Abstract
The increased use of smartphones and tablets sets the stage for new mobile-based educational programs that seek to increase student learning and engagement in school and at home. This study examines the effectiveness of Escribo Play, a game-enhanced educational program, on preschool students’ phonological awareness, word reading, and writing skills. The randomized controlled trial was conducted with 749 students from 62 classrooms from 17 schools located in five cities. Quantitative pretesting and posttesting procedures using standardized instruments were employed. The results indicate that the experimental classrooms that used the 20 games for 3 months gained 68% in their reading scores compared to control classrooms (d = .40). They also gained 48% more in writing scores (d = .20). Multilevel analysis indicated that these findings were statistically significant.
Keywords
The low level of literacy achievement is one of the most relevant challenges faced by low-income children in many countries. In Brazil, 50% of third-grade public school students do not meet the national reading and writing standards (Instituto Nacional de Estudos e Pesquisas Educacionais Anísio Teixeira, 2017). Prior research established that language and reading skills in preschool and kindergarten are strongly correlated to reading achievement in the first and second grades (Catts et al., 2001; Duncan et al., 2007; Schatschneider et al., 2004). For example, verbal skills in kindergarten are significant predictors of reading achievement in fourth grade (Kurdek & Sinclair, 2001). These results indicate that improving literacy skills in preschool and kindergarten may contribute to students’ later reading achievement.
An essential component of effective preschool and kindergarten programs is instruction in phonological awareness (PA) (Chambers et al., 2016). Unfortunately, most Brazilian schools do not yet provide PA instruction (Morais, 2012). Nevertheless, the first Brazilian common core, which will be in effect in 2020, requires PA instruction (Ministério da Educação, 2018).
Complementary evidence has suggested that PA programs applied in early childhood education have lasting effects on word recognition skills as well as reading comprehension skills, even in the final years of elementary (Kjeldsen et al., 2014). Previous studies have found that small-group PA instruction generated higher gains in reading skills than whole-class and individual instruction (Cheung & Slavin, 2013; Ehri et al., 2001). Unfortunately, differentiating instruction and conducting small-group activities becomes extremely difficult in early childhood centers with child-staff ratios that can be as high as 1 to 14.5 (Perlman et al., 2017). Given such context, developers of technology-enhanced programs have been trying to provide means to facilitate differential and individualized instruction (Pokorni et al., 2004).
The first era of technology-enhanced PA programs sought to assist students in building their PA, reading, and writing skills (Cheung & Slavin, 2013). Although many endeavors were developed from the mid-1990s until 2010, employing computer-based multimedia applications, many early-grade students in this period coming from low-income families were not familiar with computers. These children first had to master basic technological skills, such as how to control the mouse, so that later they could try to benefit from computer-assisted instruction.
Today we are living in a new technological scenario. Mobile devices such as smartphones and tablets are more intuitive and easy to use with their touchscreen interfaces. Children are using mobile devices early in their development, spending an average of 58 minutes per day when they are between 2 and 4 years old (Common Sense Media, 2017). As of 2018, inexpensive mobile devices and quality tablets can be found for as low as US$50. Smartphone penetration in low-income families is growing in developing countries like Brazil, where 77% of the inhabitants older than 10 years old have smartphones (Instituto Brasileiro de Geografia e Estatística, 2018). The widespread dissemination of mobile technology sets the stage for new mobile-based instructional programs that seek to increase student learning and engagement, at home and in the school.
The present study is part of a larger project that included the conceptualization, design, and development of a game-enhanced instructional program. The program was created to be used by preschool and kindergarten teachers and students to foster the development of their PA, reading, and writing skills. The program was developed using research-based evidence from the fields of language development, instructional design, and video game design. This study examines the effectiveness of this newly developed game-enhanced program on preschool students’ PA, word reading, and writing skills using a randomized controlled trial. We employed inexpensive tablets that ran offline to maximize the number of schools that could afford to receive the intervention.
Literature Review
This work was built on the evidence that PA is a key component of reading development and that PA instruction can benefit 4- and 5-year-old students to build their word reading and word writing skills (Bus & van IJzendoorn, 1999; Chambers et al., 2016; Ehri et al., 2001). We examined previous PA intervention research to identify the instructional programs that generated higher effect sizes in PA, word reading, and writing skills. Our focus was to identify the key components of the most effective programs such as specific PA activities and weekly and total hours of instruction, seeking to build an instructional program that could assist teachers to conduct small-group instruction capable of building students’ PA, word reading, and word writing skills.
Ball and Blachman (1991) found that phoneme awareness training that included letter-sound correspondences significantly enhanced early reading and spelling. A meta-analysis of 52 studies also indicated that phonemic awareness instruction had a moderate impact on reading (d = .53) and spelling (d = .59) for students from kindergarten to sixth grade (Ehri et al., 2001). Phonemic awareness was found to be the strongest predictor of reading achievement (Høien et al., 1995), but PA also includes syllabic and rhyming skills.
Bus and van Ijzendoorn (1999) developed a meta-analysis of 32 PA training studies and identified that the impact of PA on reading achievement was substantial (d = .70, r = .33, N = 745). The authors also argued that preschoolers benefit more from PA programs than do kindergarten or primary school students. The results of this body of research point to the importance of PA and the effectiveness of instructional programs that foster its development in the early grades.
A computer-based intervention can be used to address the various components of PA (Piquette et al., 2014; Segers & Verhoeven, 2005). Applications can develop syllabic, rhyming, and phonemic awareness (Dias, 2006). Interventions that combine PA with letter-sound correspondence instruction were found to be more effective than interventions that delivered only PA without printed letters and texts (Bus & van IJzendoorn, 1999).
Another aspect of computer-based PA interventions regards the pitch, speed, and transitions of the speech. Segers and Verhoeven (2004) found no significant results for instruction with reduced speech rate and enhanced transitions. One implication of this research is that computer-based instruction shall employ natural oral language rather than computer synthetic voices.
Cheung and Slavin (2013) reviewed 20 randomized and quasi-experimental studies that involved 7,000 children to evaluate the impact of using technological applications and electronic tools developed to provide content and foster learning such as computer aided instruction, video, multimedia content, and integrated learning systems. Their results indicated a small beneficial effect size (d = .14) coming from the adoption of technology when compared with traditional reading interventions. Nevertheless, tutorial applications that are integrated with the curriculum and used by small groups of students generated the strongest benefits (d = .32). The authors also found evidence that individual instructional programs were more effective for struggling students. Studies that focused on primary grade students had larger effects (d = .36) than those targeted to upper elementary grades (d = .07), which reinforces the focus on providing PA activities to younger students in preschool and kindergarten.
Cheung and Slavin’s (2013) meta-analysis found that high-intensity programs that included more than 75 minutes of instruction per week (d = .19) yielded larger benefits than did low-intensity programs (d = .08). This finding was not statistically significant due to low statistical power (QB = 1.20, p < .27) but suggests that computer-based interventions shall provide an intense dosage of instructional activities. Based on the parameters identified in previous research that generated larger gains in PA, reading, and writing, an intervention was designed to provide two sessions of 45 minutes of instruction per week employing digital games, during 10 weeks.
Other kinds of technology-enhanced interventions were found to be beneficial for early literacy development. A meta-analysis of 43 articles involving 2,147 students identified that technology-enhanced stories were beneficial for story comprehension (g+ = .17) and expressive vocabulary (g+ = .20) (Takacs et al., 2015). Another meta-analysis of 122 studies assessed the influence of technology applications on elementary students’ learning and identified significant gains in language (ES = .44) and other subjects (Chauhan, 2017). Digital games were found to improve reading for second-grade poor readers (Gorp et al., 2017) and for children with Down syndrome (Felix et al., 2017). Moreover, evidence suggests that digital games can be beneficial for second language learners (Zhang, 2018) and that preservice teachers see game-based learning as a positive approach for reading and writing instruction (Karadag, 2015).
Although there is evidence about the effectiveness of educational technologies for PA, reading, and writing instruction, an effective program that employs mobile games to increase learning in early childhood education has yet to be developed and studied. To address this gap, we created Escribo Play, an evidence-based, gamified, early literacy program that is compatible with inexpensive mobile devices, which has the potential to improve teaching and learning. This study seeks to explore the following hypotheses:
Hypothesis 1: Children studying with Escribo Play will display improved PA skills when compared to students who were exposed to only the regular instructional strategies employed by the schools.
Hypothesis 2: Children studying with Escribo Play will display improved word reading ability when compared to students exposed to only the regular instructional strategies employed by the schools.
Hypothesis 3: Children studying with Escribo Play will display improved word writing ability when compared to students exposed to only the regular instructional strategies employed by the schools.
Method
A cluster randomized controlled trial was conducted to evaluate how the effects of the program manifested in students’ PA, word reading, and word writing skills (Shadish et al., 2002). The project employed a design with experimental and control groups with pretests and posttests. We randomized at the classroom level, not at the student level, to reduce the risk of contamination between students within the same classroom (Pierre, 2004; Shadish et al., 2002; Torgerson et al., 2015). Figure 1 displays the design of the study.

Study design.
Participants
We conducted the study in the metropolitan area of Fumaça (pseudonym). The city, located in the northeast region of Brazil, has 30,591 4- and 5-year-old children in 678 schools (Instituto Nacional de Estudos e Pesquisas Educacionais Anísio Teixeira, 2016). All students in Brazil speak Portuguese, and the games were developed in Portuguese. Within this population, 448 private schools had decision autonomy to participate in research.
We sent invitation letters to 25 private schools that would probably have the proper technological resources and autonomy to fit the intervention into their academic plans. The list of schools was provided by a local company that had visited 365 institutions in 2016 and assessed their resources. The first author met each school principal to explain the research goals and procedures.
A total of 17 schools agreed to participate in the study. Together, they had 1,089 students in the 4-year-old grade. The research team delivered study packets to each school that included an invitation letter and a consent form for each student/parent. Beyond asking for contact information, the consent form asked for parents’ educational attainment. The packets were delivered to the schools in the first days of August 2017, when the second semester of the Brazilian academic year starts. Two weeks after we delivered the letters, we asked the schools to send friendly reminders about the deadline, and we received a total of 749 signed consent forms from the families. Although in Brazil only 15% of citizens attend higher education (Organisation for Economic Co-operation and Development, 2015), our sample had 93.9% of the parents in this category, suggesting that the families were high achievers. Supplementary Table S2, available on the journal website, presents the sample composition details.
After finishing the pretests, we conducted a cluster randomization procedure to assign its units, the classrooms, to the control and experimental groups (Torgerson et al., 2015). The goal was to allocate half of the 62 classrooms to each group. During the randomization, four classrooms that were assigned to the experimental group had to be converted to the control group because the classrooms had their schedule already established, and the limited number of research staff could not adjust the schedule. These four classrooms did not receive instructional sessions; neither were their teachers participants in the professional development sessions. The final experimental sample included 27 classrooms.
The teachers in the experimental classrooms had an average of 14 years of teaching experience. All teachers had obtained their undergraduate degrees, and on average, they graduated 4.6 years before the intervention. About 71% of the teachers finished lato sensu specialization programs beyond their undergraduate degrees, and 5% obtained master’s degrees. Regarding professional development experiences, 47% of the teachers reported that they participated in courses that covered reading and writing instruction.
The 35 control classrooms had 418 (55.8%) students, and the 27 experimental classrooms had 331 children (44.2%). The mean age for the participants was 56 months (SD = 3.8) during the pretests, equivalent for both groups. In the experimental group, 48.6% were boys, and 51.4% were girls. In the control group, 51.3% were boys, and 48.7% were girls. The average teacher to student ratios were 1:11.6 in the control classrooms (SD = 4.5) and 1:12.1 in the experimental classrooms (SD = 3.3).
In Brazil, schools usually function in the morning and afternoon, each shift with a different set of students. In our sample, 68.1% of the students attended school in the morning, whereas 31.9% attended in the afternoon. This distribution reflected an overall trend observed in the cities of Fumaça, where most families prefer that their children attend in the morning.
All 4-year-old students in the participating schools were eligible to participate in the study. After finishing the intervention and the posttests, we found that 12 (1.6%) out of 749 students had some level of learning disability, according to the schools’ documents and teachers’ reports. Although those students participated in all research activities because we wanted to give the opportunity for all children who wanted to participate to do so, we excluded those 12 students’ data from the statistical procedures.
Study Measures
We employed the Phonological Awareness Test by Oral Production (PAT-OP) assessment in pretests and posttests (Seabra et al., 2013). The test includes 10 subtests to assess awareness of syllables, rhymes, alliterations, and phonemes in Portuguese. PAT-OP scores were positively and significantly correlated with student age, grade level, and other tests such as the word and pseudoword repetition test and the phonological discrimination test (Dias et al., 2013). Psychometric data from PAT-OP revealed a Cronbach’s alpha of .91 and a Spearman-Brown’s coefficient of .86, with normative data available for Brazilian children (Dias et al., 2013). In this study, PAT-OP had adequate Cronbach’s alphas of .86 in the pretest and .91 in the posttest.
The Word Reading and Writing Test (RWT) was employed in pretests and posttests to determine students’ word reading and writing skills in Portuguese (Pazeto et al., 2014; Pazeto et al., 2017). The first part of this assessment consists of eight words and two pseudowords that shall be read by the student. In the second part, the researcher dictates eight words and two pseudowords for the student to write. Each word is more complex than previous ones. Words are classified by their regularity and frequency of usage. The test score is calculated by the percentage of the letters that were correctly answered by the student so that the performance ranges from 0% to 100% for each skill.
The same group of researchers who created PAT-OP started to develop the RWT instrument in 2012 (Pazeto, 2012) and released its normative data in 2019 (León et al., 2019). Pazeto et al. (2014) identified robust and significant correlations between RWT and PAT-OP scores (r = .76 and .80, p < .01), letter-sound knowledge (r = .73 and .87, p < .01), and letter-name knowledge (r = .49 and .57, p < .01) and between its own reading and writing measures (r = .81, p < .01). In this study, the reading scores displayed a Cronbach’s alpha of .94 in pretests and .95 in posttests. The writing scores had a Cronbach’s alpha of .94 in pretests and .92 in posttests: excellent consistency levels.
To assess interrater reliability between coders, we calculated the intraclass correlation coefficients (ICCs) using a one-way random effects model in SPSS 25. For the reading pretest, ICC was .95 (p = .00), with 20.7% of the participants’ being rated by a second rater. ICC for the writing pretest was .97 (p = .00) and for PAT-OP was .90 (p = .00). The three ICCs were high, indicating that PAT-OP, reading, and writing pretest ratings were reliable across the raters (Graham et al., 2012).
For the posttests, the reading assessment ICC was .97 (p = .00), calculated with 20.6% of the participants’ being rated by a second rater. ICC for the writing posttest was .94 (p = .00), and for PAT-OP the posttest was .85 (p = .00). ICCs for the reading and writing posttests were high. The PAT-OP posttest was greater than the minimum desired level of .75, indicating that ratings were reliable (Graham et al., 2012).
The Instructional Program
The goal of the intervention was to develop the students’ PA, word reading, and writing in Brazilian Portuguese. The intervention was designed to provide two instructional sessions of 45 minutes per week, during the period of 10 weeks, a total of 20 sessions for each classroom (a total of 900 minutes). All classrooms completed 20 sessions. We conducted four syllable awareness sessions so that the learners could master syllable segmentation and blending as well as adding, removing, and inverting syllables to create new words (Mesmer & Williams, 2015). Then, four sessions were conducted to enhance rhyming and alliteration awareness. The phoneme awareness activities started in Session 9 with the identification of the first and last phonemes in given words. In Sessions 10 and 11, phonemic blending tasks stimulated students to build words from phonemes. We developed letter-sound activities in Sessions 12 to 14.
Word reading and writing activities were developed from Sessions 15 to 20 so that students could exercise their grapheme-phoneme correspondence skills. Such practice is essential to consolidate the ability to write any word spoken in Portuguese and to read any word written in a given text. Words that employed irregular orthographic structures were not included in the instruction because they can be learned at later ages according to the Brazilian common core (Ministério da Educação, 2018). Before the intervention, the content and designs of the 20 mobile games were improved based on feedback from field tests and doctoral-level experts from the fields of education, cognitive psychology, and speech therapy. More information about the design, development, and implementation of the games is presented in the supplementary online material available on the journal website.
Instructional sessions
We conducted the intervention with four research assistants from September to November of 2017. To compensate for holidays and school events, some classrooms had to implement three sessions per week, and others had to extend the last sessions into the first 2 weeks of December.
Each session followed Gagné’s (1965) instructional events, lasting 45 minutes and being divided into three components. In the first component, lasting from 10 to 15 minutes, the attention of the student was gained when the teacher explained the target skill on the whiteboard (e.g., cutting words into small parts—syllable segmentation) to the whole group and developed three exercises using the whiteboard or the movable alphabet. The second part, lasting around 5 minutes, started when the teacher explained how to develop the target operation in the game, playing the first three levels together with the whole group.
The third component started when the research assistant randomly assigned the students to their pairs so that two students would work cooperatively with one tablet. The Amazon Fire 8-inch tablet (2017 edition) with an EVA protective case designed for children was used to deliver the intervention. Amazon Fire tablets are inexpensive devices that cost as low as US$50 each but offer performance and battery duration superior to the requirements of the Escribo Play app.
As the students played in pairs, 10 tablets were enough as the classrooms had up to 20 students and the tablets were rotated from classroom to classroom. This strategy was employed because previous studies found that small-group instruction generated higher PA gains than large-group and individual instruction (Cheung & Slavin, 2013; Ehri et al., 2001). The goal was that each student would play cooperatively with a different peer in each of the 20 sessions of the intervention. The researcher assistants paired the children in each session to accomplish this goal. Random allocation was used to minimize the eventual effect that specific students might have in inhibiting or supporting another peer. Having different pairs of students playing in one tablet was facilitated because the Escribo Play mobile application provided the names of students so that the RA could easily call them to sit and work together.
The teacher and the research assistant were instructed to act as facilitators while the students were playing. They should move from pair to pair, observing how the children were using the game. When the students were having difficulty with a task, the teacher or the research assistant would remind them about the clues that were provided by the games. If students were still unable to advance, the facilitators would perform the same activity with the students orally, or with printed support, using a word different than the one provided by the game.
As students finished the game, they received virtual medals and were congratulated by the teacher or the research assistant. When students finished the game before the time was up, the teacher suggested they play again to make more points or to win a virtual gold medal, when they had gotten silver or bronze. After 20 sessions, the students received the established dosage for each ingredient of the treatment (e.g., rhyming, phoneme awareness). The supplementary online material available on the journal website presents more details about the sessions, content, and dosage. During the 10 weeks of the treatment, control classes kept doing their regular activities.
Data Collection
Pretests
After we received the signed consent forms, trained research assistants conducted the pretests. Because the randomization was completed after the pretests, the research assistants conducted the pretests without bias to allocation of classrooms to control or experimental groups (Torgerson et al., 2015). One of the authors of the assessments trained the assistants. Each assessment started with PAT-OP, which was followed by the reading and the writing tests. All tests were audio-recorded so that we could calculate interrater reliability later. On average, pretests lasted a total of 20 to 30 minutes. The goal was to finish the assessments with one seating, but when students lost their focus, the assistants stopped the procedure and continued later.
Professional development
We conducted one professional development session in each school. During these 1-hour meetings with the experimental group teachers, the first author followed a strict instructional protocol so that all teachers received the same content before starting the instructional sessions with their students. The content included some background about the importance of PA instruction and an explanation of how the experiment would be developed, including how each instructional session would be implemented. In the end, the teachers defined the days and times for the instruction to be delivered to each classroom.
Posttests
Posttests using the same instruments (PAT-OP and the reading and writing assessment) started in the last week of November 2017 for the control classrooms. As the experimental classrooms finished the instructional sessions on different days, when a classroom finished the 20th session, we conducted the posttests in the following days.
We hired two additional research assistants at this stage due to the urgent timeline (the winter vacation began on December 15 for most of the schools), and a total of six research assistants conducted the posttests. We sent the research assistants to schools different from the ones in which they had implemented the intervention. This was done to reduce bias, as they could not know which students were from the experimental and control groups. Students were assessed individually with the same instruments used in the pretests.
Findings
Pretest Findings
Pretests revealed that students in the control group had a similar mean score on PAT-OP of 9.3 (SD = 5.2) compared to the experimental group, which had a mean of 9.4 (SD = 5.1). The reading mean score for the control group was 10.66 (SD = 20.84), whereas the experimental group had 9.41 (SD = 19.59). Finally, the writing mean score for the control group was 15.06 (SD = 20.36), whereas the experimental group had 14.95 (SD = 19.76). Independent samples t tests did not show significant differences between the control and experimental groups for PAT-OP, reading, and writing scores. As noted in Table 1, when controlling for the student shift, the results indicated that students in the morning had better scores on the three tests, but a significant difference was found only in the writing score (p = .044).
Descriptive Statistics for Pretests and Posttests
Note. PAT-OP = Phonological Awareness Test by Oral Production.
Intervention Effects
The posttests revealed that students in the control group had a mean score on PAT-OP of 11.5 (SD = 6.2), lower than the experimental group, which had a mean of 12.4 (SD = 6.6), but this difference was not significant according to the independent samples t test (p = .08). The reading mean score for the control group was 19.84 (SD = 27.88), significantly different (p = .02) from the experimental group, which had 24.86 (SD = 30.94). The writing mean score for the control group was 23.57 (SD = 26.47), and the experimental group was 27.55 (SD = 26.63). The test of writing approached statistical significance (p = .05).
A correlational analysis with the posttest data indicated that the correlations between the tests for the experimental group were stronger than for the control group, with the higher difference between the reading and writing assessments, as can be seen in Table 2. Regarding the student shift, the significant difference that was found in the writing pretests (p = .04) was not observed in the posttests (p = .58). The lack of difference in posttests can indicate that the intervention contributed to equalizing the writing levels of the morning and afternoon students.
Correlations Between Posttests for Both Groups
Note. All correlations were significant at the p < .01 level. PAT-OP = Phonological Awareness Test by Oral Production.
To compare the growth patterns between the experimental and control groups, we calculated the differences between each student’s pretest and posttest mean scores. On average, control classes gained 9.18 points in reading between pretest and posttest. Experimental classes gained 15.45 points between pretest and posttest. We subtracted the experimental gain from the control gain (15.45 – 9.18 = 6.27) to find the difference in gain. By dividing the difference in gain by the control gain (6.27 / 9.18), the relative gain was obtained. Experimental classes gained 68% more in the reading score when compared to the gains of control classrooms. The same procedure was done with the writing score, which revealed that experimental classrooms gained 48% more in the writing assessments when compared to the control classes’ gains.
Multilevel Analysis to Determine Intervention Significance
Because the randomization was conducted at the classroom level, it is expected that students nested in the classrooms may display similar properties as they interacted and were exposed to the same instruction. To adequately account for such hierarchical data, we employed multilevel analyses (Wears, 2002). The focus of the multilevel analysis was to determine the effectiveness of the intervention on the student’s PA, reading, and writing skills while adjusting for the nested data structure. The multilevel parameters are presented in Table 3.
Multilevel Analyses of Reading, Writing, and Phonological Awareness (PA) Outcomes
p < .08. *p < .05. **p < .01. ***p < .001.
The model included the student’s pretest PA, reading, and writing scores; gender; shift; age at pretest; and parents’ educational attainment as covariates, besides the experimental condition. The model revealed a significant relationship between the experimental treatment and the reading (p = .04) and writing (p = .03) posttests. This model explained 54.15% of the variance in reading, 55.22% of the variance in writing, and 62.36% of the variance in PA posttests.
The results of the multilevel analysis confirmed Hypotheses 2 and 3. The children who studied using Escribo Play displayed improved word reading (d = .40) and word writing (d = .20) skills when compared to the students who were exposed to only the regular instructional strategies employed by the schools. Nevertheless, the multilevel analysis did not reveal a significant relationship between the treatment and the PA posttests (p = .33). This lack of significance does not allow us to confirm Hypothesis 1, which was that the children studying with Escribo Play would display improved PA skills when compared to the students who were exposed to only the regular instructional strategies.
Discussion
The findings show that the intervention led to a significant gain in the experimental classrooms’ reading and writing scores when compared to the gains of the control classrooms. The effect of PA instruction on reading and writing development that was detected in this research is in line with international research (Bus & van IJzendoorn, 1999; Chambers et al., 2016; Ehri et al., 2001). There also is evidence of reading gains as an effect of PA instruction for Brazilian Portuguese (Morais, 2015; Paula et al., 2005; Santos & Maluf, 2010).
We consistently communicated with the teachers and their coordinators during the intervention to support their instruction. From these observations and conversations along with the previous literature review, we suspect that the learning gains might have come from three factors. The first is that most of the teachers had heard about PA activities but had never implemented them systematically. With the intervention, they received professional development support and adequate tools to develop PA activities. The second factor is that students were highly engaged with the games, which may have served to increase their desire to learn how to read and write. The third factor is the design of the program, with all games employing text together with spoken sound and some games providing visual cues. Such features probably increased students’ exposure to printed words and reduced the cognitive load required to build and automatize print-to-sound relationship skills.
The lack of effect in the PA posttests can be explained by the fact that the instructional activities employed sound and print, as this is the most effective strategy to enhance reading and writing skills (Ehri et al., 2001), but PAT-OP assesses oral skills without print support. Another possibility for such lack of significance comes from the nature of the PA activities. When we observed the PAT-OP syllabic awareness subscale, which includes syllabic segmentation, blending, addition, removal, and transposition tasks, we identified a treatment effect (d = .22) like the one we found for the writing assessment. We observed no effect among the other subscales of the PAT-OP instrument. This finding may indicate that the reading and writing gains stemmed primarily from the increased syllabic awareness skills, as syllables are very apparent in Brazilian Portuguese. It is possible that the 4-year-old students were not ready to develop more advanced phonemic awareness skills.
The innovative aspect of this research is that it employed mobile games to scaffold students’ skills and to increase motivation. The quantitative data from pre and post assessments revealed that experimental classes gained 68% more in their reading scores and 48% more in their writing scores than the gains of the control classrooms. These gains suggest that the games and the instructional strategies were beneficial for students’ reading development. This finding is in line with previous research that detected the beneficial effects of using technology to develop reading (Cheung & Slavin, 2013; Piquette et al., 2014).
One relevant finding for practice is that this game-enhanced intervention was delivered using the inexpensive Amazon Fire tablets, which cost as low as US$50 each. These tablets met all of our expectations in terms of performance and durability. They ran for the full day of classes and were recharged only at night. In schools that shut down the power grid at night, it would be advisable to charge the tablets during the morning, lunch, and afternoon breaks.
Limitations
The main limitation of this study regards the generalizability of the sample, which included 17 schools from five cities in the metropolitan area of Fumaça. Although this sample is an adequate composition when considering the private school sector that serves 30% of the students (Instituto Nacional de Estudos e Pesquisas Educacionais Anísio Teixeira, 2016), it did not include public schools. As Brazilian public schools display lower student achievement levels than private schools’, this intervention may display different results in public schools. For example, in the RWT normative sample with 5-year-old students (1 year older than our sample), private school students’ reading scores were 22.7 times higher than those of the public school students (León et al., 2019). Such a high degree of inequality may indicate that public school preschool students may not have the prerequisite skills for the intervention to be effective, such as reduced vocabulary or print knowledge, but also suggests that this intervention may generate stronger learning gains for low-income children as students who perform below average often benefit greatly when they receive high-quality instruction.
Another variable that may increase or reduce the effectiveness of the program is that in our sample the students were from families of higher educational attainment. There was probably increased access to books and incentives may have been provided for these children to learn how to read and write when compared to low-income students.
Despite the limitations described above, it is worth noting that this is the first large-scale randomized trial conducted to evaluate an intervention that employed mobile games as an instructional tool to develop PA, word reading, and word writing skills in preschool. The trial revealed significant gains in word reading and writing: early literacy skills that are very important as children progress toward primary school. The adoption of gamification mechanics such as the ones that were employed in this intervention has the potential to increase student motivation and learning at scale, as mobile devices such as tablets and smartphones are significantly cheaper and easier to use and maintain when compared to the traditional computer-based laboratories.
Future Research
During the conversations with the teachers, we noticed that several teachers experienced high levels of anxiety during phoneme awareness activities. They reported that learning how to pronounce individual phonemes was a challenge as they had never received such instruction. It would be interesting to investigate whether conducting a second professional development meeting focused on letter-sound knowledge before the phoneme awareness sessions could increase the intervention effectiveness, as teachers could be more confident when delivering the instruction.
Some teachers reported that they were surprised by the results of the intervention. One said that she never had “imagined that a 4-year-old student could read” and that as she was delivering the intervention, she saw her students beginning to read. This increased awareness about the students’ capabilities led her to set higher expectations and to enhance her reading instruction with additional PA activities. It would be worthwhile to investigate the dosage of such additional PA activities and how they affected the effectiveness of the Escribo Play intervention.
An additional avenue for future research is to replicate this intervention in public schools. A cost-effective approach might include developing a prior needs assessment to determine the level of instruction and student skills in public schools. Changes in the instructional strategies and games may be needed to maximize the effectiveness potential in public schools. For example, we suspect that students with less exposure to print text may lack prerequisite knowledge to play the games, so we may need to design new games to fulfill this gap.
Another research direction is to seek more evidence about how the process of the intervention works in different circumstances. One interesting question is how the existing students’ skills are related to the intervention’s effectiveness. Furthermore, it would be helpful to understand how the technology mediation provided by the teacher, or the parents, influences reading and writing development.
Supplemental Material
AmorimEtAl._ONLINESupplement – Supplemental material for Using Escribo Play Video Games to Improve Phonological Awareness, Early Reading, and Writing in Preschool
Supplemental material, AmorimEtAl._ONLINESupplement for Using Escribo Play Video Games to Improve Phonological Awareness, Early Reading, and Writing in Preschool by Americo N. Amorim, Lieny Jeon, Yolanda Abel, Eduardo F. Felisberto, Leopoldo N. F. Barbosa and Natália Martins Dias in Educational Researcher
Footnotes
Notes
Authors
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
