According to South African Curriculum Assessment Policy Statements, learners’ first language should be the language of learning and teaching in Grade R. However, there is a mismatch between policy and practice since English is the language of Grade R learning and teaching in rural schools in Mpumalanga. The ECD manager should provide evidence-based guidance on the best facilitation approach to follow in a particular context. The study is underpinned by: ‐ Cummins’ distinction between Cognitive Academic Language Proficiency and Basic Interpersonal Communicative Skills (Cummins, 2000) ‐ Cummins’ interdependence hypothesis (Cummins, 2000) ‐ Distinction between Implicit versus Explicit learning (Dekeyser, 2003) ‐ The Mediated Learning Experience Theory within the poverty context of South Africa where the role of the teacher is seen as the mediator (Feuerstein, 1980) The literature overview focuses on the development of the Grade R curriculum, and two prominent facilitation approaches used in Grade R, the play-based and formal instruction. Research on the effectiveness of each method and language debates both internationally (Wong-Fillmore (1991), Bialystok (2006), Cummins (2000) and nationally (MacDonald (1990), Heugh (2000), Alexander (2005), Balfour (2007) and Jordaan (2011) are provided. The aim of the study was to determine the effect of facilitation on Grade R performance scores in E-L2 learning in rural schools in Mpumalanga. Teachers’ first language, teachers’ qualifications, learners’ first language, learners’ gender, teachers’ age and teachers’ experience on Grade R learners’ performance scores were also tested for interaction effects. Research was conducted in ten randomly selected schools, equally divided between the play-based and formal instruction approaches, and five different languages used in the province. There were 175 Grade R learners and ten teachers in the study sample. The English Language Proficiency standards assessment tool (ELP) was used to collect data and is reported to have no cultural bias. A quantitative methodology was followed, using a two-group comparison design. Participants were matched according to learners’ age, similar exposure period to E-L2 learning, similar rural upbringing, culture, poverty level and mainstream learners. A one-way and two-way ANOVA was used to analyse the data. It was found that the formal approach contributes to better E-L2 learner scores when compared to the play-based approach. The differences in participants’ performance scores were mostly observed in the listening scores and not so much in the speaking scores. IsiNdebele speaking teachers and younger qualified teachers, who were better trained, achieved better learner E-L2 scores than other teachers. Although learners in formal instruction classrooms achieved better results, they still did not achieve competency in basic interpersonal communicative skills in English after three months in Grade R. There was no interaction effect between the gender of the learners and the facilitation approach that was used. A hybrid model, i.e. combination of the play-based and the formal instructional approach is proposed to be implemented in rural Grade R classrooms in Mpumalanga. Further recommendations are to train teachers in educational linguistics, teach listening skills to learners and inform parents about the benefits of first language proficiency before a second language is acquired. / Thesis (DPhil)--University of Pretoria, 2013. / gm2014 / Speech-Language Pathology and Audiology / Unrestricted