Introduction and Background

Dual language learners (DLLs), or children simultaneously acquiring English alongside another home language, are one of the fastest growing populations of students in the United States (NCES, 2022). More than 11.2 million U.S. children under the age of 9 are DLLs, including approximately 87% of all 3- to 4-year-olds enrolled in Head Start preschool classrooms (Early Childhood Learning and Knowledge Center [ECLKC], 2017a; Migration Policy Institute, 2023). There are many benefits to learning two or more languages in childhood (National Academies of Sciences, Engineering and Medicine [NASEM], 2017). Preschool bilingual children generally perform better than monolingual children on measures of executive function (Bialystok, 2017; Prevoo et al., 2016). Children’s fluency in their home or heritage language allows for higher rates of language nutrition, or quality language input, from multiple caregivers while ensuring maintained relationships with family members who do not speak English (Ramírez-Esparza et al., 2017). A strong foundation in children’s home language also facilitates later academic achievement in both languages (Barnett et al., 2007; Steele et al., 2017). Despite the known benefits of bilingualism, national reading assessment results continue to show that overall, DLLs are more likely than their monolingual English peers to have lower 4th grade and 8th grade reading comprehension and writing scores (Reardon & Galindo, 2006; NCES, 2023). However, DLLs who receive bilingual education (e.g., Albuquerque, NM and Fort Worth, TX) made significant gains on National Assessment of Educational Progress (NAEP) Reading scores from 2019 to 2022 (NCES, 2023; Shoichet, 2023).

The purpose of bilingual education, or dual language instruction, is to provide students with exposure to both languages, so students do not lose proficiency in their home language (i.e., English does not become the dominant language), thus interrupting the development of both languages (Olivia-Olson, 2019; Partika et al., 2021). In a study of a dual language immersion program, Limlingan et al. (2020) found that teachers’ use of children’s home language during instruction is positively associated with English receptive language skills. Since one of the best ways to prepare young students for success in school is to build their oral language purposefully and intensely before they enter kindergarten (Strickland & Riley-Ayers, 2006), DLLs provided with bilingual education in early childhood should be able to strengthen their home language skills and improve their overall language abilities (Barnett et al., 2007; Collier & Thomas, 2017).

In 2016, the Office of Head Start published guidance (ECLKC, 2017b) intended to help preschool teachers implement one of two classroom language models that viewed language differences as social and linguistic capital: (1) dual language and (2) English-only instruction with home language support provided through continuous engagement with families and opportunities for exposure to the home language in school (EHLS). To examine the extent to which either or both models supported DLLs’ language outcomes, Olivia-Olson (2019) studied 841 Spanish–English DLLs aged 3- to 4-years old who attended Head Start programs in California and Florida. Results indicated that preschoolers enrolled in dual language model classrooms had higher pretest–posttest gains in English and Spanish oral language proficiency than did students in the EHLS classrooms. However, if preschool teachers operating in either classroom model demonstrated low support for the home language, there was little difference in students’ oral English language proficiency from pretest to posttest. The report recommended use of the dual language model as the best method for promoting the early development of both Spanish and English language skills, as both are critical for later reading and writing outcomes.

Bilingual Literacy Development

The development of early literacy can be conceptualized by the Simple View of Reading (Hoover & Gough, 1990), where reading comprehension or true reading is the product of both language comprehension and word identification skills. Successful reading comprehension relies on students’ proficiency in both skill sets. However, DLLs do not need to master language comprehension and word reading skills in each language exclusively. The Interdependence Continuum Theory (Chung et al., 2019; Cummins, 1979; Proctor et al., 2017) posits that bilingualism/biliteracy results from the interdependent development of specific literacy skills in one language that influences literacy development in another. In other words, DLLs who have strong reading skills in their home language are more likely to have higher English reading scores, and vice versa (Relyea & Amendum, 2019; Winsler et al., 1999). Yet, while several researchers have found that phonological and orthographic awareness are positively correlated across Spanish and English languages (Domínguez De Ramírez & Shapiro, 2007; Wagley et al., 2022), there is less evidence of a cross-linguistic relationship for language comprehension (Melby-Lervåg & Lervåg, 2011). On average, young students from linguistically diverse homes are slower to develop listening and reading comprehension skills in either or both languages (Han et al., 2014; Mancilla-Martinez & Lesaux, 2017; Nakamoto et al., 2008; Roijas-Cortez, 2010). Given that early systematic and explicit oral language instruction supports preschoolers’ acquisition of language comprehension (Collins, 2010; Lugo-Neris et al., 2010; Matera & Gerber, 2008), preschool DLLs may benefit from an enhanced dual oral language instruction that uses their home language as a bridge to develop both Spanish and English language comprehension skills.

Oral Narrative Language Instruction

Narratives are useful literacy tools because they are pervasive in academic and social contexts, relevant to children regardless of their culture and language, and facilitate learning the complex oral language needed in school (Barton-Hulsey et al., 2017; Curenton, 2006; McCabe, 2017; Westby, 1994). It is well established that narrative abilities at school entry (i.e., kindergarten) predict later reading comprehension and overall academic performance (Bishop & Edmondson, 1987; Fazio et al., 1996; Griffin et al., 2004; Mehta et al., 2005). It stands to reason that DLLs who enter kindergarten with strong oral narrative skills in their home language—Spanish—are likely to have higher scores on Spanish and English listening and reading comprehension measures in the later grades (Miller et al., 2006).

Narratives contain story grammar elements that together create a structural frame for exchangeable content (Stein & Glenn, 1979). This frame is referred to as discourse or text structure and is critical for literacy development (Duke et al., 2011; Scarborough, 2001). As children tell or retell stories (spoken or written), they are more likely to use complex sentences and precise vocabulary to ensure their listener understands what they did not experience (Gillam & Johnston, 1992; Westby, 1985). The integrated nature of multiple levels of academic language (e.g., word, sentence, and discourse) within narratives makes them excellent teaching targets and versatile teaching tools. For example, several researchers have delivered story grammar instruction and then used storytelling and retelling as mechanisms to promote the understanding and use of syntactically complex sentences and/or less common vocabulary words (Adlof et al., 2014; Armon-Lotem et al., 2021; Gillam et al., 2014; Petersen et al., 2020). It is hypothesized that to impact reading comprehension, it is necessary to take an integrated approach to oral language instruction—that is promote more than one aspect of academic language at the same time (e.g., more than just text structure or vocabulary alone; Cervetti et al., 2020; Wright & Cervetti, 2017).

Family Engagement in Early Literacy Development

Caregivers’ involvement in a child’s oral language and early literacy development predicts later language and literacy skills (Adamson et al., 2021; Ramírez-Esparza et al., 2017; Van Tonder et al., 2019). Caregiver-implemented, home language interventions such as family engagement activities (FEAs) may compensate for some of the barriers to comprehensive dual language instruction in school because caregivers speak children’s home languages and teachers may not. FEAs are culturally and linguistically responsive materials that facilitate academic language learning experiences and reinforce relationships between families and children learning two languages (Durán et al., 2016; Early Head Start National Resource Center, n.d.). In relation to their promise for improving DLLs’ oral language skills, Melzi et al. (2022) found a positive and significant correlation between Latine children’s oral narrative skills and family engagement practices. Narrative interventions that include FEAs provided in the caregiver’s preferred language offer the opportunity to extend school-based early literacy instruction into a child’s home and community, “providing the necessary stimulation for children to produce longer narratives” (Melzi et al., 2022, p. 91). However, few studies have specifically examined the effects of preschool dual language and literacy interventions with FEAs, and none to date have investigated the unique contribution the FEA intervention ingredient on DLL preschoolers’ oral language improvement.

Dual Language Oral Narrative Instruction with FEAs

In a recent systematic review, Pico et al. (2021) summarized 28 studies that addressed narrative language skills, including story grammar, complex sentences, and vocabulary. They concluded that narrative interventions positively impacted a large range of language outcomes for children with diverse learner characteristics. Nine of the studies included Spanish-speaking DLL children as participants (described in the literature as Spanish–English bilinguals, Spanish-speaking ELLs, Spanish-speaking children, or culturally and linguistically diverse Latino children). However, only four studies implemented the narrative intervention in two languages, and two included preschool-aged children as participants.

In a 2019 study, Spencer et al. investigated the effect of the Spanish–English dual language classroom instruction on Spanish and English narrative and receptive vocabulary outcomes using a multiple baseline research design. Teachers rarely used Spanish for classroom instruction because most of the students spoke English at home, but in their study, both teachers and research assistants worked together to provide the DLLs with small group narrative language intervention (with embedded vocabulary instruction) for 15–20 min a day. The language of intervention alternated every day (Spanish then English) and children received four intervention sessions per week. Results suggested children made important gains in English narrative skills while maintaining their age-appropriate Spanish narrative skills, but receptive vocabulary improvements were only observed in English. The authors concluded that the Spanish components were insufficiently dosed to improve children’s acquisition of Spanish vocabulary. Rather than increasing the Spanish classroom instruction, which teachers did not think was feasible, the authors proposed creating Spanish FEAs to increase the amount time children were engaged in activities that involved the practice of target Spanish vocabulary words.

In a cluster randomized control trial that followed the 2019 study, Spencer and colleagues (2020) examined the efficacy of multi-tiered (i.e., large and small group) Spanish–English narrative and vocabulary instruction paired with caregiver implemented Spanish FEAs to improve the language comprehension skills of 81 Head Start preschoolers. Specifically, the Spanish–English narrative and vocabulary instruction led to statistically significant improvements in proximal and distal measures of English and Spanish narrative retells, receptive vocabulary, and a general oral language measure (Clinical Evaluation of Language Fundamentals—Preschool [CELF-P]; Wiig et al., 2004). Although the results were favorable, the authors did not examine the degree to which the FEA component uniquely contributed to the preschoolers’ improvements because it was part of the whole package. It is unknown whether the Spanish FEAs were responsible for the observed changes in Spanish receptive vocabulary or not.

If DLLs are less likely to receive early intensive language instruction in both English and their home language, family-implemented components in children’s first language might be a viable option. To support the language abilities of young DLLs, it is essential to identify the most salient features of dual language interventions that promote their effectiveness (e.g., FEAs), as well as their associated costs (e.g., materials, time). Even if intervention components contribute to the overall effectiveness of the entire instructional package, the costs of implementing them may be prohibitive. For this reason, we examined the following questions in the current study.

  1. 1.

    Do Spanish FEAs increase the effect of dual language instruction on the English and Spanish narrative and vocabulary skills of preschoolers?

  2. 2.

    Do caregivers perceive the FEAs to be feasible?

  3. 3.

    To what extent are the FEAs cost-effective?

Method

Setting and Participants

The study took place in Head Start classrooms in three different elementary schools within a large public school district in southwestern Florida. Two classrooms were housed in the same school in an urban area (N = 385, 430), with 26.2% of the residents identifying as Black or African American, 23.1% Hispanic or Latino of any race, and 3.4% Asian. The remaining two classrooms were each located in separate elementary schools within the same county/district, but in a rural area (population: 6839; 74.3% identifying as Hispanic or Latino of any race).

With the help of a Head Start regional director who recommended classrooms serving large numbers of Spanish–English DLLs, four Head Start preschool teachers volunteered to participate in the study. Two of the teachers fluently spoke Spanish and English (Cuban and Puerto Rican) and two teachers spoke only English. The teachers delivered the dual language lessons in their classrooms with support from bilingual research assistants (RAs). All teachers used The Creative Curriculum (Dodge et al., 2002) as their core instructional approach.

To study the effectiveness of the Spanish FEAs on children’s language outcomes, we purposefully recruited only Spanish–English DLLs whose caregivers primarily spoke Spanish at home. The teachers identified these families and provided them with a letter detailing the study, a consent form, and a demographic questionnaire, all written in Spanish. Bilingual RAs were available at pick-up and drop-off periods at each school to answer caregivers’ questions or clarify study procedures. Caregivers who were interested in participating completed and returned the study-related forms to their child’s Head Start teachers. Across the four classrooms, a total of 33 caregiver-child dyads agreed to participate in the study. Table 1 provides details about the caregivers who participated in the study.

Table 1 Demographic information about children and caregivers
Full size table

Child Participant Screening

In line with the goal of Head Start as a school readiness program, the purpose of the dual language narrative and vocabulary intervention is primarily to promote young DLLs’ home language development in spirit of advancing their English language skills needed for English kindergarten instruction. The Spanish FEAs was one aspect of the entire intervention package that was specifically designed to enhance children’s Spanish vocabulary, the promotion of which was limited in the classroom for feasibility reasons. Since facilitation of English language skills is the ultimate goal of the intervention, we had to exclude participants with average or above average English language skills. Therefore, we used two English measures to determine inclusion/exclusion: the English version of Clinical Evaluation of Language Fundamentals (CELF-P; Wiig et al., 2004) and the English version of the preschool Narrative Language Measures (NLM) Listening, which is a subtest of the CUBED (Petersen & Spencer, 2016a, 2016b; Spencer et al., 2023). Preschoolers who had an English NLM Listening Retell score of 8 or higher were excluded from the study. A retell score of eight presupposes the use of key story grammar features and places a preschool student above the 20th percentile based on normative data from 281 preschool students from across the U.S. (Petersen & Spencer, 2016a, 2016b). We then administered the CELF-P to the remaining students (those with a NLM Listening Retell score between 0 and 7). We excluded students with CELF-P English Vocabulary (English EV) scores within the normal range from participating in the study. In other words, scores within age expectations for English on either screening measure disqualified students from the study. To characterize students’ Spanish language comprehension skills at pretest, we also administered the Spanish version of the preschool NLM Listening. However, these scores were not used to qualify students for study participation. The screening process resulted in 22 participants (22 caregiver-child dyads) enrolled in the study. Table 1 provides details about child demographics.

Research Team and Training

The research team was composed of five RAs who collected data and provided support to teachers and caregivers. Three RAs spoke fluent Spanish, so they administered the Spanish language outcome measures, as well as delivered Spanish small group sessions in the two classrooms where the Head Start teachers spoke only English. The Spanish-speaking RAs were also responsible for communicating with caregivers and training them to use the FEAs. One of the authors trained the RAs in assessment administration and intervention delivery procedures prior to the start of the study. All RAs demonstrated 100% fidelity on the procedures during role play opportunities.

Research Design

We used a pre/posttest randomized wait-list control group design to examine the degree to which the Spanish FEAs had an additive effect on the preschoolers’ language skills when used in conjunction with the dual language classroom instruction. We randomly assigned caregiver-child dyads to either the treatment or waitlist control (“control”) group. All child participants received the same dose of classroom dual language instruction in Spanish and English, but only those in the treatment group received the supplemental caregiver-delivered Spanish FEAs at home. We collected the child-level assessments before and after the 6–8-week intervention phase. The number of weeks required to complete that phase was determined by the school teacher, and caregiver schedules. The caregivers completed the feasibility questionnaire after the intervention phase. Upon study completion, we provided all caregivers in the control group with the FEAs and training in their use.

Dependent Measures

To measure the effects of the narrative components of the intervention delivered in both Spanish and English, we used the Spanish and English preschool NLM Listening Retell scores obtained at time of screening as pretest scores and administered the measures again at the end of the intervention phase. We also administered a receptive picture vocabulary assessment in English and Spanish to each participant at each time point. This assessment measured knowledge of the specific words taught during instruction in school and at home. Given the brevity of the intervention phase and the narrow focus on the contribution of the Spanish FEAs, it was unlikely that vocabulary improvements would be potent enough to be detected by norm-referenced standardized tests of vocabulary (e.g., Spencer et al., 2020). Hence, we only used the receptive vocabulary assessments that accompany the dual language intervention. To prevent student testing fatigue and language confusion, we administered each individual assessment across four brief sessions on different days at pretest and again at posttest.

Narrative Language Measures (NLM) Listening

We used the NLM Listening (Petersen & Spencer, 2016a, 2016b) to elicit and score the students’ Spanish and English narrative retell language samples. The Spanish NLM Listening stories were unique and not translations of the English stories (Spencer et al., 2023). While the NLM Listening includes 22–25 parallel forms (or stories) available at each grade level (PreK-3), we only used a total of 6 of the 25 Spanish preschool stories and 6 of the 25 English preschool stories in the current study (3 of each at pretest and 3 at posttest). Both the English and Spanish NLM Listening Retell sessions lasted 3–4 min each. To collect the narrative retell language samples, an RA presented a panel of 5 illustrations and read aloud an accompanying story. After reading, the RA elicited the retell by saying, “Thanks for listening, now you tell me the story.” The pictures remained in front of the participant as they retold the story. Examiners used only neutral statements to encourage the children (e.g., “It’s okay. Just do your best.” or “I can’t help, but you can just tell the parts you remember.”). The child’s retell was scored in real time by the RA, who awarded 0–2 points for each story grammar element (i.e., Character, Setting, Problem, Feeling, Action, Consequence, and Ending), based on its completeness and clarity. For example, if the child said, “eating at a restaurant,” they received 2 points for the setting. However, if the child said, “eating” or “restaurant,” the student received 1 point. The RA awarded additional points if the child included clear and complete episodic features (i.e., Problem, Action, Consequence, and Ending) in their story that were relevant to the plot. This ensures the most critical plot elements are weighted properly in the total scores. Finally, to quantify the child’s use of complex language (e.g., causal and temporal subordination) during the retell, examiners awarded 1 point for each use of the words because, when, or after (3 points maximum for each word). Because it is an earlier developing temporal marker, only 1 point was awarded for any use of the word then, regardless of total count. We summed the points earned across the story grammar, language complexity, and episode sub-scores to calculate a total NLM Listening Retell score. The NLM Listening in English and Spanish has adequate alternate-form reliability and evidence of validity (Spencer et al., 2023). Replicating the procedures of two prior studies of the dual language program (Spencer et al., 2019, 2020), we used the highest total NLM Listening Retell score for analysis.

Reliability

Each NLM Listening session was audio recorded to conduct reliability analyses after the study. We randomly selected 30% of the recorded Spanish and English retells at pretest and posttest for reliability scoring, which was completed by a trained independent RA. Scoring agreement between the initial examiner and the independent RA was calculated by dividing the total number of scores in agreement by the total number of items scored on each assessment; this number was multiplied by 100 to yield a percent agreement per retell. The mean percent agreement across all English retells was 88.1% at pretest (range: 85–100%) and 80% at posttest (71–100%). The pretest and posttest Spanish NLM Listening Retell scoring agreement was 83.04% (71–100%) and 80% (85–100%), respectively.

Receptive Picture Vocabulary Assessment

We measured preschoolers’ immediate improvements in taught English and Spanish vocabulary using receptive picture vocabulary assessments that have shown sensitivity to change in the two previous studies (Spencer et al., 2019, 2020). In the current study, we assessed only the words taught in one unit (N = 36): 12 nouns, 12 adjectives, and 12 verbs. The Spanish and English versions were parallel assessments, but the order of the words and the location of the targets within a 2 × 2 array of black-and-white line drawings were intentionally different. Each array contained three distractors and one illustration corresponding to the target noun, adjective, or verb.

RAs delivered the Spanish and English assessments individually across separate sessions at a small table within the classroom. Sessions lasted approximately 5 min each. The RAs presented each student with a field of four black-and-white line drawings and said, “I am going to show you some pictures, and I want you to point to the picture that is the same as the word I say.” If the child did not respond within 5 s, they prompted the student to, “Point to (word).” Correctly identified words received 1 point, with all other responses scored as 0. We used the total points earned (out of 36) per assessment at pretest and at posttest for the analysis. While this assessment has not undergone psychometric evaluation, similar receptive picture vocabulary tests have yielded high internal consistency correlations (Brownell, 2000; Dunn & Dunn, 2007). Reliability was not evaluated because responses were non-verbal (pointing), and scoring was straightforward.

Feasibility Questionnaire

To provide a measure of social validity, we developed a 10-item caregiver survey focused on their experience using the FEAs. We asked caregivers to provide a rating of activity completion demands, how well the activities fit into their family’s daily routines, and whether they and their children enjoyed the activities. They rated agreement with each item using a six-point Likert scale, where 0 represented disagreement and 5 represented agreement with the statement presented. Additional questions solicited information about the child’s responsiveness and engagement during the FEAs.

Independent Variable

The primary independent variable of interest was the FEAs that supplemented the Puente de Cuentos (Spencer et al., 2017) dual language classroom curriculum. We first describe the Puente de Cuentos program, which both groups received, to establish an understanding of the instructional context. Then we describe the FEAs and procedures unique to the treatment group. Table 2 provides a general outline of the instructional procedures used in the entire package, which included English large group, Spanish and English small group, and Spanish FEA lessons.

Table 2 Puente de Cuentos lesson outline
Full size table

Puente de Cuentos

Puente de Cuentos is a supplemental multi-tiered preschool academic language curriculum designed intentionally to capitalize on young Spanish–English DLLs’ home language while hastening the development of English. The scope of Puente de Cuentos focuses on oral language skills like story retelling, vocabulary, and complex sentences, taught across a series of three units (A–C). In this study, we only used Unit B, but every Puente de Cuentos unit includes 12 English large group lessons, 12 English small group lessons, and 12 Spanish small group lessons. Depending on school and teacher schedules, each unit takes about 6–8 weeks to implement. Both large and small group lessons feature a brief, culturally relevant, and developmentally appropriate Spanish or English story. The lessons also involve the systematic presentation and fading out of colorful icons that represent the key parts of every story (i.e., Character, Problem, Feeling, Action, and Ending), helping students identify patterns in stories. In addition to having the same discourse structure, each narrative also includes the same number of causal and temporal subordinate clauses, as well as three target words: one noun, one adjective, and one verb. The lessons provide multiple opportunities for students to hear the target words in English and Spanish. Each story also has an accompanying 5-panel set of illustrations, always depicting the target noun. Context clues embedded in the narratives support the targeted verbs and adjectives, which were selected based on the availability of a child-friendly definition in both English and Spanish, their relative infrequency in child speech, and utility in academic contexts (Beck et al., 2002). A picture book contains photos for extra practice of the target words in contexts that differ from the lesson story.

Large Group Instruction

Head Start teachers delivered 12 15–20-min Unit B English large group lessons. The planned schedule was twice weekly for 6 weeks, but due to varying schedules and teacher absences, some classrooms required 8 weeks to complete them all. Each lesson contained five explicit instructional activities and five extension activities. The activities required the teachers to tell a story, identify main parts of the story, emphasize vocabulary words, and prompt children to retell the story using a variety of teaching procedures. The teachers selected classroom routines (e.g., circle time, outdoor, book reading) and used those contexts to embed extension activities into natural environment teaching opportunities.

Small Group Instruction

We also provided each classroom teacher with a calendar specifying the lessons they would deliver weekly. On days when large group instruction did not occur, teachers/assistants implemented either a Spanish or English small group lesson (3 students per group). To promote students’ cross-linguistic transfer from Spanish to English, we purposefully arranged for the Spanish small group lesson to come before the corresponding English small group lesson. For example, the Spanish small group lessons may have occurred every Tuesday and small group English lessons every Thursday. Lessons took 15–20 min each and involved the teacher telling a story, highlighting the target vocabulary words, teaching the students to identify story components, playing games to reinforce the students’ storytelling, and having the students practice using the target words in sentences based on the photos in the picture book. Across the lesson steps, the teachers systematically faded use of the story illustrations and icons to encourage increasingly independent story retell. As needed, instructors used a two-part prompting procedure to help children retell the story and use the target words (i.e., ask a question, model a response). In the two classrooms with only English-speaking staff, we assigned a Spanish-speaking RA to deliver the Spanish small group lessons.

Teacher/Assistant Training and Implementation Fidelity

We asked the teachers and assistants to watch video modules that detailed how to conduct the lessons and examples of instructors implementing the procedures. We then delivered an in-person training to review the procedures and materials and answer their questions. We trained the teachers/assistants to conduct large group lessons to fidelity first. RAs assigned to the classrooms conducted the small group lessons until the teacher felt comfortable implementing the lessons independently. At such time, the RAs trained the teachers to implement the small group lessons with fidelity. Because the teachers/RAs followed the same explicit teaching procedures across all lessons, and lessons were largely scripted and easy to use, it was unnecessary to document adherence to specific lessons steps. Rather, we defined fidelity to the classroom protocol as the number (e.g., dosage) of large group and small group lessons delivered each week according to the schedules we provided. All large and small group lessons were delivered as planned, however, one class needed 7 weeks and one class needed 8 weeks to complete the planned lessons. Every participant in the treatment and control groups had the same number of opportunities to receive large group (n = 12) and small group Spanish (n = 12) and English (n = 12) lessons.

Family Engagement Activities

There were 24 Spanish FEAs, one for each Puente de Cuentos Unit B story (12 for large group, 12 for small group). Only the caregivers in the treatment group received an 8.5 × 11″ booklet with FEAs printed in color. Each FEA contained a story and corresponding 5-panel illustration, icons, and seven questions to guide caregivers’ interactions with their children. The FEA included directions for caregivers to help their child label the parts of the story using the icons, retell the story, define and practice using the target words, and tell a personal story related to the story theme (i.e., getting clothes wet, needing help, playing a game). Depending on the level of child engagement, FEAs took less than 5 min to complete. Caregivers used a provided log to track the number of FEAs completed, dates of completion, family member(s) involved, and language(s) used.

Caregiver Training

A bilingual RA visited each family’s home and trained the caregiver on FEA use. The RA showed each caregiver a 3-min video of a mother explaining, in Spanish, how to use the FEAs and then demonstrating how to use a FEA with her daughter. The RA and caregiver role-played the procedures, with the researcher playing the role of a child and providing feedback in-situ, while addressing caregiver questions as they arose. Once caregivers started the FEAs, the RA followed up with them weekly to answer caregiver questions and provide additional supports, alternating between phone calls and home visits which lasted approximately 15 min. All communication occurred in Spanish. The RA kept a detailed log of all contacts with caregivers that included the date and duration of contacts (home and phone), travel time to and from the family’s home, vehicle mileage, attempts to contact caregivers, and cancelations.

FEA Implementation Fidelity

We assessed FEA implementation fidelity according to adherence and exposure (Dane & Schneider, 1998). The FEA fidelity data are based on eight caregiver-child dyads, as three caregiver-child dyads were unable to supply the research team with the planned data. Dyad 4’s caregiver was hospitalized due to a car accident, the child in Dyad 5 engaged in severe problem behavior with his caregiver, which is considered dissent in our IRB protocol, and Dyad 6’s caregiver was unreachable (i.e., disconnected phone) when we attempted to pick up the FEA completion log.

Adherence

On 10% of all home visits, Spanish-speaking RAs observed caregivers as they implemented a FEA with their child. The RAs used a fidelity checklist to document the extent to which caregivers adhered to the FEA steps as trained, independent of their child’s response. They assigned 1 point for each of the following items completed by the caregiver correctly: (1) pointing to icons and asking the child what the icons mean, (2) asking the child to retell the story using the vocabulary words, and (3) asking the child to define the new vocabulary words. Each caregiver demonstrated 100% fidelity to those FEA procedures.

Exposure

We defined exposure, or dosage, as the number of FEAs completed by the participants (Dane & Schneider, 1998). Caregivers completed logs to document the total number of FEAs implemented, number of stories the children retold, total time spent on FEAs, and average FEA duration. Caregivers in the treatment group were asked to complete a total of 24 FEAs (1 per classroom lesson) with their child. Although each FEA included one opportunity for the child to practice story retelling, caregivers had the option to help their child retell the story more than once. Three caregiver-child dyads went beyond the one retell per FEA, completing a cumulative total of 35 additional retells among them. Thus, to calculate a measure of exposure, we converted count outcomes into percents based on the total number of FEAs and story retell opportunities per FEA (n = 24). Only one caregiver (Dyad 11) completed 22 FEAs, while the remainder met or exceeded the expectation to complete all 24 FEAs, resulting in a 98.95% average completion rate. We also calculated the total time each caregiver reportedly spent completing the FEAs by dividing the cumulative duration of FEA engagement across all caregiver sessions by the total number of dyads (N = 8). Caregivers spent a combined average of 178.5 min completing all 24 FEAs, at an average of 12.32 min per FEA.

Data Analysis

Out of the 22 preschoolers randomly assigned to either treatment or control groups, two students did not complete the study; one participant in the control group withdrew from school and a treatment group participant (Dyad 4) withdrew from the study following a caregiver’s hospitalization. Because of the small sample size and attrition assumed to occur at random, we used a multiple imputation procedure to account for missing data (< 10%) based on group assignment (Rubin, 1987; Yuan, 2011). We assumed independence in observations between groups and did not use a nested model because randomization occurred at the child level and delivery of FEAs occurred in separate households (no participants shared a home).

We used linear regression to test if the supplemental FEAs significantly predicted improvements in the preschoolers’ language outcomes, both in English and Spanish. To specify the model, we conducted a two-sample t-test to rule (out) the assumption of equal variance in group pretest means. As shown in Table 3, we were unable to reject the null hypothesis of equal group means. Therefore, we conducted multiple linear regression and included pretest scores as a covariate in our estimation of the relative contributions of FEAs on English and Spanish language skills. We allowed intercepts to vary in the model, and specified the control group means as the intercept parameter estimate (referent) so that results would report the expected treatment group’s estimates above or below the control group performance. To account for the small sample size, we performed a power estimate at 0.8 and alpha of 0.05, resulting in only the detection of effect sizes (Hedges g) above 0.48.

Table 3 Descriptive statistics of mean student outcomes at pretest and posttest by group
Full size table

Cost and Cost Effectiveness Analyses

We used the ingredients method to document the cost of the FEAs as necessary to complete a cost-effectiveness analysis. We identified, described, and assigned market values to all variables considered necessary for replication (Levin & McEwan, 2001; Levin et al., 2018). We obtained our data through direct observations, record reviews, and interviews with caregivers and research staff during the study. For example, we calculated the time variable using RA schedules, time sheets, and phone records of contacts with participants and study team members. Based on 2018 U.S. average prices in the southeastern U.S., we report monetized ingredients in Table 5. The unique cost of the FEAs was approximately $203.33 per participant, including participant incentives (i.e., gift cards), which accounted for 63% of the total study budget. Excluding the gift cards, the estimated actual market value of FEAs was $128.33 per student.

For the cost-effectiveness analysis, we compared the differences in cost and treatment effects between groups. To estimate the cost-effectiveness of the Family Engagement Activities, the \(Cost Effectiveness Ratio \left(CER\right):\) \(\frac{{C}_{ T }-{ C }_{C} \, }{{E}_{ T }- {E}_{ C}}\) = \(\frac{\Delta C}{\Delta E}\) formula was used (Levin et al., 2018).

Results

We divide the results of this study into two sections. We first present the results of the regression analyses that examine the additive effects of the FEAs on preschoolers’ Spanish and English oral language skills. Then, we report on caregivers’ feedback regarding the feasibility of the FEAs and provide a summary of FEA cost-effectiveness.

FEA Effects on Child Language Outcomes

To address the first research question regarding the additive effect of FEAs on preschool DLLs’ Spanish and English narrative retell and vocabulary skills, we examined the extent to which students improved performance on the oral narrative and vocabulary language measures from pre- to posttest. Table 3 reports the descriptive statistics for both pretest and posttest outcomes by group across measures, to include documentation that the preschoolers had stronger Spanish language skills than English skills at pretest. Further, students in the control group had significantly higher Spanish (M = 13.18, SD = 4.47) and English (M = 12.73, SD = 3.07) receptive vocabulary scores than did students in the treatment group, who had a mean Spanish vocabulary score of 10.91 (SD = 3.99) and English vocabulary mean of 10.27 (SD = 3.74). In Table 4, we report the results of two multiple regression models predicting FEA treatment effects on the Spanish (A models) and English (B models) Narrative Retell and Receptive Vocabulary outcomes, with (Model 2) and without (Model 1) pretest scores as a covariate.

Table 4 Regression models of FEA predicting narrative language and vocabulary outcomes
Full size table

Narrative Retell

Consistent with previous studies (Spencer et al., 2019, 2020), the multi-tiered dual oral language Puente de Cuentos program was associated with preschool DLLs’ significant improvements in both English and Spanish oral narrative retell skills. However, we found mixed effects of the supplemental FEAs on narrative retell outcomes. On average, the treatment group saw a raw score effect size for Spanish NLM Listening Retell scores of 3.74 points, whereas the control group’s raw score mean improvement from pretest to posttest was 2.72 points. In Table 4 [see Narrative Retell model 2A], after controlling for pretest differences across groups, the unstandardized effect of the FEAs on Spanish narrative retells was estimated to be 1.36 points (SE = 1.38), but the FEAs were not a significant predictor of the outcome (β = 0.13, p = 0.34). Similarly, students’ pretest scores predicted their posttest English NLM Listening Retell outcome (β = 0.89, p < 0.0001), whereas the FEAs did not (β = − 0.97, p = 0.45). Therefore, we cannot conclude that the FEAs had a significant effect on either the Spanish or English NLM Listening Retell scores.

Receptive Picture Vocabulary

We were also interested in examining the extent to which the caregiver-delivered Spanish FEAs improved the preschoolers’ taught English and Spanish vocabulary skills, beyond any gains made from classroom dual language instruction alone. At posttest, the students in the treatment group had higher English receptive picture vocabulary scores than their peers in the control group (M = 16.31, SD = 1.93). However, neither pretest (β = 0.42, p = 0.08) nor FEA (β = 0.29, p = 0.21) variables significantly predicted students’ scores on the English vocabulary posttest. Instead, the results in Table 4 (see Model 2A) suggest that the supplemental FEAs only had a moderate and significant additive effect on the preschoolers’ Spanish receptive picture vocabulary. The treatment group’s average posttest Spanish vocabulary score (M = 17.39, SD = 3.96) was larger than the control group’s Spanish vocabulary posttest score (M = 16.38, SD = 4.23), and the standardized FEA effect was significant (β = 0.42, p = 0.01). The estimated FEA effect size was moderate, g = 0.49 (Cohen, 1988).

Feasibility

At the conclusion of the study, eight caregivers completed and returned a feasibility questionnaire. All eight caregivers reported 100% agreement (i.e., rating of 5) with the following statements: The family engagement activities were easy to use. We could fit the family engagement activities to our family’s schedule. The family engagement activities improved my child’s language. I enjoyed the storytelling stories with my child. All but one caregiver reported that they would like to continue using the activities at home and that their child enjoyed telling stories with them, with each outlier rated as 4 (out of 5). Most caregivers reported that they conducted the activities even when their child did not cooperate, with one adult sharing that their child never refused to participate in the FEAs. Furthermore, all caregivers engaged their children in the FEAs in Spanish but estimated between 25 and 50% of the children’s retells occurred in English. In other words, while all caregivers delivered the FEAs in Spanish, half of the children retold stories using English sometimes or the majority of the time.

Cost Effectiveness

Given the significant impact FEAs had on preschoolers’ Spanish vocabulary, the third research question involved examining the extent to which FEAs were cost effective. Table 5 presents the estimated ingredients and associated costs. We derived a cost-effectiveness ratio using the FEAs’ effect size for Spanish vocabulary gains (g = 0.49) and an average FEA cost estimate of $128.33 USD per student.Footnote 1 Thus, the FEA cost-effectiveness is an estimated $261.90 USD per student for one standard deviation increase in Spanish vocabulary outcomes. Due to a lack of reported costs associated with FEAs in the literature, we cannot make a direct comparison of Puente de Cuentos with FEAs to other dual language interventions with supplemental family activities. However, our cost estimates are significantly lower than at least five other reading programs with estimated cost-effectiveness outcomes ranging from $1400 to $49,018 USD per unit increase in effect size of language outcomes (Hollands et al., 2016; Simon, 2011).

Table 5 Intervention costs
Full size table

Discussion

Bilingual education in early childhood helps to improve young children’s overall language abilities and prepares them for further instruction in U.S. instructional environments (Barnett et al., 2007; Collier & Thomas, 2017). In addition to strengthening students’ overall academic language development, effective early language instruction communicates to the society, children, and families that bilingualism is valued (Mancilla-Martinez et al., 2020, 2017). Only a few randomized control trials have examined the effects of supplemental home-language interventions with DLLs. Moreover, it is rare to find early childhood language intervention research using both Spanish and English measures to examine outcomes for preschool DLLs (e.g., Soto et al., 2020). Therefore, this study addressed these limitations in the literature by examining the extent to which Spanish FEAs that supplement dual language classroom instruction boosts preschoolers’ English and Spanish language skills.

In the current climate of educational research, it is no longer sufficient to establish that interventions work. It is also necessary to consider how interventions will be implemented in practice and whether or not end users (i.e., teachers and caregivers) can implement them properly to achieve the intended outcomes (Fixen et al., 2005). As newly and iteratively developed interventions approach dissemination and implementation stages, different types of research questions need to be asked (Moir, 2018). For example, the ease of use, contextual fit, and cost effectiveness are critical considerations that influence the extent to which an intervention is adopted, implemented with fidelity, and sustained over time (Klimes-Dougan et al., 2009; Lyon et al., 2018). Therefore, in this study, we added secondary research aims to investigate the perceived feasibility of the caregiver-implemented activities and their cost effectiveness.

Effects of Spanish FEA on English and Spanish Language Outcomes

We hypothesized that caregiver-implemented, home language FEAs would support comprehensive dual language instruction in situations where educators do not speak nor have other staff to deliver classroom instruction in children’s home languages. Although small in scale, the results of this randomized waitlist control design study confirm a dual language program with supplemental FEAs can lead to improvements in preschoolers’ English and Spanish story retelling and vocabulary knowledge (see also Spencer et al., 2019, 2020). By embedding the Spanish vocabulary words taught at school in the FEA stories used by the caregivers at home, students were provided with a developmentally appropriate and meaningful context for practicing the Spanish words across settings of care (Neuman & Dwyer, 2009). The supplemental Spanish instruction via FEAs uniquely contributed to the preschoolers’ moderate Spanish receptive vocabulary gains (g = 0.49). The results reinforce growing evidence that engagement of caregivers in the provision of language instruction can lead to improvements in preschoolers’ language proficiency.

The impact of this home-language, caregiver-implemented intervention on children’s Spanish vocabulary is a promising finding. Because the effect of the FEAs was isolated in this study, we were able to determine that they had a direct impact on children’s Spanish vocabulary, which supplemented the effect of the Puente de Cuentos classroom curriculum on children’s Spanish and English language. In the first Puente de Cuentos study (Spencer et al., 2019), no effects were seen on Spanish vocabulary, but when the Spanish FEAs were added (Spencer et al., 2020), statistically significant effects were observed. Our results lend confidence that the caregiver-implemented FEAs were indeed responsible for children’s improvements in Spanish vocabulary. Children need several repeated opportunities to hear and practice new words to gain deep knowledge of their meanings (Beck et al., 2002; Neuman & Dwyer, 2009). As Head Start teachers were responsible primarily for English instruction at school, they were unable to provide the necessary repetitions of the Spanish words. However, with the added practice caregivers provided, children received sufficient opportunities to learn the Spanish vocabulary words.

We examined potential group differences in retell scores, but the lack of differentiation is not surprising. Classroom instruction involved a heavy dose of storytelling practice in both languages, which was accumulated across intervention sessions. Learning in storytelling builds upon previous storytelling skills regardless of language. Storytelling abilities rely heavily on cognitive schemata (Anderson, 1984; Mandler, 1984), which are thought to transcend languages (Pearson, 2002; Petersen & Spencer, 2016a). As children learned the parts of stories such as character, problem, feeling, action, and ending, they likely experienced cross-over effects given these components are shared by Spanish and English storytelling (Fiesta & Peña, 2004). As a result, children reached near-ceiling performance in Spanish and English through classroom instruction alone and the caregiver-implemented FEAs did not amplify the storytelling outcomes. Since the FEAs did not impact storytelling above and beyond the classroom components, it begs the question of whether the storytelling steps in the FEAs are necessary. We believe they are and would not recommend eliminating them. The primary reason is that young children need developmentally appropriate and meaningful context for practicing new words. The FEA procedures align with recommended vocabulary instruction by extending the practice of the words across contexts and embedding the practice in relevant activities (Neuman & Dwyer, 2009). Therefore, removing the story retell steps of the FEAs might reduce the potency needed for deep vocabulary learning.

It was also a possibility for the Spanish FEAs to impact English vocabulary, but the results of this brief study do not suggest such an influence. Although vocabulary learning does not benefit from shared cognitive schemas as in storytelling, there is evidence to suggest bilingual vocabulary interventions enhance both Spanish and English word learning (Mendez et al., 2015; Restrepo et al., 2013). In this study, however, the English words were taught and practiced at school more than the Spanish vocabulary, and we saw that both groups acquired receptive understanding of the English words equally well. Without classroom instruction having a strong focus on the English vocabulary, a Spanish-to-English influence might have been observed. As it was, the Spanish FEAs only affected Spanish vocabulary. These findings indicate that supplementing the classroom dual language instruction with a caregiver-delivered Spanish component brings about the bilingual outcome balance we were looking for and verifies experimentally that the Spanish FEAs contribute to the broader language impacts observed in Spencer et al. (2020). Importantly though, we cannot say that the Spanish FEAs improved children’s English and Spanish language skills. We can only ascertain that to impact both narrative and vocabulary skills in both English and Spanish, the FEAs are a critical component of the larger intervention package.

Feasibility and Cost-Effectiveness of Spanish FEAs

We also sought to evaluate the feasibility of the comprehensive dual language program from the perspective of the caregivers as teachers. Nix et al. (2018) have shown parent engagement in Head Start programs to be beneficial for children’s school readiness and for parent–child relationships. Parents want to be involved in their child’s education. However, few studies examine the social validity from a likeability perspective. We documented that the FEAs were feasible to the caregivers’ who implemented them. Caregivers reported that they enjoyed implementing the activities at home and that they would continue doing so after the study ended. The simplicity and structure of the FEAs likely contributed to their utility in homes. The activities were brief, culturally relevant, and easy to implement correctly. Although the preponderance of evidence gathered during this study point to favorable feasibility, we would be remiss if we did not remind readers that at least one caregiver-child dyad experienced a great deal of behavioral challenges during the FEAs. This caregiver did not complete the feasibility questions, but it is reasonable to conclude that they would not report that FEAs were easy to do or enjoyable. However, it is also possible that the child required more significant supports that were beyond the scope of the project, in which case the FEAs would not be a good fit for the child or for the dyad. The implications of our feasibility data generally are that for most Spanish-speaking caregivers and for most children whose behavior does not require extensive support, the FEAs are sufficiently easy to do and enjoyable.

To date, no study has examined the effects and related costs of providing supplemental family engagement activities for DLLs. However, it was important for the current study because we wanted to explore the extent to which the costs to achieve the additive effect on Spanish vocabulary were reasonable. This is also an important consideration as education budgets dwindle and schools must choose between competing initiatives. The cost of one standard deviation improvement was $261.91 per student. This figure is remarkably small compared to the potential costs associated with preparing bilingual teachers and increasing the amount of Spanish language classroom instruction that would be needed to achieve comparable effects. In the previous studies of the dual language program, teachers reported that the majority of their instruction had to be in English and they could only afford to spend the two small group sessions twice a week on Spanish instruction with a subset of students. For classrooms like these (i.e., Head Start) with several Spanish-speaking DLLs, supplementing the dual language program with caregiver home language support seems viable. It should be noted that the researchers’ time was included in the cost analysis, but if this was implemented by schools, the repeated contacts could be reduced once caregivers were trained. Given that caregivers delivered the intervention with 100% fidelity, we likely provided more support than was necessary. Additionally, in the current study only one of the three Puente de Cuentos units was implemented, which involved 24 new words. If caregivers delivered FEAs for all three units, the time spent training would not increase substantially, but the number of potential words learned would increase to 72. Schools that strategically engage caregivers in home language interventions may be able to achieve critical bilingual outcomes without considerably increasing costs.

Summary of Implications

There are two major implications for researchers that can be drawn from this study. First, the careful phased-based research agenda through which the Puente de Cuentos dual language program was iteratively developed is a good example of what intervention science is all about (Fraser & Galinsky, 2010). The input of end users and results during early phases led the research team to develop different components for the intervention to ensure the package impacted all the intended outcomes (i.e., Spanish and English narratives and vocabulary). Once all the components were developed and working, it was necessary to evaluate the FEAs impact experimentally rather than assuming its causal contribution. Often, intervention scientists do not slow down to conduct component analyses. In the current study, had we not done this, we could never be sure of the FEAs’ value. Intervention developers should consider verifying the necessity of specific components while searching for the minimal set of implementable active ingredients.

The second implication for research is related to designing interventions with sustained implementation in mind (Moir, 2018). Studies that integrate questions of effectiveness and implementation can hasten the time needed to move an intervention through the phases of research and thereby help to reduce the research-to-practice gap (Landes et al., 2020). Although feasibility studies are common in early childhood education research, cost effectiveness studies are just beginning to appear (Levin et al., 2018). More researchers should partner with economic evaluators to plan for and explore the costs associated with adoption and implementation of newly developed interventions. Education scientists also have many opportunities to expand their methodological repertoires given that there are several new resources available to them (e.g., IES’s Cost Analysis: A Toolkit, 2020; Levin et al, 2018).

As for practical implications, early childhood programs will benefit from understanding the value and cost of the curriculum, interventions, and services they offer, adopt, or provide. Many early childhood programs such as Head Start, require family engagement, and have dedicated staff who work with families directly. Advocating for ongoing funds to support these positions could come from research like this one. Family engagement models that have been empirically tested and analyzed for cost effectiveness will likely enhance child and program outcomes while allowing programs to make informed choices about funding. Although these results needed to be replicated and more research of this sort is needed, this study contributes to a small literature base that early childhood education programs can use to inform dual language instructional decisions as well as the value and cost of family engagement.

The result of this study implies that researchers, developers, and early childhood leaders need to be open to creative whole-child solutions that accomplish the goals. The make-up of preschool classrooms varies drastically, including the language abilities of teaching teams and the number of DLLs. We cannot be constrained by traditional thinking or approach the unique needs of various types of programs by assuming one size fits all. Certainly, some classrooms will have the right conditions that allow for sufficient exposure to the narrative and vocabulary components of this intervention in both languages, but most won’t. The involvement of caregivers may also not work for some programs or some families. It is important to recognize that supplementing a classroom-based dual language program with a caregiver-implement home language add on helped to achieve the goals of this curriculum in these classrooms. Nonetheless, it is critical to examine all the strengths and needs of various early childhood contexts and communities, and to establish implementation protocols that align with the unique characteristics of that program. Research should support the development and validation of tailored implementation packages so that early childhood programs have versatile options.

Limitations/Future Directions

Despite the promising outcome, several study limitations should be noted. First, due to Head Start policies, child-level assessments were conducted in the classroom during instructional time. This may have affected the students’ performance on pre- and posttest assessments. However, it would have influenced both the treatment and control group performance equally because the Head Start policy applied to treatment and control classrooms. Another limitation of the study was that two parent–child dyads withdrew from the study, reducing the already small sample size. Nonetheless, we used intention to treat strategies to retain the two children in their randomized groups. However, we cannot generalize assumptions to a larger population. Replication with a larger sample of caregivers and their preschoolers is necessary to confirm our findings. As a third limitation, one caregiver reported the occurrence of problem behavior which made it difficult to implement the FEAs at home. The type and intensity of support this caregiver-child dyad needed was beyond the scope of this study. Therefore, we decided to retain the child in the study, but acknowledge that there was little to no intervention delivered to them. While it threatens the rigor of the study, and likely underestimates the potential effect on child outcomes, problem behavior is not uncommon among young children, especially those who are struggling to learn two languages (Hagan-Burke et al., 2016).

Given non-significant results and inadequate sample size needed to detect an effect less than 0.48, results of this study did not provide evidence of the intervention effect on children’s Spanish or English narrative retells. While we found only an additive effect for the FEAs on the Spanish vocabulary outcome, with a larger sample, it may also have been possible to detect evidence of a cross-linguistic transfer of instruction provided in a child’s home language on both English and Spanish language outcomes (Borman et al., 2019; Miller et al., 2006; Phillips Galloway et al., 2020; Wagley et al., 2022). Therefore, we encourage future studies of the same dual language curriculum with the supplemental FEAs to include a larger sample of caregiver-child dyads.

Another limitation was related to the personnel who provided the training and support to caregivers. For this study, bilingual researchers conducted home visits and regular phone contacts. However, for schools to do this on their own, a family advocate or parent trainer would assume that role. Future research should include these change agents in the research to determine if the costs and effects would be comparable to those documented here. Finally, this was a small-scale preliminary study. It is prudent to conduct pilot studies before fully powered studies and we designed an affordable, preliminary study with randomization to explore these fine-grained analyses. Few randomized control trials have been used to analyze the effect of supplemental home-language interventions with DLLs. Out of seven studies involving family home-language interventions, only two were experimental (Boyce et al., 2010; Roberts, 2008), four were quasi-experimental (Harper et al., 2011; Johnson et al., 2012; Leseman & van Tuijl, 2010; van Tuijl et al., 2001), and one employed a single-case research design (Huennekens & Xu, 2010). Thus, we believe the results of our study are valuable to other researchers and professionals interested in using family engagement strategies to promote DLL language acquisition, although our randomized group design was largely underpowered. Future researchers will need to increase the number of participants and duration of the study to evaluate the effects of FEAs more rigorously. Researchers should also consider employing natural change agents of the program in their studies and examine the costs associated with using them. It may also be important to examine the effect of caregiver-implemented interventions across other types of early childhood programs, not just Head Start.

Conclusions

Considering the results within the landscape of the broader context of dual language intervention research, we conclude that involving families in home-language instruction is something early childhood providers should do. The FEAs were perceived as enjoyable and useful tools and boosted children’s Spanish vocabulary learning. Costs associated with those benefits were reasonable. Small early costs may lead to enormous savings long term. With shrinking budgets, however, more work is needed to develop interventions with cost-effective, tailorable components that meet the personnel and financial resource needs of programs.