, Passage Comprehension; Wide Range Achievement Test Reading, Spelling, Arithmetic TOWRE: Word Reading Efficiency Non-Word Reading Gray Oral Reading Test	Consistent significant +ve @ post-test & 1 year f/up. Effect sizes NR but calculation from F values indicates large effects @ post-test^j	Mod	Promising
Nielsen and Friesen (2012)	QED	28	K	1 school (3 classes)	Unnamed (focus: storybook-based vocabulary & narrative) Post-graduate RA-delivered	Usual practice (NOS)	TOLD: vocabulary, semantics TNL: Narrative Comprehension, Oral Narration Treatment-aligned tests: Taught vocabulary Narrative	Sig +ve Tx effect for RD-only: vocabulary	Low	Not adequately assessed
Oostdam, Blok, and Boendermaker (2015) Experiment 2	RCT	139	2–4	8 schools (classes NR)	General approach: Group-based Guided Oral Reading TA-delivered	Usual practice (some Tier 2)	Dutch versions of standardised:	Sig +ve effect:	Mod	Promising
							Reading Fluency	p < .05^k
							Comprehension	ns
							Vocabulary (PPVT)	ns
							Other: Reading Attitude (7 items)	p < .05
Phillips et al. (2016)	RCT	42	Pre-K	5 preschools (10 classes)	Unnamed (oral language focus: syntactic & sematic features)	Usual practice (multi-tier)	CELF-P2: sentence structure OWLS: receptive language composite (sematic, syntactic & supra-linguistic features) WJIII: Picture Vocabulary Test Others (researcher-developed): Intervention Aligned Assessment (IAA: receptive & expressive language), Listening Comprehension (LC)	Sig +ve for RD measures: (IAA g = 0.88; LC Units 3–4 g = .78 Trends: OWLS g = 0.31 (p = .07), LC Units 1 & 2 g = 0.37 (p = .08).	High	Preliminary
Pollard-Durodola et al. (2011)	CRCT	125	Pre-K	6 preschools (classes NR)	Words of Oral Reading & Language Development (WORLD) Teacher-delivered	Usual practice (Tier 1)	PPVT: receptive vocabulary EOWPVT: expressive vocabulary Others (researcher-designed): Receptive Vocabulary (Tx aligned) Expressive Vocabulary	Sig +ve effects: RD measures only	Mod	Preliminary
Puhalla (2011)	RCT	44	1	5 schools (classes NR)	Read Aloud Booster Teacher-delivered	Usual practice (Tier 1: Read Aloud class curriculum)	Researcher-designed: storybook vocabulary assessment (child defines taught words)	Sig +ve effect: Tx>UP η^2= 0.46	Mod	Preliminary
Pullen and Lane (2014)	RCT	98	1	9 schools (classes NR)	Unnamed (focus: decoding with manipulative letters) Teacher-delivered (pre- & in-service)	Both usual practice (Tier 1) & Same program (no manipulatives [NM]) Componentry: manipulatives	WDRB: Sight Words Informal: real & pseudoword decoding, Phonological Awareness (PA) Sight words	Sig +ve effect RD only: Tx>NM (d = 0.42 PA; 0.56 decoding real words) Tx>UP (d = 0.43 PA; 0.70 decoding pseudo words; 0.80 informal test of sight words)	Mod	Preliminary
Pullen, Tuckwiller, Konold, Maynard, and Coyne (2010)	RCT	98 at-risk	1	3 schools (12 classes)	Unnamed (focus: vocabulary in shared storybook reading with rich instruction) Education student-delivered	Usual practice (Tier 1) & not-at-risk comparison	Researcher-designed Tx-aligned vocabulary tests:	Sig +ve effect: Post-test:	Mod	Preliminary
							Receptive	d = 0.37
							Contextual	d = 0.64
							Expressive	ns
								Follow-up (4 weeks): planned contrasts ns
Ritter, Park, Saxon, and Colson (2013)	QED	64	1–3	3 schools	Unnamed (focus: narrative-based, PA & phonics instruction) SLT-delivered	Alternative intervention (traditional language therapy – no PA)	WDRB: Letter-Word Identification, Word Attack, Passage Comprehension CTOPP: Blending Words, Blending Nonwords	Sig group x time effects all measures (Tx pre-post sig; Control ns): (Tx pre-post gain η^2 = 0.40–0.61) Tx>control (0.57–0.72 for blending; 0.51–0.58 for word-level reading skills).	Low	Not adequately assessed
Roskos and Burstein (2011)	QED	74	Pre-K	NR (12 classes)	General Approach: Say-Tell-Do vocabulary strategy Literacy-coach, Teacher, TA-delivered	Usual practice (classroom only)	PPVT: vocabulary Researcher-designed Tx-aligned: Expressive Vocabulary (EV) Receptive Vocabulary (RV)	Sig +ve effect: PPVT post-test (d = 0.58, p = .03); RD gain: (receptive d = 0.68; expressive d = 0.45)	Low	Not adequately assessed
Ryder et al. (2008)	RCT	24	2–3	1 school (4 classes)	Unnamed (focus: PA & phonics) TA-delivered	Usual practice (New Zealand whole-language analytic phonics)	At post-test:	Sig. Tx>UP	Mod	Preliminary
							Phonemic awareness	1.71
							Pseudo word decoding	1.69
							Burt Word Reading	0.88
							Neale accuracy	0.70
							Neale comprehension	0.98
							2-year follow-up:
							Burt Word Reading	d =0.72
							Neale accuracy	d =0.81
Schuele et al. (2008)	QED	36	K	3 schools (6 classes)	Intensive Phonological Awareness Program SLT-delivered Tier 2 component	Usual practice (Tier 1)	Phonological Awareness Literacy Skills for Kindergarteners test: Rhyme Awareness, Beginning Sound Awareness, Alphabet Knowledge; Letter Sounds; & Word Recognition in Isolation; & a Spelling measure	Spelling (p =.016, eta squared =0.16) NS for Letter Sound Knowledge & Word Recognition; others NR	Low	Not adequately assessed
Schwartz et al. (2012)	RCT	170	1	NR	Reading Recovery (group formats 1:2, 1:3 and 1:5 vs. 1:1) Special Ed Teacher-delivered	Same intervention (componentry: group size)	Sloson Oral Reading Test-Revised Other: Clay Observation Survey: Text Reading Level Ohio Word Test Hearing and Recording Sounds in Word Letter Identification Concepts about Print Writing Vocabulary	Sig group x time effects: SORT-R & all Clay subtests except Letter Identification On six of seven tests 1:1 > small groups combined But Post-hoc tests: 1:1 = 1:2 & 1:3 on several measures Small group comparisons: ns	Mod	Preliminary
Senechal, Ouellette, Pagan, and Lever (2012)	RCT	56	K	6 schools (classes NR)	Unnamed (focus: Invented spelling [IS]) Implementer NR	Both usual practice (storybook reading [SR]) & Alternative (phonemic awareness training – no printed letters [PA])	CTOPP: composite from Sound Matching, Elision, & Blending Words Researcher-designed: Alphabet knowledge, Word reading, Invented Spelling & Learn-to-Read	Sig +ve effects IS+PA>SR on CTOPP (0.45) IS> others combined for Learn-to-Read (0.50) & Invented Spelling (0.58).	Mod	Preliminary
Simmons et al. (2011)	CRCT	206	K	12 Schools (57 classes)	Early Reading Intervention Reading specialists, Teacher & TA-delivered	Usual practice (strong Tier 2)		Sig Tx effects	Mod	Promising
							CTOPP:
							Blending Words	0.40
							Sound Matching	0.42
							DIBELS:
							Phoneme Segmentation Fluency	0.47
							Nonsense Word Fluency	ns
							WRMT:
							Supplementary Letter Checklist	0.36–0.41
							Word Attack	0.51
							Word Identification	ns
							Passage Comprehension	ns
							TOWRE:
							Phonemic Decoding Efficiency	ns
							Sight Word Efficiency	ns
							Test of Written Spelling:
							spelling	ns
								(all ES Hedges’ deltaT)
Simmons et al. (2014)	CRCT	368	K	22 schools (classes NR)	Early Reading Intervention Reading specialists, Teacher & TA-delivered	Usual practice (Tier 2)	2-year follow-up: DIBELS: Oral Reading Fluency WRMT: Word Identification, Passage Comprehension	Regression: Intervention ns	Mod	Weak (but 2-year follow-up, not immediate post-test)
Smith et al. (2016)	CRCT	811	1	44 schools (classes NR)	Enhanced Core Reading Instruction (ECRI; multi-tiered curriculum-aligned) Reading-specialists, Teachers & TA-delivered	Usual practice (strong Tier 2 but un-aligned with class content)	Fall-Spring gain	*Sig +ve effects (g)	Mod	Promising
							SESAT:
							Total	0.21, ns
							Word Reading, Sentence Reading	0.24* 0.18, ns
							DIBELS:
							Nonsense Word Reading	0.19, ns
							Oral Reading Fluency	0.12, ns
							WRMT:
							Word Attack	0.32*
							Word Identification	0.24, p < .10
Spencer et al. (2015)	RCT	22	Pre-K	NR (3 classes)	Story Champs (focus: oral narrative) RA-delivered	Usual practice (Tier 1)	Narrative Language Measure (NLM) The Renfrew Bus Story Other: Personal Stories	Sig +ve Tx effects: NLM d = 1.05 @ PT, & 0.86@ 4 weeks f/up; Renfrew 1.04 @ PT, ns @ f/up, ns for personal stories	Mod	Preliminary
Torgesen, Wagner, Rashotte, Herron, and Lindamood (2010)	RCT	112	1	3 schools (classes NR)	Read, Write & Type (RWT) & Lindamood Phoneme Sequencing (LIPS) (data combined) Teacher & computer-delivered	Both usual practice (multi-tier) & alternative	CTOPP: elision, blending words, segmenting words, rapid digit & rapid letter naming; WRMT: Word Identification, Word Attack, Passage Comprehension TOWRE: sight word efficiency; GORT: text reading accuracy, text reading fluency, comprehension WRAT: Spelling	RWT =LIPS all measures; Sig Tx Effects RWT/LIPS>UP@ PT: phonological awareness, rapid naming, decoding, spelling, & comprehension (d =0.17–0.77) RWT/LIPS>UP @ 1 year: decoding, rapid naming, spelling (0.43–0.66)	Mod	Promising
Vadasy and Sanders (2008a)	RCT	188	2–3	13 schools (classes NR)	Quick Reads TA-delivered	Usual practice (Tier 1)	WRMT: Word Identification TOWRE: Sight Word Efficiency DIBELS: Oral Reading Fluency GORT: Fluency, Comprehension	Sig. +ve Tx effects: on all fluency measures (p <.05, ds 0.29–0.43) No effects on word reading or comprehension.	High	Strong
Vadasy and Sanders (2008b)	QRCT	86 rel.	K	13 schools (30 classes)	Unnamed (focus: integrated explicit phonemic & alphabetic instruction with decodable text) TA-delivered	Both usual practice (no treatment) & alternative (1:1 tutoring)	DIBELS: Letter Name Fluency CTOPP: Phonological Awareness composite; Phonemic Decoding & Sight Word Efficiency WRMT: Word Attack, Word Identification, Passage Comprehension WRAT: Spelling Oral reading fluency (ORF): Mac Gets Well	Relevant comparisons: 1:2 vs. 1:1 (ns) & 1:2 vs. control (not reported).	Mod	Weak
Westerveld and Gillon (2008)	QED	10	NR (7–9 years)	NR (‘2 sites’)	Unnamed (focus: oral language story structure instruction) SLT-delivered	Usual practice (waitlist, New Zealand system)	Test of Oral Narrative (scoring does not appear to be standardised procedure)	Sig +ve Tx effects for oral narrative comp’n (1.55) & production (1.89), & verbal fluency (1.65)	Low	Not adequately assessed
Wonder-mcdowell et al. (2011)	RCT	133	2	11 schools (classes NR)	Read Well-Aligned (multi-tier; Tier 1 & Tier 2 aligned) Reading specialists & Teacher-delivered	Same intervention (Tier 2 Read Well unaligned with class content)		Sig +ve Tx (η^2):	High	Strong
							DIBELS: Oral Reading Fluency	0.17
							WRMT:
							Word Attack	0.13
							Word Identification	0.08
							Passage Comprehension	0.18
Zettler-Greeley, Bailet, Murphy, DeLucca, and Branum-Martin (2018)	CRCT	476	Pre-K	114 preschools (classes NR)	Nemours BrightStart! Implementer-unclear, implies teacher	Usual practice (waitlist)		+ve Tx effects:	High	Promising
							GRTR: letter knowledge & PA	0.39
							Test of Preschool Early
							Literacy: Print Knowledge	0.74
							Elision	0.46,
							Blending	−0.13
							ALL: Rhyme Knowledge	0.37
								(Hedges’ g)
Zucker, Solari, Landry, and Swank (2013)	CRCT	125	Pre-K	NR (39 classes)	Developing Talkers: Pre-K curriculum Educator-delivered	Usual practice (some Tier 2).	Tx-aligned measure: C-PALLS: Rapid Vocabulary Naming	Sig. +ve Tx effect d = 0.81	Mod	Preliminary

^aNumber of participants relevant to Tier 2 intervention vs. comparisons only (not overall number screened or number assigned to Tier 3 interventions unless directly contrasted with Tier 2 intervention).

^bAge is provided if grade was not reported.

^cResults are from analyses of mean differences only. Results from growth curve analyses or analyses comparing % of intervention and control groups performing at specific cut-points are not reported here. Unless stated otherwise, effect sizes are Cohen’s d or Hedges’ g.

^dThe content of this column reflects only the results relevant to our research questions, not results for other research questions addressed by the studies.

^eStudy quality was categorised according to cut points applied to the risk of bias checklist described in the Method and in Table S3 (online only) (i.e. low <50%, moderate 50–74%, high 75+%).

^fThis refers to whether the intervention has a positive effect using a nuanced framework specifically designed for this project to assess the relative strength of positive evidence in each study.

^gUsing reported mean, standard deviation, and sample size information, ES is calculated as d =0.56 for RAPT, and 0.34 for Grammar.

^hThe comparison condition was school implemented Tier 2 intervention within a well-established RTI framework where several schools used alternative commercial interventions (Road to Code and Reading Mastery Plus) and dose and group size were equivalent.

ⁱPHAB =Phonological Analysis and Blending; PHAST =Phonological and Strategy Training (which is PHAB + WIST); RAVE-O =Retrieval, Automaticity, Vocabulary, Engagement with Language and Orthography; WIST =Word Identification Strategies Training.

^jContrasts show the three reading conditions combined significantly outperformed the mathematics control, the multi-component conditions combined (PHAB+WIST, PHAB+RAVE-O) significantly outperformed the PHAB-+CSS condition, and the multi-component conditions performed similarly to each other, ns.

^kEffect size not reported, but mean, standard deviation, and sample size indicate d =0.22.

ns = no significant intervention effect; SLT = Speech & Language Therapist; TA = Teacher Aide; UP =Usual Practice; ALL =Assessment of Language & Literacy; CELF =Clinical Evaluation of Language Fundamentals; C-PALLS = Center for Improving the Readiness of Children for Learning & Education (CIRCLE) Phonological Awareness, Language, & Literacy System; CRCT =cluster randomised controlled trial; CTOPP =Comprehensive Test of Phonological Processing; DIBELS = Dynamic Indicators of Basic Early Literacy Skills; EOWPVT = Expressive One-Word Picture Vocabulary Test; FU =Follow-up; GRADE = Group Reading Assessment & Diagnostic Evaluation; GRTR =Get Ready to Read screener; GORT = Gray Oral Reading Test; IGDIEL =Individual Growth & Development Indicators of Early Literacy; KTEA = Kaufman Test of Educational Achievement; NOS =Not otherwise specified; NR =Not reported; ns =not significant; OWLS: Oral & Written Language Scales; PALSB = Phonological Awareness & Literacy Screening Battery; paraprof. =paraprofessional; PIAT = Peabody Individual Achievement Test; PPVT =Peabody Picture Vocabulary Test; PT =Post-test; Q.ED =quasi-experimental design; RCT = randomised controlled trial; RD =Researcher-designed; SAT =Stanford Achievement Test; SESAT = Stanford Early Scholastic Achievement Test; SRI = Standardised Reading Inventory; SLP = Speech & Language Pathologist; SLT = Speech & Language Therapy/Therapist; strong Tier 1 = Tier 1 highly structured/supported; TD = Typically developing; TERA-3 = Test of Early Reading Ability 3^rd Edition; TNL = Test of Narrative Language; TOLD =Test of Oral Language Development; TPEL = Test of Preschool Early Literacy; TOWRE =Test of Word Reading Efficiency; Tx = Treatment; UP = Usual Practice; vocab = vocabulary; WDRB = Woodcock Diagnostic Reading Battery; WJIII = Woodcock Johnson Tests of Achievement; WRAT = Wide Range Achievement Test; WRMT = Woodcock Reading Mastery Test.

Table 3. Characteristics of the included studies.

Characteristic	k	%
Country
North America	46	84
United Kingdom	5	9
New Zealand	2	4
Australia	1	2
Netherlands	1	2
Sample age
Included grades PK-2	42	76
Preschool only	13	24
Targeted
Reading difficulties only	37	67
Oral language difficulty only	13	24
Combined risks	5	9
Intervention componentry
Maximum of five per group (range: 2–10 students)	41	75
Session duration 20–40 min (range: 10–60 min)	36	65
At least three sessions per week (range: once to five times)	40	73
Intervention duration at least 3 months (range: 6 weeks to 2 years)	28	51
Delivered by school personnel	35	64
Delivered by allied health professionals	6	11
Delivered by researchers	5	9
Study design
RCT or CRCT	45	82
Comparison with usual practice or waitlist	41	75
Classroom instruction only usual practice	13	24
Multi-tiered usual practice	15	27
Comparison with alternative intervention	20	36
Comparison to both usual practice and other^a intervention	6	11
Rated low quality	11	20
Rated moderate quality	30	55
Rated high quality	14	25
Outcomes assessed
Phonemic awareness	25	45
Phonics	29	53
Nonsense or pseudo-word decoding	26	47
Real word decoding	27	49
Sight word reading	12	22
Reading fluency	21	38
Reading comprehension	20	36
Vocabulary	14	25
Attitude to reading	1	2
Measures
Woodcock Reading Mastery Test (WRMT; Woodcock, 1998)	15	27
Woodcock Johnson III (WJIII; Woodcock & Johnson, 1989)	4	7
Dynamic Indicators of Basic Early Literacy Skills (DIBELS; Good & Kaminski, 2002)	13	24
Test of Word Reading Efficiency (TOWRE; Torgesen, Wagner, & Rashotte, 1999)	12	22
Comprehensive Test of Phonological Processing (CTOPP; Wagner, Torgesen, & Rashotte, 1999)	9	16
Clinical Evaluation of Language Fundamentals (CELF; Semel, Wiig, & Secord, 2000)	7	13
Peabody Picture Vocabulary Test (PPVT; Dunn & Dunn, 1997)	6	11
Renfrew Language Scales (Renfrew, 1988, Renfrew, 1997)	3	5
Research designed measures of vocabulary	7	13
Post-test timing
Immediate	44	80
1 month	3	5
12 months or more	8	15

^aOr variation of the same intervention.

Study findings

Overall, most studies reported at least one statistically significant positive effect of the Tier 2 oral language or literacy intervention relative to a usual practice comparison condition (whether Tier 1 whole-of-class instruction or multi-tiered usual practices). In some cases, Tier 2 interventions performed similarly to Tier 3 interventions where students received one-to-one sessions (Gilbert et al., 2013; Schwartz, Schmitt, & Lose, 2012). Effect sizes ranged from very small (i.e. < 0.20) to quite large (e.g. 2.17). As shown in Table 2, effect sizes were not obviously related to study quality (i.e. large effects were observed in both low- and high-quality studies, as were small effects). As a group, the 55 studies indicate that school-based oral language and literacy interventions for children in the preschool and early primary school grades can have a significant effect on short-term intended outcomes, though the magnitude of effects varies considerably (and the largest effects tend to emerge in studies with very small samples or on treatment-aligned, researcher-designed measures). Nevertheless, educationally meaningful effects (some considered moderate-to-large by conventional standards) were also observed in high-quality studies and on summarised measures. For example, flexible delivery of the Early Reading Intervention (Coyne et al., 2013; Little et al., 2012; Simmons et al., 2011, 2014) demonstrated positive effects on the WRMT subscales ranging from 0.54 to 0.76 at immediate post-test and 0.39 to 0.64 at 1-year follow-up; and the Responsive Reading Intervention (Denton et al., 2010) demonstrated positive effects ranging from 0.44 on the TOWRE, to 0.63 on some WJIII subscales. As might be expected, a general shift in intervention focus was observed such that preschool interventions predominantly involved oral language with particular emphasis on vocabulary, whereas primary school interventions were more focused on reading, with most emphasis on decoding.

Intervention-level evidence

The strength of evidence assessments for the 47 interventions in the included studies is summarised in Table 4. The three evidence rating systems identified a different, but overlapping, shortlist of interventions (i.e. those with the strongest evidence). The customised evidence rating system identified three interventions with Very Strong evidence. These are Early Reading Intervention (ERI; Coyne et al., 2013; Little et al., 2012; Simmons et al., 2011, 2014), PHAST/PHAB+RAVE-O (Morris et al., 2012) and Read Well-Aligned (Wonder-mcdowell, Reutzel, & Smith, 2011). Using the CEBC Scientific Rating Scale, three interventions received high (Well Supported or Supported) evidence ratings. These are the Early Reading Intervention (ERI), PHAST/PHAB+RAVE-O (Morris et al., 2012) and Ryder, Tunmer, and Greaney (2008) Unnamed Phonological Awareness and Phonics intervention. Using the NHMRC evidence matrix, no interventions met the highest level of evidence, but three aligned with the second-highest category. These are Early Reading Intervention (ERI; (Coyne et al., 2013; Little et al., 2012; Simmons et al., 2011)), Story Friends (Goldstein et al., 2016), and Lonigan and Phillips (2016) Unnamed needs-aligned intervention. As shown in Table 4, the evidence assessments show six interventions were shortlisted, only two of which were identified by two or more of the three evidence rating systems (Early Reading Intervention and PHAST/PHAB+RAVE-O).

Table 4. Strength of evidence assessments for each intervention.

Intervention	Studies	CEBC Scientific Evidence Rating	NHMRC Recommendation Grade	Customised rating
Early Reading Intervention	Coyne et al. (2013) intervention componentry analysis; Little et al. (2012) Florida study; Simmons et al. (2011); Simmons et al. (2014) Texas and Connecticut study with outcomes at end-Kinder and Grade 2, respectively	Well Supported	B	Very Strong
Enhanced Core Reading Instruction	Smith et al. (2016)	Promising	C	Promising
PHAST (PHAB+WIST), PHAB+RAVE-O	Morris et al. (2012)	Supported	C	Very Strong ^a
Read Well-Aligned	Wonder-mcdowell et al. (2011); Denton et al. (2011); independent studies	Promising	C	Very Strong
Quick Reads	Vadasy and Sanders (2008a)	Promising	C	Strong
Responsive Reading Intervention	Denton et al. (2010)	Promising	C	Strong
Group-based Guided Oral Reading	Oostdam et al. (2015)	Promising	C	Promising
Guided Reading & Explicit Instruction	Denton et al. (2014)	Promising	C	Promising
Mixed: Embedded vs. Stand-alone interventions	Foorman et al. (2018)	Promising	C	Promising
Mixed: Read, Write & Type; LIPS	Torgesen et al. (2010)	Promising	C	Promising
Nemours Bright Start	Zettler-Greeley et al. (2018)	Promising	C	Promising
Talk Boost	Lee and Pring (2016)	Promising	C	Promising
Team Alphie	Chambers et al. (2011)	Promising	C	Promising
Unnamed (multi-component reading skills intervention)	Gilbert et al. (2013)	Promising	C	Promising
Unnamed (repeated reading phonics and phonological awareness intervention)	Hudson et al. (2011)	Promising	C	Promising
Developing Talkers Pre-K Curriculum	Zucker et al. (2013)	Promising	C	Preliminary
MiniLit	Buckingham et al. (2012)	Promising	C	Preliminary
Nursery Language 4 Reading	Haley et al. (2017)	Promising	C	Preliminary
Read Aloud Booster	Puhalla (2011)	Promising	C	Preliminary
Reading Recovery (Group formats)	Schwartz et al. (2012)	Promising	C	Preliminary
Road to Code	Hund-Reid and Schneider (2013)	Promising	C	Preliminary
Story Champs	Spencer et al. (2015)	Promising	C	Preliminary
Story Friends	Goldstein et al. (2016); Kelley et al. (2015); independent studies	Promising	B	Preliminary
Triple Focus Reading Program	Lovett et al. (2017)	Promising	C	Preliminary^b
Unnamed (Phonics & Phonological Awareness)	Bailet et al. (2009)	Promising	C	Preliminary
Unnamed (SLT intervention)	Boyle et al. (2009)	Promising	C	Preliminary
Unnamed (multi-component reading skills intervention)	Case et al. (2010)	Promising	C	Preliminary
Unnamed (needs-aligned strands)	Lonigan and Phillips (2016) two independent studies	Promising	B	Preliminary
Unnamed (oral language intervention)	Phillips et al. (2016)	Promising	C	Preliminary
Unnamed (storybook reading with “rich” instruction)	Pullen et al. (2010)	Promising	C	Preliminary
Unnamed (use of decontextualised manipulatives)	Pullen and Lane (2014)	Promising	C	Preliminary
Unnamed (phonological awareness and phonics intervention)	Ryder et al. (2008)	Supported	C	Preliminary
Unnamed (invented spelling)	Senechal et al. (2012)	Promising	C	Preliminary
Words of Oral Language and Reading Development (WORLD)	Pollard-Durodola et al. (2011)	Promising	C	Preliminary
Early Reading Tutor	Helf et al. (2009); (2014), independent studies	Not Rateable	C	Weak
Unnamed (multi-component reading skills intervention)	Case et al. (2014)	Not Rateable	C	Weak
Unnamed (Nursery-based SLT)	Gallagher and Chiat (2009)	Not Rateable	C	Weak
Unnamed (phonemic and alphabetic instruction)	Vada sy and Sanders (2008b)	Not Rateable	C	Weak
Comprehension Booster	Horne (2017)	Promising	C	Not adequately assessed
Intensive Phonological Awareness Program	Schuele et al. (2008)	Promising	D	Not adequately assessed
Mixed: Reading Mastery, Early Interventions in Reading, Read Well	Kamps et al. (2008)	Promising	D	Not adequately assessed
Say-tell-do strategy	Roskos and Burstein (2011)	Promising	D	Not adequately assessed
Tutoring with Alphie	Madden and Slavin (2017), p. 2 independent studies	Promising	D	Not adequately assessed
Unnamed (phonological awareness intervention)	Kerins et al. (2010)	Not Rateable	D	Not adequately assessed
Unnamed (storybook language skills intervention)	Nielsen and Friesen (2012)	Promising	D	Not adequately assessed
Unnamed (phonological awareness and phonics intervention)	Ritter et al. (2013)	Promising	D	Not adequately assessed
Unnamed (oral language intervention)	Westerveld and Gillon (2008)	Promising	D	Not adequately assessed

^aThe strength of evidence for PHAB+WIST/PHAB+RAVE-O (each administered separately) was rated very strong because in addition to CRCT-level evidence there is converging evidence from the quasi-experimental study of the Triple Focus Program (where students received all three PHAB, WIST, and RAVE-O components together).

^bThe strength of evidence for the Triple Focus Program was rated preliminary because the combined-administration of the three core components has been tested in a quasi-experimental trial only.

Synthesis: shortlist of interventions

Componentry (procedural and content) for the six interventions shortlisted by the strength of evidence assessments is summarised in Table 5. See Table S6 (online only) for a detailed description of componentry for all programs, and Table S7 (online only) for a simplified visual representation of componentry for all programs together with evidence ratings for each intervention. Table 5 shows that the number of students in the small-group interventions were approximately 3–5 students for four of the six interventions (the remaining two had a range of 2– 4 or 2–6 students). Intervention dosage schedule was approximately 4–5 times per week for four of the interventions (sessions were 20–40 min for three of these, and 60 min for the other). For another intervention, however, dose was much lower (sessions were 9–12 min, 3 times per week), and for the final intervention, dose was 30 min per session and offered 2–5 times per week depending on the schedule for each treatment group. There was also variation regarding the duration of the shortlisted interventions with the number of implementation weeks ranging from 11 to 26. As can be seen in Table 5, intervention durations were mostly reported as an approximation, a range, or shown to vary according to the intervention session schedule.

Table 5. Componentry of shortlisted interventions.

Intervention & grade	Process components	Content focus	Pedagogical strategies	Implementer
Early Reading Intervention (4 studies covering 2 trials: Coyne et al., 2013; Little et al., 2012; Simmons et al., 2011; Simmons et al., 2014) Kinder	3–5 per group 30 min, x5 p/week ~26 weeks (126 lessons)	Phonological Awareness, Phonics, Sight Words, Listening Comprehension, Spelling & Writing	Corrective Feedback, Modelling, Explicit Instruction & Interactive	Teachers, Reading Specialists & Paraprofessionals Training: 2 days +IS
Lonigan and Phillips (2016) Unnamed needs-aligned intervention Pre-K	2–6 per group 15–40 min, x4 p/week 11 weeks	Code-aligned: Phonological Awareness & Phonics Language-aligned: Syntax & Vocabulary Both: Students deficient in both received both strands	Modelling, Mastery, Scaffolding & Interactive Other: Alignment of content to deficiency	‘Interventionists’ (university educated, discipline NR) Training: 7–20 h +IS
PHAB+WIST (PHAST)/ PHAB+RAVE-O (Morris et al., 2012) Grade 2–3	4 per group 60 min, x5 p/week ~14 weeks (70 sessions)	Phonological Awareness, Phonics, Vocabulary, Syntax & Morphology	Self-monitoring, Meta-cognitive Strategies, Mastery, Scaffolding, Explicit Instruction, Interactive Other: Attributional re-training	Teachers (mean experience: 7 years, most in graduate courses) Training: 1 week (trained to competence) +IS
Read Well-Aligned (2 trials: Wonder-mcdowell et al., 2011; Denton et al., 2011) Grades 1–2	2–4 per group 30 min, x2–5 p/week 8, 16, or 20 weeks Variations to frequency & duration: Condensed = x4, 8 weeks; Distributed = x2, 16 weeks; Extended = x4, 16 weeks	Phonological Awareness, Phonics, Sight Words, Fluency, Reading Comprehension, Vocabulary & Spelling	Modelling, Mastery, Scaffolding, Explicit Instruction & Interactive	Reading Specialist Teachers (min 5 years experience) Training: 70 h + IS Or Tutors Training: 12 h + IS
Ryder et al. (2008) Unnamed Phonological Awareness & Phonics intervention Grades 2–3	3 per group 20–30 min, x4 p/week 24 weeks (over 3 terms)	Phonological Awareness & Phonics	Modelling, Explicit Instruction & Interactive, Other: Manipulatives	Teaching Assistant Training: Yes, but NOS +IS
Story Friends (Goldstein et al., 2016; Kelley et al., 2015) Pre-K	3 per group 9–12 min, x3 p/week 14–26 weeks	Vocabulary & Narrative	Modelling, Explicit Instruction & Interactive	Automated Storybook (educator or researcher supervises) Training: NR

IS = implementation support (e.g. feedback during implementation or booster sessions or meetings once intervention being delivered); NOS = not otherwise specified; NR = not reported; ~ = approximately (based on number of sessions and frequency)

For the most part, the content of the shortlisted interventions focused on multiple language and reading components. Four of the interventions included instruction and activities focused on a combination of phonological awareness (e.g. rhyming, identifying syllables, isolating sounds for first and last letters, and blending and segmenting sounds), phonics (e.g. letter-sound correspondence), reading fluency, listening comprehension (e.g. following multi-step instructions and re-telling stories), vocabulary (e.g. targeting semantic categories and concepts in a story), syntax rules and uses, and morphological components. The other two each had a narrower focus, with one centred on instruction in phonological awareness and phonics, and the other focusing on narrative development with lessons incorporating vocabulary instruction (in unfamiliar words that occur frequently in spoken and written language), and language comprehension (using inferential questions about story predictions and character emotions and actions).

Of the pedagogical strategies included in the shortlisted interventions, use of modelling, explicit instruction, and interactive programming⁹ were reported most frequently (each reported for five interventions), followed by scaffolding and mastery (reported for three interventions). Use of self-monitoring, provision of corrective feedback, and meta-cognitive strategies were less commonly reported, featuring in one intervention each. Detailed, sequenced lessons and instructional scripts were utilised by four of the shortlisted interventions, and materials, objects, and props (e.g. letter tiles, picture cards, individual white boards) were used in three interventions to make abstract concepts more concrete. Most shortlisted interventions were implemented either by teachers or reading specialists or by teams that included teachers in addition to teaching assistants. It is difficult to identify common characteristics regarding provider qualifications, previous experience, and training as this was not well-reported. However, it appears that in most cases, the interventions were implemented by personnel with relevant tertiary-level qualifications (e.g. teachers, teaching assistants, reading specialists, paraprofessionals, or tutors). Five of the six shortlisted interventions involved formal pre-intervention training; however, the duration of training varied greatly, ranging from approximately 7 to 70 h. As these interventions were being delivered, support for implementers was characterised by some form of coaching (e.g. modelling, individualised feedback, regular meetings), and three of the interventions also included formalised peer support (e.g. regular small-group meetings, group email list/forum). However, the intensity of the implementation support was not described in detail. There was one intervention that did not involve training and support as this was an automated storybook supervised by an educator or researcher (Story Friends) and was specifically designed to minimise the need for extensive training or implementation support.

For pragmatic purposes, it is important to know not only which interventions have a strong evidence-base, but whether they are well-received by stakeholders (such as teachers, parents, and students) and are affordable. A quantitative evaluation of social validity was presented by only seven of all included studies, and only two of the shortlisted interventions (Early Reading Intervention and Story Friends). Both these interventions were rated positively by teachers, but parent and student perceptions were not assessed. Detailed information about intervention costs was available for only two interventions (Boyle, McCartney, O’Hare, & Forbes, 2009; Madden & Slavin, 2017), neither of which was shortlisted.

Discussion

In this review, we aimed to identify which primary school- and preschool-based small-group Tier 2 interventions are shown to improve oral language or reading outcomes for children, supported by the strongest levels of evidence. Information regarding the intervention focus, frequency, duration, implementer, and group size is needed to inform implementation decisions. There are important implications for who (and how many) students are ultimately referred to more intensive interventions and special education services (Wanzek et al., 2016).

Summary of evidence

Of the 55 studies included in the review, 44 (79%) reported some statistically significant positive mean differences on relevant outcomes. However, only six interventions were identified as having the strongest levels of evidence using one of the three strength of evidence assessment tools (CEBC Scientific Evidence Rating, NHMRC Recommendation Grade, and the Customised Evidence Rating) and shortlisted in this review: (a) Early Reading Intervention; (b) Lonigan and Phillips (2016) Unnamed needs-aligned intervention; (c) PHAB+WIST (PHAST)/PHAB+RAVE-O; (d) Read Well-Aligned intervention; (e) Ryder et al. (2008) Unnamed Phonological Awareness and Phonics intervention; and (f) Story Friends. (See Tables 2–5 for study overviews, strength of evidence assessments, and core componentry.)

When considering the componentry of the six shortlisted interventions, there is some consistency regarding their content focus, structure, and implementation. This information about the componentry associated with the strongest Tier 2 language and reading interventions for children in the early years of school is useful given there has been no systematic review addressing this to date. The interventions shortlisted in this review typically included content relating to phonological awareness, phonics, and vocabulary instruction (in the context of language or literacy), and there was consistency regarding the type of pedagogical strategies used (e.g. most interventions were considered “interactive”, involving modelling, explicit instruction and scaffolding). Further, shortlisted interventions are typically characterised by the following procedural components: group size of approximately 3–5 students, meeting at least four times per week for sessions lasting approximately 30 min, with an overall duration of around 3–6 months. Intervention implementation was conducted/supervised by personnel with relevant qualifications and experience, but the duration of training provided to implementers varied greatly, and the training details were not well reported overall. Information regarding support for implementers as the intervention is delivered was also not always well reported. It is important to know detailed information about training and the intensity of implementation support required to deliver interventions that improve oral language and reading because of the substantial cost implications (Hollands et al., 2016).

There are some important distinctions between the scope and methodology employed in this review compared to the three notable previous reviews (Gersten et al., 2020, 2017; Wanzek et al., 2016). The authors of the Wanzek et al., review searched just two databases (ERIC and PsychINFO) between 1995 and 2013 and included 72 studies from 69 papers. In contrast, in this review, we (a) focused on more recent research (i.e. 2008 to 2018); (b) covered more databases; (c) used more search terms; (d) encompassed a slightly different age range (i.e. included preschool studies, but not Grade 3s); (e) did not restrict selection of interventions to those with 15–99 sessions; (f) defined Tier 2 interventions in terms of group size, specifically small-group interventions, with explicit exclusion of one-to-one interventions; and critically (g) took into account study quality/risk of bias. Although information about level of confidence in causality was collected by Wanzek et al., this information does not appear to have been used for the purpose of weighting studies in the meta-analysis or in the interpretation of results.

The Gersten et al. (2017) review and meta-analysis (Gersten et al., 2020) included only studies meeting What Works Clearinghouse evidence standards (and included one-to-one interventions). Of the studies originally reviewed by Gersten et al. (2017) and Gersten et al. (2020) almost all were reported to have positive or potentially positive effects. In the Gersten et al. (2017) review, most of these interventions were implemented for 30–45 min per day, 4–5 times a week, for 12–35 weeks. The Gersten et al. (2020) meta-analysis provided less detail about intervention dosage, with hours per week described as ranging from less than 1 h to 4 h and intervention duration not reported. Although the set of interventions identified in the Gersten et al. reviews are different to those shortlisted by this review, the finding concerning intervention dosage is broadly consistent. Interestingly, the Gersten et al. reviews did not identify any interventions with positive effects on vocabulary, whereas this review identified several – including some with strong evidence.

Strengths and limitations

This review followed core steps and minimum requirements for restricted systematic reviews (Pluddermann et al., 2018). Moreover, additional steps recommended for reducing risk of bias (Pluddermann et al., 2018) were taken, including verification of study selection, data extraction, and risk of bias assessments by a second reviewer (for a random sample of papers).

Three evidence rating systems were incorporated into this review, including a customised strength of evidence assessment in order to provide a more nuanced picture of the literature. Although the CEBC and NHMRC scales each identified several interventions with high evidence ratings, the bulk of studies align to a single category (38 of 47 interventions are rated Promising for the CEBC, and 35 are rated level C – the body of evidence provides some support for the NHMRC assessment). In contrast, within the intermediate levels of evidence, the customised rating system identified 9 interventions as promising and 17 with preliminary evidence. This is a useful distinction to make when there is a need to identify interventions that have demonstrated positive effects with specific populations or on specific outcomes, and the top-rated interventions do not fit these specifications. The customised assessment is also more sensitive to interventions with weak evidence (it identifies 13 interventions as such, compared with 6–8 interventions for the CEBC and NHMRC scales, respectively).

Potential limitations should also be acknowledged. First, the search range for the review was restricted to journal articles published from 2008 onwards. It is important to note that (a) there was sufficient relevant literature published in this period to address the review aims and questions, and (b) the recency of included literature increases the applicability of findings to contemporary education systems. Second, the review focused on experimental and quasi-experimental studies only. Although it is possible that effective interventions other than those identified in this review exist, their effectiveness cannot reliably be determined without rigorous evaluation design (demonstrating that effects are indeed attributable to the intervention and not to other factors). Along a similar line, the review included several studies that explicitly investigated the combined effects of Tier 1 and Tier 2 instruction (e.g. Chambers et al., 2011; Scanlon et al., 2008; Smith et al., 2016; Wonder-mcdowell et al., 2011). In these studies, the intervention involved manipulation of both Tier 1 and Tier 2 instruction, rather than Tier 2 only. In such cases conclusions about the small-group intervention cannot be made independently from the core instruction. However, this does not necessarily negate the utility of the approach, nor school interest in the adoption of multi-tiered interventions. Finally, the analysis of componentry focused on the best supported interventions rather than a comparison of the more and less effective interventions. This approach is consistent with best-evidence synthesis approaches used by other researchers in this field (e.g. Gersten et al., 2017; Slavin et al., 2011). As Slavin and colleagues previously noted, this type of approach is appropriate when the goal is to provide educators and policymakers with a fair comparison of interventions.

Future directions

This review lays a good foundation for investigating the overall magnitude of small-group Tier 2 oral language and reading interventions in the early primary school years. Meta-analytic procedures could be used in future research to explore the magnitude of effects for different populations (what works for who), different outcomes, and critically, different componentry (e.g. implementer, dose and duration, content). The extent to which study quality is related to effect sizes should also be considered in future meta-analytic investigations of Tier 2 interventions.

Another avenue for future research will be to explore the extent to which Tier 2 interventions close oral language and literacy achievement gaps, and the proportion of students for whom they do so. To allow comparison between interventions, this review focused on examining differences in mean scores (i.e. students in the intervention groups had on average better results). This does not go to the key questions of interventions for language and reading difficulties. Future research may consider which interventions lead to improved student scores and sustained impacts for those children no longer requiring Tier 2 intervention. Analyses of this sort are not well-reported in the literature. Indeed, of the studies included in this review, approximately only one-third reported either gap-relevant data (e.g. comparison to a typically developing peer group or mean scores in percentiles) or ideographic data (e.g. proportion of students meeting benchmark at intervention completion). Further research is also needed to determine the long-term maintenance of effects for most interventions (i.e. delayed follow-up testing beyond 12 months post-intervention). Finally, there is a need for continued updating of the review as new studies are published and evidence increases for some programs (e.g. the Enhanced Core Reading Intervention; Fien et al., 2020).

Conclusion and implications

Overall, the standard of research investigating Tier 2 oral language and reading interventions is limited and of variable quality. Only 14 of 55 studies met high-quality standards. Clearly, there is a need for more rigorous evaluation of small-group Tier 2 oral language and early reading classroom interventions for preschool to Grade 2 students. Nevertheless, our analysis of the extant literature did identify several interventions for which the evidence base can be considered strong. This information should be valuable to researchers and intervention developers interested in identifying the core components of effective intervention, to educators seeking to implement effective evidence-based interventions in their schools, and policymakers responsible for guiding, and in some cases mandating, particular resource-allocation decisions.

Acknowledgement

This work was supported by the North East Victoria Region, Schools and Regional Services, Department of Education and Training, Victoria; Melbourne Archdiocese Catholic Schools; Catholic Education Sandhurst Ltd; Diocese of Sale Catholic Education Ltd; and Diocese of Ballarat Catholic Education Limited. Research at the MCRI is supported by the Victorian Government's Operational Infrastructure Support Program. SG is supported by a NHMRC Practitioner Fellowship (GNT1155290). We acknowledge the contribution of Catherine Lloyd-Johnsen to double-coding and data extraction.

Supplemental Material

Supplemental data for this article can be accessed here

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by the Catholic Education Sandhurst;Victorian Government Department of Education and Training;Diocese of Ballarat Catholic Education Limited;Melbourne Archdiocese Catholic Schools;

Notes

1. There is an important distinction between disorder and difficulty. Disorder is a categorical definition based on a diagnosis whereby an individual experiences lifelong, pervasive difficulties and does not respond quickly or significantly to intensive intervention. In contrast, an individual with a language or reading difficulty is more likely respond readily to intensive intervention (Australian Disability Clearinghouse on Education and Training, n.d.).

2. We included both preschool and school settings as the age at which children start school varies both within and across different national contexts (Krieg & Whitehead, 2015). Research shows children who start school at a younger age relative to their peers tend to be more developmentally vulnerable (Hanly et al., 2019), and we therefore expect that such children may benefit from interventions tested with preschool children.

3. This criterion is not strictly consistent with the registered study protocol which stated that the intervention must be delivered by regular school staff, and interventions delivered by allied health professionals will be considered on a case-by-case basis. In practice, this review also included researcher-delivered interventions as these were also considered to be feasible for implementation by school staff.

4. We retained studies of the same samples if comparisons were different (e.g. intervention versus usual practice in one publication, and intervention versus alternative program or variation on same program in another publication).

5. A 15-item checklist was originally devised but four items were dropped as they were not applicable to all included study designs or required complex judgments (e.g. risk of contamination taking into account multiple factors such as design, implementer, and type of intervention).

6. It was originally anticipated that established frameworks might be somewhat insensitive to the relative strength of evidence for the bulk of identified interventions (because these frameworks require either long term follow-up evaluation or low risk of bias together with cross-study consistency in findings for high ratings). Including three systems in the evidence assessment allowed for an assessment of sensitivity, consistency across rating systems, and comparison of results when different criteria are emphasised.

7. Within the CEBC ranking system, an intervention may be rated Supported if: “At least one rigorous RCT in a usual care or practice setting has found the practice to be superior to an appropriate comparison practice, and in that RCT, the practice has shown to have a sustained effect of at least six months beyond the end of treatment, when compared to a control group” (see www.cebc4cw.org/ratings/scientific-rating-scale). In the NHMRC ranking system Level B evidence requires one or two RCTs with low risk of bias or multiple pseudo-randomised controlled studies with low risk of bias demonstrating substantial clinical impact and consistent results across most studies.

8. The term “studies” is not synonymous with “samples/trials”. Some studies present different outcomes for the same trial (e.g. maintenance effects), or different comparisons for the same sample (e.g. intervention vs. usual practice in one study, and intra-intervention comparisons within another study).

9. “Interactive programming” refers to intervention involving students interacting with content/materials in some way, for example, manipulating letter tiles, flexible exploring of computer program, etc.