Brazil has a rapidly growing economy (World Bank, 2018) but there is a large disparity in the provision of quality education between public and private sectors (Akkari, 2013). The emergence of bilingual schools, particularly in the private sector in Brazil in the 1980s, was historically, socially, politically, and economi- cally weighted, and aimed to promote a globalised education through a bilingual curriculum. However, the level of language immersion varies greatly across indi- vidual schools (Fortes,2017). Previous research examining the impacts of bilingual immersion programmes have typically focused on the North American experience, for example, English-French immersion in Canada and English-Spanish immersion in the United States of America (Baker et al.,2016). Very little research has focused on the Brazilian experience. Brazil also faces numerous educational challenges: 48%
of children do not attain a basic level of mathematics (UNESCO,2011) and the influ- ence of the language of instruction on mathematics is of particular concern for young children at the start of school (Haag et al.,2015).
As theonebillionmaths app can deliver the same maths instruction in Brazilian Portuguese and English, this study examined if the app could provide an effective intervention to support maths development in bilingual classrooms. First, adequacy of the translation between the app content in English and Brazilian Portuguese was assessed by two reviewers: a native English speaker and a native speaker of Brazilian Portuguese and a second-language speaker of English. Brazilian Portuguese and English transcripts were randomly selected for maths activities within the app. All selected app transcripts were deemed to accurately convey the same meaning in both languages (except for one, where the meaning of the instruction given was changed in translation from English to Brazilian Portuguese).
To explore how best to implement the maths app within a bilingual context, we conducted a study in a Brazilian Portuguese-English bilingual immersion private school in Recife in the North-East of Brazil (Outhwaite et al.,2020). The participating school implemented a one-way dual language programme with 50/50 immersion (Gomez et al.,2005). Time allocated for instruction in each language was split equally across the school day. In the first half of the school day all subjects, including maths, were taught in English. The second half of the school day was dedicated to language skills, including reading and writing, which were taught in Brazilian Portuguese.
A total sample of 62 children aged 5–6 years in the first year of elementary school took part. As we were interested in how the language of instruction influenced children’s learning of elementary maths with or without the interactive app, children’s proficiency in Brazilian Portuguese and English was assessed with a teacher-rated, 7-item questionnaire given at pre-test. The questionnaire included four items on speaking, and one item on each of reading, writing, and listening, and was repeated for each child for Brazilian Portuguese and English. It was developed specifically for this study and was adapted from the Alberta Language and Development Questionnaire (ALDQ; Paradis et al., 2010). The ALDQ is a non-language and non-culturally specific parental questionnaire and assesses children’s competencies in their first and second language. Teacher ratings showed that, as a group, participating children were unbalanced bilinguals as their proficiency in Brazilian Portuguese was stronger than their proficiency in English.
The study followed a non-randomised, quasi-experimental design with three groups: 23 children in Class 1 received the maths app in Brazilian Portuguese, 20 children in Class 2 received the maths app in English, and 19 children in Class 3 received their regular maths teaching instruction, which in this school was delivered in English. Children allocated to receive instruction with theonebillionmaths app (Class 1 and Class 2) used the app for 20-minutes, four times a week, for 10 consec- utive weeks. The app was implemented by the class teacher and teaching assistant and was given instead of a small group embedding mathematics activities used in standard practice.
To evaluate learning with and without the maths app, all children completed the Early Grade Mathematics Assessment—EGMA (RTI International, 2009), before
and after the 10-week intervention period. Assessments were delivered on a one- to-one basis with each child in a quiet area, free from distraction, in the child’s familiar school environment. All assessment instructions were delivered using an audio recording of a standardised script in Brazilian Portuguese. For items that required a non-verbal response, children responded by pointing (e.g. pointing to which number was bigger for two presented numerical digits). For tasks that required a verbal response, children could choose to respond in either Brazilian Portuguese or English. Children’s progress through the maths app was also assessed by the number of topics completed, as this has been shown to correlate significantly with EGMA performance in a group of 116 grade 1 children (6–7 years) attending mainstream school in Malawi demonstrating that the more topics children passed the more their maths ability improved (Pitchford et al.,2018, p. 8).
Learning gains across the intervention period were calculated for each of the three instructional groups. Statistical analysis showed that in this bilingual immersion context, when children received instruction with theonebillionapp, given in either Brazilian Portuguese or English, they made significantly more progress in learning elementary maths than children who received regular classroom instruction [one-way analysis of variance:F (2, 58)=3.78,p=.029]. On average, over the course of the 10-week intervention period, children who received app-based maths instruction in Brazilian Portuguese gained 17% on EGMA, children who received app-based maths instruction in English gained 18%, whereas children who received regular mathematics instruction in English given by their class teacher gained 11%.
Interestingly, there was no significant difference in learning gains achieved between classes of children that received the maths app as measured by EGMA, although children who used the app in Brazilian Portuguese made more progress through the app, passing 14 topics on average, compared to children who used the app in English, who passed on average 12 topics. Moreover, proficiency in the language of instruction was shown to be positively associated with progress through the maths app, as indicated by the number of topics completed: children with greater profi- ciency in the language of instruction made significantly more progress through the app than those with less proficient language skills [r=0.36,p=.022] but this rela- tionship was not significant for learning gains, as measured by EGMA [r=0.25,p
=.113].
These results demonstrate that the onebillionapp can be an effective tool for learning elementary maths for bilingual children, when delivered in either the child’s first or second language. In addition, proficiency in the language of instruction was significantly associated with progress through the app (topics passed), but not with learning gains as measured by EGMA. This most likely reflects differences in the sensitivity of the two outcome measures as a positive correlation was found with both measures, demonstrating again the importance of language proficiency in accessing learning materials within the app (Strand & Hessel,2018; Cummins, 2008). Compared to standard maths instruction delivered by class teachers, app- based maths learning might be constrained by the inability of the app to provide additional supportive and contextual cues, such as gestures, intonation, and concrete aids, which facilitate learning for bilingual children.