Language of instruction transitions: factors that predict multilingual literacy acquisition and cross-linguistic skills transfer from four countries

Group Submission Type: Formal Panel Session

Proposal

Recent reviews of evidence on learning in low - and middle-income countries (LMICs) have underscored that instruction in a child’s first language (L1), or most familiar language, is one of the most significant inputs for consistent learning gains across a variety of programs and contexts (Evans & Acosta, 2020; Nag, et al., 2019). Despite this evidence, an estimated 37% of learners in LMICs are still not taught in the language they use and understand (World Bank, 2021).

While many LMICs now offer some instruction in learners’ L1 in the early grades, they almost universally require an eventual shift in the Language of Instruction (LOI) to a second language (L2) at some point school (often between grades 3-6) (Trudell, 2016). We refer to this shift in LOI and the multilingual curricular sequence through which it occurs as the LoI transition. While considerable research and policy implementation efforts have focused on expanding home language instruction, much less is understood about how education systems should effectively manage LOI transitions.

Successful LOI transition requires careful attention to a number of learning and contextual factors, including fostering learners’ L2 language skills while teaching foundational reading skills and content knowledge in L1. The success of this skill acquisition relates to a complex array of inputs, conditions, and processes.

In many LMICs, certain school and teacher characteristics have not been adequately studied, such as classroom language environment and teacher attitudes and competencies related to LOI. Additionally, there are few studies that have examined the cross-linguistic pathways through which skills are transferred or shared across languages to support literacy acquisition in the multilingual education schooling in these contexts. Without this evidence, it is difficult for decision makers to prioritize resources, design curricula, and address the elements of effective policy design and implementation as the system seeks to improve on the path towards bilingualism, biliteracy, and effective LOI transitions.

This study is sponsored by USAID as part of the SHARE activity (Sustaining Holistic and Actionable Research in Education), led by the University of Notre Dame with a set of local research partners. It seeks to address this lacuna in research by employing quantitative methods and examining the relationship between classroom conditions and student language and literacy skill development as well as the internal cognitive processes and pathways of skills transfer across languages for reading comprehension across four countries. This study focuses on two areas: 1) classroom and teacher factors tailored for multilingual literacy acquisition, specifically: teacher competency and attitudes, classroom language environment and usage, and availability and quality of teaching and learning materials; and (2) the cognitive and linguistic processes underpinning successful multilingual literacy acquisition.

The study seeks to answer the following research questions:
How do the following key classroom/teacher factors predict multilingual literacy outcomes? What is the relative strength of the relationships?
Teacher knowledge (language proficiency)
Teacher attitudes
Teacher practices (pedagogical, linguistic)
Classroom language usage patterns
Availability, quality, and use of TLMs
How do different language and reading sub-skills predict reading comprehension (within and across languages)? What are the pathways through which these variables predict RC?
How are the pathways shaping L1 and L2 decoding and reading comprehension influenced by linguistic characteristics (e.g. orthographic depth, L1/L2 distance, degree of student’s multilingual proficiency)?
The panel will present results to these questions from four countries where this research has been conducted: Kenya, Mozambique, Rwanda, and Senegal.
In each of the 4 study countries, survey data was collected in a stratified random sample of 60 schools. With the exception of Rwanda, data were collected in two regions in at least 2 local languages. Data collection took place over a period of 2 to 3 weeks (depending on the country) in May and June, 2023. 10 students were selected randomly and assessed/surveyed from both 3rd and 4th grade classrooms (20 per school) for a total of 1,200 students in each country (600 boys and 600 girls), 4,800 total across the study. In addition, one Grade 3 and one Grade 4 teacher (from the same class as the assessed students) was interviewed and their class (either a literacy lesson or L2 language lesson) observed in each school for a total of 120 interviews and classroom observations in each country, 480 across the study.

The data collection instruments include: 1) a student survey, which includes Early Grade Reading Assessment (EGRA) tasks, measures of students language skills, and student background data, 2) a teacher survey, and 3) a classroom observation protocol. All students were assessed in two languages, a local language of instruction and a target second language of instruction. In Kenya, Kiswahili, a national language, was treated as the first language of instruction in urban or peri-urban schools.

The teacher survey focused on teacher background, pedagogical competence in multilingual teaching and reading instruction, teacher reading comprehension in the languages assessed, and attitudes towards multilingual teaching. Classroom observations examined pedagogical practices, language usage in the classroom, and the availability and use of teaching and learning materials.

Analysis:
Descriptive statistics will be generated on all major variables of interest and key results shared.
The first research question on classroom/teacher factors that predict students’ multilingual and multiliteracy outcomes will utilize multi-level regression analysis.
The second research question on the sub-skills that predict reading comprehension within and across languages will use structural equation modeling to explore the pathways of cross-linguistic skills transfer.
The third research question will utilize moderation analysis to examine how the phonological proximity and orthographic proximity between different L1s and L2s moderate the relationships between L1 and L2 decoding.

Findings:
Data collection has been completed in each of the country cases and cleaning and analysis is currently underway and will be completed by the dates of the CIES conference.
This robust, multi-country study on a topic of high policy relevance promises to make a critical contribution to the fields of language in education policy and planning and foundational skill development in LMICs, with major contributions to inform policy and practice.

Sub Unit

• Language Issues SIG

Organizer

• Annie Conaghan, University of Notre Dame-Pulte Institute of Global Development

Chair

• DAMALIE KAJUMBA NSANGI, Makerere University, Resilient Africa Network

Individual Presentations

• Lessons and key descriptive findings from a cross-national study on language of instruction transitions. - Anthony Ssebagereka, ResilientAfrica Network Makerere University School of Public Health; Pooja Nakamura, American Institutes for Research; TJ D'Agostino, University of Notre Dame

• Key teacher and classroom factors that predict students’ language and literacy skills in multilingual contexts - findings from a multi-country study - Leih Maruss Sinsay-Villanueva, Philippine Institute for Development Studies

• Patterns of cross-linguistic transfer of language and reading skills in multilingual contexts - findings from a multi-country study - Evans Wafula Mahaya, University of Nairobi

• Exploring how linguistic characteristics affect reading skill development and moderate cross-linguistic skills transfer in multilingual contexts - Carlos Joaquim Manuel, Universidade Eduardo Mondlane

Discussants

• TJ D'Agostino, University of Notre Dame

• Pooja Nakamura, American Institutes for Research

• Mbarou Gassama-Mbaye, ARED

• Michael Tusiime, University of Rwanda

• Andre Kone, University of Arts and Human Sciences of Bamako