Public Disclosure Authorized   Public Disclosure Authorized                            Public Disclosure Authorized   Public Disclosure Authorized

                                                                                                                                            A World Bank Policy Approach Paper




                                            Effective Language of
                                       Instruction Policies For Learning
                                                                     LOUD AND CLEAR:
© 2021 International Bank for Reconstruction and Development / The World Bank

1818 H Street NW
Washington DC 20433
Telephone: 202-473-1000
Internet: www.worldbank.org

This work is a product of the staff of The World Bank with external contributions. The findings, interpretations, and
conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive
Directors, or the governments they represent.

The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors,
denominations, and other information shown on any map in this work do not imply any judgment on the part of
The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries.

The paper was designed by Nicole Hamam.




Rights and Permissions

The material in this work is subject to copyright. Because The World Bank encourages dissemination of its knowledge,
this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work
is given.

Any queries on rights and licenses, including subsidiary rights, should be addressed to World Bank Publications,
The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights@
worldbank.org.
LOUD AND CLEAR:
Effective Language of Instruction
Policies For Learning
A World Bank Policy Approach Paper
4   LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
TABLE OF CONTENTS




Acknowledgements .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 6


Abbreviations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7


Executive Summary .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 8


Part 1: Why Should we Care about this Problem and What Characterizes it?.  .  .  .  .  .  .  .  .  .  .                                                                        13
     Section 1: Why should we care? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
     Section 2: How big is the problem? .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   . 20
     Section 3: Political, economic, and sociological considerations influence language choices. .  .  .                                                                       24
     Section 4: Language of instruction policy should be multifaceted .  .  .  .  .  .  .  .  .  .  .  .  .  .  .                                                              29


Part 2: Promoting Solutions through More Active Engagement.  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 33
     Section 5: Effective LoI policy examples exist but are not being widely implemented .  .  .  .  .  .                                                                      34
     Section 6: A new approach to language of instruction for the World Bank .  .  .  .  .  .  .  .  .  .  .  . 52


Annexes
     Annex A – Ethiopia: Multilingual L1-Based Education .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .  58
     Annex B – Rwanda: A Changing Education System for a Changing Time .  .  .  .  .  .  .  .  .  .  .  .                                                                      63
     Annex C – Hong Kong: Language of Instruction in Secondary Schools .  .  .  .  .  .  .  .  .  .  .  .  .                                                                   68
     Annex D – Middle East and North Africa: Mastering Arabic & the Transition from Colloquial to
     Formal .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   . 74
     Annex E – Peru: Bilingual Intercultural Education .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .  77
     Annex F – India: The Language & Learning Foundation, Transforming Teaching.   .   .   .   .   .   .   .   . 79


References .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 86


Endnotes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95




TABLE OF CONTENTS                                                                                                                                                                      5
    ACKNOWLEDGEMENTS




    This paper was produced by a team led by Michael Crawford and Sergio Venegas Marin,

    under the guidance of Jaime Saavedra and Omar Arias, and including Penelope Bender,

    Barbara Trudell, Dhir Jhingran, Huma Kidwai, Elaine Ding, Laura Gregory, Lisha Almeida,

    Amberine Huda, Aishwarya Khurana, and Tihtina Zenebe Gebre. The paper was edited

    by John Steinhardt. The team benefited from comments from Reema Nayar, Jason Allen

    Weaver, Adelle Pushparatnam, Samer Al-Samarrai, Lianqin Wang, Tara Beteille, Toby

    Linden, Harry Patrinos, Hanna Katriina Alasuutari, Juan Manuel Moreno, Marguerite

    Clarke, Melissa Adelman and other members of the Education Global Practice who

    participated in discussions on the paper. The paper was art directed and designed by

    Nicole Hamam with illustrations by Margaret Flatley.




6                                              LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
ABBREVIATIONS




                                                          LIST OF ACRONYMS
     CC BY         Creative commons by [attribution]; meaning open access
      ECD          Early Childhood Development
     ECEC          Early Childhood Education and Care
    EGMA           Early Grade Math Assessment
     EGRA          Early Grade Reading Assessment
      FCV          Fragility, Conflict, and Violence
      GBA          Global Book Alliance
      GRN          Global Reading Network
      HCI          Human Capital Index
      HCP          Human Capital Project
      ILSA         International Large-Scale Assessment
       L1          Language in which children are proficient when they start school*
       L2          Language in which children are not proficient when they start school **
     LAYS          Learning-adjusted Years of Schooling
      LLF          Language and Learning Foundation (India)
     LMIC          Low- and Middle-Income Countries
       LoI         Language of Instruction
      LPP          Literacy Policy Package
      LWC          Language of Wider Communication
     MSA           Modern Standard Arabic
  MTB-MLE          Mother Tongue-based Multilingual Education
     MWL           Minority Written Language
      NYC          New York City
    PASEC          Programme d’Analyse des Systèmes Educatifs de la CONFEMEN
     PIRLS         Progress in International Reading Literacy Study
     PISA          Program for International Student Assessment
     SDGs          Sustainable Development Goals
      SES          Socioeconomic Status
    TIMSS          Trends in International Mathematics and Science Study
      TLM          Teaching and Learning Materials
     WBG           World Bank Group

*L1 is typically defined as children’s first language. In this paper, it is used to indicate a language that children speak and understand well when they
start school. It may or may not be the first language they learned.
**L2 is typically defined as children’s second language, which they often learn at school. In this paper, it is used to indicate a language that children do
not speak and understand well when they start school. It may or may not be their second language.
These definitions are acutely relevant to multilingual societies or regions with many small communities that speak different languages but use a single
regional language as a means of broader communication across communities.


LIST OF ACRONYMS                                                                                                                                               7
    Executive Summary




    HUMAN CAPITAL ACCUMULATION IS LARGELY A LANGUAGE-BASED ENDEAVOR. It is the basis
    of wealth in modern societies and is primarily acquired through schooling. Policies that maximize
    the amount and quality of human capital have a myriad of positive effects. To maximize it, countries
    must maximize learning in school, and in so doing raise the number of Learning Adjusted Years of
    Schooling (LAYS) completed by their population. The World Bank created the Human Capital Project
    and the Human Capital Index to bring policy attention to the need for countries to invest in their
    people and for people to invest in themselves. Policies that promote human capital accumulation—
    such as good language of instruction policies—should be actively championed by the World Bank and
    central to country dialogues on helping people achieve their full potential.

    SHOCKINGLY LOW LEARNING OUTCOMES MAY BE A REFLECTION OF INADEQUATE LANGUAGE
    OF INSTRUCTION POLICIES. The LAYS component of the Human Capital Index and measures of
    Learning Poverty show shockingly low learning levels. In some countries, children with three years
    of primary education cannot identify a single written word, and may only know one letter.1 Without
    consideration of language of instruction (LoI) issues, one might erroneously conclude that teachers
    lack the knowledge and skills to teach, or that students are too disadvantaged to learn. An alternative,
    plausible consideration is that teachers are required to provide instruction in a language that students
    do not speak or understand. The low test scores in some instances simply reflect this near-total lack of
    understanding of the language used for teaching and/or testing; they do not indicate any inability to
    learn under the right learning conditions.

    APPROPRIATE LANGUAGE OF INSTRUCTION POLICIES FACILITATE LEARNING AND MORE. When
    children are first taught in a language that they speak and understand wellthey learn more, are better
    placed to learn other languages, are more likely to stay in school, and enjoy a school experience appro-
    priate to their culture and local circumstances. A diverse and substantial body of research, cited
    throughout this paper, attests to this fact. Appropriate LoI policies also promote equity in schools
    and in labor markets, improve the cost-effectiveness of education, and promote inclusiveness. Good
    LoI policies should be a cornerstone of effective national human capital development strategies, and
    therefore of acute concern to national policy makers and development partners.

    GOOD LANGUAGE OF INSTRUCTION POLICIES REMAIN THE EXCEPTION, NOT THE RULE. Despite their
    many benefits, many education systems do not implement appropriate LoI policies. Instead they often
    require children to learn in languages they do not know well—and, in far too many cases, in languages they
    do not know at all. The children who are impacted by these policies are often those who are disadvantaged
    in other ways, such as socioeconomic status and distance from urban centers, that also make learning and
    progress at school challenging. Requiring teachers to provide instruction in languages that neither they nor
    the students speak is a common practice despite the evidence that 90 percent or more of students may fail
    to acquire foundational skills such as basic literacy and numeracy in schools that implement these policies.2




8                                                         LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
POOR LOI POLICIES HARM LEARNING, ACCESS, EQUITY, COST-EFFECTIVENESS, AND INCLUSION.
A substantial and growing body of research shows that children learn better in their first language (L1)
than in a second language (L2). When taught in their L1 first, they are more likely over time to become
proficient in an L2 and comfortably absorb academic content. They are also more likely to remain in
school. Children from households in the bottom 40 percent of the socioeconomic distribution are
more likely to endure instruction in a language they do not understand, and lack the family resources
to mitigate the effects of inappropriate LoI policies in the schools they attend.3 Countries pursuing LoI
policies that promote the use of languages neither spoken nor understood by teachers and students see
a disappointing return on their investment in education. These policies contribute to higher dropout
rates, repetition rates, and lower learning overall. In this sense, countries would gain financially if, by
adopting better LoI policies, they were to lower the cost of each graduate produced and each unit of
learning mastered. That would also make learning relevant to and inclusive of students in many ways,
not least through the inherent value and validation a language receives when it is used as an official
medium of instruction.

INAPPROPRIATE LOI POLICIES AFFECT AN ESTIMATED 37 PERCENT OF STUDENTS IN LOW- AND
MIDDLE-INCOME COUNTRIES. UNESCO (2016) estimates that globally four in every ten students
are taught in languages they do not know. Evidence from original analyses undertaken for this paper
largely confirm the figure and attest to the enormity of the problem: 37 percent of students in low- and
middle-income countries are not being taught in the language they speak and understand best. In
some countries, more than 90 percent of students are not taught in a language they speak and under-
stand. More than a quarter of a billion students are affected. Based on data from Ethnologue,4 12 of the
20 countries with the highest rates of learning poverty use instructional languages that few of their
students understand when they come to school. The 53 percent of children in low- and middle-in-
come countries (LMIC) who fail to read with understanding by age ten5 almost certainly includes a
large share of students who are not taught in their L1.6,7




             37 percent of students in low- and middle-income countries are
           not being taught in the language they speak and understand best.
           100

           90              87%             87%
                  80%
           80

           70
                                                   67%
                                                                                                             58%
PERCENT%




           60
                                                                                51%
           50

           40                                                                                                                                          37%
                                                                                                     30%
           30
                                                                                                                                                                 21%
           20
                                                                                                                                 13%      13%
           10                                                           6%
            0
                 Sub-Saharan Africa      Middle East and                Latin America                 South Asia                  Eastern and        East Asia and Pacific
                                          North Africa                   & Caribbean                                             Central Europe



                                      Share of children taught in a language they don't understand           Share of children in learning poverty




EXECUTIVE SUMMARY                                                                                                                                                            9
     MASSIVE PROGRESS IS FEASIBLE BY TEACHING IN A SMALL NUMBER OF ADDITIONAL
     LANGUAGES. The prospect of offering instruction in each of the world’s more than 7,000 languages8
     is daunting, but that sentiment is misguided. Policy dialogue often labors under the assumption that
     embracing mother-tongue-based multilingual education (MTB-MLE) is expensive and onerous.
     In fact, MTB-MLE is cost-effective and simpler to organize than usually thought. Indeed, the vast
     majority of students currently learning in an L2 would benefit from a rather modest global expansion
     of the number of languages used for instruction. Three-quarters (75 percent) of the problem could
     be attenuated by offering instruction in an additional 220 languages worldwide—on average about
     one new language per country. To reach 84 percent of all language minority students in all countries
     requires teaching in only about 559 languages—fewer than three additional languages per country
     on average. In both scenarios, the would-be additional languages are already “written languages.”
     They have established orthographies, and each is spoken by at least 1.5 million people. The remaining
     roughly 6,200 languages of the world are spoken by fewer than 10 percent of the world’s people. Five-
     sixths of the problem can be solved with concentrated and focused effort; solving the final one-sixth
     is then likely to require exponentially increasing time and resources.

     LOI POLICIES ARE INFLUENCED BY LARGER POLITICAL AND ECONOMIC CONSIDERATIONS. In
     light of this evidence, why are so many countries still choosing a LoI that results in significantly
     worse learning outcomes and social exclusion? One reason may be lack of knowledge of the bene-
     fits of teaching the language that students best speak and understand. Other factors are arguably less
     tractable. Language choices for education are often the result of political considerations beyond the
     education sector. Language is closely tied to national identity and political identity. Nation-building
     sometimes involves the promotion of one group’s language at the expense of the languages of other
     groups. Language proficiency has labor market value. Parents often view competence in metropolitan
     or international languages as central to their children’s career success. The ends—the goals—such as
     proficiency in an international L2, can dictate the means in a manifestly counterproductive manner
     when all instruction is in that L2 regardless of teachers’ or students’ levels of proficiency. Consideration
     of the broader political context is essential to effective policy engagement on LoI issues.

     POLICY SOLUTIONS EXIST BUT ARE NOT BEING SYSTEMATICALLY IMPLEMENTED. Advocacy for
     good LoI policies has been ongoing for decades but has seldom held the attention of key decision
     makers and development partners. A trend toward more and better research is thankfully emerging
     in some regions.9 Experience has been summarized and analyses have distilled sound policy advice
     for the issue as a whole and for virtually all relevant sub-issues. Sound advice is available for coun-
     try-level situation analyses and for planning for language use in education. Guidelines for and examples
     of good scopes and sequences for L1 instruction have proliferated. Methods for the development of
     teaching and learning materials, including software to create textbooks and storybooks for children,
     are freely available and being used more frequently than in the past. Policy options with respect to a
     range of issues affecting teachers, teaching, and training are growing. Knowledge about and concern for
     language of assessment issues is increasing. Empirical evidence on the effectiveness of “late-exit” (see
     below) and other effective curricular and pedagogical strategies is available. However, progress is too
     slow and available tools are not sufficiently known or understood.

     THIS PAPER DESCRIBES THE NEW WORLD BANK POLICY APPROACH ON LOI, AS PART OF THE
     OPERATIONALIZATION OF THE LITERACY POLICY PACKAGE IN SUPPORT OF THE BANK’S NEW
     LEARNING TARGET. This new approach aims to support progress on LoI policies and interventions,
     as LoI-based challenges are identified and addressed in light of each country’s context. Under the new
     approach, the World Bank considers it critical to observe the following principles to enhance learning
     in low- and middle-income countries:




10                                                        LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
 The new World Bank policy approach on LoI, as part of the operationalization of the Literacy Policy
        Package in support of the Bank’s new Learning Target, is based around 5 principles:

   PRINCIPLE 1              PRINCIPLE 2            PRINCIPLE 3            PRINCIPLE 4            PRINCIPLE 5




    Teach children           Use a language          Introduce an       Continue using the     Continuously plan,
     in a language        children understand    additional language     language children        develop, adapt,
   they understand          for instruction in     (if desired) as a       understand for       and improve the
     starting with         academic subjects      foreign language        instruction even       implementation
   Early Childhood           beyond reading      with a focus on oral      after a foreign      of LoI policies, in
 Education and Care           and writing.         language skills.     language becomes        line with country
   (ECEC) services         Students need to          Students can         the principal LoI.   contexts and goals.
 through at least the       master reading             master two          L1 instruction
   first six years of       and writing in a          languages in          continues to
  primary schooling.         broad range of        basic education           improve L2
    It is critical that    disciplines and in     if instruction and      performance in
   instruction be in      all school subjects.     sequencing are         important ways
 the language most                                     optimized.        even after the L2
 students speak and                                                      becomes the LoI.
  understand best.




To promote and operationalize this approach, the World Bank proposes three strands of work to ensure
the principles are routinely integrated within World Bank operations. Strand 1 focuses on analyzing and
assessing the LoI situation in each client country as well as the main policy options available. Strand 2
involves more proactive engagement with LoI issues through enhanced dialogue and planning to help
countries address priority issues. As part of Strand 3, the Bank’s Education Global Practice would work
with other partners to advocate for appropriate attention to LoI issues and in creating, collecting, and
disseminating the full range of cutting-edge knowledge and policy solutions and promoting long-term
solutions for language issues in education.

THE PURPOSE, AUDIENCE, SCOPE, AND STRUCTURE OF THIS PAPER. Its purpose is to ensure that
the WB Education Global Practice can help clients mitigate or eliminate the serious problems that
inappropriate LoI policies impose on learning and schooling. The paper is intentionally short; it is
not an exhaustive academic paper. Sufficient detail is presented to begin to explain issues in light of
current policy decisions, not to treat the subject in its entirety. The paper is selective and does not look
at LoI in all aspects of education policy. It does not consider the role of LoI in tertiary education, tech-
nical and vocational education and training, adult education, or lifelong learning. The paper recom-
mends endorsement of the new approach by WBG management. Embracing its recommendations
involves commitment to a range of actions that will support WBG client countries, including more
in-depth analyses, development and dissemination of global public goods, improved education staff
knowledge and capacity and more effective dialogue, especially in basic education. Readers should
bear in mind that:
   • We do not yet have all the answers. As we move forward, key knowledge gaps will need to be
     addressed and closed. We will need, for example, to better understand and develop recommen-
     dations to improve second-best situations, such as those that involve the use of one-way L2 im-
     mersion. Likewise, the compounding effects of poor teacher proficiency in the target language in
     combination with low student proficiency need to be disentangled. The effects of time-on-task,



EXECUTIVE SUMMARY                                                                                                     11
     the quality of teaching and learning materials, and teacher support and other issues constitute a
     future agenda which warrants urgent attention if policy advice is to be optimized.
   • Leadership requires sustained commitment. LoI issues can be managed but never fully solved.
     Policy efforts by countries should be a routine part of overall education policy. Sustained com-
     mitment on the part of development partners, with the Bank seeking to set an example as a key
     partner, should likewise be part of overall efforts to improve learning.
Part 1 addresses why we should care about LoI issues and the major challenges involved. Its four sections
are entitled: (i) why should we care? (ii) how big is the problem? (iii) the role of political economy;
and (iv) diverse LoI contexts. Part 2 presents existing solutions (in section 5) and proposes a detailed
way forward for the WB Education Global Practice (section 6). It should be noted that the paper does
not claim to possess or propose a complete set of technical solutions for the myriad of difficult policy
issues involved. By enhancing engagement and devoting adequate resources to the problem, existing
solutions will be deployed, and new solutions devised. Increased partnership and knowledge sharing
will be part of this, as will be the testing of innovative approaches. The new approach will involve
learning at the individual and institutional level, with an intensity of engagement commensurate with
the urgency of the issue.
                                  PART 1:
                                  Why should we care
                                  about this problem and
                                  what characterizes it?




SECTION 1 : WHY SHOULD WE CARE?                            13
     SECTION 1:
     Why should we care?




     HUMAN CAPITAL ACCUMULATION IS THE KEY TO DEVELOPMENT. The World Bank understands
     human development as the very center of economic and social development. Decades of research
     have confirmed that education and health are the keys to poverty reduction and the promotion of
     prosperity, for individuals and for societies. Creating conditions whereby people can invest in them-
     selves, and governments and families can invest in children’s futures, is a pillar of World Bank policy.
     The Human Capital Project and the Human Capital Index create a systematic approach to promote
     investments in people, and understanding of what works in this domain. The Learning Target—the
     goal of halving Learning Poverty—brings focus to the key role of foundational learning in facilitating
     lifelong investment of people in themselves and governments in people. Education and learning are
     the main investments. The themes and activities of the Bank’s Education Global Practice are chosen
     and organized to maximize human capital accumulation for human development.


         THE GLOBAL LEARNING CRISIS IMPEDES HUMAN CAPITAL ACCUMULATION AND PROS-
          PERITY. The WDR 2018: Learning to Realize Education’s Promise, reports that millions of chil-
           dren are struggling to learn in school.10 Near-universal enrollment in primary education has
            not led to near-universal learning.11 Projections show that, at current rates of improvement,
              the fourth sustainable development goal (SDG 4) will not be achieved.12 Fifty-three percent
                of children in low- and middle-income countries are trapped in Learning Poverty: unable
                 to read and understand a simple text by age 10.13


                      LANGUAGE IS ESSENTIAL TO LEARNING, SCHOOLING, AND HUMAN CAPITAL
                      ACCUMULATION. Instruction unfolds through language—written and spoken—in
                       foundational learning and core academic subjects. Almost all schooling is inti-
                        mately tied to language. Teachers carefully select words and themes in ways to
                         unlock or transmit the meaning of the words and the concepts that underlie them
                           for students. Some school subjects, like physical education, art, and music, may
                             draw less on verbal tools, but students cannot participate if they do not under-
                               stand the language used to support instruction. In some school settings,
                                 reading, writing, listening, and speaking with understanding and compe-
                                   tence amount to the main or only requirement for students. A first
                                     essential task of school is to have children gain command of written



14                                                      LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
language, through reading and writing. Those who fail to do this by age 10 are unlikely to succeed in
school without intensive remediation. Choosing the right LoI is an exceptionally important under-
taking. The right choice is the language that allows students to successfully begin their school careers.

GOOD LOI POLICIES MANDATE THE USE OF STUDENTS’ L1 FIRST. The essence of good LoI policy
is to provide instruction in a language that students best speak and understand—their L1—from their
participation in ECEC services through roughly their first eight years of school. A second language
can be taught too, as a foreign language at first. Gradually, if students master the basics of literacy in
their L1, a second language can become the medium of instruction, with L1 learning continuing in
some form through 12th grade. Voluminous research supports this practice.14 Providing instruction
in this way allows students to link the new concepts they learn in school to words they already know
and thereby affirms a bedrock principle of cognitive science: that new knowledge builds on existing


     CHILDREN SHOULD BE TAUGHT IN A LANGUAGE THEY UNDERSTAND.
                    When children learn in a language they understand,
                       children have a greater chance of learning:




                                           OTHER SCHOOL
                                         SUBJECTS LIKE MATH
                                            AND SCIENCE.
             THE LANGUAGE
              THEY SPEAK
               AT HOME.                                                      AS WELL
                                                                            AS OTHER
                                                                           LANGUAGES.




            While promoting the development of general cognitive abilities.

knowledge. Conversely, when children are expected to acquire language abilities and simultaneously
master new concepts via those abilities, they struggle. Using an L1 for foundational learning is consis-
tent with the science of learning. It promotes the variety of benefits discussed below. When children
are sequentially taught two languages through appropriate, high-quality LoI policies, they achieve
higher aggregate learning in both languages.15

LEARNING FIRST IN AN L1 IMPROVES FOUR TYPES OF LEARNING OUTCOMES. Literacy (speaking,
listening, reading, and writing) is typically the dominant subject and key learning task in primary
school. Becoming literate in an L1 at the start of ECEC and primary school promotes the improvement
of three types of learning outcomes and general cognitive abilities.
   • It promotes better learning outcomes in the L1. Box 1 summarizes evidence from numerous
     studies from a range of regions, contexts, ages, languages, income levels of countries, and lan-



SECTION 1 : WHY SHOULD WE CARE?                                                                              15
             Wealth of Evidence for a Variety of Positive Outcomes from Teaching in L1
     BOX 1: A
     Impact on Literacy Skills: Evidence suggests that children who are taught in a language that is not their
     L1 perform significantly worse than their peers. A study across 48 countries found that students who
     did not speak the language of the test before starting school had much lower average achievement
     on PIRLS 2011 (479 versus 516). In South Africa, SACMEC data from 2005 shows an even larger
     difference (37 percentage points between students who use their L1 versus those studying in an
     L2). Other studies show similar results across a diverse set of countries. However, mother-tongue
     based multilingual programs (MTB-MLE), which start in an L1 and later introduce an L2, can remedy
     these gaps. In Mali’s pédagogie convergente program, children who began their schooling in their L1
     scored 32% higher on French proficiency tests at the end of primary school than those in French-only
     programs. Similarly, a recent study of Somali MTB-MLE in Swedish schools found that participation in
     MTB education contributed positively to participants’ results on Somali reading comprehension, and
     these results were associated with higher results in Swedish reading comprehension.

     Impact on Mathematics: MTB-MLE instruction has positive effects on student performance across
     curricular areas, including mathematics. An experiment in Cameroon that provided a treatment
     group of schools with MTB-MLE instruction during the first three years of schooling found that these
     students exhibited gains in both English and mathematics of 1.1 to 1.4 of a standard deviation in grades
     1 and 3 as compared to control students. A study in Peru found that indigenous Quechua children
     who attended Quechua-medium schools scored 0.54 standard deviations higher in mathematics
     than indigenous children who attended Spanish-medium schools, even after controlling for variables
     such as school resources. By contrast, evidence from TIMMS 2007 shows that the mathematics
     scores of students who ‘never’ (368) or ‘sometimes’ (438) spoke the language of the test at home
     were lower than those who ‘always or almost always’ spoke the language at home (483).

     Impact on other Educational Outcomes: Beyond academic achievement, L1 instruction has effects on
     other aspects of schooling, such as attendance and overall attainment. A study of a 1994 Ethiopian
     policy change that introduced mother-tongue instruction to the largest ethnic group in the country
     found that the policy led to an increase of 0.75 to one year of primary schooling in the affected cohort,
     and increased the percentage of students completing six or more years of schooling by 31 percent.
     Similarly, a study of 23 countries representing 153 linguistic groups found that when MTB instruction
     is available at half or more of the schools frequented by members of a linguistic group, the proportion
     of out-of-school children is 10 percent lower than in groups for which MTB instruction is not available.

     Sources: Mullis, Martin, Foy, Drucker 2012; Bühmann and Trudell, 2008; Ganuza & Hedman, 2019; Hynsjö & Damon, 2015; Martin, Mullis and
     Foy 2008; Ramachandran 2012; Smits, Huisman & Kruijff, 2008




         guage characteristics. Evidence is from randomized control trials (RCTs), international large-
         scale assessments (ILSAs), and national learning student assessments (NLSAs), and published
         project reports, among other sources. Findings consistently show large effects, with effect sizes
         of one-third to one-half a standard deviation being common. Brown’s (2011) findings were typi-
         cal: students in Uganda taught in their L1 identified 20 letters per minute and read seven words
         per minute versus six letters and one word per minute respectively for the students in the con-
         trol group taught in an L2. Other studies show similar massive and consequential differences in
         learning outcomes.
      • It promotes higher learning outcomes in a subsequent L2. Students who become proficient first
        in their L1 have a greater ability to subsequently learn an L2, especially when the L2 is introduced


16                                                                      LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
     at the right time. Goldenberg’s (2008) meta-analysis of evi-
     dence from high-income countries is matched by growing
     numbers of studies from low- and middle-income coun-
     tries; in South Africa, Taylor and von Fintel (2016) con-
                                                                             L1 instruction is the
     sidered more than 9,000 primary schools and 800,000                 ultimate student-centered
     student records. They found that “MTB-MLE instruc-                  pedagogical practice, as it
     tion in the early grades significantly improves English
     as measured in grades 4, 5, and 6.”                                strengthens teacher-student
   • It promotes learning in other academic subjects—                     interactions and it allows
     such as mathematics, science and history/social stud-                   teachers to build on
     ies. Seid’s 2019 findings from the Young Lives Ethiopia
     School Survey found that “Learning in mother tongue                        students' prior
     first (in grades 1-4) increases students’ mathematics and                    knowledge.
     literacy test scores later after they transition to English in-
     struction (in grade 5) by 0.269 and 0.089 standard deviations
     respectively”; (cited in Evans and Acosta 2020).
   • It promotes the development of general cognitive abilities. Students
     who are taught in their L1 until at least grade 5 (around age 10–11) develop
     their cognitive abilities to a threshold that is not generally reached by students
     who begin school in an L2. Multiple studies conducted by Collier and Thomas from 1997 to 2014
     and replicated by other researchers (especially Lindholm and Leary, 2004) provide consistent
     evidence for this effect. Trudell and Piper (2014) provide evidence for this effect from developing
     country settings.

GOOD LOI POLICIES PROMOTE ACCESS AND RETENTION. A growing body of evidence shows that
children who are taught in their L1 first stay in school longer than those taught in an L2. Laitin et al.
(2019) found that students in Cameroon taught in their L1 were 22 percent more likely to be in school in
grade 3, and 14 percent more likely to still be in school in grade 5. Seid (2017) found MTB-MLE in Ethi-
opia increases the likelihood of enrollment in primary school and of attending the right “grade for age.”
Ramachandran (2017) similarly found L1 instruction in the early grades leads to an additional half year of
completed schooling and a five percent increase in the chance of finishing primary school. Positive impact
on enrollment and retention is particularly important, as an estimated 54 million out-of-school children
(or 72 percent of the total out-of-school population) live in countries with high linguistic diversity.16

GOOD LOI POLICIES PROMOTE EQUITY AND INCLUSION. USAID’s “Planning for Language Use
in Education” handbook cites multiple research studies confirming that (a) language of instruction
issues negatively affect students from lower socioeconomic status backgrounds more than their high-
er-income peers; and (b) gender equity is promoted by initial L1 education, as its positive effects in the
classroom are at least as likely to benefit girls and help overcome some of the bias-related educational
challenges they face (USAID 2015). Good LoI policies also allow parents in difficult socioeconomic
circumstances to participate more fully in their children's schooling.

L1 TEACHING ALLOWS FOR USE OF EFFECTIVE STUDENT-CENTERED PEDAGOGY. L1 instruction
may be the ultimate student-centered pedagogical practice. The use of L1 strengthens the student–
teacher interactions which are responsible for a large share of student learning17 and it allows teachers
to build on students‘ prior knowledge—language skills and background knowledge acquired outside
school—as the foundation for new concepts. The sound-based symbols of written language are arbi-
trary (see Sapolsky 2013) in whatever language they are found—the symbol “B” for instance has no
inherent connection to the first sound in the word “basket”. Students must learn to match abstract



SECTION 1 : WHY SHOULD WE CARE?                                                                              17
     symbols to the sounds of words they know. After some reasonable repetition, most students fully
     master these alphabetic tasks. If children are asked to map symbols to sounds in words they do not
     know (and therefore may not be able to disentangle phonemically, or even hear), this becomes the
     ultimate exercise in arbitrary memorization. Indeed, there are few purely pictographic languages, but
     where sounds do not correlate to symbols, reading progress is slower. Countries have invented sound-
     based alternatives, such as pinyin for Chinese, to make the initial stages of learning to read easier. It is
     hardly surprising that in contexts where mapping is merely an exercise in arbitrary memorization, 98
     out of 100 students can barely recognize any written words by the end of third grade. When teachers
     teach in a language that students do not understand, the teaching of basic letter/sound relationships
     is impossible, as is the kind of broader class discussion that might otherwise facilitate it. By contrast,
     when they teach through the medium of a language that their students know, teachers can reduce the
     use of “chalk and talk” instruction that largely consists of copying and memorizing. Instead, then can
     deploy a range of more effective strategies, including question and answer exchanges, wider discus-
     sion, and small group activities.18,19

     GOOD LOI POLICIES PROMOTE EFFICIENT AND COST-EFFECTIVE PROVISION OF EDUCATION.
     The evidence cited above attests indirectly to the efficiency gains and improved cost-effectiveness
     of education using good LoI policies. Other studies have measured costs and confirmed not only that
     MTB-MLE is highly cost-effective, but that its costs are consistently lower than most policy makers
     and other stakeholders believe them to be. In Guatemala, Patrinos and Velez (2009) found cost-sav-
     ings equivalent to USD five million per year from lower dropout and repetition rates in a program
     providing bilingual education to over 100,000 students. Grin (2005, cited in Heugh 2011) analyzed
     several models of provision of MTB-MLE and concluded that there would be: (a) saving in financial
     outlays due to efficiency gains in teacher training and instructional delivery and, (b) long-term bene-
     fits from lower repetition and dropout, plus increased human capital accumulation and use for the
     skills of graduates.

     THE TARGET OF REDUCING LEARNING POVERTY BY HALF IS FOR READING IN L1. The Literacy
     Policy Package that accompanies this World Bank operational target calls for students to be taught
     in the language they best speak and understand. Achieving the Learning Target requires the type of
     acceleration in progress that is associated with L1 instruction. Countries may wish to have graduates
     who can read, write, speak, and listen equally well in two or more languages; teaching in L1 first for
     all of primary school and part of lower secondary school (while providing L2 “as a foreign language”
     instruction) is the most promising way to achieve the needed progress.

     GOOD LOI POLICIES INCREASE THE EFFICIENCY OF OTHER EDUCATION INVESTMENTS. Grin
     (2005) estimated the external and internal efficiency gains likely to result from improved LoI poli-
     cies. Improved efficiency of teacher training contributes to overall internal efficiency gains through
     its potential to lower direct cost and raise effectiveness. Instruction by teachers trained in a language
     they speak and understand is synonymous with effective planning and delivery of instruction. There is
     also the likelihood of complementary improvements when LoI is integrated into an overall education
     strategy—and the danger of failure if it is left out. Interventions and investments that combine appro-
     priate approaches to LoI with strong instructional programs will have higher impact than improve-
     ments in instruction alone. Teacher training and coaching, classroom observation, early childhood
     development, and other programs exert more impact when conducted in languages children know.
     By contrast, investments in activities such as teacher training, classroom observation, and struc-
     tured lesson plans could suffer a comprehensive “O-ring” failure (Kremer 1993) when conducted in a
     language that students and teachers do not know: what would otherwise have succeeded in the right
     language is bound to fail in an L2.



18                                                        LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
   BOX 2: Teaching in L1: A Core Element of the Literacy Policy Package

  The Literacy Policy Package (LPP) identifies five core elements for helping students become literate,
  and start on the path to becoming advanced readers: (i) a political commitment to clear goals for
  making all students literate; (ii) effective instruction by supported teachers; (iii) adequate teaching
  and learning materials, including at least one teacher’s guide per teacher, one textbook per child
  and additional reading books for practice; (iv) instruction in a language that the student speaks and
  understands; and (v) promotion of love of reading.

  The elements of the LPP reinforce one another; each makes the others more effective. Clear goals
  facilitate effective lesson planning and teaching; adequate teaching and learning materials promote
  efficient use of class time and student practice; love of reading drives students to use early foundational
  skills to read with pleasure, and read to learn.

  Using the student’s L1 is the foundation for effective instruction. Instructional goals become more
  attainable, teaching becomes more effective, textbooks and other TLM can be used effectively.
  The WBG’s new approach to LoI issues is consistent with and promotes the implementation of the
  Literacy Policy Package.


CONTINUOUS ENGAGEMENT WITH LOI IS ESSENTIAL TO THE BANK’S OVERALL EDUCATION
APPROACH. The Education Global Practice regards foundational learning as the indispensable
cornerstone of an individual’s educational success, and, by extension, the underpinning of entire
education systems. This represents an evolution from earlier approaches, which paid scant atten-
tion to foundational learning or LoI issues. With task teams having to engage with LoI issues as
a way of improving foundational learning, a need has arisen for an approach to LoI issues that
can help strengthen, sustain, and increase these efforts for greater impact. The Bank’s current
goals—and more importantly the goals of students, parents, and families in client countries—are
best achieved if significant and sustained attention to LoI issues is fully integrated into World Bank
education policy.




SECTION 1 : WHY SHOULD WE CARE?                                                                                 19
     SECTION 2:
     How big is the problem?




     IN LOW- AND MIDDLE-INCOME COUNTRIES, AN ESTIMATED 37 PERCENT OF CHILDREN LEARN IN
     A LANGUAGE OTHER THAN THEIR L1. In some countries, more than 90 percent of children learn in
     a language that is not their L1. UNESCO (2016) states that “globally, close to 40 percent” of children
     are learning in a language that is not their L1. An original estimation done for this paper of the figure
     for low- and middle-income countries (LMICs) finds a very similar figure: 37 percent of all students
     are taught in a language other than their L1. In absolute numbers, this means that of about 1 billion
     students in LMICs, 370 million are not being taught in the language they best speak and understand.
     Unsurprisingly, the figure is more than half of the overall headcount of the roughly 550 million chil-
     dren in Learning Poverty.


     FIGURE 1: Share of Students in Low- and Middle-income countries by Type of Language Spoken




                                            37%                      27% speak MWLs not taught in school
                                      Non-National Language,
                                       Not Used for Teaching
                    62%                                               10% speak other less common languages
                National Languages,
                 Used for Teaching




                                                               1%
                                                       Non-National Languages,
                                                          Used for Teaching




20                                                             LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
MINORITY WRITTEN LANGUAGES: LANGUAGES SPOKEN BY LARGE LINGUISTIC MINORITIES.
Policy makers have expressed concern about the cost and difficulty of providing L1 instruction for
every child, given the great diversity of languages spoken in the world. This may be true for some
small, unwritten languages, where orthographies need to be developed before they can be used for
instruction. However, this dialogue has not been informed by data on the number of languages,
numbers of speakers, and corresponding orthographies. To overcome this, the concept of a “Minority
Written Language” (MWL) was defined as a language that (i) currently has more than 1.5 million
speakers; (ii) is written (has an existing orthography); and (iii) is not the official, national, or dominant
language of the country in question. Since, globally, one in six people is a school-aged student (1.3
billion of the world’s 7.8 billion people), we can expect that any language spoken by 1.5 million people
will be spoken by at least 250,000 school-aged children (or more—given the demographic age profiles
of many LMICs). For each MWL in which instruction is offered, at least one-quarter of a million
students would potentially be reached in their L1.

MANY CHILDREN SPEAK A RELATIVELY SMALL NUMBER OF UNTAUGHT L1S. The figure of 37
percent mentioned above divides into twenty-seven percent (27 percent) who speak a minority-written
language (MWL) and 10 percent who speak a less common language. The 10 percent represent a
“long tail” of many languages, each with relatively few speakers. The 27 percent share, by contrast,
represents a large group of students who speak about 237 languages, or only three percent of the
8,789 languages spoken in these countries. (This figure of 8,789 is arrived at by deliberately counting
each language once for every country where it is spoken, hence the total of 7,117 known languages is
exceeded.) Teaching this 27 percent of students in this three percent of all languages requires signifi-
cantly less effort than reaching the entire 37 percent. The analysis confirms a “90/10 rule” whereby 90
percent of students speak either the national (dominant) language or a common, widely spoken and
written L1. Globally, teaching in an additional 220 languages could serve the needs of approximately
270 million students. Although such an expansion of instruction shouldnot be portrayed as a trivial
undertaking, it would demonstrate the relatively easy gains on offer: the “low hanging fruit” presented
by a modest augmentation of languages taught.

TEN PERCENT OF STUDENTS IN LOW- AND MIDDLE-INCOME COUNTRIES (LMICS) SPEAK A
LANGUAGE IN THE “LONG TAIL” OF MANY LANGUAGES WITH RELATIVELY FEW SPEAKERS.
Those not included in either national L1s or MWLs are less than 10 percent of the population of
LMICs. This group is composed of fewer than 100 million students who may any one of more than
6,000 languages that are not national L1’s or MWLs. The student population in this group is likely to
be even smaller than this estimate, however, as this count includes endangered languages that are very
rarely effectively spoken or learned by young people. The value of these languages and the associated
cultural capital they represent is unquestioned. However, as countries pursue the goal of providing
equitable education for all children, when national programs cannot cover all languages in the short-
or even medium-term, consideration of language coverage and costs may factor into choices. Flex-
ible, context-specific strategies must be developed for children speaking demographically very small
languages. In many cases, when these children come to school, they are bilingual in a small language
and a larger language used across communities. In those cases, the larger language can be effectively
used for instruction. In other cases, community teaching aides and materials may be needed to support
children in learning foundational learning skills before an appropriate L2 is introduced.

WHAT WOULD IT TAKE TO REACH MORE THAN 80 PERCENT OF L1 SPEAKERS IN ALL LOW-
AND MIDDLE-INCOME COUNTRIES? The analysis above yields a total of 220 languages by
counting each individual additional language only once, even if it is a potential L1 LoI in more
than one country. Two or more countries developing learning materials in the same language



SECTION 1 : WHY SHOULD WE CARE?                                                                                 21
     could, in theory, partially share or collaborate in the creation of curricula and teaching and
     learning materials. In practice, this type of beneficial sharing would not drastically reduce
     the financial and time costs associated with curriculum and language program creation.
     If each country’s languages are counted individually, instruction in a total of an additional 559
     languages (or “country-languages”) would be needed to reach 84 percent of students in each country
     in their L1. Thus if Hausa, for example, is an L1 for both Niger and Nigeria, in this estimate it is counted
     twice. Although this number is significantly above the 220/3% estimate, one sees that teaching in 12
     percent of the world’s languages reaches five-sixths-fifths of learners.


         TABLE 1: National (Official) versus “Minority Written” Languages in WB Client Countries1

                                    National (Official) LANGUAGE                                        “Minority Written” Languages

                                                                          MWL
                                     POP. FOR                                                                            MWLS AS       AVERAGE #
                                                      POP. WHO        SPEAKERS AS                TOTAL      TOTAL #
                                      WHOM                                                                                SHARE         SPEAKERS
                          POP.2                       SPEAK AN         A SHARE OF                 # OF        OF
                                     NATIONAL                                                                             OF ALL        PER MWL
                                                        MWL1              TOTAL                  LANGS.      MWLS
                                     LANG = L1                                                                            LANG.        (MILLIONS)
                                                                      POPULATION

         EAP              2,148          1,359            653               30%                   3224         56            2%             11.7
         ECA               690           603               32                5%                   706           5            1%             6.5

         LAC               620           580               14                2%                   1058          6            1%             2.4

         MNA               414            55              316               76%                    55          39           12%              8.1

         SA               1,694          1,192            400               24%                   757          53            7%              7.5
         SSA-
                           309            66              182               59%                   1158         22            2%             8.3
         Central
         SSA-East          324            67              155               48%                   636           41           6%             3.8

         SSA-West          175            26               59               34%                   944           15           2%             3.9
         WB Client
                          6,371         3,947             1,811             28%                  8,798         237           3%             7.6
         Countries

     1
      Source: Estimations based on 2020 Ethnologue data – total number of languages exceeds 7,117 from double-counting the same language used in more
     than one country (i.e., Spanish counted once per Spanish-speaking country in the 8,798 figure but only once in the 7,117 figure.)
     2
         Counts shown in millions
     3
         17 of these languages are currently used as official LoI; 237-17 = 220 untaught MWLs.




     MOST LEARNING POVERTY IS IN LANGUAGES THAT TOGETHER USE ONLY A FEW WRITING
     SYSTEMS. One should consider not only the number of languages but also the number of writing
     systems required for instruction, since foundational literacy involves mastery of the use of written
     language. The congruence between languages and their writing systems varies. Few languages are
     purely either sound- or symbol/image-based, and a small number of alphabets are very widely used.
     Among alphabetic languages, some more easily (“transparently”) represent the sounds of the spoken



22                                                                                LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
language. Learning to “decode” in these language/writing system pairs generally takes less time and
effort than for language/writing system pairs where the sound-symbol relationships are less straight-
forward. English is an example of a language that takes longer to learn. Languages whose current
writing systems have been developed or adopted more recently, and/or whose , orthographies may
have been developed by linguists for ease of use, will often have greater sound-symbol alignment and
be highly transparent. This is the case for many languages in sub-Saharan Africa. The preliminary
analysis conducted for this paper shows that: (i) Learning Poverty is concentrated in languages that
use a small number of writing systems; and (ii) alphabetic languages are very heavily represented, with
50 percent of Learning Poverty in languages written with an alphabet derived from Latin alphabet.
This issue is important as approaches to foundational learning are affected by language difficulty and
the relative complexity of the language/writing system combination. Countries that teach multiple
L1s and countries that teach L1s and an L2 with a simpler writing system like those with Latin-derived
alphabets—with its 23 letters20 —may be able to standardize certain aspects of orthography and gram-
matical conventions across languages. This makes it easier for children to learn L2s, provided that
instruction explicitly teaches the similarities and differences between languages.

LEARNING POVERTY IS CONCENTRATED IN LANGUAGES USING NO MORE THAN 14 WRITING
SYSTEMS. The 20 countries with the highest percentages of students in Learning Poverty use only
three writing systems: Latin, Arabic, and Ethiopic. Consideration of absolute numbers of students in
Learning poverty involves more writing systems, but still a relatively small total number: the 20 coun-
tries with the highest Learning Poverty by headcount use a total of 14 writing systems. Of these, 12
are alphabetic and more than half are used in. India, which uses eight writing systems to write widely
spoken regional languages. China’s two widely used writing systems (Bopomofo and Han) are both
non-alphabetic. Developing tools for instruction in additional languages can concentrate on a small
number of writing systems and a small share of the world’s languages.

MORE AND BETTER EMPIRICAL MEASURES OF THE PROBLEM ARE NEEDED. The analyses presented
above are preliminary, but they demonstrate that: (a) at the global level, coverage of L1 speakers can be
vastly increased with a relatively modest increase in the number of LoI; and (b) quantification of the
problem with at least first-order estimates is possible. This work must be done at the country level.

QUANTIFYING THE PROBLEMS BRINGS MULTIPLE ADVANTAGES. First, governments and develop-
ment partners, including the World Bank, have known that LoI is “an issue” but have generally not
known how significant an issue it is or what it would take to make progress in solving it. These esti-
mates allow for better dialogue. Second, as data are improved, the relationships between Learning
Poverty and language diversity and proficiency can be analyzed and investigated. Third, other actors,
potentially including private developers of teaching and learning materials and education technolo-
gies, can access information about potential investment opportunities.

REALITY CHECK: MULTIPLE PERMUTATIONS OF LANGUAGES, PROFICIENCY LEVELS, AND USE.
These initial analyses tend to paint the differences among language speakers as uniform and even
binary. For example: this student speaks an L1 but not an L2. In fact, students often fall somewhere
along a continuum of proficiency levels for their L1 and any other languages spoken. Policies should
pay special attention to where proficiency differences are greatest between L1s and LoI for both
students and teachers, as learning outcomes are at the highest risk in these situations.




SECTION 2: : HOW BIG IS THE PROBLEM?                                                                        23
     SECTION 3:
     Political, economic, and sociological
     considerations influence language
     choices



     LANGUAGE POLICIES ARE SHAPED BY HISTORY AND SERVE POLITICAL AND ECONOMIC GOALS.
     LoI policies are related to national language policies. These are influenced in turn by history and by
     political and economic goals such as political identity, cultural identity, economic modernization and
     integration into global trading communities. National language policies routinely seek to facilitate
     communication among all citizens. Some languages are part of regional or global communities such
     as “la francophonie.” These goals can be contentious. Their ends may be in direct conflict with educa-
     tion policies that promote equitable learning. Attempting to create or preserve national unity through
     a common language of communication may confer advantages to speakers of the chosen language
     while linguistic minorities must learn a new language. Using an official language that is only spoken
     by a small number of citizens may preserve the advantages of the elite. Governments have also used
     language policies and LoI policies to signal a break with past political groups, eras, and practices.

     POLITICAL LEADERS HAVE USED LANGUAGE POLICY TO BUILD NATIONAL UNITY. Languages can
     define cultural identity and create communities. For some countries, national identity, a common
     language, and a common culture and heritage are synonymous. The original source of linguistic unity
     may be in the remote past. The language may have been imposed as a result of a long-forgotten mili-
     tary conquest, a period of colonialism, or as the result of one linguistic group gaining hegemony within
     national territory. Nonetheless, citizens see themselves as Italians, Armenians, Koreans, or Russians
     because they speak that national language. In other cases, the efforts to create this unity are more
     explicit and deliberate. When Tanzania gained independence, Julius Nyerere promoted Kiswahili to
     increase communication among citizens who spoke more than 100 indigenous languages. He linked it
     to social and cultural policies under the banner of “ujamaa.” The connection of Kiswahili to Pan-Af-
     ricanism helped consolidate its position beyond Tanzania’s borders. Singapore’s leaders balanced
     pragmatism in communications with cultural policies that have sought to recognize the linguistic
     heritage of its main ethnic groups, and promote the continued use of several L1s. Canada’s Constitu-
     tional Act of 1867 defines French and English as its official languages, but it was not until 1969 that it
     became a bilingual country where all public services and education would be made available in both
     languages. In some cases, the struggle for national unity is conflictual, and languages may be imposed
     by force through internal or external strife and subjugation. Linguistic minorities are often marginal-
     ized socially, economically, and politically.21 Metropolitan prestige languages often exclude linguistic
     communities, even when these constitute the large majority of citizens. In Haiti, for example, a small
     French-speaking elite based primarily in the capital controls social, economic and political oppor-



24                                                       LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
tunity. Meanwhile, Haitian Creole is spoken by more than 99 percent of Haitians, but it has been a
struggle to have it used for education.

PARTICULAR CONTEXTS MAY FAVOR THE INTRODUCTION OF IMPROVED LOI POLICIES. The links
between language and history, politics and conflict do not prevent policy discussions on LoI. Experi-
enced observers have identified four contextual conditions that may increase the likelihood of adop-
tion of sound LoI policies. Box 3 outlines these and provides examples of countries that have success-
fully introduced and implemented improved LoI policies.22



   BOX 3: Enabling Contexts and Conditions for Adoption and Implementation of
           Education Language policies
  LoI policy, regardless of its content, thrives when the following criteria are met:
      1. It is seen as reflecting a new national direction (e.g. post-revolution policy), and/or
      2. The national policy reflects a stronger commitment to national identity.
      3. The state is stable and strong enough to resource and align other policies to support it.
      4. Local appropriation of language policy aligns with national policy, even if it is not enforced.

  Examples:

  Ethiopia’s post-revolution constitutional support for the development and use of all Ethiopian
  languages, instead of only Amharic (the language of Ethiopia’s dominant ethnic community for
  generations before). Criteria met: 1, 2, 3, to some extent 4.

  Eritrea’s policy statement of 1991, reflecting its new independence from Ethiopia, supporting the use
  of all nine Eritrean languages as languages of instruction as a fundamental democratic right. At least
  seven of those languages are currently in use in Eritrean classrooms. Criteria met: 1, 2, 3, 4.

  South Africa’s post-apartheid policy includes 11 languages of instruction. The list includes English
  and Afrikaans, as well as nine additional South African languages, indicating the government’s
  determination to provide opportunity for all. Criteria met: 1, 2, 3, 4.

  Rwanda’s move away from French towards English in the education system, beginning after the
  genocide and change of regime in the early 1990s. Kinyarwanda, the first language of more than 90
  percent of the population, has been included as part of this evolution in language policy. However,
  the recent policy change from Kinyarwanda-plus-English to 100 percent English-medium instruction
  demonstrates how policy changes may advance or retard the cause of L1 instruction for all through at
  least the end of primary school. Criteria met: 1, 2, 3.




PERCEIVED OR REAL ECONOMIC VALUE OF LANGUAGE PROFICIENCY SHAPES POLICIES AND
POLITICS. Language policies reflect economic aspirations. Economists have shed light on the
economic value of different languages as well as the real and perceived effects of linguistic attributes
on earnings.23 These are intrinsically linked to language preferences and policy decisions. Countries
want to take advantage of global trade and economic integration, and therefore seek to build skills in



SECTION 2: : HOW BIG IS THE PROBLEM?                                                                       25
     international linguae francae. An international language, used
     for commerce and diplomacy, may acquire or retain status
     because of perceived high value outside the economic
                                                                                           Especially in low
     sphere.24 A study of the economics of language has docu-                             resource contexts,
     mented the labor market value of language abilities in a                             children who start
     variety of contexts.25 A persistent empirical finding is the
     disadvantage of being in a linguistic minority or failing to                      school in an L1 learn an
     speak and understand the dominant language.26 Expla-                              L2 more effectively than
     nations of wage premia include factors directly related to
                                                                                         students who start
     productivity such as communication costs and language-me-
     diated access to networks as well as factors such as discrim-                         school in the L2.
     ination. The literature on the economic value of language
     pertains to their use by adults in the labor market.

     THE PERCEIVED VALUE OF L1 PROFICIENCY AND USE VARIES.
     In many contexts, speakers of an L1 take pride in their language and the
     culture and values they feel it represents. In others, biases against L1 instruction
     exist, often among a range of stakeholders: teachers, parents, community leaders,
     school authorities, and students.27 In many countries, policies implemented during the colonial period
     and continued since independence have created and perpetuated an aversion to L1s and a preference
     for metropolitan, generally European, L2s as languages of education and government more broadly.28
     These historical considerations are often mixed with perceived labor market biases, especially in the
     formal economy.29 A strong attachment to L1s by community stakeholders may be present alongside
     a preference for L2s in education. This becomes especially problematic when education policy is
     communicated as either L1 or L2 instruction, not the L1 and L2 additive bilingualism endorsed by
     evidence and this paper. Kalenjin-speaking parents in rural Kenya, for example, told researchers that
     L1 was “a distraction from learning”.30

     LABOR MARKET VALUE DOES NOT INDICATE THE BEST WAY TO ACQUIRE LANGUAGE PROFI-
     CIENCY. Especially in low resource contexts, children who start school in an L1 learn an L2 more
     effectively than students who start school in the L2. The perception of labor market value for specific
     languages in business, commerce, and administration strongly influences preferences with regard to
     school language among parents and policy makers. However, strong disconnects are common between
     a legitimate desire for schools to provide graduates with specific valuable language skills and the best
     means of structuring learning trajectories to arrive at this end point. Selection of an inefficient means
     of teaching languages—such as early use of an L2 as the medium of instruction for all school subjects—
     decreases the chances of a student ever obtaining the labor market premia. Policy makers are well
     served to separate goals for school language policies from the means for obtaining these goals. Other
     stakeholders, especially parents, should be well-informed that MTB-MLE provides the best path to
     their children becoming bilingual, preserving their L1 and becoming stronger speakers of the L2.

     INCOMPLETE INFORMATION ABOUT THE COST-EFFECTIVENESS OF LOI POLICIES CREATES
     CONCERN. Many governments and development partners raise concerns about costs during LoI policy
     discussions.31 However, estimates show that the cost of implementing appropriate LoI policies aver-
     ages only four to five percent of additional expenditure in countries where an additional language is
     added.32 At the same time, successfully implementing strong LoI policies tends to result in cost-saving
     benefits such as a decline in the repetition rate, a decline in the dropout rate, an increase in graduation
     rates, and better learning outcomes. These promote higher productivity and growth, which in turn
     generate higher tax revenue. Even under conservative assumptions, the value of the benefits (lower



26                                                        LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
costs and higher productivity) consistently exceeds the additional investment costs.33 These identi-
fied benefits result in cost-savings not just in countries with a small number of languages, but also
in linguistically diverse countries. For instance, in Mali, a World Bank study found that French-only
programs cost about eight percent less per year than mother-tongue schooling, but the total cost of
educating a student through the six-year primary cycle was about 27 percent more, largely because
of the difference in repetition and dropout rates.34 Studies looking at the provision of MTB-MLE
in Guatemala, Senegal, and elsewhere confirm these conclusions.35 Net benefits are positive even
without including any valuations of considerations wholly external to the labor market such as the
preservation of languages as living cultural assets.36 Figure 2 compares the impacts and cost-effective-
ness of different teaching interventions. Among them, MTB-MLE stands out as the one that yields the
most learning gains per equivalent investment costs. For more information regarding the cost impli-
cations of MTB-MLE, refer to Chapter 6 in Optimising Learning, Education and Publishing in Africa:
The Language Factor.37


FIGURE 2: Highly Effective Practices to Increase Access and Learning Outcomes

         Mother-tongue instruction                                                                                                                         $

           Better teaching methods                                                                                                                        $$$

       Providing remedial education                                                                                                                       $$

             Doubling learning time                                                                                                                        $

          Group by learning activity                                                                                                                       $

Update teachers on student progress                                                                                                                        $

    Teacher performance incentives                                                                                                                        $$


                                       0%                          5%                                10%                               15%                  20%


                                               $ Interventions cost 0-3% of standard (basic) costs         $$ Interventions cost 4-9% of standard costs
                                            $$$ Interventions cost 10% or more of standard costs               Impact on learning outcomes



Source: Education Commission analysis (2016). Note: The improvements are calculated from a baseline of 50 percent (of enrollment, completion, or
reaching learning targets) and measured as percentage points gained. The costs are estimated relative to average baseline costs—with average class
size, materials, support, and salaries.




RATIONAL GOALS LEAD TO SEEMINGLY IRRATIONAL EDUCATION POLICIES FOR LOI. The high
prevalence of LoI practices that ignore evidence of the effectiveness of L1 instruction point to a crit-
ical disconnect between the end goals of language policies and the means of achieving these goals.
Reducing or eliminating this disconnect should be a main goal of national policies to issues of LoI.
Examples abound of both negative and positive experiences. Whereas Box 3 presented some posi-
tive experiences, Box 4 presents the Rwandan case, which exemplifies the notion of rational goals
leading to LoI decisions that go against the evidence (please see case study on Rwanda, Annex B). This
paper recommends that governments: (a) clarify the goals of LoI policies; (b) review their past and
current experience with LoI policies; (c) analyze their language landscape from technical and polit-
ical perspectives; and (d) systematically plan and employ the most effective means for achieving high
learning outcomes across school subjects.



SECTION 2: : HOW BIG IS THE PROBLEM?                                                                                                                              27
        BOX 4: Rwanda: A Changing Education System for a Changing Time

       Rwanda has made impressive progress in education over the past few decades, having achieved a gross
       enrollment rate of over 100 percent and maintaining a low rate of out-of-school children. Learning,
       on the other hand, has not advanced at the same pace. While Rwanda has not yet participated in any
       international assessments, it implemented several Early Grade Reading Assessments (EGRAs) in
       Kinyarwanda. A 2011 EGRA found that only 32 percent of grade 4 students and 47 percent of grade
       6 students could read and understand a grade 2 text. At the core of this learning crisis is the issue of
       language.

       In Rwanda, Kinyarwanda is the national language, spoken by all ethnic groups. English is spoken
       by only 190,000 people in a population of nearly 13 million. Yet, for the past 25 years, Rwanda has
       consistently introduced reforms to LoI policies that have almost exclusively given priority to the use
       of English for instruction instead of Kinyarwanda beyond grade 3. The latest policy change came in
       December 2019, when the Government of Rwanda announced a new LoI policy, mandating the use of
       English as the primary language in all grades in basic education.

       This shift can be explained by the pursuit of political and economic goals. The Government of Rwanda
       has ambitious economic goals for the future of the country, and a reliable education system is a key
       part of that plan. In 2007, Rwanda joined the East African Community. In 2009, Rwanda joined the
       Commonwealth. These factors, plus a decision to cease using the language associated with the
       country’s Franco-Belgian colonial period,, provided a strong impetus to favor English.

       Societal pressures have a role to play as well. With the goal of an increasingly globalized economy,
       Rwandan parents want their children to gain English proficiency. This push is driven by wealthier
       parents who can send their children to private schools better able to support English instruction. In
       both the short and the long term, children from poorer households bear more of the brunt of hasty
       decisions that prove difficult to implement.

       Rwanda is an example of how political and economic goals requiring educated adults are often at odds
       with best practice to improve learning for young children. Achieving desired L2 proficiency by the end
       of basic education is best done through initial instruction in L1 in primary school for at least 5–7 years,
       accompanied by L2 “foreign language” instruction and a well-planned, skill-based transition to L2
       instruction in secondary school. (Annex B considers the case of Rwanda in more detail).




     THE COVID-19 PANDEMIC IS A CHANCE TO ADVANCE IMPROVED LOI POLICIES. The global
     COVID-19 pandemic will exacerbate the learning crisis and worsen inequality in education. School
     closures have affected almost all children worldwide. It is expected that Learning Poverty will rise
     10 percentage points in LMICs (from 52 percent to 63 percent).38 The combined shocks of school
     closures and the economic impact of the pandemic are expected to result in higher dropout rates.
     Adjustment requires schools and school systems to quickly develop or adapt new curricula for new
     modes of delivery, especially distance learning. Sound policy engagement can mitigate negative effects
     and allow countries to “build back better”,39 with evidence-based LoI policies and programming.




28                                                         LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
SECTION 4:
Language of instruction policy
should be multifaceted




LOI POLICY CONTEXTS ARE NOT MONOLITHIC. Language, cultural and political contexts vary
within and across countries. In formulating effective policy, there are several key factors to consider
(see Table 2):
   • Absolute number of languages spoken. Overall language diversity is first a function of the num-
     ber of languages within national territory. Low- and middle-income countries (LMICs) span a
     spectrum from being monolingual to having several hundred languages. The sizes of language
     communities vary, as do the number of languages and dialects spoken in them and the range of
     proficiency levels.
   • Number of major languages used in education. Countries may seek to have gradu-
     ates of basic education who are proficient in several national languages. It is com-
     mon for one of these to be an international or colonial language that may share
     few characteristics with local languages. A typical situation in Sub-Saharan Af-
     rica, for example, may include local L1s, a Language of Wider Communication
     (LWC—sometimes referred to as a lingua franca), an official language or lan-
     guages, and various additional national or official languages.
   • Education model for language instruction and schooling. Countries may
     use “L2 only from the start of primary school (and even in ECEC),” which is
     the least effective model. Slightly, but only slightly more effective, are “ear-
     ly-exit” bilingual models, which provide a minimal amount of L1 instruction,
     often in parallel with L2 instruction. More effective for learning are models such
     as two-way (language learners and language-proficient students in the same class-
     room), dual language education, and late-exit models whereby language and academ-
     ic content in other school subjects are taught in L1 first, then gradually in L2 under
     explicit transition plans that are informed by research.
   • Numbers and distribution of students and teachers. Countries may not have enough
     teachers who are literate in L1s or they may not be posted appropriately around the
     country. Teachers in larger cities may be proficient in the official languages
     but lack L1 language skills for teaching. Teachers in rural areas
     may have the opposite skills, namely proficiency in L1s, but not
     in the official language.
   • Homogeneity of speech communities. Rural communities
     may be characterized by a single widely spoken L1 or by



SECTION 4: LANGUAGE OF INSTRUCTION POLICY SHOULD BE MULTIFACETED                                          29
          several. Geographically, the boundaries of speech communities may be sharply or weakly de-
          fined. Section 6 suggests the best ways to develop policies according to the number of languages
          spoken in a classroom, school, or school system.
        • Proficiency levels of students and teachers. Since L2-only education leads to lower graduation
          rates, assessments of upper secondary or lower secondary students will understate the learning
          gaps between L1 and L2 speakers. L1 speakers who persist in school to upper grades will have
          better L1 and L2 abilities than a randomly chosen peer from their age cohort, due to more school-
          ing and likely interaction effects associated with having remained in school.
        • Similarity of languages, writing systems, or both. LoI policies will be affected by levels of sim-
          ilarity among languages in question. It must also be borne in mind that dialects can vary widely.
          In some cases, such as formal Arabic versus colloquial Arabic, the distinction is between versions
          (and registers) of a single language (Box 5). In other cases, languages may be closely related and
          share common word roots and mutually intelligible vocabulary (as do Spanish and Portuguese).
          At the other end of the spectrum, languages—even those found within the same country or com-
          munity—may share very few characteristics. Similarity between languages and writing systems
          facilitates learning. Achieving bilingual proficiency is easier when languages and writing systems
          share characteristics. Overlapping sounds and symbols in L1 and L2 will shorten the instruction-
          al time required for proficiency in L2, if instruction is designed to facilitate bridging between
          languages.40 In Kenya, studies found that reading skills in L1 (Gikuyu, Dholuo, and Kiswahili)
          and L2 (English) were strongly correlated.41 Similar findings have been found in myriad other
          studies in low-, middle-, and high-income settings.42
        • Ages of language learning. LoI policies need to be contextualized to reflect the languages chil-
          dren know when they start school. In some contexts, children are bilingual or even multilingual
          at a very young age. In other countries, children are monolingual until they leave home to go
          to school. In yet other countries, there are differences by gender due to differences in gender
          roles. Ages of language learning can also vary according to parental occupation, and other fac-
          tors. When children first arrive in school, typically at age five or younger for ECEC and six or
          seven for primary school, their instruction needs to be in a language they speak at that time in
          their lives.
        • Special considerations. Refugees and internally displaced school-age children require in-
          struction in a language they speak and understand. Continuity with the LoI of their earlier
          schooling is ideal, but circumstances may not permit it. Language of instruction for learners
          with disabilities should also be considered. This may involve accessible learning materials,
          along with additional support, such as specialized staff, or community volunteers. For exam-
          ple, deaf children’s language of instruction should be a sign language in which instruction may
          be offered. A multiplicity of other language-related considerations applies to those with learn-
          ing difficulties and other communication challenges (but lies beyond the remit of this paper).


     CONTEXTS AND CONSULTATIONS ARE ESSENTIAL TO POLICY FORMULATION AND IMPLEMENTA-
     TION. Since so many factors influence LoI policies, it is critically important for all involved to (a) under-
     stand the context; and (b) consult widely with stakeholders during policy development and imple-
     mentation. Stakeholder consultations help policy makers understand the diversity of the language
     context and allow those most affected by policy to articulate their goals and concerns. Understanding
     and respecting diverse perspectives makes it more likely that LoI policies will align with stakeholders’
     goals, which in turn increases the likelihood of a policy’s long-term acceptance and viability




30                                                         LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
TABLE 2: Factors for Consideration in Language of Instruction Policy Formulation and Implementation


                                                                                   DUAL BI, SMALL
                                                          SINGLE/ MONO                                      MULTIPLE /TRI OR MORE
                                                                                    DIFFERENCE,
                                                         NO DIFFERENCES                                     GREAT HETEROGENEITY
                                                                               LITTLE HETEROGENEITY

                               Number of
                                                     1                     2                               More than 2
                               languages spoken
    LANGUAGE




                               Number of
                               languages to be       1                     2                               More than 2
                               learned

                                                                           Bilingual- dual +               Trilingual
                               Language
                                                     Mono-lingual +        Duration?                       Duration?
                               education model
                                                                           Language only or subjects?      Language only or subjects?
    DISTRIBUTION OF SPEAKERS




                                                                           Geographically dispersed, but   Diversity and numbers of
                               Distribution of
                                                     Uniform               concentrated in communities     languages spoken & heteroge-
                               speakers (learners)
                                                                           of homogenous speakers          neity in linguistic communities


                               Distribution of                                                             Does not match number and
                                                                           Matches distribution of
                               teachers by target    N/A                                                   distribution of speakers/
                                                                           speakers
                               language skill                                                              learners


                               Proficiency levels                                                          High within and across learner
                                                                           Moderate variation within
                               of learners in        Little variation                                      groups (starting with no
    PROFICIENCY




                                                                           and across learner groups
                               target language                                                             proficiency)

                               Proficiency levels                                                          High within and across teacher
                                                                           Moderate variation within and
                               of teachers in        Little variation                                      groups (starting with no
                                                                           across teaching groups
                               target language                                                             proficiency)
 CHARACTERISTICS




                               Diversity of
    LANGUAGE




                               language                                    Similar                         Very different
                                                     N/A or Dialect only
                               characteristics by                          Same language family            Multiple language families
                               language pairs
 LEARNERS WITH




                                                                           FCV
   DISABILITIES




                               Additional                                  Refugee populations             Learners with disabilities in
                                                     N/A
                               Considerations                              Marginalized groups             linguistic minorities
                                                                           Deaf children




SECTION 4: LANGUAGE OF INSTRUCTION POLICY SHOULD BE MULTIFACETED                                                                             31
     BOX 4: Mastering Arabic: The Transition from Colloquial to Formal
     Arabic is spoken by more than 420 million people, making it the sixth most spoken language across
     the world. Arabic exists in a continuum of forms, from Classical Arabic—the language of the Qur’an—
     to the many informal, colloquial versions used in everyday communication. Modern Standard Arabic
     (MSA) lies near the formal end of the continuum, and is described by some linguists as a version that
     dates back to the 19th century, having dropped some words and phrases over the intervening decades,
     and added new technical vocabulary. MSA is the written form of Arabic that is used across Arabic-
     speaking countries in literature, official documentation, mass media, and educational institutions.
     However, MSA is not an L1; it is learned through formal education.

     Formal Arabic (Classical or MSA) differs significantly from colloquial Arabic. There are many types
     of colloquial Arabic, including Iraqi, Gulf Arabic, Levantine, Yemeni, Sudanese, Egyptian, Maghrebi
     (North Africa excluding Egypt), and Hassaniya (mainly Mauritania) and even differences within
     some of these. The strictly formal version of Arabic has barely changed over time, unlike the rapid
     development of colloquial Arabic, resulting in widening gaps between the two.43

     Children grow up hearing colloquial Arabic in their homes and communities, but are expected to
     learn to read and write in formal Arabic as soon as they enter school. Given that oral language
     proficiency is a prerequisite for reading acquisition, this results in delays in reading. Parents in Arabic-
     speaking countries are less likely to have children’s books in the home and are less likely to read to
     their children when they are young than parents in other countries (See Annex D). This is a concern,
     as children transition more easily from their L1 colloquial to formal Arabic if they are exposed to
     children’s literature.44

     Low quality instructional materials and ineffective teaching methods make it difficult for children
     to learn formal Arabic. Teaching and learning materials in schools use MSA. Children are often taught
     to read and write in a rigid way, with a focus on grammar and accuracy and a lack of playfulness and
     inquiry.45 Teachers correct children as they learn to read and write, while inventive or spontaneous
     engagement with words or spellings is considered inappropriate.46 Likewise, phonics-based methods
     are rarely used as they rely on breaking words down and assessing using non-words. Teachers do
     however sometimes use colloquial Arabic to help to explain features of the formal Arabic language,
     but this tends to add a level of complication.47 In fact, teachers are themselves frequently the
     products of poor Arabic language education, meaning they rarely have the knowledge and skills in
     formal Arabic to be comfortable using it as a medium of instruction. Furthermore, those training to
     become Arabic language teachers rarely experience the in-depth training they need on how to teach
     Arabic, with very few university teacher preparation courses in the region including Arabic pedagogy.

     Teaching and learning approaches for Arabic have not enabled all students to reach their potential.
     Improving Arabic language teaching and learning would boost overall learning outcomes. This should
     include (1) encouraging parents to read to children from an early age to help develop vocabulary,
     (2) developing more, and more engaging, children’s literature, (3) using evidence-based pedagogical
     methods in the early grades, and (4) devoting sufficient time within the curriculum to the development
     of Arabic language skills. (Annex D considers the case of Arabic in more detail).




32                                                        LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
PART 2:
Promoting Solutions
through More Active
Engagement
     GREATER USE OF EXISTING SOLUTIONS AND MORE PROACTIVE ENGAGEMENT BY THE BANK.
     Although there are examples of excellent engagement with LoI issues within Bank teams, the Educa-
     tion Global Practice has too often found itself reacting to unhelpful LoI policy changes instead of
     engaging proactively. It is not unusual for the Bank to confront the LoI question with considerable
     verve in the wake of a client government’s announcement of a new policy to make an (often inter-
     national) L2 the official LoI for all of primary education. The Bank then coordinates development
     partners to marshal evidence illuminating the policy’s shortcomings. But this is generally too late—
     the decision has been taken, and the government is already implementing the intended change. In
     other cases, the Bank simply omits analysis of LoI from lending or country dialogue. In practice, few
     impending projects have been slowed down for lack of a comprehensive LoI plan. This pattern of
     reactivity and neglect is clearly unsustainable: the Education Global Practice needs to work proac-
     tively on LoI policy, building on existing solutions. Section 5 reviews existing policy solutions and
     proposes ways to increase their use. Section 6 proposes a roadmap for the Bank to move from reac-
     tivity to proactivity and then to leadership on LoI in education.




       BOX 6: Use Existing Resources to Accelerate Progress

       Many development partners have worked on LoI policy and implementation. Familiarity with their
       policy discussions, use of their resources, and an analysis of lessons learned from the implementation
       history of their programs can accelerate reading progress for children in World Bank Group client
       countries. A great many organizations and initiatives have produced policy documents, programing
       guidance, teaching and learning materials, and assessments in hundreds of languages. Below are
       some suggested resources:

       USAID – USAID has focused intensively on early grade reading since 2011, with programs in more
       than 40 countries. Most produced student books, teacher’s guides, and supplementary materials,
       often in multiple languages. These materials can be accessed through USAID in many World Bank
       client countries. Storybooks are also often available through the Global Digital Library48 and other
       digital platforms. Other resources include research reports on language topics and language maps.




34                                                        LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
  BOX 6: Use Existing Resources to Accelerate Progress cont.

 The Global Reading Network (GRN) – The GRN is hosted on USAID's Leading Through Learning
 Global Platform. Many GRN-hosted resources49 support work on LoI policy and programing, including
 the Handbook on Language of Instruction Issues in Reading Programs (Pflepsen 2019).

 The Global Book Alliance (GBA) – The GBA is a partnership of donor agencies, multilateral
 institutions, and civil society organizations committed to bringing books to every child in the world
 by 2030. Its initiatives include the Global Digital Library, tools for analysis of the book supply chain in
 a given country, support for publishers, and Begin with Books, a program to create thousands of open
 source books in underserved spoken and signed languages.50 The GBA also hosts extensive resources
 on the book supply chain, copyright and licensing, and other key topics.

 Eddata II – Eddata (funded by USAID and implemented by RTI International) produced the Early
 Grade Reading Assessment (EGRA), the Early Grade Math Assessment (EGMA), conducted the
 assessments around the world in multiple languages, and produced other important research, such
 as the Survey of Children’s Reading Materials in Eleven African Languages (RTI 2016).51

 SIL International – SIL has developed a host of resources to support LoI issues.52 These include
 Good Answers to Tough Questions in MTB-MLE (Trudell and Young, Eds. 2016) and the MTB-MLE
 Resource Guide (Malone 2018). SIL also runs the Ethnologue, the primary source of information for
 languages around the world.53 SIL’s technology resources include:
 • Bloom: Bloom includes software to quickly and easily develop books for young readers, hundreds
   of titles in hundreds of languages, and a large image bank. 54 SIL also facilitates author & publisher
   workshops using Bloom, which can produce multiple titles over the course of a workshop.
 • Primer Pro: Primer Pro can analyze text in a given language and suggest a linguistic scope and
   sequence for reading instruction.55 It can also search for particular parts of speech and other
   linguistic features. From this information, it can build word lists that authors can use for creating
   decodable text.

 People’s Action for Learning Network (PAL) – PAL is a partnership of organizations working on
 three sets of material to implement learning assessments, and programs to improve learning.56 PAL
 members, including Pratham in India and Uwezo in East Africa, have developed citizen-led learning
 assessments (often called ASER assessments) in multiple Indian and African languages.57

 The following additional sources are for storybooks and versioning tools:
 • African Storybook provides open access (CC BY) to thousands of storybooks in African languag-
   es for reading online, downloading for print, and adapting.58 Its software tools also allow authors
   to create new titles.
 • Pratham Book’s Storyweaver59 provides open access (CC BY) to thousands of storybooks from
   around the world, as well as software to create and version books in other languages.




PART 2: PROMOTING SOLUTIONS THROUGH MORE ACTIVE ENGAGEMENT                                                     35
     SECTION 5:
     Effective LoI policy examples
     exist but are not being widely
     implemented



     EFFECTIVE APPROACHES EXIST FOR ALL FOUR PHASES OF LOI POLICY DEVELOPMENT AND
     IMPLEMENTATION. The cycle for LoI policy and programing can be divided into four phases: (i) plan-
     ning for language use; (ii) development of curriculum and materials, plus the allocation and training
     of personnel; (iii) implementation and support; and (iv) assessment and adjustment. Much has been
     learned in the past decades about the most successful approaches for each of these phases.




     Planning for language use
     PLANNING REQUIRES INFORMATION ABOUT LANGUAGE USE, ATTITUDES, GOALS, RESOURCES,
     AND CHALLENGES. Countries would be well-advised to begin their planning processes with a strong
     understanding of their linguistic context. Educational goals must take into account the language(s)
     of instruction, including carefully planned, mastery-based language transitions. Some countries and
     territories, such as Hong Kong (Box 7), have comprehensive and sound policies for LoI. Inventories
     of human, educational, and financial resources facilitate decision making, as does familiarity with
     frequently encountered obstacles to implementation. Stakeholders working towards effective LoI
     policies tend to succeed when they: understand the sociolinguistic and educational context, including
     prevalent perceptions of language use; support the development of appropriate teaching and learning
     materials (TLM); attract, select, and support motivated teachers; match teachers’ language skills with
     their instructional mandate; and creatively harness technology in ways that respect the context. Infor-
     mation gathering needs to prioritize local knowledge and consider stakeholder attitudes and beliefs
     alongside language data, and the implementation landscape (including available resources).

     MAPPING THE SOCIOLINGUISTIC CONTEXT. Language use is constantly evolving, so this informa-
     tion requires continuous updating. What languages are spoken by students, teachers, parents and
     community members? How well? By how many people? Do the languages have standardized orthog-
     raphies? What are people’s attitudes toward the different languages and their use? How are these
     languages used in classrooms and elsewhere? To all these questions, language mapping will provide
     the kind of answers needed to fundamentally inform the planning process from the outset. Ethiopia
     did this to its benefit, as illustrated in Annex A. Mapping exercises showed that of the country’s more
     than 80 languages, five are spoken by nearly 80 percent of the population. These were the appropriate
     starting point for the development of assessments, materials, and curriculum. Good language mapping



36                                                      LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
  BOX 7: Hong Kong China’s Highly Successful Model for Language of Instruction

  For primary schools, owing to demographic necessity and the alignment of policies to evidence-
  based practices, the topic of LoI has been wholly uncontroversial as policy, largely unproblematic in
  practice, and has resulted in high levels of learning. Teaching students in a language they understand
  has paid significant dividends for learning outcomes. Hong Kong China (HKC) grade 4 students
  scored second, first, and third in the world in the Progress in International Literacy Study (PIRLS)
  assessments in 2006, 2011, and 2016, respectively.

  Meanwhile, secondary schools have been at the center of LoI policy debate in Hong Kong.60 The LoI
  in public primary schools in HKCong is Cantonese Chinese, whereas at university it is predominantly
  English. It is the crucial transition from Chinese to English at secondary school that has given policy
  makers pause for thought.

  Policy changes over the past five decades have mirrored historical, economic, and political changes.
  Language policy in secondary schools in Hong Kong has changed three times. At first, schools were
  allowed to choose their LoI. Then in 1994 schools were categorized as Chinese-medium, English-
  medium, or dual-medium. Then there was a period, from 1998, when Chinese was the only permitted
  medium. The latest change, in 2010, saw the abolition of the above-mentioned classification system
  that used to attach a single language-medium label to an entire school. Instead, a school can now
  offer English-, dual-, and Chinese-medium classes.

  The current LoI model in HKC is characterized by research-based decisions. The model successfully
  promotes L1 instruction in the early grades, builds a bridge from L1 to L2, and supports the use of L1
  even after L2 has been introduced in secondary school. To ensure that this model works in practice,
  the Education Bureau of Hong Kong supports many aspects of the policy such as the alignment of
  assessments to LoI, the provision of TLMs in Cantonese and English, and other interventions to
  ensure the effective use of English as LoI in the secondary grades. The learning outcomes in Hong
  Kong are an exemplary example of the power of effectively addressing language in education. (Annex
  G considers Hong Kong in greater detail).




also considers similarities across languages to facilitate potential economies of scale in development of
curricula and TLM where appropriate. Language mapping lays the groundwork for effective strategies
to communicate language policy.

DETERMINING LANGUAGE LITERACY FOR TEACHER ALLOCATION. Data on the language profi-
ciency of students and teachers underpin several types of policy decision. Centralized teacher posting
systems often result in teachers being sent to work outside their region of language proficiency. In
some countries, and for certain languages, systems for the hiring and allocation of teachers need to
ensure there is a match between the language knowledge of student and teacher respectively. Effective
communication with teachers and teachers’ unions or associations will be required to build wide-
spread support among teachers for any changes in teacher employment. Language mapping exer-
cises can gather information on the students’ and teachers’ level of proficiency in their primary and
secondary language (assessed proficiency, rather than reported proficiency, as there can be significant
differences between the two), and provide information for matching students and teachers by area,
school, and classroom. USAID’s language mapping exercises in Mozambique61 and Ghana62 are good
examples of how this information can be gathered and used.



SECTION 5: EFFECTIVE LOI POLICY EXAMPLES EXIST BUT ARE NOT BEING WIDELY IMPLEMENTED                         37
     COORDINATING LANGUAGE POLICIES WITH OTHER EDUCATIONAL GOALS. LoI policies serve
     larger educational goals and must be integrated with these. Changing the LoI to students’ L1 will not
     improve learning outcomes if instruction is ineffective. Addressing Learning Poverty requires improving
     overall instruction as well as using appropriate languages. Improved instructional approaches explic-
     itly consider and coordinate educational goals, curricula, instructional time, teaching methods,
     teacher preparation and support, teaching and learning materials, parent/community engagement,
     assessment, and other considerations. Coordinating LoI with other aspects of instruction is essential.

     EFFECTIVE LOI MODELS START WITH L1 AND DEVELOP PROFICIENCY IN MULTIPLE LANGUAGES.
     A language of instruction model sets the basic plan for matching instruction in target languages to
     linguistic aims, and educational goals writ large. Of the five following models, two have been shown to
     promote proficiency in L1 and an L2 by the end of basic education.
        • The L1 model provides instruction without a transition to an L2. Countries like Canada that
          offer parents a choice of instruction in one of two national languages rely on this type of model.
          In practice, this model is often supplemented with instruction in the L2 as a foreign language.
        • The immersion model may provide minimal or no L1 instruction. Two-way immersion pro-
          grams build learning around substantial interactions between native speakers of the target lan-
          guage and those who are still learning it (freely using more than one language whenever neces-
          sary). One-way immersion programs prohibit or severely restrict the use of anything but the L2
          target language. One-way immersion has some record of success in high-resource settings with
          highly trained teachers, but has not been documented to be effective in a wider range of settings.
          This is partly the case because in high-resource countries, one-way immersion involves intro-
          ducing the L2 in a structured manner, with sufficient reinforcement of new knowledge and skills.
          Meanwhile, in low-resource settings, the less effective63 submersion approach—sometimes de-
          scribed as “sink or swim”—is more common.
        • Early-exit transitional models use L1 as the LoI in the early grades of primary school with a
          transition— prior to the end of primary school—into learning in an L2 and cessation of instruc-
          tion in the L1. This model can improve learning outcomes, but it generally fails to maximize the
          potential of L1 instruction, especially in low-resource contexts with low learning levels overall.64
        • Late-exit transitional models involve the use of L1 as the LoI through at least lower secondary
          school for literacy and academic subjects. L1 instruction may end in secondary school. Academ-
          ic attainment is generally higher than in early-exit programs, typically with faster development
          of general cognitive abilities.
        • The additive approach maintains the use of the language students best speak and understand
          throughout schooling, from kindergarten to grade 12, while adding instruction in an L2 as a
          foreign language at some point in the middle grades.

     The models are a point of departure for coherent policy choices that take account of other essen-
     tial factors. However, clarity on choices between models seems to be lacking in policy dialogue and
     policy formulation processes. Without such clarity, it is difficult to adopt a consistent approach to the
     following four evidence-based elements of effective policy making: (a) maintaining L1 through the
     first six years of schooling, (b) using L1 for instruction in other academic subjects beyond reading/
     writing, (c) introducing L2 literacy and content area instruction only after students have achieved a
     level of proficiency in L1 literacy and L2 oral language, and (d) continuing L1 instruction after an L2
     becomes the principal LoI. During the policy development process, cogent discussion of language
     models will be essential in order to garner the support of all relevant stakeholders.




38                                                       LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
 FIGURE 3: Language Models 1

          MODEL                              PRIMARY SCHOOL                                SECONDARY SCHOOL


                                1   2    3       4          5         6   7        8       9    10    11      12


       L1 Based Instruction


                 Immersion


      Transitional Early Exit


       Transitional Late Exit


                    Additive




                                                     L1 Instruction       L2 Instruction




MAINTAINING THE L1 AS THE LOI FROM ECEC THROUGH AT LEAST THE FIRST SIX YEARS OF
SCHOOLING. L1 instruction is critical when children are learning to read, write, understand introduc-
tory mathematics, and acquire initial content knowledge. Using L1 in ECEC supports the transition
from home to school, strengthens relationships between children and caregivers, and furthers chil-
dren's oral language development, which is the foundation of literacy. L1 ECEC programs more effec-
tively support the development of motivation to read, involving an enjoyable introduction of "concepts
of print” (the way print functions, such as left to right and top to bottom, depending on language),
letter knowledge, and phonological awareness. Even in monolingual contexts with strong institutional
capacity, as many as one-third of children have trouble mastering literacy subskills such as segmenting
sounds in words, even when they know several thousand words in their L1.65 Children learning to read
in a foreign language may know only a few words, or none at all, and so the ability to discern (in effect,
to hear) word sounds and map them to letters becomes substantially more difficult. Decoding, which
should allow students to turn written words into “spoken language in their heads,” instead devolves
to rote memorization of arbitrary and meaningless relationships. Then reading and literacy skills fail
to develop sufficiently, and students are left unable to learn content or progress to “reading-to-learn.”
Building sufficient literacy and cognitive capacity to progress in school66 requires a minimum of six
years (the precise duration required will be dependent on instructional quality, instructional time, and
other factors).67 Figure 4 shows how children taught and assessed in their L1 outperform students
taught and assessed in an L2. Both reading and mathematics scores are higher for the L1-taught chil-
dren. Burundi was the only one of 10 countries participating in PASEC 2014 to use L1 as the language
of the grade 2 assessment; scores were more than 100 points above the test average. By contrast, on
their grade 6 assessment, conducted in French, Burundian students’ performance is much closer to
the average of the participating countries.




SECTION 5: EFFECTIVE LOI POLICY EXAMPLES EXIST BUT ARE NOT BEING WIDELY IMPLEMENTED                                39
     FIGURE 4: Percentage Point Difference in Share of Children Achieving Level 1 in Reading and
     Mathematics between Students Who always or Sometimes Spoke the Language of Instruction at
     Home and Those Who Never Did— assessment by PASEC, 2014

                                            GRADE 2                                                                                      GRADE 6
           Congo                                                                                     Congo
            Niger                                                                                     Niger

       Cameroon                                                                                  Cameroon
     Côte d'Ivoire                                                                             Côte d'Ivoire

             Togo                                                                                      Togo
         Senegal                                                                                   Senegal
            Chad                                                                                      Chad

     Burkina Faso                                                                              Burkina Faso
            Benin                                                                                     Benin


                     0%         10%          20%          30%          40%          50%                        0%        5%        10%        15%        20%        25%        30%


           % point difference in the share of children achieving Level 1 in PASEC Reading            % point difference in the share of children achieving Level 1 in PASEC Reading
           % point difference in the share of children achieving Level 1 in PASEC Math               % point difference in the share of children achieving Level 1 in PASEC Math




     USING L1 IN OTHER ACADEMIC SUBJECTS. STUDENTS NEED TO LEARN TO USE LANGUAGE IN
     MANY DIFFERENT CONTEXTS AND FOR LEARNING DIFFERENT SUBJECTS. After mastering the
     first steps in literacy, children who receive L1 instruction in subject-specific content areas are better
     prepared to build vocabulary and expand comprehension to new domains. Numerous studies have
     shown that learner ability in the LoI is often too low for learners to achieve acceptable levels of content
     mastery.68 L1 instruction in all subjects enhances not just literacy, but learning outcomes in the broader
     curricular content.69 The near-perfect correlation between a country’s reading score on the Progress
     in International Literacy Study (PIRLS) assessment and its math score on the Trends in International
     Mathematics and Science Study (TIMSS) assessment70 is consistent with the higher performance in
     mathematics and science for those who speak the language in which they were tested.71

     CONTINUING L1 INSTRUCTION AFTER TRANSITION TO L2 AS DESIGNATED LOI. The longer chil-
     dren learn in a language they know, the more they learn. Numerous examples in low-, middle-, and
     high-income countries have demonstrated that transitioning children from L1 instruction to learning
     exclusively in an L2 is less effective in supporting children’s literacy development, second language
     acquisition, and subject-specific learning.72 By contrast, continuing to use the L1 after the L2 has been
     introduced allows students to further develop literacy skills and learn content area knowledge. These
     students continue to strengthen their foundation in their L1 and transfer those skills to the L2.73 It is
     talking about new ideas and concepts that truly helps students comprehend them.74 Thus, even if an
     L2 is already the designated LoI, explaining content in L1 when students are at risk of not compre-
     hending fully, or as a complement to L2, will yield better results than exclusive L2 use in instruction.75

     INTRODUCING L2 AS A FOREIGN LANGUAGE WITH EMPHASIS ON ORAL LANGUAGE ABILITIES IN
     L2. Planning for language use means planning to build all the abilities students are meant to acquire
     by the end of basic education. Research conducted in the United States consistently finds that students



40                                                                                          LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
can spend up to half of their instruction time in an L2 with no negative effects on L1 learning (Collier
and Thomas, 2017). An overall plan for multilingual proficiency must first of all build students’ oral
fluency and vocabulary in an L2 (and sense of ease) as a foreign language.76 Children who receive
reading instruction in their L1, have better L2 literacy scores77 thanks to the transfer of skills from L1
to L2. However, students who do not reach sufficient skill levels in L1 and oral fluency in the L2 have
difficulty transferring their reading skills from L1 to L2.78,79 Whatever the schedule for L2 introduc-
tion, teachers need to have a high level of oral and written proficiency in the L2 to teach it effectively.

DEVELOPING AND IMPLEMENTING EFFECTIVE COMMUNICATION STRATEGIES FOR ALL STAKE-
HOLDER GROUPS. In contexts where the use of L1 in education has been neglected and devalued, effec-
tive communication on the importance of L1 instruction will be required for every stakeholder group,
from policy makers to teachers’ unions to individual teachers to parents to students. Outreach beyond
the Ministry of Education to other government departments will build political consensus around the
simple necessity in basic education—in terms of human capital, and indeed finance—of using languages
that children understand. This outreach and advocacy is critical for policy uptake and sustainability.
Communications strategies should use multiple approaches, including direct messaging from one stake-
holder group to another (such as Ministry of Finance officials to Ministry of Education officials, and
education officials to parents), media (including radio, television, and print, as appropriate), and stake-
holder discussion groups (particularly at the community level). For more information and resources, see
the Handbook on Language of Instruction Issues in Reading Programs (USAID, 2019) and Good Answers
to Tough Questions in Mother Tongue-Based Multilingual Education (Young and Trudell, 2016).




Language and TLM development
STANDARDIZATION OF ORTHOGRAPHIES FOR WRITING SMALL LANGUAGES. Some languages
will require orthography standardization as part of planning for their use. Occasionally, when working
in (demographically) very small languages that have not yet been used for any kind of education or
written communication, a writing system will not have been devised. Others will have competing
orthographies within or across national borders. Before the curriculum and supporting TLM can be
developed, it is vital to assess the maturity of the orthography and technical vocabulary of each L1 to
be used for instruction. If these are not sufficiently standardized, supporting language development
on these two fronts will need to be part of the implementation process.

“ACADEMIC VOCABULARY” DEVELOPMENT FOR CERTAIN SCHOOL SUBJECTS WILL BE NECES-
SARY. Languages used for instruction require an adequate academic vocabulary for use in subject-spe-
cific academic classes. Specialized words and phrases such as “long division”, “denominator” or
“reciprocal” will need to be selected or even created in a particular L1. This is true for both written
and unwritten languages. Support may be needed to develop the technical vocabulary necessary to
use students’ L1 for content in all academic subjects taught.80 This may be done with “loan words”
from another language, the creation or selection of terms in the student’s home language, or a mix.
Classroom teachers who toggle back and forth between an L1 and L2 as they seek to explain curric-
ular concepts in multilingual classrooms are an ideal resource for academic vocabulary development.
The standardization of terms paves the way for development of high-quality L1 teaching and learning
materials, lesson plans, and instruction.

L1 INSTRUCTION REQUIRES HIGH-QUALITY, CONTEXTUALIZED TEACHING AND LEARNING
MATERIALS (TLM). Strong curricula require adequate teaching and learning materials. Teacher



SECTION 5: EFFECTIVE LOI POLICY EXAMPLES EXIST BUT ARE NOT BEING WIDELY IMPLEMENTED                           41
     guides with structured lesson plans, textbooks, and supplementary reading materials for reading prac-
     tice and content knowledge development are just a few of the inputs needed. Teacher guides must
     use language that teachers will comprehend, and include content and teaching methods appropriate
     to the context and available resources. Content difficulty and instructional pacing must be aimed at
     the “right level” so that instruction focuses on students’ learning challenges. TLM developers should
     use existing evidence to determine the knowledge and skills of children when they begin to learn, and
     how quickly they can and should progress. Too many curricula are aimed at the top 10–15 percent of
     learners and leave more than four out of five students in Learning Poverty. TLM development can
     draw on existing materials, global and cross-border resources (for cross-border languages). Trans-
     lation or adaptation of existing material can be mixed with fully local development. Sources could
     include Ministries of Education, local publishers, local bookstores, local markets, and multilingual
     digital libraries such as African Storybook, the Global Digital Library, Pratham’s Storyweaver, the
     Bloom Library, or the Asia Foundation resource platform.81

     THE CREATION OF TEACHING AND LEARNING MATERIALS CAN DRAW UPON LOCAL RESOURCES
     AND GLOBAL PUBLIC GOODS OR PLATFORMS. Using existing market survey approaches or devel-
     oping country-specific instruments, teams can review existing resources in each language of instruc-
     tion, including braille and sign language, to assess whether new TLM are needed. The review should
     consider all languages of instruction, the quality of the writing, the quality of the pedagogical approach,
     cultural relevance, inclusion, and all subjects of the curriculum. Resources within the relevant country,
     cross-border resources (in the case of cross-border languages), and resources available through global
     platforms should be considered. If new TLM are needed, a decision will need to be made on whether it
     is best to fill those gaps by versioning (translating and adapting) existing open source (open licensed)
     resources or whether materials should be newly developed. ICT tools can be particularly helpful for
     materials development (these are discussed in a later part of this section).

     CONSULTATION IS CRITICALLY IMPORTANT TO ORTHOGRAPHY AND TLM DEVELOPMENT. A
     diverse range of stakeholder perspectives should inform discussions on new resource development for
     an L1 language. Wide consultations, for example, help a new orthography make its way into the written
     culture of a language community, in and outside of school. The process requires buy-in from influen-
     tial stakeholders and institutions that use written language, including the education system, religious
     and civic institutions, and the publishing and broadcasting industries. Developing language-specific
     and technically sound TLM requires technical expertise, and knowledge of languages and culture.
     Successful examples from Ethiopia and Ghana showed that the key to success was to involve (and
     building the capacity of ) government counterparts.82 Good planning includes stakeholder consulta-
     tions throughout all stages of TLM development, from planning to use. In some countries, the textbook
     industry (international or national) is a powerful partner of the Ministry of Education and should be
     included in stakeholder consultations. This will ensure that the rationale for effective LoI policy is well
     understood, and that plans for materials development and procurement are supported accordingly.

     LOI POLICIES CAN HELP BUILD MORE INCLUSIVE EDUCATION SYSTEMS. Appropriate LoI policies
     ensure that teaching methods and instructional materials represent and support learning for all chil-
     dren, including children with diverse educational needs. Expanding the use of braille, developing sign
     languages, and developing additional instructional tools for students who need support will allow the
     improved LoI policy to benefit all students. Effective policies on language create educational environ-
     ments that pay attention to the needs of all students, starting with the need to understand instruc-
     tion and to be able to participate in the classroom. The use of L1 instruction increases the use of
     child-centered teaching methods, which build environments that are conducive to learning.83 The
     choice of L1 instruction empowers teachers to use strong pedagogical practices and allows students



42                                                        LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
to fully engage with the material, ultimately creating a learning envi-
ronment in which all students can participate. The use of L1 also
                                                                                     Careful teacher
allows parents and communities to support children's learning
more effectively.                                                                      recruitment,
                                                                                  selection,support, and
                                                                                  matching (of student
Teacher allocation, professional development,                                     and teacher language
and recruitment as seen through a LoI lens                                       competencies) is critical
TEACHERS ARE A CRITICALLY IMPORTANT DETERMINANT OF                                    to effective L1
STUDENT LEARNING IN ANY LANGUAGE.84 Teachers are the most                               instruction.
important factor in student learning. The bilingual or multilingual
classroom is no exception, and careful teacher recruitment, selection,
support, and matching (of student and teacher language competencies) is
critical to effective L1 instruction. Good LoI policies make teachers’ jobs easier
and ensure that teachers with the right skills are in the right classrooms. Good poli-
cies in effect support and develop all teachers in all contexts.

THE PROFICIENCY LEVELS, ATTITUDES, AND BELIEFS OF TEACHERS AND STUDENTS SHOULD
BE DOCUMENTED. The primary language competency of effective teachers in a bilingual learning
environment is biliteracy: oral and written fluency in both languages. Some teachers will have oral
competence in L1s, but they may lack corresponding written L1 competence and L2 competence.
Other teachers are more likely to be fluent in oral and written L2, but less able or willing to use L1 for
classroom teaching. Identifying teacher competencies is an important step for appropriately planning
teacher recruitment and posting. This should be done according to language needs and language-spe-
cific professional development for teachers. In some countries, particularly those with centralized
posting systems, this will entail a significant cultural and political shift and needs to be well-planned,
with extensive stakeholder engagement, particularly on the part of teachers’ unions. Teachers’
beliefs and preferences about L1 instruction will influence the implementation of language policies.
Conducting a thorough assessment of these factors and involving stakeholders in policy discussions
will provide policy makers with a good understanding of the gaps that need to be addressed to ensure
effective language of instruction policy implementation.

LANGUAGE PROFICIENCY SHOULD INFLUENCE TEACHER RECRUITMENT, DEPLOYMENT AND
PROFESSIONAL DEVELOPMENT. Language mapping exercises help to identify where teachers lack
relevant language proficiency. Short-term solutions have involved contracting community members
as teaching assistants or translators. Political and professional sensitivities tend to beset recruit-
ment. Proficiency assessment may make teachers uncomfortable for fear of demotion or demission.
Teachers may fear loss of advancement opportunities if their skills are associated with languages used
in remote regions with less desirable professional conditions. Development partners, governments,
and teachers’ unions should participate to determine effective and acceptable approaches to teacher
deployment.

TEACHERS NEED LOI-SPECIFIC SUPPORT INTEGRATED INTO LARGER SUPPORT SYSTEMS. Effec-
tive teaching in L1, and later L2, requires specific pedagogical skills and tools.85 Support systems
must help teachers to acquire the necessary linguistic, pedagogical and other expertise.86 Language
considerations need to inform professional development before and during employment.87 This can
be as simple in the short term as ensuring that training is delivered in a language that teachers speak



SECTION 5: EFFECTIVE LOI POLICY EXAMPLES EXIST BUT ARE NOT BEING WIDELY IMPLEMENTED                          43
     and understand, or become a longer term approach, such as helping teachers strengthen their L1 and/
     or L2 literacy skills. In Ethiopia, the Reading for Ethiopia’s Achievement Developed (READ) program
     revised preservice training to align it with the new curricula and to support seven regional languages
     of instruction.88 Subject-specific pedagogical training, especially for early grade reading instruc-
     tion, should integrate language and other pedagogical considerations.89 Adapting in-service training
     should be a priority, as it is probably the fastest way to reach large numbers of active teachers, but
     pre-service training is also important, especially in countries with high teacher turnover and rapidly
     increasing enrollment. Teachers are not generally opposed to upgrading their skillset, but it must be
     understood that their teaching techniques will have been acquired over the course of several years,
     through studying and later teaching. The literature highlights the effectiveness of explicit instruction
     techniques and of coaching. Kenya,90 Malawi,91 and the Gambia92 have documented their success in
     using structured lessons plans in an L1 as has Pratham in India. Coaching programs likewise have well
     documented successes.93 Box 8 and Annex F describe the success in India of the Language & Learning
     Foundation’s comprehensive support model.




     Implementation and its challenges
     POLICIES SHOULD BE MATCHED TO THE SPECIFIC IMPLEMENTATION CHALLENGES. Too often,
     the policy dialogue around LoI policies fails to take sufficient account of country contexts. Low- and
     middle-income countries present a range of contexts, including:
        • Context 1: Homogeneous L1 speech communities learning in a single L2. Often the challenge for
          LoI policy is to bring a mostly or completely linguistically homogeneous group of students (and
          teachers) who speak an L1 to proficiency in a single L2. This may be under “additive” or more
          likely subtractive dual language instruction. It is often pursued by simply mandating instruction
          in an L2. Fortunately, homogeneous speech communities are more likely to have teachers who
          are L1 speakers. They can follow the evidence-based approach of (a) teaching in L1 in the pri-
          mary grades and focusing on establishing literacy and numeracy in L1; (b) introducing L2 as a
          foreign language once L1 threshold competencies have been achieved and developing initial oral
          language abilities in L2, also in the primary grades; and (c) developing a transition to L2 instruc-
          tion once threshold competencies are achieved in L1 and L2 with continued instruction in L1
          throughout the primary and preferably lower secondary years. This will maximize literacy levels
          in L1 and L2, as well as content learning.
        • Context 2: Moderately linguistically diverse speech communities learning a single L2. The sec-
          ond context exists when students within a single classroom speak several languages. The correct
          policy solution depends on the number of languages spoken, the existence or absence of a com-
          mon language between students, and the proficiency levels of students and teachers. There may
          be an L1 that is spoken by many students and learned early by others. There may be two main
          L1s and teachers may have to teach in both. There may be a regional language that all students
          will learn as they grow, and with which they are already somewhat familiar when they begin to
          attend school. Schools can also be so microdiverse that community assistants are required for
          small groups of children. These situations demand contextual analysis and communication to
          develop solutions acceptable to all.
        • Context 3: Homogeneous L1 speech communities seeking to learn two L2 languages. Some
          countries pursue policies that seek to have children master two L2 languages in addition to their
          L1. One of these is often a Language of Wider Communication, while the other might be an in-
          ternational language whose use is connected to a colonial past. In Kenya, for example, students




44                                                       LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
   BOX 8: Language & Learning Foundation (LLF) - Transforming language teaching in India

  Founded in 2015 with the goal of improving language learning outcomes of children in grades 1 to 3
  in home and school languages, LLF presently works with seven state governments and will reach 17
  million children through their programs by 2021. To build government commitment and capacity to
  plan and implement early language learning programs at scale, LLF focuses on the three dimensions
  shown in the figure below.

  As part of continuous professional                                          LLF’s Three-Pronged Strategy
  development courses, LLF designs and
  implements blended capacity building
  courses of varied duration on early
                                                                 1                               2                                 3
                                                           CONTINUOUS                      SCHOOL LEVEL                        SYSTEMIC
  literacy and multilingual education                     PROFESSIONAL                    INTERVENTIONS                        REFORMS
                                                          DEVELOPMENT                   School demonstration                 Systemic reform
  for teachers and teacher educators.                        COURSES                   projects that showcase              related to effective
                                                      Continuous preofessional         transformed teaching-            implementatioon of high
  The courses provide conceptual                      development of teachers,         learning processes at a            quality early learning
  understanding of early language                      teacher educators, and
                                                     educational administrators
                                                                                      reasonable scale and the
                                                                                         support government
                                                                                                                            programs of scale.

  learning as well as classroom teaching                on early literacy and           agencies  to scale the
                                                       multilingual education                 program.
  and learning activities and materials.
  Designed with the aim of creating
  conviction and capacity within the                                            TWO MODES OF INTERVENTION

  education system at all levels, these
  courses play a key role in supporting                       Direct Implementation                      Technical Support to State Governments

  large-scale literacy and multilingual
                                                                    INTERNAL MONITORING, PROCESS STUDIES AND
  programs and sustaining them over the                          RESEARCH TO GENERATE EVIDENCE AND KOWLEDGE
  long term. The courses and workshops
  precede implementation of school-level
  projects and range from one-year flagship courses to one-month courses on themes within early
  literacy and multilingual education. Almost always oversubscribed, the courses have helped LLF build
  credibility around its leadership for early literacy and multilingual education. Some of the courses are
  certified by Indian universities. External evaluations
                                                 !             of the nine-month course have shown significant
  gains in teacher knowledge and implementation of positive changes in teaching–learning practice,
  especially the strategic inclusion of children’s home languages.

  For school-level interventions, LLF designs and implements projects that demonstrate transformative
  changes in the teaching and learning of languages and improve student learning. These comprehensive
  projects aim at building capacity of teachers and school heads through training workshops,
  orientation of government staff responsible for school supervision and support, regular academic
  support to teachers through classroom visits, and review meetings. They also offer mobile phone-
  based sharing of resources, provision of instructional materials like workbooks, grids for decoding
  and reading materials (such as reading cards and storybooks), the creation of print-rich classrooms
  and regular student assessments. Each LLF mentor provides academic support to teachers in 50
  schools, in collaboration with existing government monitoring staff. External evaluations of LLF’s
  school programs have shown significant learning gains for children.

  Lastly, as part of systemic reforms, LLF works with state education institutions to influence the
  state-wide annual in-service training of teachers, school-based assessments, review and revision of
  textbooks, and development of an improved curriculum for pre-service teacher training, with a focus
  on multilingual education.




SECTION 5: EFFECTIVE LOI POLICY EXAMPLES EXIST BUT ARE NOT BEING WIDELY IMPLEMENTED                                                                45
         learn in Kiswahili and English in primary school. For many Kenyan students, neither of these
         is their L1. Good research on success in more than two foreign languages in the primary grades
         is limited, but experience suggests that L1 instruction with L2 and L3 taught first as foreign
         languages produces best results. Clearly, the amount of time needed to teach foreign languages,
         especially two foreign languages, in the primary grades, will reduce already limited instructional
         time for the teaching of foundational skills and content knowledge.
       • Context 4: Highly diverse speech communities seeking to learn at least one L2. Traditional set-
         tlement patterns, urbanization, migration, displacement, and a number of other factors make
         multilanguage communities, schools and classrooms a reality. Although they are not as prevalent
         as sometimes suggested,94 multilanguage communities in many countries are expanding.95 Stu-
         dents in multilingual settings still need good LoI policies, with some creativity and flexibility in
         implementation. Specific needs include the identification of languages, teachers with proficiency
         to match that of students, linguistically and culturally relevant teaching materials, along with
         assessment methods and curricula that follow the principles outlined above in terms of align-
         ment with L1 instruction. Strategies for working effectively in these settings include conducting
         language mapping, using a language with which the majority of students are familiar (often a
         regional language) instead of each student’s L1, using L1 speakers from the community as assis-
         tant teachers, involving parents and other community members in classroom support, and pro-
         viding multilanguage materials. For a more detailed discussion on these strategies, refer to Good
         Answers to Tough Questions in Mother Tongue-Based Multilingual Discussion.96 Ultimately, while
         diversity of language in the classroom does obtain in some settings, the approach to successfully
         ensuring learning for all students relies on appropriate language and educational services. This is
         particularly the case when there are high levels of linguistic diversity in education systems, often
         in urban settings, such as the New York City Public School System (see Box 9).


     ONE SIZE DOES NOT FIT ALL: DIFFERENTIATED IMPLEMENTATION FOR SPECIAL CIRCUM-
     STANCES: Additional challenges to implementation exist when situations or student needs are not
     those of a traditional school system:
       • Fragile and Conflict-Affected (FCV) Settings: FCV settings require a unique approach to LoI. Fra-
         gility and conflict often result in long-term population displacement. On average, refugees spend 10
         to 25 years98 outside their home country or region. Their displacement is not a transitory human-
         itarian crisis, but a long-term one that requires a long-term approach to service delivery. A large
         influx of displaced populations into host communities requires strategic planning to ensure that
         children participate in instruction in a language they know. Additional challenges can include gaps
         in schooling resulting in lower than expected L1 literacy levels, lack of alignment between LoI in
         the host country and the displaced children’s country of origin, socioemotional challenges requir-
         ing additional support, and a lack of teachers fluent in children’s L1.99 It is important to consider
         political sensitivities as well, including community perspectives regarding newly arrived children,
         the perceived responsibility of host communities to offer and customize education services to the
         displaced population, and the roles of international actors assisting with the provision of schooling.
       • Children with Diverse Educational Needs: Good LoI policies allow all children to learn in a lan-
         guage they know well. However, attention is needed to ensure that no children are left behind.
         Many children who do not attend primary school have a disability.100 In developing countries, 42
         percent of children with disabilities are out of school at the primary level and 56 percent at lower
         secondary level.101 Estimates from 19 low- and middle-income countries show that children with
         disabilities are 13 percentage points more likely never to be enrolled in school, 15–18 percentage
         points less likely to finish primary school, and about 16 percentage points less likely than their
         peers to be literate.102 When developing LoI policy, it is critical to consider how learners with
         disabilities will be supported. Blind children use braille, and thus the LoI policy must address the



46                                                      LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
  BOX 9: New York City’s English Language Learner Program97 Screens and Allocates
  Students among Three Main Interventions
  The New York City (NYC) public education system serves a very diverse student body; its more
  than one million students speak 176 different languages. More than 42 percent of students enrolled
  in the city’s public schools come from households in which English is not the primary language of
  communication. To meet the need of its student body, the NYC Department of Education (NYC DOE)
  has put in place an English Language Learner (ELL) program in all public schools.

  To identify students who qualify for ELL, parents or guardians are asked to fill out a Home Language
  Identification Survey (HLIS). If a language other than English is indicated, the child will take the New
  York State Identification Test for English Language Learners (NYSITELL) to determine their English
  language proficiency. Students who score below a State-determined level on the assessment are
  identified as ELLs and entitled to ELL programs and services.

  The NYC DOE offers three modalities of ELL programs aimed at promoting proficiency in the students’
  home language and English. The modalities are Dual Language (DL), Transitional Bilingual Education
  (TBE), and English as a New Language (ENL). Parents or guardians of ELLs choose one of these
  modalities.
      1. English as a New Language (ENL) programs are offered in all New York City public schools.
         These programs aim to develop English language proficiency and are taught in English as
         stand-alone ENL classes and/or integrated ENL instruction in subject area classes.
      2. Transitional Bilingual Education (TBE) programs are designed so that students develop content
         knowledge in their home language as they learn English. Schools provide English language
         arts (ELA), home language arts (HLA), and subject-area classes in students’ home languages
         and English. Students’ home language is used until the student scores at or above a State-
         determined level of English proficiency, whereupon they are integrated into English classrooms.
      3. Dual Language (DL) programs are designed to develop bilingualism, biliteracy, and cross-
         cultural understanding. Through these programs, English-proficient students are also given
         the opportunity to learn a new language. DL programs are offered from pre-kindergarten to
         high school.


  ELL Program Modality

                            NO. OF SCHOOLS              LANGUAGES                                    ENROLLMENT AMONG ELL
                                                        (NUMBER OF PROGRAMS)                         STUDENTS (%)
   ENL                      All public schools          English                                      125,271 (81.2%)
   TBE                       291                        Arabic (4); Bengali (4); Chinese (41);       18,047 (11.7%)
                                                        Haitian Creole (2); Polish (1); Russian
                                                        (1); Spanish (231); Yiddish (7)
   DL                       267                         Arabic (1); Bengali (1); Chinese (26);       9,583 (6.2%)
                                                        French (9); Haitian Creole (2); Italian
                                                        (2); Japanese (1); Polish (1); Russian
                                                        (4); Spanish (219)

  Among ELL qualifying students 1,375 (0.89%) are not served. DL programs also serve non-ELL students, whose enrollment figures are not
  reflected here.




SECTION 5: EFFECTIVE LOI POLICY EXAMPLES EXIST BUT ARE NOT BEING WIDELY IMPLEMENTED                                                       47
          question of whether braille is available for the L1 in question or whether it needs to be developed.
          Similarly, deaf children who sign will need to learn sign language for L1 and L2. To ensure learning
          for all, schools, teachers, and students need resources to address specific educational needs such
          as differentiated instruction, braille, sign language, hearing loops and speech-to-text technology.103


     ANTICIPATION OF IMPLEMENTATION CHALLENGES INCREASES EFFECTIVENESS. Successfully
     implementing LoI policies requires early consideration of key implementation challenges. Ethiopia’s
     experience (Box 10) speaks to the importance of persistence when teaching in many languages and
     sustaining policy implementation over the long term.


        BOX 10: Multilingual Education in Ethiopia
       Ethiopia is very ethnically and linguistically diverse, with over 80 officially recognized languages and
       over 200 dialects.104 The five largest ethnic groups account for about 78 percent of the population (34.4
       percent Oromo, 27 percent Amhara, 6.2 percent Somali, 6.1 percent Tigray and 4 percent Sidama),105
       with ethnic and linguistic diversity greatest in the western and southwestern regions of the country.

       In 1994, this ethnolinguistic diversity was recognized by the Ethiopian Educational and Training Policy
       (EETP). The EETP explicitly recognized that L1 education has pedagogical advantages for children’s
       learning and made primary education in ‘nationality languages’ compulsory, to be followed by a
       transition to English as a medium of instruction in secondary and higher education. Amharic was to
       be taught as a language of countrywide communication.

       In the early years of policy implementation, larger languages were quickly adopted as a medium of
       instruction in almost all regions. Gradually, the number of languages used increased and, as of 2015, about
       30 languages were being used as a medium of instruction for part or all of instruction in the primary grades,
       with about 51 languages being offered as a subject.106 In all regions of the country, English is introduced as
       a subject in grade 1, while Amharic is introduced as a third language, often starting in grade 3 or 5.

       However, despite strong commitment to the policy, implementation continues to face challenges.
       Three to highlight are:
       • Significant variation across languages in their level of development and readiness for use in educa-
         tion: A lack of focus on language development could lead to widening inequality across groups.107
       • Lack of trained teachers: In the early years of implementation, finding trained and qualified teach-
         ers who spoke the needed languages was challenging.108 However, Ethiopia has made gradual but
         significant progress in addressing this gap. Pre-service teacher training programs for lower-pri-
         mary teachers in many of these languages are gradually being introduced in Colleges of Teacher
         Education; in-service teacher training programs have been expanded to upgrade the qualification
         of unqualified teachers; and teachers are also recruited locally by woredas (districts).109
       • High costs of providing textbooks and other teaching and learning materials: For regions where
         there are many demographically small languages, the cost of printing textbooks and TLM contin-
         ues to be high, due to fragmented, more localized procurement and smaller print runs. Further,
         for many of these languages, written materials other than textbooks remain scarce.

       Despite the challenges, EETP is yielding promising results. Evidence suggests that L1 instruction is
       having positive impacts on years of schooling as well as learning outcomes.110 Nevertheless, the
       Ethiopian case highlights the issues that must be addressed for linguistically diverse students and
       countries to thrive. (For further detail on the case of Ethiopia, see Annex A).



48                                                           LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
Assessing progress and adjusting policies
LANGUAGE(S) OF ASSESSMENT AND INSTRUCTION SHOULD BE CAREFULLY INTEGRATED. Assess-
ment is key if students, families, teachers and policy makers are to understand learning progress. To
be effective, the language(s) of assessment should reflect the language(s) of instruction. If it does not,
assessments that are meant to reveal levels of mastery of specific material will be stymied by inade-
quate knowledge of the language of assessment. A mathematics problem expressed in words (rather
than figures and symbols) designed to find out if a student can, for example, answer “What does two
plus two make?” is meant to reveal a student’s knowledge of arithmetic. But many students get such
questions wrong even when they do have that knowledge, and could in fact answer correctly in an L1
or a language they speak and understand better. When tests are de facto testing language proficiency,
then the ostensible subject-matter tends to be obscured, and the results biased. This is true for a range
of assessments, including national large-scale assessments and exams (of much greater consequence
for a student’s future than regional or international large-scale assessments).

ASSESSMENT OF L1 LITERACY IS FUNDAMENTAL TO PLANNING FOR TIMING OF ANY L2 TRAN-
SITIONS. Children’s ability to acquire literacy skills in an L2 depends on their L1 reading skills111 and
their L2 oral language skills.112 Thus, for L1 to L2 transition to succeed, children’s fluency in both
L1 and L2 must be assessed. Along with the development of a language-specific scope and sequence
for reading instruction and other instructional approaches needed for successful bilingual education
programs, emphasis should be placed on consistency across assessments as well as alignment with the
curriculum and the LoI model.113

THE LANGUAGE OF EXAMINATIONS SENDS STRONG SIGNALS ABOUT WHAT LANGUAGE ABIL-
ITIES ARE VALUED. Subjects that are tested are subjects that are taught; effective LoI implementa-
tion requires the inclusion of L1 competencies in classroom testing and on national exams, particu-
larly high stakes exams. In many countries, exam results dictate access to higher levels of education
and employment. As such, exam success is the highest priority for teachers, school administrators,
students, and parents. Fair, transparent, and consistent language policies should govern examinations,
and these should be clearly communicated to stakeholders.

MISMATCHES BETWEEN LANGUAGES OF ASSESSMENT AND INSTRUCTION MAY INVALIDATE
ILSA SCORES. International large-scale assessments are less important to local stakeholders, but still
important to policy makers. As Figure 4 showed, millions of children are being assessed in a language
that they do not understand and in which they are not taught. This results in assessment outcomes
that give an unfair advantage to the few who use the assessment language at home. These assess-
ments serve better as a measure of foreign language learning than of skills or content knowledge.




Technology could greatly improve implementation effectiveness if used well
LEVERAGING TECHNOLOGY TO SUPPORT IMPROVED LOI POLICIES AND PROGRAMING. Various
forms of technology—educational, communications, and others—are powerful tools to help students
reach proficiency in their L1 and any L2s. Becoming literate, learning a language as a foreign language,
and mastering academic content in a given LoI can all be aided by astute use of technology. Perhaps
more importantly for the near term, technology can help provide the information needed to make
and implement good LoI policies, develop effective teaching and learning materials, and aid in their
monitoring and evaluation.



SECTION 5: EFFECTIVE LOI POLICY EXAMPLES EXIST BUT ARE NOT BEING WIDELY IMPLEMENTED                          49
     TECHNOLOGY CAN HELP IMMEDIATELY, AND IN THE MEDIUM AND LONG TERM. Technology is
     available to help solve immediate problems in LoI policy and programing. It also holds great promise
     for providing additional tools. Successful harnessing of ICT for LoI will proceed on two fronts: a
     greater ability to use existing technology effectively, and the ever-growing power of technology to
     provide tailored solutions for diverse education needs. Discussions of technology’s use for LoI will


       BOX 11: How does Language of Instruction affect Test Scores?

       Students are normally tested in the official language of instruction, especially for international large-
       scale assessments (ILSAs) and national large-scale assessments. This works well in monolingual
       contexts but creates numerous problems in multilingual ones. Students for whom the language of
       instruction is not L1 systematically score lower than students for whom it is.

       The literature documenting lower scores in L2s is voluminous and consistent. Mullins et al. (2012)
       found that 92 percent of students from the 48 countries tested in the 2011 PIRLS assessment spoke
       the language of the test at home. Those not tested in their L1s scored significantly lower: 479 versus
       516, or more than one-third of a standard deviation. Glewwe et al. found that linguistic minority
       students fall behind very early in their school experience and have a hard time catching up.114 Other
       authors have documented the strong correlation between being taught in one’s L1 and persisting in
       primary school.115

       Do those lower scores indicate lower achievement, test bias, or both? If there is test bias, is it due to
       bad translation, or to deeper “configural problems” when the constructs themselves fundamentally
       differ between languages? These are the key questions. Translation problems are relatively easy to
       spot and fix with enough resources, but configural problems pose a greater challenge.

       At the heart of the problem is the fact that tests in a single language cannot distinguish between
       students who answer incorrectly because they truly do not know the construct and others who could
       have answered correctly if only the question had been posed in their L1. When students are grouped
       and tested in their respective L1s, post-test analyses can in theory estimate the extent of bias between
       test versions. In practice, this way of testing and analyzing results is rare outside the main languages
       spoken in OECD countries.116

       One part of the story is clear: the problem is not the inability of bilinguals or multilinguals to achieve
       as well or better than monolinguals. Collier and Thomas117 found that when students receive enough
       high-quality instruction in both languages, bilinguals will eventually outperform monolinguals even
       in the monolinguals’ L1. But instruction is almost invariably suboptimal, and so it remains of crucial
       importance to find out by how much a linguistic minority is lagging behind: and why.

       The problem is becoming more important as testing coverage expands globally. ILSAs were initially
       designed for—and first undertaken in—OECD member countries, which tend to be more linguistically
       homogeneous than non-OECD countries.118 In 2000, PISA (Program for International Student
       Assessment) had 41 national test “versions” in 25 languages for 30 participating (OECD member)
       countries; by 2006, 77 versions in 42 languages were given (the increase accounted for by non-OECD
       member countries). The expansion “added considerably to the challenge of ensuring equivalence and
       fairness of instruments across all participating countries.”119 The challenge may well be formidable,
       but it is only by testing students in their L1s, and appropriately analyzing differences between language
       groups, that stakeholders can be confident that assessments are measuring learning.



50                                                         LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
be most effective if they focus on the following questions: (a) what
can we use right now and what can we use it for? (b) what can help
us in the medium term and what would we need to do for it to                     Technology is available
be helpful? (c) what exists now but can only be used effectively                to help solve immediate
if traditional barriers to technology use are overcome? and (d)
what new technologies will further improve learning opportu-                      problems in LoI policy
nities if their potential can be actualized to help students every-              and programing. It also
where to learn at the right level?
                                                                                   holds great promise
EXISTING TECHNOLOGICAL RESOURCES CAN ALREADY MAKE                                for providing additional
THE CREATION AND ADAPTATION OF TLM EASIER. Existing
software programs and other tools make the development of TLM
                                                                                             tools.
in target languages much easier, faster, and cheaper. Numerous tools,
including many that are freely available public goods, expand and accel-
erate the creation of key TLM resources. This is most true right now for
early grade reading materials: critical for literacy and lacking in many countries
and languages. Existing tools to create or adapt reading books for practice include
Bloom software from SIL, African Storybook, Pratham’s Storyweaver, and the Global
Digital Library. Tools to develop and adapt student textbooks and teacher’s guides include Primer Pro
and SynPhony. A wide range of linguistic information and materials is available on the Ethnologue
database. These programs help determine appropriate linguistic scopes and sequences and produce
“decodable” word banks and then help create and adapt “decodable story books” that children need
to practice the skills they learn during early grade reading instruction. Bloom, African Storybook, and
Storyweaver have tools for versioning thousands of existing open licensed reading books into new
languages. Early-grade reading programs supported in more than 40 countries by USAID, such as
the Uganda School Health and Reading Program (SHRP)120 or the Ghana Learning program,121 have
used these software programs to produce TLM in more than 80 languages. An additional USAID-sup-
ported initiative used Bloom to develop books through author workshops in six countries, a process
that resulted in 3,000 decodable and leveled books in 15 languages.

EXISTING SOFTWARE CAN IMPROVE LINGUISTIC SCOPES AND SEQUENCES FOR CURRICULA.
Good early grade reading instruction follows a scope and sequence consistent with the characteris-
tics of the language of instruction. More common and easier sounds and letters (symbols) are taught
before less common and more complex ones. These differ by language, and existing software can iden-
tify them. Once identified, they can be properly integrated into effective “scopes and sequences” for
the introduction of letters and sounds in reading instruction in alphabetic languages. For alphabetic
languages (those that use a set of letters to compose words), Primer Pro122 and SynPhony123 analyze
selected written texts to determine the frequency of individual letters, syllables, and other linguistic
information that can then be used to develop a scope and sequence with materials for teaching reading
in that language. Quality control processes for TLM and scope and sequence development can also be
strengthened with technology, as was the case in Ghana, where an e-filing system was key to devel-
oping materials in 11 languages.124

ADDITIONAL RESOURCES CAN AID LANGUAGE MAPPING AND LANGUAGE DEVELOPMENT. The
Ethnologue database is the most authoritative source of information on languages around the world.
It is not, however, a substitute for full language mapping exercises; it lacks granularity and depth of
information on speaker numbers and characteristics. It can nevertheless serve as a valuable point of
departure for governments to learn or confirm linguistic information and what gaps in knowledge need
to be filled through full-scale language mapping. Additional resources exist for languages that require



SECTION 5: EFFECTIVE LOI POLICY EXAMPLES EXIST BUT ARE NOT BEING WIDELY IMPLEMENTED                         51
         orthography and/or vocabulary development or standardization. The World Atlas of Language Struc-
         tures is a large database of phonological, grammatical, and lexical properties of languages.125 Manuals
         for orthography development126 or standardization127 also exist. No technological resources currently
         exist to support the development of academic vocabulary, but these could easily be developed as a
         database of words in mathematics, science, social studies, history, art, music, foreign languages, phys-
         ical education, and so forth, for use in instruction of all school subjects, through the medium of any
         given language.128

         DATA MINING COULD MAP LANGUAGE USE AND PROFICIENCY LEVELS IN THE MEDIUM TERM.
         Countries need to know who speaks what languages and to what levels of proficiency. This informa-
         tion needs to be periodically updated. Mining data from electronic communications represents the
         best option for doing this on an ongoing basis. Smartphones with social media and communications
         apps are becoming ubiquitous in many countries. With appropriate safeguards, the data they produce
         could be used to analyze what languages are used and how they are used. Similar to how cellphone
         data is revolutionizing “contact tracing” for public health purposes, language use online could become
         the cost-effective source of information on language use needed for LoI policy formulation and imple-
         mentation. Analysis and processing of this data could be aided by, and in turn help strengthen, existing
         networks of linguists and language experts and planners such as the African Linguistic Network.129
         To collect information about community perspectives on L1 as well as educational policies, a variety
         of cost-effective tools such as Survey Solutions, Geo-Enabling initiative for Monitoring and Evalua-
         tion, Survey CTO, among many others, can be used as appropriate.130 Adaptation and use of tools for
         automation of phone surveys in multiple languages, such as Viamo, could and should be expanded.131
         Existing and new resources to address issues related to cost-effective language mapping are emerging
         almost daily.132

         TECHNOLOGY CAN TEACH INDIVIDUAL STUDENTS “AT THE RIGHT LEVEL” IF USED WELL. This
         paper advocates that countries adopt policies under which each student first learns in their L1
         throughout primary school and most of lower secondary school. It argues that instruction in roughly
         550 additional languages worldwide would serve 84 percent of all students in their L1s. The problem
         is massive, and solutions will need to be conceived on a commensurate scale. Existing software
                            programs such as Duolingo’s “Duo ABC” are designed to make children literate;
                                 such programs will continue to proliferate. Their use is constrained by the
                                    myriad of challenges affecting the use of technology in classrooms, espe-
                                        cially for young learners, and in low-income or generally under-re-
                                          sourced settings. As these all too familiar challenges are overcome,
                                           more tools will be available for students for individualized learning
                                             in L1 and for transition to L2 proficiency.
         The pandemic
      heightens the need                        ARTIFICIAL INTELLIGENCE (AI) FOR LANGUAGE LEARNING
                                                WILL CONTINUE TO IMPROVE. Technology is gaining the ability
       for language and                        to analyze spoken language and compare it to “correct” pronun-
     technology solutions.                     ciation, and to quiz students on the meaning of words, phrases,
                                              and spoken or written discourses. It can automatically review
                                            any interaction and has the potential to present a learner with a
                                          tailored next task. All aspects of language learning, literacy instruc-
                                        tion, and other school tasks are likely to be improved by AI technology.
                                     The equivalent of more than USD five trillion per year is spent on educa-
                                  tion (four percent of the world’s 142 trillion GDP); by contrast, in 2012,
                                  Facebook used the labor of fewer than 4,000 employees to create a network



52                                                          LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
reaching one billion people. This means technology in theory at least allows solutions to be devel-
oped and delivered at global scale with massive cost-effectiveness and economies of scale. The cost
of developing and delivering tailored, effective instruction to every student globally would in theory
cost a minuscule share of annual spending on education. The obstacles to effective use of AI may
continue to be organizational and sociological—or, to put it more bluntly, it cannot be assumed that
just because it can be done, it will be done well (or done at all). Nonetheless, it is vitally important
to maximize the power of AI and new technologies to expand individualized language- appropriate
learning.

THE PANDEMIC HEIGHTENS THE NEED FOR LANGUAGE AND TECHNOLOGY SOLUTIONS. The
COVID-19 pandemic has significantly increased demand for distance learning. At its best, remote
learning allows students to engage with material in an individualized way, according to the learning
levels and languages of each student. While programs to date have primarily focused on mitigating
the impact of school closures, programs that work well during the pandemic could be retained and
integrated into education on a permanent basis. Distance printing and virtual approaches could also
support in-service training and coaching programs for teachers.133




SECTION 5: EFFECTIVE LOI POLICY EXAMPLES EXIST BUT ARE NOT BEING WIDELY IMPLEMENTED                        53
     SECTION 6:
     A new approach to language of
     instruction for the World Bank




     ATTENTION SHOULD BE COMMENSURATE WITH THE IMPORTANCE OF THE CHALLENGE. Issues
     linked to LoI are at the heart of many global education policy considerations. Addressing them will
     present short-term challenges around retooling and retraining, but will provide long-term benefits in
     improved learning outcomes. For too long, the existence of multiple languages within a country has
     been perceived, by development partners and national stakeholders alike, as an unfortunate problem
     for education service delivery. National linguistic realities have simply been ignored, leaving teachers
     and children without the communication skills needed for successful learning. The Bank and devel-
     opment partners should lead the way to ensure, not only that LoI issues receive the attention they
     deserve, but that they are seen as a welcome opportunity for unleashing learning potential. The first
     thing the Bank should “do differently” regarding LoI is to address it consistently as integral to an
     overall strategy to improve learning.

     IMPROVING LOI AND LEARNING OUTCOMES WITH FIVE FUNDAMENTAL PRINCIPLES. The losses of
     learning and foregone human capital can be mitigated with good evidence-based policies on language
     of instruction. Evidence accumulated over the past several decades provides a clear and consistent
     picture: students learn more when taught in their L1. In fact, there are very specific insights on the
     type of LoI model that will yield the best results for students to achieve L1 literacy, L2 literacy, and
     improved overall learning outcomes. The evidence presented in this paper points specifically to five
     key principles that together comprise a new approach to enhance learning in low- and middle-income
     countries:
        • Principle 1: Teach children in their L1 from ECEC through at least the first six years of primary
          schooling. It is critical that instruction be in the language most students speak and understand
          best.
        • Principle 2: Use L1 for instruction in academic subjects beyond reading/writing. Students need
          to master reading and writing in a broad range of disciplines and in all school subjects.
        • Principle 3: Introduce L2 as a foreign language with a focus on oral language skills. Advance to
          L2 literacy and content area instruction only after students have achieved a level of proficiency
          in L1 literacy and L2 oral language. Students can master two languages in basic education if in-
          struction and sequencing are optimized.
        • Principle 4: Continue L1 instruction after an L2 becomes the principal LoI. Some L1 instruction
          continues to improve L2 performance in important ways even after the L2 becomes the language
          of instruction.



54                                                      LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
     • Principle 5: Continuously plan, develop, adapt, and improve the implementation approach for
       LoI policies according to country contexts and goals. The World Bank should work closely with
       client countries and development partners to create and share resources as well as advocate for
       the equitable and high-quality schooling opportunities that appropriate LoI policies provide.

THREE STRANDS OF WORK TO OPERATIONALIZE THE NEW WB APPROACH TO LOI.
It is proposed that the Bank adopts a three-strand approach to addressing LoI issues and promoting policy dialogue:

            Getting up
 STRAND 1




            to speed                 WB staff will work together to assess the current situation with respect to LoI in a
            for proac-               given country, the related political economy issues, and the ways this affects overall
            tive client              education goals and strategies;
            engagement.

            Supporting               World Bank will move from being reactive to being proactive in engaging countries in
                                     policy dialogue on LoI issues. Rather than efforts to mitigate the impact of previous
 STRAND 2




            proactive
                                     policy decisions (such as those that ask teachers to teach in a language they do not
            dialogue
                                     know) the Bank will engage client governments and other stakeholders early to “get
            and actions              ahead” on LoI issues and policies. It will seek to work with governments on a range
            to improve               of measures that address language policy as an opportunity to improve instructional
            LoI policies.            practice and learning, consistent with the existing evidence base; and,

                                     The World Bank will work with other partners to focus attention on LoI issues while
                                     creating, collecting, and disseminating the full range of cutting-edge knowledge and
 STRAND 3




                                     policy solutions and promoting long-term solutions for language issues in education.
            Leading on               Building on existing work, the Bank will define priority areas for research, convene
            LoI issues.              research networks and sponsors, conduct new analytic work, and disseminate work
                                     done by others related to LoI. This will include the engagement of private sector
                                     partners and foundations, along with the full range of stakeholders and development
                                     partners.


THREE STRANDS OF WORK TO OPERATIONALIZE THE NEW WB APPROACH TO LOI. It is proposed
that the Bank adopts a three-strand approach to addressing LoI issues and promoting policy dialogue:
     • Under Strand 1, WB staff will work together to assess the current situation with respect to LoI in
       a given country, the related political economy issues, and the ways this affects overall education
       goals and strategies;
     • Under Strand 2, the Bank will move from being reactive to being proactive in engaging countries
       in policy dialogue on LoI issues. Rather than efforts to mitigate the impact of previous policy
       decisions (such as those that ask teachers to teach in a language they do not know) the Bank will
       engage client governments and other stakeholders early to “get ahead” on LoI issues and poli-
       cies. It will seek to work with governments on a range of measures that address language policy
       as an opportunity to improve instructional practice and learning, consistent with the existing
       evidence base; and
     • Under Strand 3, which will be parallel to Strands 1 and 2 and not sequential, the Bank’s Education
       Global Practice will work with other partners to focus attention on LoI issues while creating,
       collecting, and disseminating the full range of cutting-edge knowledge and policy solutions and
       promoting long-term solutions for language issues in education. Building on existing work, the
       Bank will define priority areas for research, convene research networks and sponsors, conduct
       new analytic work, and disseminate work done by others related to LoI. This will include the
       engagement of private sector partners and foundations, along with the full range of stakeholders
       and development partners.
Each strand of work would be appropriately staffed and funded. Specific products and tasks would be
agreed with management in annual work programs. Program progress would be regularly assessed by
the management team as part of regular Education Global Practice management.



SECTION 6: A NEW APPROACH TO LANGUAGE OF INSTRUCTION FOR THE WORLD BANK                                                       55
     STRAND 1: GETTING UP TO SPEED FOR PROACTIVE CLIENT ENGAGEMENT. The consensus in
     the Education Global Practice is that to date we are too often in the position of addressing LoI
     issues when the government is already implementing or has already announced a policy to use an
     L2 exclusively from the start of primary school. Some of our most vigorous policy engagements
     have been to dissuade governments from pursuing goals they have already publicly announced.
     Operations do occasionally include LoI issues directly in lending related activities and in technical
     dialogue for foundational learning where opportunities present themselves, on an ad hoc basis.
     However, Projects and Project Appraisal Documents on basic education routinely ignore LoI issues
     even in countries where poor LoI policies are clearly an obstacle to learning. Promoting a new
     approach to LoI will reverse this reactive stance and replace it with a concerted effort to “get up to
     speed” on LoI issues. This involves the following activities:
        • Understand how language is being used in education at the country level, including the rea-
          sons (historical and current) for language choice, number of languages spoken in and outside
          of school, teachers’ linguistic skills in L1 and L2, teacher and community attitudes towards
          L1 and L2, policies and stakeholders (national and international) that govern school language
          use, and the state of policy implementation.
        • Collect, review, and understand relevant existing LoI-related work, especially from Min-
          istries of Education, national civil society organizations, and development partners. This
          would include existing language mapping data, assessment results, and all L1 TLM materi-
          als, as a start.
                              • Review the main challenges for every country in light of the stakeholder
                                issues and approaches to good practice described in this document and
                                key related documents.
                              • Undertake a systematic review of World Bank experience supporting
                                improved LoI policies and implementation in lending and analytical
                                work.
                              • Create regional focal points for LoI issues, potentially with overlapping
                                responsibilities for foundational learning.
                              • Expand technical contacts with development partners and national
                                civil society organizations to explore their activities, tools, and lessons
                                learned.
                              • Include LoI dialogue in country, regional, and global network discus-
                                sions around learning, programing, and funding.
                              • Assess LoI issues per country using a scale similar to the one proposed
                                in the chart below.
                              • Initiate work on consultant rosters and TOR templates for key LoI im-
                                provement tasks.
                              • Prepare an annual work program for LoI issues with prioritized actions
                                and regular progress reporting.


                        STRAND 2: PROACTIVE, SUPPORTED DIALOGUE AND ACTIONS TO IMPROVE
                        LOI POLICIES. The second strand would be characterized by support for policy
                        dialogue with clients on LoI issues. The prioritization exercises undertaken
                        in Strand 1 would influence the intensity of engagement. Activities in Strand 2
                        would include but not be limited to:




56                                                       LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
   • Strengthening analytical work at the global and country level, including language mapping for
     countries that lack clear information on languages spoken by school-age children and teachers.
   • Strengthening analysis of the political economy and diverse stakeholders who affect LoI policy
     development and implementation.
   • Increasing support to improve LoI policy implementation in lending, with an expectation that
     basic education lending in all medium- and high-priority countries would address LoI.
   • Developing and disseminating resources and other tools for effective communication and advo-
     cacy with diverse groups of stakeholders.
   • Engagement on language of assessment/examination issues at the national, regional, and global
     level.
   • Developing specific advice for countries with “multiple L2’s” or those that expect students to
     master a Language of Wider Communication and an L2 along with an L1 by the end of basic
     education.
   • Developing mastery-based approaches to transitions from L1 to L2, including minimal thresholds
     for literacy subskills in L1 and oral language ability in L2 and agreement on related assessment.
   • Supporting the use of L1 in ECEC (with a focus on oral language development and print awareness).
   • Advancement of joint activities with development partners, such as the support of L1 instruc-
     tion in foundational learning.
   • Supporting improved classroom observation, teacher training, professional
     development, and coaching with LoI concerns fully integrated.
   • In collaboration with other development partners and organiza-
     tions, determining gaps in existing instructional materials and
     supporting the development of sets of materials (textbooks,
     teacher’s guides, and supplemental reading books) in languag-
                                                                                 More efficient and
     es needed to reach at least 80 percent of students.                     effective use of resources
   • Determining contextually appropriate strategies to reach the             and improved learning
     additional 20 percent of students (working in collaboration
                                                                             outcomes will be among
     with other development partners and organizations).
   • Ensuring that support for children with disabilities is fully
                                                                              the positive results of a
     integrated in all materials and instructional strategies.                   stronger LoI policy
   • Disseminating or developing global public goods for academic                   environment.
     vocabulary development by subject area where needed (mathe-
     matics, science, social studies, and so forth) to allow countries to
     benefit from previous experience.
   • Initiating work on issues of LoI for tertiary education, adult and lifelong
     education, nonformal education, and technical and vocational education.
   • Fully developing analytical and operational advice for the use of dialects in instruction.
   • Deepening policy formulation strategies for FCV, and education for refugees and displaced
     persons.
   • Using open licensing (such as Creative Commons) for all materials developed to ensure wide-
     spread dissemination.
   • Sharing materials developed on global platforms such as the Global Digital Library (GDL).




SECTION 6: A NEW APPROACH TO LANGUAGE OF INSTRUCTION FOR THE WORLD BANK                                   57
     STRAND 3: LEADING ON LOI ISSUES. GIVEN THE CRITICAL IMPORTANCE OF APPROPRIATE LOI
     POLICY IN ADDRESSING LEARNING POVERTY, THE BANK WILL WORK WITH PARTNERS TO ADVO-
     CATE FOR EVIDENCE-BASED POLICY AND PROGRAMING. Ongoing research, country contexts,
     political economy, and student needs are constantly changing, and policy and programing need to be
     responsive. Advances in technology continue to make it easier and less costly to develop, produce, and
     distribute instructional materials in multiple languages. Technology is also making teacher coaching
     and formative assessment more effective and efficient. Economies of scale and collaborative learning
     within and across languages and countries will enhance success. In order to support country, regional,
     and global efforts to develop and implement appropriate LoI policy and programing, the Bank, in
     collaboration with other development partners, will:
        • Promote the use of effective data mining tools for monitoring and assessing language proficiency
          and use by teachers, students, and parents/families as needed for informed policy formulation
          and implementation.
        • Expand support to the private sector for the development, production, and distribution of open
          source instructional materials in L1 languages.
        • Facilitate the rapid and thorough analysis of language characteristics and language family group-
          ings as a means of improving L1 materials, instruction, and assessment.
        • Expand and implement technical work in support of assessment as a way to eliminate system-
          atic measurement distortions, in exams, national assessments, and in international large-scale
          assessments.
        • Promote effective policy and programing solutions, including adaptive materials for blind and
          deaf students, students with developmental language and other delays, and those who use
          non-written L1 languages.
        • Share all materials, strategies, and learning through multiple national and international plat-
          forms and channels.


     WHAT SUCCESS WILL LOOK LIKE. The Bank’s new approach to LoI issues should be organized around
     a clear principle: to enhance the ability of children to learn throughout their school experience, based
     on early acquisition of foundational literacy skills. This means that all children will learn in a language
     they understand, and that all teachers will receive appropriate professional development, including L1
     literacy acquisition if necessary and adequate teaching and learning materials. World Bank assistance
     will result in students who have fully developed their abilities through L1 instruction over appropriate
     timeframes, and acquired additional languages fully and efficiently. More efficient and effective use of
     resources and improved learning outcomes will be among the positive results of a stronger LoI policy
     environment.




58                                                        LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
ANNEXES
     ANNEX A:
     Ethiopia:
     Multilingual
     L1-Based Education



     Context

     Ethiopia is a large and diverse land-locked country in the horn of Africa. The country has a population
     of over 109 million people, which is projected to reach 190 million by 2050. The country has a federal
     structure consisting of nine regional states and two federally chartered cities,134 which are administra-
     tively divided into zones and woredas (equivalent to districts).

     Ethiopia is ethnically and linguistically very diverse, with more than 80 officially recognized languages
     and more than 200 dialects.135 The five largest ethnic groups account for about 78 percent of the popu-
     lation (34.4 percent Oromo, 27 percent Amhara, 6.2 percent Somali, 6.1 percent Tigray and 4 percent
     Sidama).136 Ethnic and linguistic diversity is greatest in the western and southwestern parts of the
     country. The Southern National and Nationalities and Peoples (SNNP) regional state has about 56
     languages, far more than any other state. Historically, Amharic has been used as a lingua franca and as
     the national language. Ethiopia does not have a former colonial language as it was never colonized by
     a European country (apart from a five-year occupation by Italy).



     Ethiopia’s language policy in education
     Primary and secondary education in Ethiopia follows a 4–4–2–2 structure, with eight years of
     primary education (split into a lower cycle covering grades 1–4 and an upper cycle covering grades
     5–8) followed by four years of secondary education (split into a first cycle, grades 9–10; and second
     cycle, grades 11–12). At the end of primary school, a regional examination is administered to deter-
     mine secondary entrance.137 At the end of tenth grade, students take a national examination, which
     determines whether they pursue an academic or a technical and vocational education track. For those
     who move onto the academic track, they must take another national examination at the end of twelfth
     grade to qualify for university enrollment.138

     Despite the presence of extensive linguistic diversity in the country, Amharic, a language that origi-
     nated from the Amhara region, served as the sole medium of instruction in primary schools throughout
     much of the country's modern history.139 English was taught as a subject in all grades and used as a
     language of instruction at upper primary, secondary and post-secondary levels.




60                                                       LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
In 1991, a new government led by the Ethiopian People's Revolutionary Democratic Front (EPRDF)
took power. The EPRDF enacted policies significantly different from previous regimes. It acknowl-
edged the ethno-linguistic diversity of the country and established an ethnic federalist structure
that decentralized power to regional governments. The EPRDF promoted the use of local languages
for administrative and educational purposes. Amharic continued to be the working language of the
federal government.140 The biggest change in educational language policy came with the introduction
of the Ethiopian Educational and Training Policy (EETP) in 1994. The EETP explicitly recognized the
advantages of L1 education for children’s learning. The EETP made primary education in ‘nationality
languages’ compulsory followed by transition to English as a medium of instruction in secondary and
higher education. Amharic is to be taught as a language of countrywide communication. ETTP also
mandated that in any given area the language of teacher training for kindergarten and primary educa-
tion be the nationality language used there.



Implementation
In the early years of policy implementation, larger languages were quickly adopted as languages of
instruction in almost all regions. Gradually, the number of languages used in schools increased and
as of 2015, about 30 languages were being used for part or all of the primary grades, with about 51
languages being offered as a subject. 141 In all regions of the country, English is introduced as a subject
in grade 1, while Amharic is introduced as a third language, starting in grade 3 or 5.

Implementation in urban areas: Most urban areas of Ethiopia are ethnically and linguistically diverse
due to internal migration. Amharic is often used as a common language. In most urban areas, in addi-
tion to the regional or local languages, Amharic is offered as a medium of instruction, sometimes within
the same school. By contrast, in Addis Ababa, the capital and largest city, only Amharic is offered as the
medium of instruction up to sixth grade.


TABLE A-1: Coverage of instruction in L1

                                               % PUPILS LEARNING IN A LANGUAGE THEY SPEAK AT HOME

 REGION                                               Grade 4                         Grade 5

 Addis Ababa                                           86.3                             82.7
 Amhara                                                96.8                             94.7
 Oromia                                                88.8                            88.8
 SNNP                                                  68.2                             0.1
 Tigray                                                99.9                            99.8
 Somali                                                76.7                             61.6
 Afar                                                  47.0                             47.2
 Sample average                                        80.5                             59.8

Source: Young Lives Ethiopia survey 2012–13.




ANNEXES                                                                                                       61
     Variation in length of L1 instruction: There is significant variation in the duration of mother-tongue
     instruction across regions. Some of the major languages follow a late-exit approach, in line with
     the EETP, using L1 languages as the medium of instruction for six to eight years for all subjects (in
     Amhara, Oromia and Tigray regions) followed by English. On the other hand, most languages in Benis-
     hangul-Gumuz, Gambella and SSNP are used only for three to five years, followed by English for all
     higher grades. While census-based information on the number of students learning in their L1 is not
     available, a survey by Young Lives (2014) conducted in the seven regions of the country shows that
     80.5 percent of grade 4 students and 60 percent of grade 5 students are learning in a language that they
     speak, with variation by region as illustrated in Table A1.

     While there is strong commitment to MTB-MLE from the federal government and regions, the shift
     from policy to implementation involves challenges, some of which are highlighted below:
        • Selection of language in diverse communities: An important challenge stems from the selection
          of language of instruction in contexts where there is high linguistic diversity with many demo-
          graphically small languages concentrated in a small geographic area. This type of situation is
          particularly apparent in many parts of the SNNP region. In these contexts, a single language (for
          example, the language of administration at the zone level) is chosen to serve as the medium of
          instruction, while in reality it might not be the L1 language of many students. Therefore, “moth-
          er-tongue education” for many students continues to be education in a regional or zonal admin-
          istrative language.142 To address this challenge, the region attempted to harmonize languages
          with similar linguistic features to create a common hybrid language (‘Wogagoda,’ created from
          the harmonization of Wolayta, Gamo, Gofa and Dawro). This effort was very unpopular and was
          ultimately abandoned.143
        • Selection of language in diverse communities: An important challenge stems from the selec-
          tion of language of instruction in contexts where there is high linguistic diversity with many
          demographically small languages concentrated in a small geographic area. This type of situation
          is particularly apparent in many parts of the SNNP region. In these contexts, a single language
          (for example, the language of administration at the zone level) is chosen to serve as the medi-
          um of instruction, while in reality it might not be the L1 language of many students. Therefore,
          “mother-tongue education” for many students continues to be education in a regional or zonal
          administrative language. To address this challenge, the region attempted to harmonize languag-
          es with similar linguistic features to create a common hybrid language (‘Wogagoda,’ created from
          the harmonization of Wolayta, Gamo, Gofa and Dawro). This effort was very unpopular and was
          ultimately abandoned.
        • Development of languages to be used in education: Another challenge relates to the level of de-
          velopment of languages to meet the demand of the primary education curriculum. Cohen (2007)
          argues that there is significant variation across languages in their level of development and ad-
          equacy for use in education, and the current implementation process can lead to widening in-
          equality across groups.
        • Lack of trained teachers: In the early years of implementation, for many languages, trained and
          qualified teachers who spoke them were not available, and regions were frequently obliged to
          use unqualified teachers.144 However, Ethiopia has made significant progress here. Pre-service
          teacher training programs for lower-primary teachers in many of these languages are gradual-
          ly being introduced at Colleges of Teacher Education, which are managed by regions. In-ser-
          vice teacher training programs have been expanded to upgrade the qualifications of unqualified
          teachers. Primary school teachers are also recruited locally by woredas (equivalent to districts),
          which is gradually improving the situation.145




62                                                      LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
     • High cost of providing textbooks and other teaching and learning materials: For regions where
       there are many demographically small languages, the cost of printing textbooks and TLM re-
       mains high as current approaches to procurement do not permit economies of scale. Further,
       for many of these languages, written material other than textbooks remains scarce. Despite
       these challenges, Ethiopia has made considerable progress in improving provision of textbooks
       and teacher guides, in large part with support from development partners including the World
       Bank.146 By 2013, the country had achieved a 1:1 student to textbook ratio for many subjects, nota-
       bly English-medium textbooks for the secondary grades, and for most subjects in the four major
       regional languages for the primary grades.147
     • Difficulty transitioning to English-medium instruction in higher grades: For many students,
       the transition to English-based instruction in the upper primary or secondary phases is abrupt
       and problematic. Most students only have a limited grasp of English, and low reading proficiency
       in their L1 when they transition to English-medium instruction.148 The use of English in Ethio-
       pia is very limited in students’ lives. As Heugh et al. (2007) describe, students rarely encounter
       any kind of English-speaking environment outside the classroom. Teachers also lack competen-
       cy in English, making it very challenging for them to teach in English. 149


Impact of the mother-tongue language policy on student outcomes
There are very few studies that have rigorously examined the impact of the L1 language policy in Ethi-
opia on educational attainment and learning. Studies that have looked at impact on years of schooling
in the early years of implementation find mixed results. Ramachandran (2012) assessed the effect
of the 1994 language policy change on educational outcomes and found that L1 instruction had had
a positive effect at all levels of schooling, leading to a 12 percent increase in the number of students
completing six years or more of schooling. Chicoine (2019) and Gebre (2015) found a negative impact
on enrollment and years of schooling among the first cohorts of students undergoing the policy. These
negative results are partly explained by the implementation challenges in the early years of the reform.

FIGURE A-1 Percentage of non-readers at the end of grade 2 or beginning of grade 3
Ethiopia - Addis Ababa - Amharic              10
     Ethiopia - Harari - Hararigna                 18
     Ethiopia - Amhara - Amharic                        27
      Ethiopia - Tigray - Tigrigna                           30
   Ethiopia - Oromiya - Oromiffa                                  34
 Ethiopia - SNNPR - Sidamu Afoo                                                                      69
                 Gambia - English                                           44
                Liberia - English                            30
              Mali - Bamanankan                                                                                81
                   Mali - Songhoi                                                                                   84
                     Mali - Bomu                                                                                              91
                   Mali - Fulfulde                                                                                            91
                    Mali - French                                                                                             92
                  Nigeria - Hausa                                                                         79
      Uganda - Central - Luganda                                                      51
       Uganda - Central - English                                                          53
         Uganda - Lango -Lango                                                                                 82
        Uganda - Lango - English                                                                                         88
                Zambia - English                                                                                              91


                                     0   10        20   30             40        50             60   70   80             90        100

Source: RTI International, 2014




ANNEXES                                                                                                                                  63
     The evidence on student learning is limited; however, the few existing studies show promising results.
     Findings from the analysis of the national learning assessment at the end of grade eight in 2000 and
     2004 show that students taught in their L1 for eight years of primary school perform better than
     those who were taught in the L1 for five years.150 Further, comparison of Early Grade Reading Assess-
     ment (EGRA) results across Sub-Saharan African countries provides evidence that the percentage of
     non-readers151 was lower for several language groups in Ethiopia, compared to other countries that
     have multiple languages but are not using local languages in education.

     Within Ethiopia, there is also evidence that students who learn in their L1 longer have better learning
     outcomes than those who transition early to English. Heugh et al. (2007) find that in the 2004 Grade
     8 regional examinations, the mean scores of students who are taught in a national language through
     eighth grade and take the exam in that language have higher scores (except in English) than those
     students attending primary schools where English is the LoI.

     TABLE A-2: Mean Grade 8 scores (%) by subject and medium of instruction (2004)

                                                                         SUBJECTS

      LOI                               English              Math           Biology               Chemistry                Physics

      Oromiffa                            39.1               44.4             49.1                   43.0                    39.5
      Amharic                             39.1               41.3             48.3                   44.6                    41.8
      Tigrigna                            41.6               42.8             48.4                    43.6                   39.3
      Somali                              42.4               42.6             36.3                    37.6                   34.5
      English                             39.4               35.9             35.9                    37.3                    31.5

     Source: Adapted from Heugh et al. (2007: Table 5.2.4)




64                                                                  LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
ANNEX B:
Rwanda:
A Changing
Education System
for a Changing Time

Between 2006 and 2016, Rwanda’s per capita gross domestic product (GDP) grew at five percent per
year, one of the fastest rates in the world, and second only to Ethiopia’s among African countries. In
2018, per capita GDP was about four times greater than it had been 25 years earlier. Rwanda is now
implementing its National Strategy for Transformation (NST) covering the period 2017–2024. This
follows the earlier poverty reduction strategy known as EDPRS2, and is a further step toward Vision
2050, which calls for Rwanda’s attainment of upper-middle-income status by 2035 and a much higher
quality of life for Rwandans by the middle of the 21st century.152 Due to Rwanda’s tumultuous past,
there is national unity around these ambitious goals. Among the key drivers identified for sustained,
inclusive growth is human capital development, making education a key component in this plan.


Current education context
As of 2013, Rwanda had achieved a gross enrollment rate to over 100 percent, and has since main-
tained a low rate of out-of-school children. However, the primary school retention rate remains below
80 percent. The gross enrollment rate in upper secondary education increased from 21 percent in 2010
to 30 percent in 2017. 153 Although Rwanda does not participate in international assessments, an Early
Grade Reading Assessment was administered to a sample of students in Kinyarwanda. The results
clearly leave ample room for improvement: among grade 4 students, 32 percent could read a grade 2
text with comprehension, as could 47 percent of grade 6 students.

While language remains a fraught subject in Rwanda, Kinyarwanda has been given national language
status, given that it is spoken by 99 percent of Rwandans, according to the 2002 census. The number
of English speakers in Rwanda was estimated to be five percent.

Rwanda currently operates a 6–3–3 exam-based system for public education. The first six years of
primary (P1–P6) are a universal right. Students obtain access to three years of junior secondary (S1–S3;
also called O Level, “ordinary level”) and three years of senior secondary (S4–S6; also called A Level,
“advanced level”) upon successful completion of qualifying examinations. At the end of sixth grade,
students take an exam to qualify for secondary school and obtain their Primary Leaving Certificate. In
2018, 250,000 students sat for this exam and 81 percent passed.

Currently, Rwanda’s primary completion rate is about 61 percent, while other low-income coun-
tries have a rate of about 66 percent. Likewise, completion rates for lower-secondary school are 34
percent in Rwanda versus 37 percent in low-income countries on average. There are stark urban/rural


ANNEXES                                                                                                    65
     disparities. Recent measures indicate that students in P1 (first grade) can read 7.7 words per minute in
     Kinyarwanda, and students in P2 can read 25 words per minute. As a comparison, norms for English-
     speaking students learning to read in English in the United States are 60 for the end of first grade and
     100 for the end of second grade (fiftieth percentile) (Hasbrouck-Tindal 2017).

     A survey by the Rwandan Ministry of Education (MINEDUC) in 2005 found that 74 percent of all
     Rwandan adults were able to read and 96 percent were able to write—regardless of the level of education
     attained, including ‘never attended.’ To achieve its Education Sector Strategic Plan-II goals, the country
     is investing to enhance teachers’ professional competence while also expanding the teaching force.

     While examinations remain an important element of the Rwandan education system, a recent shift incor-
     porated competency-based assessments to encourage critical thinking and deprioritize rote learning.

     Rwanda is in a quandary comparable to that of several other nearby countries in terms of its dearth of
     qualified, appropriately trained teachers who are proficient in English and Kinyarwanda. With more
     and more students enrolling in free, basic education, as well as the new LoI policy, there is an even
     more urgent need for qualified teachers. In order to improve teacher education standards as well as
     build capacity within the system, MINEDUC established the Rwanda Teacher Service Commission in
     2008 and is hoping to put in place an official licensing process. However, despite these measures, and
     sufficient funding, there are still substantial shortcomings with regard to teacher quality and efficacy.



     Language of instruction policy
     Rwanda has long struggled with its LoI in primary and secondary schools, experimenting with trilin-
     gual and bilingual models.

     Early language policies after independence:Shortly after gaining independence from Belgium in
     1962, the Rwandan government designated Kinyarwanda the medium of instruction for grades 1, 2
     and 3. Beyond grade 3, the schools shifted to French.

     A shift towards the L1: In 1978 the policy of ‘Rwandization’ was introduced as a decolonizing effort
     in the country. As part of this initiative, the Ministry of Education (MINEDUC) extended primary
     education to grade 8 and made Kinyarwanda the LoI for all eight years. This was intended to allow
     students to gain proficiency in their primary language before moving to an international language.

     Additional language policies have since been enacted, including a 2008 Cabinet resolution shifting the
     LoI in all public schools from Kinyarwanda and French to English (see Table B.1).


     TABLE B-1: LoI Changes for Different Year Groups in 1996 and 2008

                                                     LOI + LANGUAGES AS A SUBJECT
      LEVEL                     1996 MINEDUC reform                                    2008 Cabinet change
      P1 – P3              Kinyarwanda + French & English                         English + Kinyarwanda & French
      P4 – P6              French & English + Kinyarwanda                         English + Kinyarwanda & French
      S1 – S3            French or English + English or French                    English + Kinyarwanda & French
      S4 – S6            French or English + English or French                    English + Kinyarwanda & French


66                                                               LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
Does more English mean better English?

In 1996, as part of a rehabilitation effort after the Rwandan genocide of 1994, the government desig-
nated Kinyarwanda as LoI from grades 1 to 3, with English joining French as LoI for grades 4 to 6.
In October 2008, the Rwandan Cabinet issued a resolution requiring the Minister of Education to
“put in place an intensive program for using English in all public and Government-sponsored primary
and secondary schools and higher learning institutions”.155 This decision meant that English would
replace Kinyarwanda, the language of Rwanda and LoI for the first three years of school, and French,
the LoI since colonial times for Primary 4 onward. English would be the sole LoI across all grades
and levels, primary through tertiary. This new policy was put into effect immediately. Teachers were
provided with additional training in English, and were threatened with penalties if they missed training
sessions. In rural areas where schools were already burdened with high teacher:student ratios and
double shifts, teachers who were primarily graduates from Francophone teacher training institutes
faced termination of their employment unless they improved their English skills quickly. In 2010, the
MINEDUC retracted this decision to vigorously promote English in early grades following a growing
concern that young Rwandans were uncomfortable speaking Kinyarwanda, as well as growing aware-
ness among African academics about the importance of L1 instruction in early years.


The new policy

In December 2019 the government of Rwanda announced a new LoI policy, mandating the use of
English as the primary language for all years of education. An official communiqué released by the
MINEDUC stated “This is an important requirement, and schools are required to strengthen pupil/
students debates, language competitions, appropriate measures to learn and use English language,
and supporting and monitoring measures to ensure use of English language in class, on the school
premises, and to ensure mastery of the language, so that it enables pupils/students to understand
better and obtain higher scores in other subjects. All schools are required to strengthen teaching of
Kinyarwanda to promote Rwandan cultural values, and teaching of French language to promote trilin-
gualism for global competitiveness.” The new policy overturns a 2015 reform that mandated teachers
deliver lessons in Kinyarwanda for the first three years of primary school, and switch to English in
year four. The government changed the LoI from French to English in 2008.

These Rwandan developments can be viewed as a reflection of ongoing tensions between global educa-
tion academics and politicians over LoI. The effective renunciation of L1 is also likely to stem from pres-
sure from parents, many of whom were reportedly against teaching in Kinyarwanda in the first place.


What prompted the shift?

As outlined above, the government of Rwanda harbors ambitious goals with regard to the future of
the country, and a robust and global education system is a key lever in that plan. In 2007, Rwanda
joined the East African Community, a majority Anglophone conglomerate. In 2009, Rwanda joined
the Commonwealth, as part of a conscious move to place the country firmly within several powerful
bodies. These factors provide a strong impetus to favor English as the language of business—over and
above the post-genocide push to distance the country from its Franco-Belgian colonial roots.

Societal pressures have a role to play as well. With an increasingly global economy, Rwandan parents
desire an English education for their children, to arm them with the skills needed to compete in today’s
world. This desire is particularly evident among parents able to send their children to private schools,


ANNEXES                                                                                                       67
     where teachers might possess a stronger grasp of the language as well as more resources to deal with
     the radical change in policy described above. Children from poorer households tend to experience the
     adverse aspects of the change, in both the short and the long term.


     Causes of concern

     There is a growing consensus among the global community that a learning gap in the first few years
     of school has far-reaching ramifications on all future educational goals. Students who do not learn to
     read in the early years struggle to learn mathematics, science and social sciences. L1 instruction in
     the first three years of primary school (as a minimum) is crucial to the attainment of language skills.
     Students who are taught in their L1 in the early years learn to read faster than those who are taught in
     an unfamiliar language.

     A 2017, a Joint Review of the Education Sector recommended emphasizing quality of education not
     only in languages but also mathematics, making it crucial to provide teachers with resources on effec-
     tive mathematics instruction in English.

     The swift nature of this decision as well as its execution have left room for concern among several
     stakeholders, including development partners familiar with local (regional) circumstances. The
     following issues warrant particular attention:
        • Many teachers lack the required competence to provide a good model of English usage.156 The
          Building Learning Foundation’s baseline study found that only 37 percent of P1–P3 teachers meet
          the assigned benchmark of English competence.
        • Despite several international and national interventions, the quality of language teaching in pri-
          mary education remains weak.157
        • Instructional materials in both English and Kinyarwanda remain limited. As of 2018, the average
          pupil to textbook ratio was 5:1.158
        • Widespread use of double shifts and large class sizes limit the amount and quality of instruction-
          al time available for language and literacy teaching.
        • The Competence Based Curriculum159 (CBC) demands a relatively rich English vocabulary
          which is beyond the command of most learners, and even many teachers.
        • As a result of changing language policies, Rwanda does not have a history of using English as a
          language of wider communication. Most children, particularly those not in urban schools, en-
          counter English predominantly in schools and rarely in their homes.

     These concerns have been widely echoed in discussions within communities and on social media, by
     families of children impacted as well as by development partners. Local media outlets have questioned
     the haste of this move as well as the government’s ability to efficiently execute this decision, and at the
     same time follow through on its promise of a high-quality education.

     Within days of the announcement, the MINEDUC issued a statement with regard to its decision:
        1. All private schools that were previously using English or French as a medium of instruction at lower
           primary level are allowed to maintain this practice.
        2. All public and government-aided schools currently using Kinyarwanda as medium of instruction
           at lower primary level will gradually transition to using English as medium of instruction, within a
           determined period to be communicated by the Ministry of Education.
        3. Kinyarwanda will be taught as a mandatory subject in both private and public schools


68                                                        LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
This was followed by the implementation of this decision solely in public schools, with a phased-out
approach in private schools.

While it remains to be seen how this decision impacts the education quality of students in schools
today, the following four recommendations, derived from the World Bank literacy policy package,
should lie at the heart of any future iterations of the policy:
   1. Ensure political and technical commitment to clear goals, pathways, and measures for literacy;
   2. Ensure effective teaching for literacy;
   3. Ensure timely access to more and better texts that are appropriate to age and skill; and
   4. First teach children in the language they speak and understand.

In order to adapt to the new policy change proposed, bearing in mind those recommendations, the
following steps could be taken:
   1. In addition to developing tools for English language proficiency, materials could also be devel-
      oped to aid early childhood ESL education. Teachers should be trained on effective strategies to
      engage learners and build language skills in multilingual settings.
   2. Age and context appropriate literature should be created (or sourced) and provided to the schools
      as they transition to locally developing textbooks and other materials in English.


As mentioned in the Ending Learning Poverty report, in many countries, shifting to teaching in the
L1 in the early primary years proves to be essential to improving performance. Avenues for further
research and analysis include finding out if mother-tongue literacy instruction has positive effects
on student learning in other core subjects, including English, Kiswahili and mathematics. It is also
prudent to explore the role of a teacher’s own proficiency in the language and what kind of training
they receive on learning outcomes of students. Finally, if the government of Rwanda seeks to go
forward with its ‘English-first’ policy, it would be wise to explore and learn from successful—and suit-
ably paced—implementation of such a move in analogous contexts elsewhere.




ANNEXES                                                                                                    69
     ANNEX C:
     Hong Kong:
     Language of Instruction
     in Secondary Schools



     Introduction and Language Context

     The majority of the population of Hong Kong SAR, China (hereafter “Hong Kong”) is bilingual
     or trilingual. Hong Kong has a population of 7.5 million,160 95 percent of whom are ethnic Chinese
     and speak Cantonese—a southern Chinese language—as their L1.161 Chinese and English162 are official
     languages in Hong Kong, with Cantonese being acknowledged as the de facto official spoken variety
     of Chinese, while Mandarin (also known as Putonghua) is also accepted. Following the Handover
     of Hong Kong in 1997 from the British to the Chinese, the Education Bureau of Hong Kong (EDB)
     promulgated the explicit aim for all students “to be biliterate (master written Chinese163 and English)
     and trilingual (speak fluent Cantonese, Mandarin Chinese, and English).”164 As of the 2016 Govern-
     ment Census, 77 percent of the population aged five to 59 were at least bilingual, and 51 percent trilin-
     gual. For the age groups that have attended compulsory education since the enactment of the Educa-
     tion Bureau’s “biliterate and trilingual” policy (ages five to 29), 87 percent are bilingual, and 63 percent
     are trilingual.165


     Language of Instruction in Primary Schools: Historical Context and Demo-
     graphic Necessity
     L1 instruction began as a historical and demographic necessity. Language policy in the British
     colony prior to World War Two favored English language instruction. Given that education was largely
     limited to the elite classes, English language education was by and large the norm. Very few local
     Chinese attended school at this time, and the British were largely unconcerned about the educational
     attainment of the local population,166 demonstrated by the fact that, in 1950, only about one-quarter
     of the population over the age of 15 had completed primary education.167 Interestingly, for the small
     population of local Chinese who did attend school, primary education was in the local language
     (Cantonese), rather than the national language (Mandarin Chinese). This led to Hong Kong being the
     only place in East and Southeast Asia where a local rather than a national language functioned, and
     continues to function, as the LoI in primary education.168

     The end of World War Two brought about immense demographic changes that had a major
     impact on Hong Kong’s language of instruction policy. The influx of Chinese-speaking immigrants
     and refugees from Mainland China after 1949 was a major factor that propelled the government in



70                                                        LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
1950 to permit the establishment of primary schools using Chinese as the language of instruction
(LoI).169 Hong Kong had to quickly train a large number of teachers from the local population to
teach the Chinese-speaking migrants from the Mainland 170. Primary education was recognized as an
explicit goal in 1965, and became both free and compulsory in 1971. The decision to adopt Chinese as
the medium of instruction in primary education was born out of necessity, both in terms of teacher
ability and student profile.

With the education sector growing at an unprecedented pace,171 there was a powerful incentive
to align educational policies with best practices emerging from research. For more than half a
century, the Government has recognized—and institutionalized—the benefits of L1 instruction on
student learning outcomes. The 1974 Report of the Board of Education on the Proposed Expansion
of Secondary Education recommended “Chinese should become the usual medium of instruction
in lower forms of secondary schools; every effort should be made to develop good textbooks for all
subjects written in Chinese, to train teachers capable of instructing through the medium of Chinese.”
172
     The 1990 Education Commission explicitly stated that students can learn better in their L1, as
the use of mixed-code, “can lead to time being wasted on translation of English texts in class and,
worse still, learning being reduced to rote memorization of facts in English.”173 In 1994, the Standing
Committee for Language Education and Research (SCOLAR) was formed, with the explicit aim of
realizing the Education Bureau’s ambition for the population to become biliterate and trilingual. The
1997 “Medium of Instruction Guidance” issued to schools by the Education Department described the
educational benefits of L1 teaching, whereby the use of Chinese as the LoI would remove language
barriers impeding the study of most subjects, and students “…will be better able to understand what
is taught, analyze problems, express views, develop an enquiring mind and cultivate critical thinking.
Mother-tongue teaching thus leads to better cognitive and academic development.”174 The imple-
mentation of L1 instruction was therefore enabled by a policy environment that drew heavily from
research and continuous internal evaluation.

Due to demographic necessity and alignment of policies to best practices emerging from the
evidence, the topic of language of instruction (LoI) at primary school has been “wholly uncontro-
versial as policy and largely unproblematic in practice.” 175 The teaching of students in a language
they understand—and in which they can best express themselves—has paid significant dividends
for the learning outcomes of the population. Indeed, Hong Kong has some of the highest learning
outcomes in the world. Grade 4 students in Hong Kong scored second, first, and third in the world in
PIRLS 2006, 2011, and 2016, respectively.


LoI in Secondary Schools: Politics, the Economy, and National Identity
Secondary—not primary—schools have since been at the center of LoI policy debate over the
decades.176 With the LoI in all public primary schools being Chinese, and the LoI in universities being
predominantly English, the transitional period of secondary school has subsequently been the main
locus of change and focus of debate over the years. The subsequent section will examine LoI policies
in secondary schools in Hong Kong.

Contemporary LoI policies in secondary schools in Hong Kong have mirrored the historical,
economic, and political changes experienced by the city state over the past five decades. Language
policy in secondary schools in Hong Kong has seen four different and distinct LoI stages.177 They can
be briefly characterized as follows:




ANNEXES                                                                                                   71
         1. The decades prior to the Handover (pre-1994);

         2. The immediate years before the Handover (1994-97);

         3. The decade following the Handover (1997-2010); and

         4. The current policy (2010 to present).

     The pre-1994 period saw a rather laissez-faire approach to LoI, whereby primary schools were taught
     in Cantonese, while secondary schools were free to select their own LoI. This led, however, to a
     preponderance of schools claiming to be English-medium schools but which, in effect, used mixed
     language teaching. This was seen as a consequence of inadequate English language skills on the part
     of both teachers and students, and it led to the 1994 streaming policy that split secondary schools
     into three categories: Chinese-medium, English-medium, or dual-medium schools. Although an indi-
     vidual school could still make a decision regarding its LoI, English-medium schools had to demon-
     strate that their students had sufficient language ability to be able to learn effectively in English. 178
     The change in sovereignty of Hong Kong in 1997 led to a subsequent change in LoI policy, whereby all
     secondary schools where required to teach through the medium of Chinese only, with the exception
     of 114 schools granted exemptions to continue with English-medium instruction179 (see Figure C1).
     These schools had to meet criteria that showed that their teaching staff and students were capable of
     teaching and learning effectively through English. 180 At the start of the 1998–99 school year, 307 public
     secondary schools (the remaining 70 percent of schools) switched to Chinese-medium instruction. 181
     The latest change came in 2010, when the Education Bureau abolished the classification system, such
     that a school could no longer be labelled an English-, dual-, or Chinese-medium institution. Instead,
     schools could offer English-, dual-, and Chinese-medium classes. This arrangement was brought in at
     the start of the 2010–11 school year, and still obtains today.182

      FIGURE C-1: Percentage of students in English- and Chinese-medium secondary schools, 1960–1998,
     (taken from Evans and Morrison, 2017).




72                                                       LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
FIGURE C-2: Medium of Instruction in Hong Kong over time (adapted from Evans and Morrison, 2017).

2010 to present                   !"#$%"&
                                    Primary                                   Secondary                            Tertiary


 1998 to 2009                       Primary
                                  !"#$%"&                                    Secondary
                                                                          '()*+,%"&/                             Tertiary
                                                                                                                -(".#%"&

    Pre 1998                        Primary
                                  !"#$%"&                                      Secondary
                                                                             '()*+,%"&                           Tertiary
                                                                                                                -(".#%"&

                  0   1       2     3      4    5      6      7      8   9        10       11   12     13     14         15   16


                          Chinese Medium       Mixed Model Schools       English Medium         Mixed Model Classes




Strengths of Choice/Implementation:
For secondary schools, the government attempted to offer autonomy of choice regarding LoI
decisions during different phases, while providing appropriate support and resources (including
research). Secondary schools have been, and are, supported to make autonomous decisions on the
LoI in their schools as a whole (before 1997) and in their classes (after 2010). Even during the inter-
vening period of Chinese-medium instruction (between 1998 and 2010), the government allowed
exemptions if certain criteria were met (teaching staff had to prove capable of teaching effectively
through English, while students, conversely, had to be demonstrably capable of benefiting from the
content imparted through English). To this day, criteria continue to be applied to schools that host
English-medium classes to ensure the effective use of English as a LoI for both students and teachers.
These criteria include: ensuring that student ability within the school is such that they can learn
effectively in English; maintaining teaching capability, to be based on the principal's assessment and
certification; and setting language benchmarks for teachers (with EDB inspectors visiting schools to
gather information on their aptitude for using English as LoI).

To support effective English-, Chinese-, and dual-language instruction, the EDB’s Curriculum
Development Institute (CDI) publishes all curriculum support materials bilingually. Materials
include resources such as teaching syllabi, curriculum guides, and subject teaching guidelines. The
CDI has produced English–Chinese glossaries of terms commonly used in the teaching of Secondary 1
to Secondary 7 syllabi. Best practices are collected and disseminated to the public in documents such
as “The Language Teaching Album: A Collection of School-based Practices”, supported by various
other resources such as online publications, lesson plans, and sample worksheets for teachers of
English and Chinese, available online. A Task Force on Language Support was set up by the Education
Bureau (EDB) in 2004 with the aim of raising the standard of language proficiency in Hong Kong. It
consists of teaching consultants who provide Chinese (including Putonghua) and English language
curriculum development support services to all primary and secondary schools. Language support
officers work closely and intensively with teachers in a variety of school contexts on projects initiated
by the schools.

Since the early 1990s, public examinations to determine university places have been offered in
both English and Cantonese.183 Students are able to select the language of assessment by subject
(with the exception of subjects in which the language used is axiomatic, for example Chinese
language) which facilitates alignment between class and assessment languages. Furthermore, to rule
out potential inferences or implications that could flow from the fact of having taken a given subject
through the medium of one particular language—Chinese, for example—in contradistinction to the



ANNEXES                                                                                                                            73
     other, the examinations board has mandated that the language of examination is not indicated on
     student certificates.184 In 2015, the overall percentage of students who opted to take examinations in
     English (composite number, derived from all non-axiomatic subjects) was 42 percent.

     The EDB has and continues to make research-based decisions on LoI. Over the past few decades,
     beginning as early at 1982, the EDB has commissioned more than 15 major research studies on outcomes
     and impact of LoI in schools.185 The resulting findings have been used to make decisions on LoI policy,
     with the most significant being the roll-back of English- and Chinese-medium schools in 2010. A seven-
     year longitudinal study in 2008 (Tsang), found that Chinese Medium of Instruction students were
     disadvantaged in gaining entrance to universities. Roughly two years later, the Government introduced
     the policy that replaced the distinction between English-, Chinese-, and dual-medium schools, in favor
     of English- and Chinese-medium classes (within a single school, as appropriate).


     Weaknesses of Choice/Implementation:
     Despite the ostensible designation of English-medium classes (and earlier, of English-medium
     schools), teachers generally use Cantonese to communicate with their students.186 This code-
     switching and code-mixing reflect a larger phenomenon that can be seen elsewhere in bilingual (or
     trilingual) education systems, where many teachers lack the requisite language proficiency to deliver
     content in a second language. A set of qualitative interviews conducted by Evans and Morrison with
     students in English-medium classrooms indicated that many of them sit in classes with teachers who
     “close the door” before embarking upon clandestine explanations in L1.187 As recently as last year,
     the Education Bureau conducted an English Language Proficiency Assessment for Teachers in Hong
     Kong. The results showed a high degree of proficiency in Reading, Listening, and Classroom Language
     (with attainments rates of 84 percent, 87 percent, and 92 percent respectively). However, Writing and
     Speaking attainment was significantly lower (42 percent and 58 percent, respectively, EDB, 2019).

     The previous retention of an English-medium instruction (EMI) stream was seen to have advan-
     taged students in English-medium secondary schools in gaining admission to local and foreign
     universities. The 1997 mother-tongue instruction policy was unpopular when first introduced, as
     parents thought that the English language held more prestige and utility internationally.188 When the
     exemptions were granted to 114 schools to continue with EMI, administrators from the remaining
     (and majority) Chinese Medium Instruction (CMI) schools felt as if they had been relegated to
     “second-class schools and their students, second class students.” 189 A set of questionnaires sent to
     all CMI schools showed 53 percent of administrators 190 indicated that their schools had suffered the
     loss of high-ability students, with 32 percent indicating that the academic standards of their students
     had subsequently dropped. The public perception regarding the utility of English-language instruc-
     tion for academic and economic outcomes turned out to be well founded. Despite research in the
     early 2000s that found students in Chinese-medium schools boasted higher learning outcomes191
     and that Chinese-medium schools had higher quality teaching-learning processes, 192 a longitudinal
     study of students who attended English- and Chinese-medium schools published in 2008 showed
     “significant negative effects on CMI students’ chances of success in educational attainments in the
     HKCEE and HKALE (public examinations)”, with the major obstacle being the lower achievements
     in the subject of English language in CMI schools as compared to EMI students.193 The public outcry
     over the misalignment of CMI secondary schools with EMI tertiary institutions led to the subsequent
     roll-back of English- and Chinese-medium school designations in favor of a class-based approach. As
     alluded to above, the examinations board continues to combat the stigma attached to Chinese-me-
     dium schools by not specifying the language used in a student’s examination record.



74                                                      LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
Related to the issues of alignment between CMI secondary schools and EMI tertiary institutions
is the significant impact of students’ English abilities on their adjustment to subsequent univer-
sity study. Attempts to align secondary school LoI with pre-university public assessments belies the
fact that all tertiary institutions in Hong Kong are English-medium. There is evidence that students
who attend majority English-medium classes enjoy a distinct and substantial advantage in English
language over their peers in majority Chinese-medium classes.194 The same study showed also that
graduates who had been taught mainly in English and who sat exams in English self-reported that
they had adapted better to local university study.


Lessons Learned
The Hong Kong experience yields the following noteworthy lessons:

    1. The necessity to rapidly universalize primary education meant that Hong Kong aligned its
       policies to two factors: what was available (Chinese-speakers who they could quickly train to
       become teachers), and what was most likely to be effective (emerging evidence on the effec-
       tiveness of teaching children in their mother-tongue). L1 education at primary levels benefited
       from demographic and historical circumstance: the rapid influx of Chinese-speaking refugees
       and the universalization of primary education necessitated the use of local Chinese-speaking
       teachers. In addition, the universalization of primary education was also a time when policy
       makers drew from international research and best practice to further solidify the policy of L1
       instruction as the effective basis for forming cognitive abilities and human capital. At that time,
       L1 instruction in primary schools amounted to an uncontroversial decision. If it has subse-
       quently evolved, then this has merely to become a largely self-evident policy.

    2. At the secondary level, students, schools, and parents were given the autonomy to make LoI
       decisions, which were supported through a variety of technical means and subject to trans-
       parent validation processes to ensure adherence to a predefined level (that is, of teacher and
       student language proficiency). Supportive policies such as bilingual curriculum materials and
       examination policies enabled the effective and consistent implementation of these throughout
       public schools. Explicit policy goals—biliteracy alongside trilingualism—set the stage for the
       formulation, implementation, and adjustment of comprehensive policies.

    3. Economic considerations and the social capital of learning in a more “international” language
       will be the driving factor behind parents’ and students’ decisions—and the alignment of assess-
       ment and higher education with LoI policies in secondary school should be of paramount
       importance. Language policies that evolve with the frontiers of research-based best prac-
       tices, which fit national goals, and are sensitive to economic considerations, are most likely to
       survive, and hold sway, over future decades.




ANNEXES                                                                                                      75
     ANNEX D:
     Middle East and North
     Africa: Mastering Arabic
     & the Transition from
     Colloquial to Formal


     Arabic is spoken by more than 420 million people, making it the sixth most spoken language across the
     world. It is the official or co-official language of 25 countries,195 and is used by more than 1.8 billion
     Muslims. Arabic can be considered as a continuum of forms, from the formal Classical Arabic—the
     language of the Qur’an—to the many informal, colloquial versions used in everyday communication.
     Toward the formal end of the continuum, some linguists describe Modern Standard Arabic (MSA) as a
     version that dates back to the 19th century, and has since dropped some words and phrases and added
     some new technical vocabulary in response to the times.196 MSA is the written form that is used across
     Arabic-speaking countries for literary works, official documentation, mass media, and educational
     institutions. However, MSA is not anyone’s L1; it is acquired through formal education.197

     Formal Arabic (Classical or MSA) differs significantly from the various versions of colloquial Arabic
     spoken daily and acquired in homes as the L1. There are many varieties of colloquial Arabic spoken across
     countries or groups of countries, including (proceeding very approximately from east to west): Iraqi,
     Gulf Arabic, Yemeni, Levantine, Sudanese, Egyptian, Maghrebi (North Africa), and Hassaniya (mainly
     Mauritania). The formal version of Arabic has not changed over time, unlike the fast development of
     colloquial Arabic (particularly through the media), resulting in widening gaps between the two.198

     Children grow up hearing a colloquial form of Arabic in their homes and communities. As soon as they
     enter school, children are expected to learn to read and write in formal, standard Arabic. Their expe-
     rience with this form of Arabic is very limited until then. Parents in Arabic-speaking countries are less
     likely to have children’s books in the home (Figure D1) and less likely to read to their children when
     they are young (Figure D2) compared to parents in other countries. For example, only around 20–25
     percent of grade 4 students had often been read to as a child before they began school, compared to
     around 80 percent in Northern Ireland, New Zealand, Australia, and the Netherlands, and to around
     54 percent for the average of PIRLS 2016 participating countries. In Morocco and Saudi Arabia, more
     than 60 percent of grade 4 students have fewer than 10 children’s books in their homes, a far more
     worrying statistic than the PIRLS 2016 participating country average of 19 percent. The concern here
     derives in large part from evidence that children can transition much more easily from their L1 dialect
     to formal Arabic if children’s literature is read to them early on. 199

     In school, written materials—such as textbooks and children’s literature—are all in formal Arabic. Chil-
     dren are taught to read and write, with a focus on grammar and accuracy and a lack of playfulness and
     inquiry. 200 Teachers correct children as they learn to read and write, while a spirit of linguistic inven-



76                                                        LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
                                                                                                                         0%
                                                                                                                               10%
                                                                                                                                      20%
                                                                                                                                             30%
                                                                                                                                                    50%
                                                                                                                                                           50%
                                                                                                                                                           60%
                                                                                                                                                                  70%
                                                                                                                                                                  80%
                                                                                                                                                                          90%
                                                                                                                                                                         100%




                                                                                                                          0%
                                                                                                                                10%
                                                                                                                                       20%
                                                                                                                                              30%
                                                                                                                                                     50%
                                                                                                                                                            50%
                                                                                                                                                            60%
                                                                                                                                                                   70%
                                                                                                                                                                   80%
                                                                                                                                                                           90%
                                                                                                                                                                          100%
                                                                                                                                                                                                                                                                                                                                                                                                                             0%
                                                                                                                                                                                                                                                                                                                                                                                                                            10%
                                                                                                                                                                                                                                                                                                                                                                                                                            20%
                                                                                                                                                                                                                                                                                                                                                                                                                            30%
                                                                                                                                                                                                                                                                                                                                                                                                                            50%
                                                                                                                                                                                                                                                                                                                                                                                                                            50%
                                                                                                                                                                                                                                                                                                                                                                                                                            60%
                                                                                                                                                                                                                                                                                                                                                                                                                            70%
                                                                                                                                                                                                                                                                                                                                                                                                                            80%
                                                                                                                                                                                                                                                                                                                                                                                                                            90%
                                                                                                                                                                                                                                                                                                                                                                                                                           100%




                                                                                                                                                                                                                                                                                                                                                                                                                              0%
                                                                                                                                                                                                                                                                                                                                                                                                                             10%
                                                                                                                                                                                                                                                                                                                                                                                                                             20%
                                                                                                                                                                                                                                                                                                                                                                                                                             30%
                                                                                                                                                                                                                                                                                                                                                                                                                             50%
                                                                                                                                                                                                                                                                                                                                                                                                                             50%
                                                                                                                                                                                                                                                                                                                                                                                                                             60%
                                                                                                                                                                                                                                                                                                                                                                                                                             70%
                                                                                                                                                                                                                                                                                                                                                                                                                             80%
                                                                                                                                                                                                                                                                                                                                                                                                                             90%
                                                                                                                                                                                                                                                                                                                                                                                                                            100%
                                                                                                    Northern    Ireland
                                                                                                     NorthernIreland                                                                                                                                                                                                                                                                                    Northern     Ireland
                                                                                                                                                                                                                                                                                                                                                                                                     Northern    Ireland




ANNEXES
                                                                                                             Australia
                                                                                                             Australia                                                                                                                                                                                                                                                                                             Zealand
                                                                                                                                                                                                                                                                                                                                                                                                         NewNewZealand
                                                                                                        New    Zealand
                                                                                                        NewZealand                                                                                                                                                                                                                                                                                               Australia
                                                                                                                                                                                                                                                                                                                                                                                                              Australia
                                                                                                               Sweden
                                                                                                              Sweden                                                                                                                                                                                                                                                                                         Netherlands
                                                                                                                                                                                                                                                                                                                                                                                                          Netherlands
                                                                                                               Canada
                                                                                                               Canada                                                                                                                                                                                                                                                                                            Germany




                                    About
                                                                                                                                                                                                                                                                                                                                                                                                              Germany




          Source: IEA PIRLS 2016.
                                                                                                                Ireland
                                                                                                               Ireland                                                                                                                                                                                                                                                                                              Finland
                                                                                                                                                                                                                                                                                                                                                                                                                 Finland




                                    Abouthow
                                                                                                               Norway
                                                                                                              Norway                                                                                                                                                                                                                                                                                              Slovenia
                                                                                                                                                                                                                                                                                                                                                                                                               Slovenia
                                                                                                                  Malta




                                          howmany
                                                                                                                 Malta                                                                                                                                                                                                                                                                                               Austria
                                                                                                                                                                                                                                                                                                                                                                                                                 Austria
                                                                                                             Germany
                                                                                                            Germany                                                                                                                                                                                                                                                                                                  Ireland
                                                                                                                                                                                                                                                                                                                                                                                                                 Ireland
                                                                                                             Denmark
                                                                                                            Denmark                                                                                                                                                                                                                                                                                                Sweden
                                                                                                                                                                                                                                                                                                                                                                                                                Sweden




                                             manychildren's
                                                                                                                Finland
                                                                                                               Finland                                                                                                                                                                                                                                                                            Russian      Federation
                                                                                                                                                                                                                                                                                                                                                                                                     RussianFederation
                                                                                                                Austria
                                                                                                               Austria                                                                                                                                                                                                                                                                                              Canada
                                                                                                                                                                                                                                                                                                                                                                                                                Canada




                                                  children's books
                                                                                                   Belgium    (French)
                                                                                                    Belgium (French)                                                                                                                                                                                                                                                                                             Denmark
                                                                                                                                                                                                                                                                                                                                                                                                              Denmark




                                                             books are
                                                                                                              Hungary
                                                                                                             Hungary                                                                                                                                                                                                                                                                                               Norway
                                                                                                                                                                                                                                                                                                                                                                                                                Norway
                                                                                                            Singapore
                                                                                                           Singapore                                                                                                                                                                                                                                                                                   Czech      Republic
                                                                                                                                                                                                                                                                                                                                                                                                         CzechRepublic
                                                                                                       ChineseTaipei
                                                                                                       Chinese    Taipei                                                                                                                                                                                                                                                                                          Hungary
                                                                                                                                                                                                                                                                                                                                                                                                               Hungary
                                                                                                                 France
                                                                                                                France                                                                                                                                                                                                                                                                                             Georgia
                                                                                                                                                                                                                                                                                                                                                                                                                Georgia
                                                                                                                   Israel
                                                                                                                  Israel                                                                                                                                                                                                                                                                                      Kazakhstan
                                                                                                                                                                                                                                                                                                                                                                                                           Kazakhstan
                                                                                                                                                                                                                                                                                                                                                                                                                     Poland
                                                                                                                                                                                                                                                                                                                                                                                                                 Poland




                                                                   are there in your home?
                                                                                                     Hong Kong
                                                                                                     Hong   KongSAR  SAR




                                                                                                                                                                                                                                                                                                                           you or someone else in your home read books to him or her?
                                                                                                                                                                                                                                                                                                                          you or someone else in your home read books to him or her?
                                                                                                         Netherlands
                                                                                                         Netherlands                                                                                                                                                                                                                                                                                               Latvia
                                                                                                                                                                                                                                                                                                                                                                                                                       Latvia
                                                                                                                                                                                                                                                                                                                                                                                                      Slovak      Republic
                                                                                                                                                                                                                                                                                                                                                                                                        SlovakRepublic




                                                                                                                                                                                                                                                                                                                           Before your child began primary/elementary school, how often did
                                                                                                                                                                                                                                                                                                                          Before your child began primary/elementary school, how often did
                                                                                                      Czech   Republic
                                                                                                      Czech Republic
                                                                                                  Belgium                                                                                                                                                                                                                                                                                                         France
                                                                                                                                                                                                                                                                                                                                                                                                                      France
                                                                                                            (Flemish)
                                                                                                   Belgium (Flemish)




                                      0-10
                                                                                                                                                                                                                                                                                                                                                                                                 Trinidad    andTobago
                                                                                                                                                                                                                                                                                                                                                                                                   Trinidadand      Tobago
                                                                                                               Poland
                                                                                                                Poland
                                                                                                                                                                                                                                                                                                                                                                                                              Lithuania
                                                                                                                                                                                                                                                                                                                                                                                                                 Lithuania
                                                                                                             Slovenia




                                                                                                                                                                                                                                                                     FIGURE D.2: Number of Children’s Books in the Home
                                                                                                              Slovenia
                                                                                                                                                                                                                                                                                                                                                                                                                   Malta
                                                                                                                                                                                                                                                                                                                                                                                                                       Malta
                                                                                                                 Latvia
                                                                                                                  Latvia
                                                                                                                                                                                                                                                                                                                                                                                                                    Israel
                                                                                                                                                                                                                                                                                                                                                                                                                        Israel
                                                                                                        Dubai,
                                                                                                         Dubai,UAE  UAE
                                                                                                                                                                                                                                                                                                                                                                                                                     Italy
                                                                                                                                                                                                                                                                                                                                                                                                                         Italy
                                                                                                             Portugal
                                                                                                              Portugal




                                      11-25
                                      11-25
                                                                                                                                                                                                                                                                                                                                                                                                    Belgium       (French)
                                                                                                                                                                                                                                                                                                                                                                                                       Belgium(French)
                                                                                                                 Spain
                                                                                                                   Spain                                                                                                                                                                                                                                                                                           Spain
                                                                                                                                                                                                                                                                                                                                                                                                                       Spain
                                                                                                Trinidad




                                                                                                                                                                                                                                                                                                                                      Never or almost
                                                                                                          andTobago
                                                                                                Trinidad and   Tobago




                                                                                                                                                                                                                                                                                                                                      Never or almost
                                                                                                                                                                                                                                                                                                                                                                                                   Belgium      (Flemish)
                                                                                                                                                                                                                                                                                                                                                                                                     Belgium(Flemish)
                                                                                                 Russian   Federation
                                                                                                  Russian Federation                                                                                                                                                                                                                                                                                           Bulgaria
                                                                                                                                                                                                                                                                                                                                                                                                                  Bulgaria
                                                                                                     Slovak  Republic
                                                                                                     Slovak Republic                                                                                                                                                                                                                                                                                           Portugal
                                                                                                                                                                                                                                                                                                                                                                                                                  Portugal
                                                                                                            Lithuania




                                      26-50
                                                                                                             Lithuania




                                      26-50
                                                                                                                                                                                                                                                                                                                                                                                                            Azerbaijan
                                                                                                                                                                                                                                                                                                                                                                                                               Azerbaijan
                                                                                                         Macao
                                                                                                          MacaoSAR   SAR                                                                                                                                                                                                                                                                                     Singapore
                                                                                                                                                                                                                                                                                                                                                                                                                Singapore
                                                                                                                    Italy
                                                                                                                     Italy                                                                                                                                                                                                                                                                               Dubai,      UAE
                                                                                                                                                                                                                                                                                                                                                                                                            Dubai,       UAE
                                                                                                         Kazakhstan
                                                                                                          Kazakhstan                                                                                                                                                                                                                                                                                  Hong            SAR




                                                                                                                                                                                 Number of children’s books in the home as reported by parents of grade 4 students
                                                                                                                                                                                                                                                                                                                                                                                                        HongKongKong      SAR




                                                                                                                                                                                                                                                                                                                                      Sometimes
                                                                                                             Bulgaria



                                                                                                                                                                                                                                                                                                                                       Sometimes
                                                                                                              Bulgaria                                                                                                                                                                                                                                                                                               Chile
                                                                                                                                                                                                                                                                                                                                                                                                                         Chile




                                      51-100
                                                                                                                Qatar




                                       51-100
                                                                                                                 Qatar                                                                                                                                                                                                                                                                                            Qatar
                                                                                                                                                                                                                                                                                                                                                                                                                      Qatar
                                                                                             United
                                                                                             United Arab    Emirates
                                                                                                     Arab Emirates                                                                                                                                                                                                                                                                                      Chinese
                                                                                                                                                                                                                                                                                                                                                                                                          Chinese  Taipei
                                                                                                                                                                                                                                                                                                                                                                                                                       Taipei
                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                    FIGURE D.1: Frequency of Parents Reading Books to Children Before Primary School




                                                                                                              Georgia
                                                                                                               Georgia                                                                                                                                                                                                                                                                                            OmanOman
                                                                                                   Abu
                                                                                                   Abu Dhabi,
                                                                                                         Dhabi,UAE  UAE                                                                                                                                                                                                                                                                       United    ArabEmirates
                                                                                                                                                                                                                                                                                                                                                                                               UnitedArab       Emirates
                                                                                                             Bahrain                                                                                                                                                                                                                                                                                           Bahrain
                                                                                                                                                                                                                                                                                                                                      Often




                                      More
                                                                                                              Bahrain                                                                                                                                                                                                                                                                                             Bahrain
                                                                                                                                                                                                                                                                                                                                       Often
                                                                                                                                                                                                                                                                                                                                                                                                                                   Percentage of grade 4 students whose parents report reading to them often, sometimes, or never




                                                                                              Iran, Islamic   Rep.
                                                                                              Iran, Islamic Rep.of     of                                                                                                                                                                                                                                                                           Abu  Dhabi,
                                                                                                                                                                                                                                                                                                                                                                                                      Abu Dhabi,     UAE UAE




                                       Morethan
                                                                                                                   Chile                                                                                                                                                                                                                                                                       Iran, Islamic    Rep.
                                                                                                                                                                                                                                                                                                                                                                                                 Iran, Islamic Rep.      ofof
                                                                                                                    Chile




                                            than100
                                                                                                                Oman                                                                                                                                                                                                                                                                                       Macao      SAR




                                                 100
                                                                                                                 Oman                                                                                                                                                                                                                                                                                         Macao SAR
                                                                                                          Azerbaijan                                                                                                                                                                                                                                                                                    Saudi       Arabia
                                                                                                                                                                                                                                                                                                                                                                                                          SaudiArabia
                                                                                                           Azerbaijan
                                                                                                                                                                                                                                                                                                                                                                                                             Morocco
                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                       to teach Arabic, as very few university teacher training courses in the region cover Arabic pedagogy.




                                                                                                           Morocco
                                                                                                            Morocco                                                                                                                                                                                                                                                                                             Morocco
                                                                                                       Saudi Arabia
                                                                                                       Saudi Arabia
                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                       and skills in formal Arabic to be comfortable using it as a medium of instruction. Furthermore, those
                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                       features of the formal Arabic language, but this adds a level of complication.202 In fact, teachers are
                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                       are rarely used on the grounds that they rely on breaking words down to teach sound–symbol correspon-
                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                       tion—making up words or spellings—is considered inappropriate. 201 Likewise, phonics-based methods




                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                       training to become Arabic language teachers rarely experience the in-depth training they need on how
                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                       themselves the products of poor Arabic language education, meaning they do not have the knowledge
                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                       dences and often assess using non-words. Teachers sometimes use colloquial Arabic to help to explain




77
     Students frequently dislike Arabic language studies, feel insecure about their abilities, and do not
     enjoy reading and writing owing to: (1) a lack of exposure to formal Arabic vocabulary and litera-
     ture in the early years, (2) rigid and didactic teaching methods, and (3) an inability to use play and
     creativity in language learning experiences.203 To compound the problems, countries in the region
     have low instructional hours and overcrowded curricula, resulting in less than adequate time for
     Arabic learning. Preschool enrollment is very low across the region, although some countries are
     seeing a rise in non-Arabic-based preschool programs (such as English). Some of the visual features
     of the Arabic language can exacerbate the difficulties, including differentiating letters, which can be
     more difficult in Arabic than in Roman script, the use of small font sizes, and the deletion of vowels
     at grades 3-4, which results in reliance on context and memorization.204 Taken together, these issues
     mean that the fundamental skills that make up reading are slow to develop in Arabic. The result can
     be seen in the poor outcomes of Arabic-speaking countries in international student assessments such
     as the PISA, TIMSS, and PIRLS.

     There is no reason to assume that any student would fail to reach their potential if only they were
     enabled to benefit from the insights gained over the past few decades into teaching and learning prac-
     tices for Arabic. Strengthening Arabic language teaching and learning, while allowing flexibility for
     students to engage with the language in creative ways, would provide the foundations needed to boost
     overall learning outcomes. This should include: (1) encouraging parents to read to children from an
     early age, (2) developing a broader library of more engaging literature for children, especially stories
     that use both colloquial and standard Arabic as appropriate, (3) using evidence-based pedagogical
     methods in the early grades, and (4) devoting sufficient time in the curriculum to the development of
     Arabic language skills.




78                                                      LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
ANNEX E:
Peru:
Bilingual Intercultural
Education


In Peru, a country with a population of nearly 32 million, approximately four million people speak
an indigenous language as their L1. Of these, approximately 1.5 million are school-age children and
youth who speak one of 40 indigenous languages and thus have a need for educational resources in
those languages.205 Estimates suggest that 20% of the schools in Peru should be offering these services,
but only 40% of them are estimated to de facto provide them appropriately (with TLMs, teachers who
speak the languages, specific pedagogical strategies, and pedagogical support).206

Peru has a long history of gradually elevating the status of indigenous languages and integrating
bilingual education into its education system. The notion of bilingual education was first introduced
in 1952, but early programs focused primarily on isolated indigenous communities in the Peruvian
Amazon rainforest. These programs were heavily supported by international NGOs and did not present
a comprehensive national approach to language issues since the much larger Quechua speaking popu-
lations were excluded from these programs. Subsequent actions like the Education Reform of 1972
that expanded educational services to indigenous populations in the Andean and Amazonian regions,
the recognition of Quechua as a language in 1975, the creation of a national Directorate of Bilingual
Education within the Ministry of Education in 1988, and the establishing of Articles 2 and 17 in the
1993 Constitution helped create the right environment for effective bilingual education policies.
Article 2 of the 1993 Constitution gave official language status to all indigenous languages in Peru, and
Article 17 outlined the obligation to promote bilingual and intercultural education in accordance with
the individual characteristics of each zone.

Following the enactment of the 1993 Constitution, significant efforts made bilingual education in
Peru a reality. Bilingual intercultural education (Educación Intercultural Bilingüe, EIB in Spanish) in
Peru is characterized by the use of L1 in the early years, to learn how to read and write, while intro-
ducing Spanish as a second language. Over time, Spanish becomes the language of instruction (LoI).
Peru’s EIB specifies that the shift from the L1 to Spanish should depend on the child’s proficiency in
Spanish. For example, children who enter school as monolingual in an indigenous language use their
L1 as the LoI and study Spanish as a second language during grades 1 and 2. In grade 3, Spanish is
introduced as LoI for 20 percent of class time, and this proportion increases by 10 percent per year,
to reach 50 per cent by grade 6. 207 An important aspect of the Peruvian model is the emphasis on the
culture of the students, which is captured in the curriculum as well as teaching and learning materials.
The EIB program does not change the curriculum for children, since all children in Peru are expected
to learn the same content, but it does introduce culturally sensitive and indigenous-language mate-
rials to promote learning among indigenous communities.208



ANNEXES                                                                                                     79
     Starting in 1996, various programs have worked to develop and distribute high-quality TLM for
     bilingual education and to train teachers in the pedagogical skills required. From 1996 to 2000, 94
     bilingual teaching manuals were produced, more than 10,000 teachers were trained in bilingual teaching,
     and other relevant materials were distributed.209 Starting in 2004, the Rural Education and Teacher
     Development Project, which was funded by the World Bank Group, also worked to strengthen the EIB
     program. As part of this project, TLM was developed including: (i) workbooks for the six primary grades
     in 10 Peruvian languages and five dialects of Quechua; (ii) teachers' guides on how to maximize the
     use of classroom instructional materials; (iii) teachers' guides on effectively teaching a multigrade class;
     (iv) play kits to strengthen language acquisition and use both in Spanish and the Peruvian languages;
     and (v) an outreach package for teachers to use in introducing parents and community members to
     the advantages of bilingual instruction.210 The practice of developing suitable TLM to strengthen and
     facilitate bilingual language education continues today. As the COVID-19 pandemic has led to school
     closures around the country and the Ministry of Education manages the continuity of learning through
     the school closures, high-quality TLM enables children all around the country to continue learning in
     the language that they know and understand best. Aprendo en Casa offers online resources for children
     to continue learning remotely in their L1. 211 The site currently offers online lessons in nine languages. In
     addition to this, a network of 35 radio stations broadcast structured educational material in L1 to support
     remote learning. 212

     A variety of factors hinder implementation of bilingual intercultural education in Peru. While large
     amounts of materials have been developed, many schools still lack appropriate materials to meet their
     language-grade needs.213 Finding qualified teachers who have the right language skills has also proven
     to be difficult in some areas, and even when those are available, the level of support they receive may be
     insufficient.214 Another complication relates to multi-grade classrooms, in which it can be difficult to
     follow LoI guidelines.215 A further challenge involves community perspectives. Many indigenous parents
     do not support the idea of bilingual education because they fear that it will interfere with Spanish acqui-
     sition, and ultimately keep their children from gaining access to social and economic opportunity.216

     Despite challenges, bilingual intercultural Education in Peru has shown promising results. Evidence
     suggests that the EIB program results in positive cognitive and emotional impacts on indigenous chil-
     dren. A meta-analysis of the various evaluations highlighted improved overall learning outcomes, higher
     numeracy outcomes, more self-confidence in the use of Spanish, more student participation in the class-
     room, and decreased (albeit still significant) disparities in learning outcomes between rural and urban
     students. 217 When used to investigate the effect of Quechua-medium instruction on academic achieve-
     ment, the Peruvian school-level data from the Young Lives international study of childhood poverty
     showed that indigenous children who attend Quechua-medium schools achieve mathematics scores
     0.54 standard deviations higher than indigenous children who attend Spanish-medium schools.218 The
     same data, however, fails to demonstrate conclusively whether or not literacy scores reflect a compa-
     rable effect.




80                                                         LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
ANNEX E:
India:
The Language &
Learning Foundation,
Transforming Teaching

Despite massive gains in access to education (97 percent of children enrolled in primary schools),
learning assessments reveal that by the end of grade 5 almost half the children in India have not
achieved minimum proficiency levels in literacy and numeracy. 219 About 50 million children enrolled
in government schools 220 in grades 1 to 3 are at the highest risk of not acquiring basic literacy skills.221
Children who study in an unfamiliar language face a huge learning disadvantage. It is estimated that
25 percent of primary school children in India face a moderate to severe learning disadvantage as a
consequence of the fact that the language used at school is not their home language.222


Language & Learning Foundation’s (LLF) vision for change
Goal: Language and Learning Foundation was founded in 2015 with the goal of improving the language
learning outcomes of children in grades 1 to 3 in their home and school languages by working in collab-
oration with state governments. LLF is currently working with seven state governments to reach
roughly 17 million children through its programs, thus helping more than one-third of the at-risk chil-
dren in these states.

Objectives: To reach this goal, LLF focuses on three strategies:
    1. Enhancing state governments’ commitment to improving language and literacy instruction;
    2. Persuading governments to include children’s home languages (L1) in school; and through
       multilingual approaches; and
    3. Developing capacity in the public education ecosystem to implement and sustain high quality
       interventions in early literacy and multilingual education at scale.


LLF’s core operating principles
Program Design
   • Transform classroom teaching and learning processes for language.
   • Incorporate children’s home languages in varied and pragmatic ways, such as using these
     non-dominant languages as language of instruction (LoI) in the early primary years or strategic
     use of children’s L1 to support instruction through a different LoI.
   • Focus on equitable learning for all children, without exception, through active engagement strat-
     egies, frequent review of lessons (within a carefully structured “spiraling” curriculum), differ-


SECTION 4: LANGUAGE OF INSTRUCTION POLICY SHOULD BE MULTIFACETED                                                81
          entiated instruction, and extra support for struggling learners as a part of the regular teaching–
          learning process.

     Functionality
        • Target educationally disadvantaged areas, especially where children’s home languages are differ-
          ent from the LOI at school.
        • Work at scale in collaboration with state governments. Focus on system-oriented work through
          continuous professional development and support for basic reforms.
        • Contextualize and co-create instructional design, materials, and training at each location, build-
          ing on research and evidence-based principles and strategies.
        • Generate and share knowledge and evidence through internal monitoring, evaluation, and process
          studies.


     LLF’s three-pronged approach for implementing early language learning
     programs at scale
     As LLF’s goal is to build commitment and capacity within the government education system to plan
     and implement early language learning programs at scale, LLF works in collaboration with state
     governments from the beginning on the three main dimensions shown in Figure F1.


     FIGURE F.1: LLF’s three-pronged strategy
                                       LLF’s Three-Pronged Strategy



                            1                               2                                      3
                       CONTINUOUS                      SCHOOL LEVEL                            SYSTEMIC
                      PROFESSIONAL                    INTERVENTIONS                            REFORMS
                      DEVELOPMENT                   School demonstration                     Systemic reform
                         COURSES                   projects that showcase                  related to effective
                  Continuous preofessional         transformed teaching-                implementatioon of high
                  development of teachers,         learning processes at a                quality early learning
                   teacher educators, and         reasonable scale and the                 programs of scale.
                 educational administrators          support government
                    on early literacy and           agencies to scale the
                   multilingual education                 program.




                                              TWO MODES OF INTERVENTION


                         Direct Implementation                      Technical Support to State Governments



                             INTERNAL MONITORING, PROCESS STUDIES AND
                            RESEARCH TO GENERATE EVIDENCE AND KOWLEDGE




82                                                           LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
Continuous professional development courses: LLF designs and implements blended capacity
building courses on early literacy and multilingual education for teachers and teacher educators. The
courses provide conceptual understanding of early language learning as well as classroom teaching
and learning activities and materials. Designed with the aim of creating conviction and capacity within
    education
the 1:          systemlanguage,
       Understanding,   at all levels, these          play a key
                                              courses teaching
                                         2: Language              role in in
                                                               practices   supporting  large-scale
                                                                                  3: Principles      literacyand
                                                                                                 of teaching  and
multilingual   programs
    literacy and          and sustaining
                 multilingualism                  over the long
                                             them classrooms:
                                         primary                   term. The following
                                                               A reflective               list in
                                                                                     learning   shows    the main
                                                                                                  early grades
themes covered in the courses of longer durationanalysis(Figure F2).
         4: Importance of using                  5: Balanced approach for language             6: Oral language development
  children’s home languages and                      teaching in primary grades
   BOX      F.2: Major
        introducing        themes included in LLF courses
                       multilingual
                education
  1: Understanding, language, literacy and multilingualism          8: Learning to decode
      7: Emergent literacy, print                       8: Learning to  decode                   9: Reading with fluency and
  2:awareness      and phonological
     L anguage teaching practices in primary classrooms:                                                comprehension
                                                                    9: Reading with fluency and comprehension
     A reflective analysis
                awareness                                           10: Teaching–learning strategies for reading
   3: Principles of teaching and learning in early grades
 10: Teaching–learning strategies                                ofDevelopment
                                                 11: Development11:             of writing skills
                                                                    writing skills         12: Approaches and strategies
  4: I mportance of using children’s home languages and intro- 12: Approaches and             for multilingual
                                                                                   strategiesfor                education
               for  reading
     ducing multilingual education                                                                 multilingual    education
                                                                  13: Framework for grades 1 to 3 for language teaching–
  13:  Framework
   5: B                   grades
                      forfor
        alanced approach           1 to
                             language    3 14:
                                      teaching      Addressing multi-grade
                                               in primary              learningteaching        15: Providing regular academic
      grades
  for language teaching–learning                and multiple learning    levels   in the             support
                                                                   14: Addressing multi-grade teaching        to teachers
                                                                                                        and multiple learning
   6: Oral language development                              classroom levels in the classroom
   7: Emergent literacy, print awareness and phonological        15: Providing regular academic support to teachers
      awareness
The courses and workshops precede implementation of school-level projects. They range from month-
                                                       long courses on specific themes within early
                                                      The    courses
                                                       literacy        and workshops
                                                                 and multilingual            precede
                                                                                     education          imple-
                                                                                                  to a year-
                                                       long   flagship   course.    These    courses
                                                      mentation of school-level projects. have           They
                                                                LLF build
                                                       helpedfrom
                                                      range                            coursesitson
                                                                          credibility around
                                                                      month-long                   academic
                                                                                                       specific
                                                       leadership
                                                      themes               earlyliteracy
                                                                    for early
                                                                 within                   andand
                                                                                   literacy     multilingual
                                                                                                     multilin-
                                                       education.   Some   of
                                                      gual education to a year-long flagshiptend
                                                                              the  courses   (which       to
                                                                                                       course.
                                                       be    oversubscribed)      are   also   certified
                                                      These courses have helped LLF build credi-          by
                                                       universities in the country. External evaluations
                                                      bility around its academic leadership for early
                                                       of the nine-month course have shown
                                                      literacy and multilingual education. Some of
                                                       significant learning gains for the participant
                                                      the  coursesand
                                                       teachers      (which   tend to be oversubscribed)
                                                                          implementation        of positive
                                                      are  also certified  by  universities
                                                       changes in teaching–learning practice  in theincountry.
                                                                                                        their
                                                      External
                                                       classrooms.evaluations   of  the  nine-month     course
have shown significant learning gains for the participant teachers and implementation of positive
changes in teaching–learning practice in their classrooms.

School-level    interventions:
School-level interventions:     LLFLLF   designs
                                     designs        and implements
                                             and implements             projects
                                                                 projects         that demonstrate
                                                                          that demonstrate               transformative
                                                                                               transformative    change
in the teaching
change            and learning
         in the teaching    and of  languages
                                 learning       as well as improved
                                            of languages   as well asstudent
                                                                       improvedlearning.
                                                                                  student Their  programs
                                                                                            learning.       areprograms
                                                                                                                 initially
                                                                                                        Their
implemented
are              in 150–200 schools
    initially implemented               and are
                                in 150–200        later and
                                              schools   scaled
                                                             areuplater scaledmore
                                                                    to cover    up toschools  morestate
                                                                                       cover with         government
                                                                                                     schools   with state
support.   These   comprehensive      projects   aim at  building
government support. These comprehensive projects aim at building the capacity of teachers on
                                                                   the  capacity of  teachers   and  school   heads     and
English-language
school               learning or multilingual
        heads on English-language               education
                                          learning          (ELL or MLE)
                                                     or multilingual      through: (ELL
                                                                        education   training
                                                                                           or workshops;
                                                                                              MLE) through:orientation
                                                                                                                  training
of government
workshops;        staff responsible
               orientation                        supervision
                                      for schoolstaff
                              of government                    and support;
                                                       responsible            regular
                                                                      for school       academicand
                                                                                  supervision     support   to teachers
                                                                                                       support;    regular
through    classroom    visits; review   meetings    and   mobile    phone-based    resource
academic support to teachers through classroom visits; review meetings and mobile phone-based   sharing;  provision    of
instructional materials (such as workbooks); decoding and reading materials such as reading cards and
resource sharing; provision of instructional materials (such as workbooks); decoding and reading
storybooks; and the creation of print-rich classrooms and regular student assessments. One LLF mentor
materials such as reading cards and storybooks; and the creation of print-rich classrooms and regular
provides academic support to teachers in 50 schools, in collaboration with existing government
student
monitoringassessments.          LLF mentor
                          Oneevaluations
              staff. External                of provides  academic
                                                LLF’s school  programs support  to teachers
                                                                          have shown           in 50learning
                                                                                                     schools,gains
                                                                                        significant             in collab-
                                                                                                                      for
oration
children. with  existing   government      monitoring     staff. External   evaluations    of  LLF’s  school    programs
have shown significant learning gains for children.

                                                             69

ANNEXES                                                                                                                         83
         Systemic reforms: LLF works with state education institutions to influence the state-wide annual
         in-service teacher training, school-based assessments, textbook review and revision and the develop-
               of improved
         ment reforms:
     Systemic                 works withteacher
                         LLF preservice           educationinstitutions
                                           state education   curricula. LLF  also supports
                                                                         to influence       state governments
                                                                                       the state-wide          for
                                                                                                       annual in-service
         formulating   policies on first-language-based  multilingual education.  LLF  develops  and
     teacher training, school-based assessments, textbook review and revision and the development of disseminates
         public preservice
     improved   goods like instructional    design and
                             teacher education         assessment
                                                   curricula.      frameworks,
                                                              LLF also  supportsteaching   and learning for
                                                                                                         practices
                                                                                   state governments        formulating
         videos, handbooks for teachers and mentors, course materials, resource packs and playbooks that can
     policies on first-language-based multilingual education. LLF develops and disseminates public goods like
         support implementation.
     instructional design and assessment frameworks, teaching and learning practices videos, handbooks for
     teachers and mentors, course materials, resource packs and playbooks that can support implementation.
          Technical support to state governments: In the technical support model, LLF works with state level
          institutions to co-create instructional design; develop children’s learning materials, training manuals
     Technical    support
          and teacher        to state
                          guides;               state-levelInmaster
                                         governments:
                                   and orient                 the technical
                                                                     trainers support
                                                                               for trainingmodel,    LLF works
                                                                                               of teachers.       with
                                                                                                              In this   state level
                                                                                                                      model,
     institutions
          LLF does  tonot
                       co-create    instructional
                           post its staff            design;
                                          in the field.      develop children’s
                                                        The responsibility         learning
                                                                            for local          materials,
                                                                                      implementation         training
                                                                                                          lies        manuals
                                                                                                                        state and
                                                                                                               with the
     teacher   guides;   and  orient   state-level   master  trainers  for training   of  teachers.   In this
          governments, with only technical support from LLF. This model has the potential to influence the    model,  LLF  does not
           its staff in the field.
     postteaching–learning          The   responsibility  for
                                process and student learning outcomes in about 200,000 primary schools across with
                                                              local implementation       lies with  the  state  governments,
           technical
     onlythree           startingfrom
                 states,support          LLF. This model has the potential to influence the teaching–learning process
                                   in 2021.
     and student learning outcomes in about 200,000 primary schools across three states, starting in 2021.
         Pedagogical
     Pedagogical             Underpinning
                    Underpinning
         1. 1.Transforming
               Transforming the
                              the teaching
                                  teachingand
                                            andlearning   processLLF believes that sustainable improvement in
                                                 learningprocess:
               learning  outcomes   happens  only when  the
     LLF believes that sustainable improvement in learning  teaching and learning
                                                                  outcomes        process
                                                                             happens      is transformed
                                                                                       only              along and
                                                                                             when the teaching
               the dimensions
     learning process                     F3. the dimensions in Figure F3.
                                in Figurealong
                        is transformed

                                 FigureinF.3:
          FIGURE F.3: Vision of change        Visionand
                                          teaching   of learning
                                                        change in teaching and learning




          2. Including children’s home languages
              2. Including children’s home languages
     The best way to support children from non-dominant language backgrounds is to use their language as
              The best
     the medium            way to support
                      of instruction         children
                                      for several        from
                                                    years,     non-dominant
                                                             while               language backgrounds
                                                                    gradually introducing                is to use their
                                                                                             additional languages.    However,
              language as the medium of instruction for several years, while gradually introducing additional
     sociolinguistic    situations in India are varied and complex and a ‘one-size-fits-all’ approach for using
              languages. However, sociolinguistic situations in India are varied and complex and a ‘one-size-
     children’s    languages
              fits-all’ approachis not  appropriate.
                                    for using  children’s Anlanguages
                                                              understanding      of the language
                                                                         is not appropriate.        contexts of of
                                                                                              An understanding    schools
                                                                                                                      the and
     classrooms            prerequisite
                     is a contexts
              language                     for developing
                                    of schools  and classrooms appropriate     strategies
                                                                   is a prerequisite        for inclusion
                                                                                      for developing        of children’s
                                                                                                                  strate- first
                                                                                                     appropriate
              giesAfor
     languages.       simple  sociolinguistic
                         inclusion of children’smapping     exercise
                                                 first languages.     is necessary
                                                                   A simple          to identify
                                                                             sociolinguistic     the approach
                                                                                              mapping           and
                                                                                                      exercise is    strategies
                                                                                                                  neces-
              sary   to identify the approach
     for including children’s languages.        and  strategies for  including  children’s languages.

     Multilingual Education


                                                                  LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
84
                                                                  70
Multilingual Education
LLF supports the following three approaches for the use and sequencing of languages in bilingual
or multilingual programs. The proportion of schools included under each of these approaches (as a
percentage of total schools where LLF is implementing programs or providing technical support) is
indicated in brackets:
    1. Multilingual education based on L1, whereby children’s L1 is used as LoI for the primary
       years; with L2 taught as a subject before being introduced as LoI using bilingual methodology.
       (15%)
    2. L2 used as LoI with development and extensive use of L1 in the oral domain. Purposeful, stra-
       tegic and flexible use of languages as part of teaching and learning. (80%)
    3. Using multiple L1s in the classroom as a resource to develop competence in L2 and additional
       languages through strategies of translation, multilingual communication, comparing and
       contrasting languages, and so forth. Few working models are available currently. A local lingua
       franca to which most children are already exposed could be used as a means of communica-
       tion, or even as LoI. (5%)


Essential elements of multilingual education
LLF has identified the following essential elements for their work:
   • Children’s first languages (L1s) are used formally in the classroom.
   • Languages are not taught and learnt in water-tight compartments; fluid and flexible use of lan-
     guages is encouraged.
   • An MLE classroom reflects tolerance and mutual respect for all languages and cultures.
   • A multilingual approach to teaching and learning is used across the curriculum.
   • Appropriate second-language teaching and learning methods for gradual learning of unfamiliar
     languages.
   • A multilingual approach is also multicultural, bringing children’s cultural practices and experi-
     ences into the classroom.

Ajuvaroo (light), LLF’s Multilingual Education Program, has been implemented since 2019 in 40
schools in the Dungarpur district of Rajasthan in Western India. Most children in these schools belong
to indigenous tribes and speak a language called Wagdi and its dialects. They have little or no under-
standing of Hindi, the language used as the LoI, when they start grade 1. The program aims to main-
tain and develop Wagdi (children’s L1), while gradually building basic vocabulary and understanding
of Hindi. The main components of the program are shown in Figure F4.

The emphasis is on the use of children’s orality and developing Wagdi for higher order thinking,
reasoning and oral expression through activities like interactive read aloud and conversations. During
this period, strategies like Total Physical Response (TPR) are used to help children pick up about 200
words, phrases, and simple sentences. L1 is used extensively and strategically to scaffold L2 learning in
the first two years; for example, L1 is used for new or difficult concepts and for higher order compre-
hension, while L2 is used for simple texts with recall-type comprehension. Teachers use and promote
purposeful mixing of languages (“translanguaging”), to enable full comprehension and expression.
The classrooms have seen a transformation in terms of children’s language skills and their active
engagement in the classroom process. Learning assessments at the end of the first year show signifi-
cant gains in learning outcomes.




ANNEXES                                                                                                     85
                 FIGURE F.4: Main components of Rajasthan’s MLE program




         The emphasis is on the use of children’s orality and developing Wagdi for higher order thinking, reasoning
         and oral expression through activities like interactive read aloud and conversations. During this period,
         strategies like Total Physical Response (TPR) are used to help children pick up about 200 words, phrases,
         and simple sentences. L1 is used extensively and strategically to scaffold L2 learning in the first two years;
         for example, L1 is used for new or difficult concepts and for higher order comprehension, while L2 is used
         for simple texts with recall-type comprehension. Teachers use and promote purposeful mixing of
         languages (“translanguaging”), to enable full comprehension and expression. The classrooms have seen a
         transformation in terms of children’s language skills and their active engagement in the classroom
          emphasis
     Theprocess.       is on the
                Strategies
                   Learning       use building
                                 for   of children’s
                               assessments             orality
                                                    strong
                                               at the  end      andfirst
                                                                the  developing
                                                            of partnerships      Wagdi
                                                                                 with
                                                                         year show               gainsorder
                                                                                         for higher
                                                                                        governments
                                                                                    significant              thinking,
                                                                                                       in learning       reasoning
                                                                                                                     outcomes.
     and oral expression through activities like interactive read aloud and conversations. During this period,
                To ensure sustainable change, LLF’s work is focused on partnership with governments. Figure F5
     strategies        Total
                 likefor
         Strategies
                outlines   the Physical
                           building
                               key leversResponse
                                     strong           (TPR) are
                                          for partnerships     with
                                              strong partnership  used  to help children pick up about 200 words, phrases,
                                                                    governments
                                                                   with governments.
     andTosimple
            ensure sentences.
                      sustainableL1 change,
                                     is used extensively
                                               LLF’s work isand   strategically
                                                               focused  on partnership  withL2governments.
                                                                                                learning in the
                                                                                to scaffold                       first F5
                                                                                                               Figure   two  years;
                                                                                                                           outlines
         example,
     for the         L1  is used  for  new   or difficult concepts
             key levers for strong partnership with governments.      and for higher  order  comprehension,      while   L2 is used
                    textsF.5:
     for simpleFIGURE       with   recall-type
                               Strategies         comprehension.
                                           for building                           useownership
                                                                       Teachers and
                                                        government partnership         and promote purposeful mixing of
     languages (“translanguaging”),         to enable
                            Figure F.5: Strategies    forfull comprehension
                                                          building  government  and expression.and
                                                                                  partnership      Theownership
                                                                                                        classrooms have seen a
     transformation in terms of children’s language skills and their active engagement in the classroom
     process. Learning assessments at the end of the first year show significant gains in learning outcomes.

     Strategies for building strong partnerships with governments
     To ensure sustainable change, LLF’s work is focused on partnership with governments. Figure F5 outlines
     the key levers for strong partnership with governments.

                      Figure F.5: Strategies for building government partnership and ownership




                                                                    72



86                                                                   LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
Lessons Learned and the Way Forward
Here are some challenges and suggestions based on LLF’s experience:
     1. Extend language and literacy development programs to later primary grades: Large-scale
        early grade reading projects need to evolve into systematic, holistically designed bilingual or
        multilingual programs for the entire primary cycle. Early grade reading programs cannot be
        effective unless children’s first languages are formally included and used for oral language
        development, including thinking and reasoning and early literacy. Additionally, the develop-
        ment of bilingual or multilingual abilities and multiliteracies require a commitment to building
        a scope and sequence that spans several years in the child’s primary language.
     2. Program designs must include multiple grades of curriculum development and instruction
        to support multigrade teaching situations and multiple learning levels within classrooms:
        70 percent of primary schools in India have multigrade teaching situations and students with
        a wide range of learning levels. However, most NGO and government programs to improve
        learning are implemented with monograde teaching designs. Program designs must include
        ongoing assessment, differentiated instruction and extra support for children falling behind as
        integral parts of the instructional strategy.
     3. School preparedness must be prioritized by all stakeholders: Investment in high-quality early
        childhood education programs in children’s home languages will significantly help to improve
        learning outcomes in early primary grades.
     4. Working with governments enables scale and sustainability: It is important to work with
        several levels of the government education system, from the middle level of field supervisors
        all the way up to the senior education leadership at the state and national level.


Overview of Program Costs
LLF has managed to keep program costs within reasonable limits due to the significant contribution
made by the state governments (see Table F1).

TABLE F.1: Overview of LLF program costs

                                                                                                GOVERNMENT CONTRIBU-
                                     PARTICIPANT EFFORT             LLF COST PER PARTICIPANT/
        PROGRAM                                                                                  TION PER PARTICIPANT/
                                         (IN HOURS)                        CHILD (USD)
                                                                                                      CHILD (USD)
                                              Professional Development Courses
 1-year blended course* 300                                        400                          125
 3-month course*              85                                   125                          135
 1-month course*              20                                   28                           n/a

                                                   School Level Interventions
 Demonstrated
 projects (150-200
                              n/a                                  13                           9.5
 schools) including
 developmental costs
 Programs at scale**          n/a                                  6.2                          3.5


*One mentor for 20-25 participants
**In scaled programs, one LLF academic mentor is available for 5o primary schools.


ANNEXES                                                                                                                  87
     REFERENCES




     Abadzi, Helen. 2012. “Developing Cross-Language Metrics for Reading Fluency Measurement: Some
          Issues and Options (English).” Global Partnership for Education (GPE) Working Paper 6, World
          Bank, Washington DC.
     Abu-Rabia, S. 2000. “Effects of Exposure to Literary Arabic on Reading Comprehension in a Diglossic
          Situation.” Reading and Writing: An Interdisciplinary Journal 13: 147-157.
     Adegbija, E., 1994. Language Attitudes in Sub-Saharan Africa: A Sociolinguistic Overview. Multilin-
          gual Matters Ltd., Clevedon.
     Akmal, M., M. Musa, and M. Silverman. 2020. Diaries from the Frontline: Distance Learning for
          Offline Children. Center for Global Development. May 1, 2020.
     Al-Huri, Ibrahim. 2015. “Arabic Language: Historic and Sociolinguistic Characteristics.” English
          Literature and Language Review 1 (4): 28–56.
     Alidou, H., A. Boly, B. Brock-Utne, Y. Diallo, K. Heugh, and H. Wolff. 2006. Optimizing learning and
          education in Africa: The language factor. Human Sciences Research Council, Cape Town.
     Alidou, H., & Brock-Utne, B. (2011). Teaching practices: Teaching in a familiar language. In A. Ouane
          and C. Glanz (Eds.), Optimising learning, education and publishing in Africa: The language factor
          (pp. 159–185). UNESCO Institute for Lifelong Learning (UIL), Germany.
     Andina. 2019. Aprendo en Casa: estas 35 radios regionales transmiten contenidos en lenguas originarias.
          https://andina.pe/agencia/noticia-aprendo-casa-acompanamiento-docente-llega-al-859-las-famili-
          as-peru-810473.aspx
     ASER. 2018. Annual Status of Education Report.
     August, D., & Shanahan, T. (Eds.). (2006). Executive summary: Developing literacy in second-lan-
          guage learners. Report of the National Literacy Panel on Language-Minority Children and Youth.
          Mahwah, NJ: Lawrence Erlbaum Associates.
     Azevedo, J. P. 2020. Learning Poverty in the time of COVID-19: A crisis within a crisis. World Bank
          Brief.
     Azubuike, O.B., Gayatri Vaidya and Rhiannon Moore. 2017. ‘Functional English’ Skills in Ethiopia,
          India and Vietnam: Comparing English Ability and Use Among 15 Year Olds in Three Countries.
     Baker, C. 2001. Foundations of bilingual education and bilingualism. Tonawanda, NY: Multilingual
          Matters.
     Ball, J. (2011). Enhancing learning of children from diverse language backgrounds: Mother tongue-
          based bilingual or multilingual education in the early years. Paris, France: United Nations Educa-
          tional, Scientific and Cultural Organization (UNESCO).
     Banerjee, A. & E. Duflo. 2006. Addressing absence. Journal of Economic Persepctives, 20(1), 117–132.
     Barro, R. & J. Lee. 2013. A New Data Set of Educational Attainment in the World, 1950–2010. Journal
          of Development Economics, vol 104, pp.184–198.
     Batibo, H. M. 2014. Searching for an optimal national language policy for sustainable development. In
          H. McIlwraith (ed.), The Cape Town Language and Development Conference: Looking beyond
          2015, 16–20. London: British Council.


88                                                      LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
Benson, C. 2000. The primary bilingual education experiment in Mozambique, 1993 to 1997. Interna-
    tional Journal of Bilingual Education and Bilingualism 3(3):149–166.
Benson, C. 2002. Real and Potential Benefits of Bilingual Programmes in Developing Countries, Inter-
    national Journal of Bilingual Education and Bilingualism, 5:6, 303–317.
Benson, C. 2003. Possibilities for educational language choice in multilingual Guinea-Bissau. In Huss,
    Lena, Antoinette Camilleri and Kendall King (eds.), Transcending monolingualism: Family, school
    and society, 67–87. Lisse, the Netherlands: Swets and Zeitlinger.
Benson, C. 2004. Bilingual schooling in Mozambique and Bolivia: From experimentation to imple-
    mentation. Language Policy, 3(1), 47–66.
Benson, C. 2005. Girls, Educational Equity and Mother Tongue-based Teaching, Bangkok, Thailand:
    UNESCO, 2005.
Benson, C. 2008. Summary overview: Mother tongue-based education in multi-lingual contexts. In
    UNESCO (Ed.), Improving the quality of mother tongue-based literacy and learning: Case studies
    from Asia, Africa and South America. UNESCO Asia and Pacific Regional Bureau for Education.
    Bangkok.
Beteille, T, and D. Evans. 2019. Successful Teachers, Successful Students: Recruiting and Supporting
    Society’s Most Crucial Profession. The World Bank Group.
Boakye-Yiadom, F, M. Matthews, E. Miksic, and B. Trudell. 2018. Developing Early Grade Reading
    Materials in 11 Languages: Learning in Ghana. Presentation at the Comparative and International
    Education Society (CIES) annual conference, Mexico City.
Boateng, Pearl. 2019. Managing transitions from mother tongue instruction to English as the medium
    of instruction.
Bloch, C. 2014. Growing young readers and writers: Underpinnings of the Nal’ibali National Read-
    ing-for-Enjoyment Campaign. In H. McIlwraith (ed.), The Cape Town Language and Develop-
    ment Conference: Looking beyond 2015, 50–56. London: British Council.
Brenzinger, M., 1997. An evaluative account of Ethiopia’s new language policy. Language choices:
    Conditions, constraints, and consequences, pp.207–221.
Brown, V. 2011. Mango Tree Lao Literacy Project: Early Grade Reading Assessment end of year
    survey. Final report. Kampala, Uganda: Mango Tree. Retrieved from:                               :
    http://mangotreeuganda.org/downloads/NULP%202011%20EGRA%20Report.pdf
Brunette, T., Piper, B., Jordan, R., King, S., and Nabacwa, R. 2019. The Impact of Mother Tongue
    Reading Instruction In Twelve Ugandan Languages and the Role of Language Complexity, Socio-
    economic Factors, and Program Implementation. Comparative Education Review, 63(4), 591–612.
    http://doi.org/10.1086/705426
Bühmann, D. and B. Trudell. 2008. Mother Tongue Matters: Local Language as a Key to Effective
    Learning. UNESCO.
Central Statistics Agency (CSA). 2008. Summary and Statistical Report of the 2007 Population and
    Housing Census. Addis Ababa.
Chicoine, L., 2019. Schooling with learning: The effect of free primary education and mother tongue
    instruction reforms in Ethiopia. Economics of Education Review, 69, pp.94–107.
Clark, M. 1979. Reading in Spanish and English: Evidence from adult ESL students. Language Learning,
    29(1), 121–150.
Clegg, J. 2005. Moving towards bilingual education in Africa (Presentation). 7th International
    Language Development Conference, Addis Ababa, 2015. [Online]
Clegg, J. and J. Simpson. 2016. Improving the effectiveness of English as a medium of instruction in
    sub-Saharan Africa. Comparative Education, 52(3), pp. 359–374.
Cohen, G.P., 2007. Mother tongue and other tongue in primary education: Can equity be achieved with
    the use of different languages? Language and development: Africa and beyond, pp.62–75.
Collier, V. and W. Thomas. 2004. The astounding effectiveness of dual language education for all.
    NABE Journal of Research and Practice 2(1):1–20.


REFERENCES                                                                                               89
     Collier, V. and Wayne P. Thomas. 2017. Validating the Power of Bilingual Schooling: Thirty-two Years
         of Large Scale, Longitudinal Research. Annual Review of Applied Linguistics, 37, 1–15.
     Comings, J. 2014. An evidence based model for early-grade reading programmes.
     Crawford, N., J. Cosgrave, S. Haysom, and N. Walicki. 2015. Protracted Displacement: Uncertain Paths
         to Self-Reliance in Exile. London: Overseas Development Institute.
     Cummins, J. 1979. Linguistic interdependence and the educational development of bilingual children.
         Review of Educational Research, 49(2), 222–251.
     Cummins, J. 1981. The role of primary language development in promoting educational success for
         language minority students. In California State Department of Education (Ed.), Schooling and
         language minority students: A theoretical framework (pp. 3–49). Los Angeles: Evaluation, Dissem-
         ination and Assessment Center, California State University.
     Cummins, J. (2009). Fundamental psycholinguistic and sociological principles underlying educational
         success for linguistic minority students. In T. Skutnabb-Kangas, R. Phillipson, A. Mohanty, A., &
         M. Panda, (Eds.), Social justice through multilingual education (pp. 19–35). Bristol: Multilingual
         Matters.
     Cummins, J. (2012). The intersection of cognitive and sociological factors in the development of
         reading comprehension among immigrant students. Reading and Writing, 25, 1973–1990.
     Denton, C. A., Fletcher, J. M., Taylor, W. P., Barth, A. E., & Vaughn, S. (2014). An experimental evalua-
         tion of Guided Reading and explicit interventions for primary-grade students at-risk for reading
         difficulties. Journal of Research on Educational Effectiveness, 7(3), 268–293.
     Demilew, F.A. and Sekeroglu, B., 2019. Ancient Geez script recognition using deep learning. SN Applied
         Sciences, 1(11), p.1315.
     Devictor, X. and Q. Do. 2016. How Many Years Have Refugees Been in Exile? Policy Research Working
         Paper No. WPS 7810. Washington, DC: World Bank.
     Diarra, E. 2003. Choice and description of national languages with regard to their utility in literacy
         and education in Angola. In A. Ouane (ed.), Toward a multilingual culture of education, 333–348.
         Hamburg: UIE.
     DIGEIBIR – Dirección General de Educación Intercultural Bilingüe y Rural (2013) Hacia una
         Educación Intercultural Bilingüe de Calidad Propuesta Pedagógica, Lima: Ministry of Education,
         Peru.
     Dobbie, W. and Fryer Jr, R. G. (2013). “Getting beneath the veil of effective schools: Evidence from
         New York City.” American Economic Journal: Applied Economics, 5(4), 28–60.
     Dryden-Peterson, Sarah. 2006. “The Present Is Local, the Future Is Global? Reconciling Current and
         Future Livelihood Strategies in the Education of Congolese Refugees in Uganda.” Refugee Survey
         Quarterly 25 (2): 81–92.
     Dryden-Peterson, Sarah. 2015. The educational experiences of refugee children in countries of first
         asylum. Washington DC: Migration Policy Institute.
     Dryden-Peterson, Sarah. 2016. “Refugee Education in Countries of First Asylum: Breaking Open the
         Black Box of Pre-Resettlement Experiences.” Theory and Research in Education 14 (2): 131–48.
     Eble, A., C. Frost, A. Camara, B. Bouy et al., 2019. How much can we remedy very low learning levels in
         rural parts of low-income countries? Impact and generalizability of a multi-pronged para-teacher
         intervention from a cluster-randomized trial in The Gambia. (EdWorkingPaper: 19–157).
     Education Bureau (EDB) 1997a. Medium of Instruction – Guidance for Secondary Schools.
     EBD 1997b. Medium of Instruction – Guidance for Secondary Schools – Why should we teach in the
         mother tongue?
     EDB 2002. Evaluation on the Implementation of Medium of Instruction Guidance for Secondary
         Schools.
     EDB 2012. Major Studies in the Use of Different Medium of Instruction Mode in Hong Kong.
     EDB 2019. Language Proficiency Assessment for Teachers (English Language) 2019 Assessment
         Report.


90                                                       LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
EdQual. 2010. The impact of language of instruction, teacher training and textbooks on quality of
    learning in Africa. Language of Instruction and Quality of Learning in Tanzania and Ghana.
    EdQual Policy Brief No. 2.
Education Commission 1990. Education Commission: Report 4.
Ethnologue database. 22nd Edition. Ethnologue.com.
Evans, David K., and Amina Mendez Acosta. 2020. Education in Africa: What Are We Learning?
    Working Paper 542. Center for Global Development. Washington, DC.
Evans, S. 2013. The Long March to Biliteracy and Trilingualism: Language Policy in Hong Kong
    Education Since the Handover. Annual Review of Applied Linguistics, 33, 302–324. doi:10.1017/
    S0267190513000019
Evans, S. & B. Morrison. 2017. English-medium instruction in Hong Kong: illuminating a grey area in
    school policies and classroom practices, Current Issues in Language Planning, 18:3, 303–322.
Fazio, L. and Lyster, R. 1998. Immersion and submersion in classrooms: A comparison of instructional
    practices in language arts. In: Journal of Multilingual and Multicultural Development 19 (4):
    303–317.
Fietelson, D. Z. Goldstein, J. Iraqi, and D. L. Share. 1993. “Effects of Listening to Story Reading on
    Aspects of Literacy Acquisition in a Diglossic Situation.” Reading Research Quarterly 28: 71‒79.
Florida Center for Reading Research (FCRR). 2009. Curriculum Review Team, Frequently Asked
    Questions about Reading Instruction (Tallahassee, FL: FCRR, 2009).
Garcia, M. E. 2010. Rethinking Bilingual Education in Peru: Intercultural Politics, State Policy and
    Indigenous Rights, International Journal of Bilingual Education and Bilingualism 7.5: 348–67.
Gebre, T.Z., 2015. Essays on the Effects of Education Policy Reforms in Developing Countries (Doctoral
    dissertation, University of Notre Dame).
Gebre, Y., 2010. Cultural contact and change in naming practices among the Aari of southwest Ethi-
    opia. Journal of African Cultural Studies, 22(2), pp.183–194.
Glewwe, P., Qihui Chen, and Bhagyashree Katare. 2012. “What Determines Learning Among Kinh and
    Ethnic Minority Students in Vietnam: An Analysis of Round 2 Young Lives Data.” Young Lives
    Working Paper No. 80. Oxford, UK.
Goldenberg, C. (2008). Teaching English language learners: What the research does—and does not—
    say. American Educator, Summer, 8–44.
Goldenberg, C. (2012). Research on English learner instruction. In M. Calderon (Ed.), Breaking
    through: Effective instruction and assessment for reading English learners. Bloomington, IN:
    Solution Tree Press.
Government Secretariat. 1981. Overall review of the Hong Kong education system. Hong Kong:
    Government Printer.
Grin, F. 2003. Language Planning and Economics
Grin, F. 2004. On the costs of cultural diversity. In: Cultural Diversity versus Economic Solidarity.
    Brussels : [s.n.]. p. 189–202.
Grisay, A., John H.A.L de Jong, Evaline Gebhardt, Alla Berezner, and Beatrice Halleux-Monseur.
    2007 “Translation Equivalence Across PISA Countries.” Journal of Applied Measurement, 8(3),
    249–266.
Hamre, B. K. (2014). Teachers’ daily interactions with children: an essential ingredient in effective
    early childhood programs. Child Development Perspectives, 8(4), 223–230.
Harris, P. 2011. Language in schools in Namibia: The missing link in educational achievement? Mono-
    graph 1. Windhoek: The Urban Trust of Namibia.
Heugh, K. 2011. Cost implications of the provision of mother-tongue and strong bilingual models
    of education in Africa. In A. Ouane and C. Glanz (Eds.), Optimising learning, education and
    publishing in Africa: The language factor. UNESCO Institute for Lifelong Learning.
Heugh, K., Benson, C., Bogale, B., and Yohannes, M. A. G. (2007). Final report: Study on medium of
    instruction in primary schools in Ethiopia. Addis Ababa: Ministry of Education.


REFERENCES                                                                                               91
     HK Census 2016. Accessed at: https://www.bycensus2016.gov.hk/en/bc-mt.html
     HK Census 2019. Accessed at: https://www.censtatd.gov.hk/hkstat/sub/so20.jsp
     Hovens, M. (2002). Bilingual education in West Africa: Does it work? International Journal of Bilin-
          gual Education and Bilingualism, 5(5), 249–266.
     Huisman, J., Smits, J., and K. Kruijff. 2008. Home Language and Education in the Developing World.
          Background paper prepared for the Education for All Global Monitoring Report 2009 (Nijmegen,
          Netherlands: Radboud University Nijmegen).
     Hynsjö, D. & Damon, A. 2015. Bilingual Education in Latin America: Does Quechua-Medium Educa-
          tion Improve Peruvian Indigenous Children’s Academic Achievement? Young Lives, Oxford, UK.
     Kan, V. 2011. Can Language Policy Outcome be Predicted – A Matrix Approach to Language Policy
          Analysis Using Hong Kong from 1997 to 2010 as a Test Case. Thesis Submitted to the Hong Kong
          Institute of Education.
     Kan, V., K. C. Lai, T. A. Kirkpatrick, and A. Law. 2011. Fine-tuning Hong Kong's medium of instruction
          policy. Hong Kong: The Hong Kong Institute of Education.
     Kerwin, J.T., and R.L. Thornton. 2020. Making the Grade: The Sensitivity of Education Program Effec-
          tiveness to Input Choices and Outcome Measures. The Review of Economics and Statistics, 1–45.
          http://doi.org/10.1162/ rest a 00911
     Kone, A’ame. 2010. Politics of Language: The Struggle for Power in Schools in Mali and Burkina Faso.
          International Education, Vol. 39 Issue (2).
     Kotz, J., B. Fleisch, and S. Taylor. 2018. Alternative forms of early grade instructional coaching:
          Emerging evidence from field experiments in South Africa. International Journal of Education
          Development. Volume 66, April 2019, Pages 203–213
     Kremer, Michael. 1993. The O-Ring Theory of Economic Development. The Quarterly Journal of
          Economics. Vol. 108, No.3. pp. 551–575. UK. Oxford University Press.
     Kudó, I. 2004. La Educación Indígena en el Perú. Cuando la Oportunidad Habla Una Sola Lengua, in
          Donald R. Winkler and Santiago Cueto (eds.) Etnicidad, Raza, Género y Educación en América
          Latina, Santiago: PREAL
     Kulbir Singh, Ruchi; Abu Alghaib, Ola. 2019. Every Learner Matters: Unpacking the Learning Crisis
          for Children With Disabilities (English). Washington, D.C.: World Bank Group.
     Lai, Mee-Ling 1999. Hong Kong: Language and Education in a Post-colonial Era, Language, Culture
          and Curriculum, 12:3, 191–195.
     Laitin, D.D., R. Ramachandran, and S. L. Walter. 2015. Language of instruction and student learning:
          Evidence from an experimental program in Cameroon. Paper presented at the 15th SAET Confer-
          ence on Trends in Economics. Cambridge, UK, 27–31 July 2015.
     Laitin, D.D., Ramachandran, R., and S.L. Walter. 2019. The Legacy of Colonial Language Policies
          and Their Impact on Student Learning: Evidence from an Experimental Program in Cameroon.
          Economic Development and Cultural Change, 68(1), 239–272. http://doi.org/10.1086/700617
     Lang, K. (1986) A language theory of discrimination. Quarterly Journal of Economics 101, 363–82.
     Lopez, L. E. and Küper, W. 2000. Intercultural Bilingual Education in Latin America: Balance and
          Perspectives.
     Maamouri, Mohamed. 1998. “Language Education and Human Development: Arabic Diglossia and
          its Impact on the Quality of Education in the Arab Region.” Mediterranean Development Forum,
          World Bank, Washington, DC.
     Male, C. and Q. Wodon. 2017. Disability gaps in educational attainment and literacy (English). The
          price of exclusion: disability and education. Washington, D.C.: World Bank Group.
     Malone, Denise, and Susan Malone. 2011. Teacher Education for Mother Tongue-Based Education
          Programs. SIL International.
     Martin, P. 2005. ‘Safe’ language practices in two rural schools in Malaysia. In Lin and Martin (eds.),
          Decolonisation, Globalisation. Clevedon: Multilingual Matters. pp. 74–97.



92                                                     LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
Martin, Michael, Ina V.S. Mullis, and Pierre Foy. 2008. “Students’ Backgrounds and Attitudes Towards
    Science,” TIMSS 2007 International Science Report: Findings from IEA’s Trends in International
    Mathematics and Science Study at the Fourth and Eighth Grades (Chapter 4). Chestnut Hill, MA:
    TIMSS & PIRLS International Study Center, Boston College.
McBrien, J. 2005. “Educational Needs and Barriers for Refugee Students in the United States: A
    Review of the Literature.” Review of Educational Research 75 (3): 329–64.
Mendenhall, M., L. Bartlett, and A. Ghaffar-Kucher. 2017. “‘If You Need Help, They Are Always There
    for Us’: Education for Refugees in an International High School in NYC.” The Urban Review 49:
    1–25.
Migué, J.-L. (1970) Le nationalisme, l’unité nationale et la théorie économique de l’information [Nation-
    alism, national unity and the economic theory of information]. Revue Canadienne d’Économique,
    3, 183–98.
Milligan, L., J. Clegg, and L. Tikly. 2016. Exploring the potential for language supportive learning in
    English medium instruction: a Rwandan case study, Comparative Education, 52(3), pp. 328–342.
Milner, James, and Gil Loescher. 2011. Responding to Protracted Refugee Situations: Lessons from
    a Decade of Discussion. Oxford, UK: University of Oxford, Oxford Department of International
    Development, Refugee Studies Centre.
Ministerio de Education de Peru (MINEDU). 2018. Política de educación intercultural y educación
    intercultural bilingüe.
Mizunoya, S., S. Mitra and I. Yamasaki, ‘Towards Inclusive Education: The impact of disability on
    school attendance in developing countries’, Innocenti Working Paper No. 2016–03, UNICEF
    Office of Research, Florence, May 2016, 24–25.
Montoya Rojas, R. 2001. Limites e Possibilidades da Educação Bilíngüe Intercultural no Peru. Revista
    Letras 56: 13–31.
Muijs, D., Kyriakides, L., van der Werf, G., Creemers, B., Timerley, H., and Earl, L. (2014). State of the
    art – teacher effectiveness and professional learning. School Effectiveness and School Improve-
    ment, 25 (2), 231–256
Mullis, I. V. S., Martin, M. O., Foy, P., & Drucker, K. T. (2012). The PIRLS 2011 international results in
    reading. Chestnut Hill, MA: Trends in International Mathematics and Science Study (TIMSS)
    and Progress in International Reading Literacy Study (PIRLS), Boston College.
Mullis, I. V. S., M. O. Martin, P. Foy, and M. Hooper. 2017. PIRLS 2016 International Results in Reading.
Muthwii, M.J., 2004. Language of instruction: a qualitative analysis of the perceptions of parents,
    pupils and teachers among the Kalenjin in Kenya. Language, Culture and Curriculum 17 (1), 15–32.
New York State Education Department. 2020. 2018–2019 English Language Learner Demographic
    Report
Ng, Dorothy F. P. 2007. Medium and learning in Chinese and English in Hong Kong classrooms.
    Language Policy, 6, 163–183
Patrinos, H. A., & Velez, E. 2009. Costs and benefits of bilingual education in Guatemala: A partial
    analysis. International Journal of Educational Development, 29(6), 594–598.
Pflepsen, Alison, with Amy Pallangyo. Handbook on Language of Instruction Issues in Reading
    Programs. A Global Reading Network Resource. Prepared by University Research Co., LLC. (URC)
    under the Reading within REACH initiative for USAID’s Building Evidence and Supporting Inno-
    vation to Improve Primary Grade Assistance for the Office of Education (E3/ED). Chevy Chase,
    MD: URC, 2019.
Pinnock, Helen. 2009. Language and education: the missing link. Save the Children.
Piper, B., Schroeder, L., & Trudell, B. (2015). Oral reading fluency and comprehension in Kenya:
    Reading acquisition in a multilingual environment. Journal of Research in Reading.
Piper, B., S. S. Zuilkowski, D. Kwayumba, and C. Strigel. 2016. Does technology improve reading
    outcomes? Comparing the effectiveness and cost-effectiveness of ICT interventions for early



SECTION 4: LANGUAGE OF INSTRUCTION POLICY SHOULD BE MULTIFACETED                                             93
         grade reading in Kenya. International Journal of Education Development. Volume 49, July 2016,
         pp. 204–214.
     Piper B., S. S. Zuilkowski, M. Dubeck, E. Jepkemei, S. J. King. 2018. Identifying the essential ingredi-
         ents to literacy and numeracy improvement: Teacher professional development and coaching,
         student textbooks, and structured teachers’ guides. World Development, Volume 106, 2018, pp.
         324–336
     Poon, Anita Y. K. and Connie M. Y. Lau. 2016. Fine-tuning medium-of instruction policy in Hong Kong:
         Acquisition of language and content-based subject knowledge. Journal of Pan-Pacific Association
         of Applied Linguistics, 20(1), 135–155.
     Pritchett, Lant. 2013. The Rebirth of Education. Schooling Ain’t Learning. Center for Global Develop-
         ment
     Rackham, N. (2001). The coaching controversy. Purcellville, VA: Huthwaite, Inc.
     Ramachandran, R., 2012. Language use in education and primary schooling attainment: Evidence
         from a natural experiment in Ethiopia. Working Papers 2012/34, Institut d’Economia de Barce-
         lona (IEB).
     Ramirez, J., & Merino, B.J. (1990). ‘Classroom talk in English immersion, early-exit and late-exit tran-
         sitional bilingual education programs.’ In Jacobson and C. Faltis (Eds.), Language Distribution
         Issues in Bilingual Schooling (pp. 61–103) Clevedon, England: Multilingual Matters.
     Raynauld, A. and Marion, P. (1972) Une analyse économique de la disparité inter-ethnique des revenus
         [An economic analysis of the interethnic earnings differentials]. Revue Économique 23, 1–19.
     Rodríguez Lozano, E. 2012. Barreras Lingüísticas en la Educación? La Influencia de la Lengua Materna
         en la Deserción Escolar, Economia 35.69: 83–151
     RTI International. 2014. Assessing Early Grade Reading Skills in Africa. Washington, DC: USAID.
     RTI International. 2015. Planning for Language Use in Education: Best Practices and Practical Steps to
         improve Learning Outcomes. Washington, DC: U.S. Agency for International Development
     Sampa, F. (2005). Zambia’s Primary Reading Programme (PRP): Improving access and quality educa-
         tion in basic schools. African Experiences Country Case Studies 4. Paper commissioned by ADEA
         for its Biennial Meeting, December 2003.
     Schroeder, L. 2010. The Bantu Orthography Manual (Rev. ed.). SIL E-Books 9.
     Setati, M., Adler, J., Reed, Y., Bapoo, A., 2002. Incomplete journeys: code-switching and other language
         practices in Mathematics, Science and English language classrooms in South Africa. Language
         and Education 16, 128–149.
     Schroeder, L., M. Mercado, and B. Trudell. 2020 (forthcoming). Research in multilingual learning in
         Africa: Assessing the effectiveness of multilingual education programming. In Erling, Clegg, and
         Reilly (eds.), Multilingual Learning in Sub-Saharan Africa: Critical Insights and Practical Appli-
         cations. Language and Content Integrated Teaching & Plurilingual Education Series. London:
         Routledge.
     Seid, Y. 2019. The impact of learning first in mother tongue: evidence from a natural experiment in
         Ethiopia. Applied Economics, 51(6), 577–593.
     Shin, S., 2005. Developing in Two Languages: Korean Children in America. Multilingual Matters,
         Clevedon.
     Skutnabb-Kangas, T., & Heugh, K. (Eds.). (2012). Multilingual education and sustainable diversity
         work from periphery to centre. London, UK: Routledge
     Spolsky, B. 2009. Language Management. Cambridge: Cambridge University Press
     Spratt, J., King, S., and J. Bulat. 2013. Independent Evaluation of the Effectiveness of the Institute
         pour l’Education Populaire’s “Read-Learn-Lead” (RLL) Program in Mali. Endline Report. https://
         shared.rti.org/content/independent-evaluation-effectiveness-institut-pour-l%E2%80%99educa-
         tion-populaire%E2%80%99s-%E2%80%9Cread-learn-lead%E2%80%9D
     Taha-Thomure, H. 2008. “The Status of Arabic Language Today.” Journal of Education, Business and
         Society: Contemporary Middle Eastern Issues. 1(3): 186‒192.


94                                                      LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
Taha-Thomure, H. 2019. “Arabic Language Education in the UAE.” In K. Gallagher (ed.) Education in
    the UAE: Innovation and transformation. Springer.
Taylor, S., and M. von Fintel. 2016. Estimating the impacts of language of instruction in South African
    primary schools: a Fixed Effects approach. Economics of Education Review, 50, 75–89. http://doi.
    org/10.1016/j.econedurev. 2016.01.003
Thomas, W., & Collier, V. (2002). A national study of school effectiveness for language minority
    students’ long-term academic achievement. Berkeley, CA: Center for Research on Education,
    Diversity, and Excellence (CREDE).
Trudell, B. 2005. Language choice, education and community identity. International Journal of Educa-
    tional Development 25(3):237–251.
Trudell, Barbara. 2007. Local community perspectives and language of education in sub-Saharan
    African communities. International Journal of Educational Development 27 (2007) 552–563.
Trudell, B., and B. Piper. 2014. Whatever the law says: Language policy implementation and early-
    grade literacy achievement in Kenya. Current Issues in Language Planning 15(1):4 –21.
Trudell, B.; C. Young; and S. Nyaga. 2015. Language, education and development: Implications of
    language choice for learning. In S. McGrath and Qing Gu (eds.), Routledge Handbook of Interna-
    tional Education and Development, pp. 133–149.
Trudell, B., 2016. The impact of language policy and practice on children’s learning: Evidence from
    Eastern and Southern Africa. UNICEF, Paris.
Trudell, B. and C. Young. 2016. Good Answers to Tough Questions in Mother Tongue-Based Multilin-
    gual Education. SIL International.
Trudell, B. 2020. Language in Education Policy [title subject to change]. Unpublished manuscript.
Tsang, Wing, K. 2008. Evaluation research on the implementation of the medium of instruction guid-
    ance for secondary schools. Hong Kong Institute of Educational Research Newsletter, 24, 1E7.
Tsui, Amy B. M., Mark S.K. Shum, Chi Kin Wong, Shek Kam Tse and Wing Wah Ki. 2011. Which
    Agenda? Medium of Instruction Policy in Post-1997 Hong Kong. University of Hong Kong.
UNESCO. 2006. Education for people and planet: creating sustainable futures for all, Global education
    monitoring report, 2016.
UNESCO. 2016. If you don’t understand, how can you learn? Global Education Monitoring Report
    Policy Paper 24. http://unesco.org/ark/48223/pf0000243713
UNESCO. 2019. Building Knowledge Societies in the Arab Region: Arabic Language as a Gateway to
    Knowledge. UNESCO, Paris.
USAID. 2013. Baseline Assessment of Pre-Service Teacher Education Programs at Colleges of Teacher
    Education in Ethiopia.
USAID 2015. Performance Evaluation of the USAID/Malawi Early Grade Reading Activity (EGRA)
USAID. 2017. Language Mapping Study in Mozambique – Final Report.
USAID. 2018. Language mapping study: Analysis report.
Vawda, A.Y., & Patrinos, H. A. (1999). Producing educational materials in local languages: Costs from
    Guatemala and Senegal. International Journal of Educational Development, 4(5), 287–299.
Versteegh, Kees. The Arabic Language. New York: Columbia University Press.
Vygotsky, L. 1986. Thought and language. Trans. Alex Kozulin. Cambridge, MA: MIT Press.
Walter, S. L., & Chuo, K. G. (2012). The Kom experimental mother tongue education pilot project
    report for 2012. Cameroon: SIL International.
Walter, Steve and Carol Benson. 2012. Language policy and medium of instruction in formal educa-
    tion. In Spolsky (Ed.) The Cambridge Handbook of Language Policy, pp.278–300. Cambridge.
    Cambridge University Press.
Waters, T. and K. Leblanc. 2005. “Refugees and Education: Mass Public Schooling Without a Nation-
    State.” Comparative Education Review 49 (2): 129–47.
Woldemariam, Hirut. 2014. Writing both difference and similarity: Towards a more complete phonemic



REFERENCES                                                                                                95
          orthography for Wolaitta, Gamo, Gofa, and Dawuro. Journal of languages and culture. Vol. 5 (3)
          pp. 44–53.
     Woldetsadik, Girma; Raysarkar, Chandrani. 2017. Textbook Provision for All in Ethiopia: Lessons
          Learned from the General Education Quality Improvement Project. World Bank, Washington,
          DC.
     Wong, Anny. 2017. “Insights from East Asia’s High Performing Education Systems: Leadership,
          Pragmatism, and Continuous Improvement.” Background paper for East Asia and Pacific flag-
          ship report, Growing Smarter: Learning and Equitable Development in East Asia and Pacific,
          World Bank, Washington, DC.
     World Bank. 2002. Peru – Rural Education and Teacher Development Project (English). Wash-
          ington, DC: World Bank.
     World Bank. 2005. In their Own Language: Education for All. Education Notes. June 2005.
     World Bank. 2018a. The Human Capital Project. World Bank, Washington, DC.
     World Bank. 2018b. World Development Report 2018: Learning to Realize Education's Promise.
          Washington, DC: World Bank.
     World Bank. 2019a. World Development Report 2019: The Changing Nature of Work. World Devel-
          opment Report; Washington, DC: World Bank.
     World Bank. 2019b. Ending Learning Poverty: What Will It Take? World Bank, Washington, DC.
     World Bank. 2020. The COVID-19 Pandemic: Shocks to Education and Policy Responses. Wash-
          ington, DC.
     Yip, Din Y., Wing Kwong Tsang, and Sin Pui Cheung 2003. Evaluation of the Effects of Medium of
          Instruction on the Science Learning of Hong Kong Secondary Students: Performance on the
          Science Achievement Test, Bilingual Research Journal, 27:2, 295–331.
     Zeng, Wei 2006. Medium of Instruction in Secondary Education in Post-Colonial Hong Kong: Why
          Chinese? Why English? Working Papers in Educational Linguistics. Vol 22, No 1.




96                                                     LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
ENDNOTES




1B
     ased on SDI data estimates.
2B
     ased on SDI data estimates.
3
     Alidou et al. 2006.
4
     Ethnologue, 23rd edition, published 2020 (https://www.ethnologue.com/about/country-info)
5
     World Bank. 2019. “Ending Learning Poverty: What Will It Take?” Washington DC: World Bank. p.6.
6
     UNESCO. 2018. Mother Tongue-based Multilingual Education [Infographic]. Bangkok, Thailand: UNESCO Asia and Pacific Regional Bureau.
7
 It is important to quickly improve the estimates of the effect of LoI on these key indicators of school effectiveness and human capital devel-
opment, as a step toward improving LoI policies, their implementation, and learning outcomes.
8
 The Ethnologue database is the most authoritative source on the number of languages and where they are spoken. It identifies 7,106 lan-
guages currently spoken worldwide.
9
     Evans and Mendez Acosta 2020.
10
     World Bank 2018b.
11
     Pritchett 2013.
12
     World Bank 2019b.
13
     World Bank 2019b.
14
     Collier and Thomas 2017.
15
     Ibid.
16
     Pinnock 2009.
17
     Dobbie & Fryer 2013; Hamre 2014; Muijs et al. 2014.
18
     Malone and Malone 2011.
19
     Alidou and Brock-Utne 2011.
20
  The relationship between languages and writing systems is complex and the subject of voluminous linguistic research. The intuitive conclu-
sion is that writing systems with fewer symbols make it easier to become literate in a given language. That is true up to a point, but the level
of difficulty is also dictated by the specific sound-symbol relationships involved. Since few languages have more than about four dozen basic
sound components, it has been conjectured that the “perfect alphabet” would have one symbol for each sound and therefore roughly 48 let-
ters (Moats 2020; in press). English uses 26 letters to represent about 44 sounds and needs over 100 key letter combinations to accomplish
this—making English a relatively difficult language in which to become literate.
21
     Kone 2010.
22
      Trudell 2020.
23
     Grin 2003.
24
      Kone 2010.
25
      Grin 2003.
26
      Grin 2003.
27
     Trudell 2007.
28
      Adegbija 1994.
29
      Muthwii 2004.
30
      Alidou and Brock-Utne 2011.
31
     Grin 2004.
32
     Grin 2003.
33
     World Bank 2005.




REFERENCES                                                                                                                                         97
     34
          Patrinos and Velez 2009; Vawda and Patrinos 1999.
     35
          Grin 2004.
     36
          Heugh 2011.
     37
          Azevedo 2020.
     38
          World Bank 2020.
     39
          Alidou et al. 2006; Baker 2001.
     40
          Piper, Schroeder, and Trudell 2015.
     41
          Ball 2011; Goldenberg 2008; Brown 2011; Diarra 2003; Harris 2011; Walter & Chuo 2012.
     42
          UNESCO 2019.
     43
          Feitelson, Goldstein, Iraqi, and Share 1993; Abu-Rabia 2000.
     44
          Taha-Thomure 2019.
     45
          Taha-Thomure 2008.
     46
          Taha-Thomure 2019
     47
          https://digitallibrary.io/
     48
          GRN (https://www.globalreadingnetwork.net/)
     49
          Global Book Alliance (https://www.globalbookalliance.org/)
     50
        Resources are available at: https://shared.rti.org/content/education-data-decision-making-eddata-ii-national-early-grade-litera-
     cy-and-numeracy-survey
     51
          https://www.sil.org
     52
          https://www.ethnologue.com/
     53
          https://bloomlibrary.org/landing
     54
          https://software.sil.org/primerpro/
     55
          https://palnetwork.org/
     56
          Assessments and related tools are available at https://palnetwork.org/tools/
     57
          https://www.africanstorybook.org/
     58
          https://storyweaver.org.in/
     59
          Choi 2010; Evans and Morrison 2017.
     60
          USAID 2017.
     61
          USAID 2018.
     62
          Fazio & Lyster 1998; August & Shanahan 2006.
     63
          Heugh 2011; Ball 2011; Alidou and Brock-Utne 2011; Walter & Benson 2012; Cummins 2009.
     64
          Denton, Fletcher, Taylor, Barth, and Vaughn 2014.
     65
          Sampa 2005; Heugh, Benson, Bogale, Yohannes 2007; Walter and Chuo 2012; Skutnabb-Kansas & Heugh 2012.
     66
          Boateng 2019.
     67
          Clegg 2005; Brock-Utne et al. 2011.
     68
          Martin, Mullis, Foy 2008; Alidou and Brock-Utne 2011; Ramirez & Merino 1990.
     69
          World Bank 2019b.
     70
          Martin, Mullis, Foy 2008.
     71
          Pflepsen & Pallangyo 2019.
     72
          Cummins 2009; Cummins 2012; Thomas & Collier 2002.
     73
          Vygotsky 1986.
     74
          Trudell and Young 2016.
     75
          Pflepsen & Pallangyo 2019.
     76
          Goldenberg 2012.
     77
          Clark 1979; Cummins 1979.
     78
          Benson 2008.
     79
          Spolsky 2009.
     80
          Asia Foundation – Let’s Read (https://www.letsreadasia.org)
     81
          Pflepsen & Pallangyo 2019.
     82
          Pflepsen and Pallangyo 2019.
     83
          Beteille and Evans 2019.
     84
          Clegg 2005; EdQual 2010.
     85
          Milligan et al. 2016; Clegg and Simpson 2016.
     86
          RTI International 2015.
     87
          USAID 2013.
     88
          Pflepsen & Pallangyo 2019.
     89
          Piper et al. 2018.




98                                                                          LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
90
      USAID 2015.
91
      Eble et al. 2019.
92
      Rackham 2001.
93
      Benson 2000; Benson 2003.
94
      Dryden-Peterson 2015.
95
      Trudell and Young 2016.
96
      New York State Education Department 2020.
97
      Crawford et al. 2015; Devictor and Do 2016; Milner and Loescher 2011.
98
      Dryden-Peterson 2006, 2016b; McBrien 2005; Mendenhall, Bartlett, and Ghaffar-Kucher 2017; Waters and Leblanc 2005.
99
      UNESCO 2006.
100
       UNICEF 2016.
101
      Male and Wodon 2017.
102
       Kulbir Singh, R. and O. Abu Alghaib 2019.
103
       Gebre 2010; Demilew & Sekeroglu 2019.
104
       CSA 2007.
105
       Derash 2013; Gebre 2015; Bashir et al. 2018.
106
       Cohen 2007.
107
       Gebre 2015.
108
       Bashir et al. 2019.
109
       RTI International 2014.
110
      Goldenberg 2012.
111
      Clark 1979; Cummins 1979; Benson 2008.
112
      Pflepsen & Pallangyo 2019.
113
      Glewwe, P., et al. 2012.
114
      Ramachandran, 2012.
115
      Ibid.
116
      Collier, V., and Wayne P. Thomas, 2017.
117
      Ethnologue, 2020. Author’s analysis.
118
      Grisay, A. et al. 2007 p.250
119
      https://www.sil-lead.org/ugandashrp.
120
       https://www.sil-lead.org/ghana-learning.
121
      PrimerPro (https://software.sil.org/primerpro/)
122
       SynPhony Software (https://www.sil.org/about/news/synphony-tool-creating-sequenced-learning-resources-0)
123
       Boakye-Yiadom, F, M. Matthews, E. Miksic, and B. Trudell. 2018.
124
       World Atlas of Languages Structures (https://wals.info/)
125
       Schroeder 2010.
126
       SIL Lead 2017 (https://www.sil-lead.org/sil-lead-mtb-mle-resources)
127
      Trudell 2020.
128
       African Linguistic Network https://alnresources.wordpress.com/
  Survey Solutions (https://mysurvey.solutions/); GEMS (Geo-Enabling Initiative for Monitoring and Supervision); SurveyCTO (https://
129


www.surveycto.com/)
130
       Viamo (https://viamo.io/)
131
      E.g. Pflepsen & Pallangyo 2019.
132
       Kotz, Fleisch, and Taylor, 2018; Piper et al., 2016; Akmal, Musa, and Silverman, 2020.
  These regions and cities are Addis Ababa city, Amhara, Afar, Benishangul-Gumuz, Dire Dawa city, Gambella, Harari, Oromia, Southern
133


Nations, Nationalities and People (SNNP), Somali and Tigray.
134
       Gebre, 2010; Demilew and Sekeroglu, 2019.
135
       CSA 2007.
136
       This exam is administered in the medium of instruction used at the grade 8 level in the region.
137
      Grade 10 and 12 examinations are administered in English. Post-secondary education is also conducted in English.
138
   Before western education was introduced, the traditional system of education (religious education) in the nineteenth century was in the
Ge’ez and Amharic languages. When modern schools were opened in 1908 French and Arabic were used a medium of instruction. As modern
schooling expanded in the country, Amharic and English became the medium of instruction. The first official Ethiopian educational language
policy emerged in 1944 in the form of a directive that Amharic was to be the general language of instruction with English being used in upper
primary and higher levels (Getachew and Derib 2006).
139
       The use of Amharic as the working language of the federal government is a frequently debated issue.
140
   Derash, 2013, Gebre, 2015, Bashir et al., 2018, Heugh et al., 2007, identified 24 languages, while Gebre, 2015 identified an additional six
languages that were included as a medium of instruction in SNNPR.




REFERENCES                                                                                                                                       99
      141
            Woldemariam and Lanza 2014.
      142
             Getachew and Deriba 2006.
      143
             Gebre 2015.
      144
             Bashir et al. 2019.
      145
             Woldetsadik & Raysarkar 2017.
      146
             Woldetsadik & Raysarkar 2017.
      147
             Heugh et al. 2007, Cohen 2005.
      148
             Azubuike et al. 2017.
      149
             Heugh et al. 2007.
      150
          Non-readers are students, tested at the end of grade 2 or beginning of grade 3, who are unable to read a single word in the language of
      instruction.
        World Bank 2018, Future Drivers of Growth in Rwanda. http://documents.worldbank.org/curated/en/680671541618477651/pdf/131875-
      151


      V2-WP-PUBLIC-Disclosed-11-9-2018.pdf
      152
             UNESCO 2020. http://uis.unesco.org/en/country/rw
      153
             World Bank 2011, Rwanda Education Country Status Report.
      154
            http://documents.worldbank.org/curated/en/677741468107072073/pdf/579260SR0P11151353788B01PUBLIC10Web.pdf
      155
             Republic of Rwanda, 2008.
      156
            https://planipolis.iiep.unesco.org/sites/planipolis/files/ressources/rwanda_essp_2008-2012.pdf
      157
            DFID 2018, Learning for All Annual Review.
      158
            https://devtracker.dfid.gov.uk/projects/GB-1-204463/documents
      159
             BLF (2018), Baseline Report (unpublished).
      160
         MINEDUC/RES, Rwanda Education Statistics. http://197.243.16.104/~mineduc/newweb/fileadmin/user_upload/pdf_files/2018_Rwan-
      da_Education_Statistics.pdf
      161
            MINEDUC/REB, 2015, Competence Based Curriculum.
      162
            http://old.reb.rw/fileadmin/competence_based_curriculum/syllabi/CURRICULUM_FRAMEWORK_FINAL_PRINTED.compressed.pdf
      163
             HK Census 2019.
      164
             Poon and Lau 2016.
      165
             English is widely used in government, academic circles, tertiary institutions, and courts of law.
      166
          Written Chinese can take two forms—either Simplified or Traditional characters. The digital era has rendered the basic labor-saving dis-
      tinction between the two (number of strokes per character written) slightly less important than it used to be. Mandarin Chinese tends to
      utilize Simplified characters. Cantonese Chinese is a dialect—also described, correctly, as a Sinitic language—without its own formal writing
      scheme. However, it does informally and straightforwardly borrow either written script, with the addition of a few ad hoc, non-standard or-
      thographic innovations. The history of Hong Kong as a British colony during the 1950s and 1960s (the period when Mainland China shifted
      to Simplified characters), has resulted in the widespread retention of Traditional Chinese (likewise retained in Taiwan). The “biliterate and
      trilingual” policy does not specify whether written Chinese pertains to Traditional or Simplified characters.
      167
             EDB 1997a.
      168
             HK Census 2016.
      169
         Wong, 2017; An important point to note is that the British colonization of Hong Kong was not, in its essence, a proselytizing mission, and
      local Chinese students generally did not attend religious schools where they were required to learn English.
      170
             Barro-Lee 2011.
      171
            Kirkpatrick et al. 2011.
      172
             Kan 2011.
      173
             Wong 2017.
      174
            The 1950s and 60s saw the doubling of primary school enrollment.
      175
            Government Secretariat, 1981; Wang and Kirkpatrick, 2015.
      176
             Education Commission, 1990.
      177
             EDB 1997b.
      178
             Evans 2013.
      179
             Choi 2010; Evans and Morrison 2017.
      180
             Poon and Lau 2016.
      181
         Students’ scores on average had to be no less than 85% of the 3-year average of Secondary 1 students grouped in Medium of Instruction
      Grouping Assessment (MIGA) Groups I and III. Group I: able to learn effectively in either Chinese or English; Group II: able to learn more
      effectively in Chinese; Group III: able to learn better in Chinese but may also learn effectively in English. The groupings are based on students’
      performance in their primary 5 and 6 internal school assessments.
      182
             After widespread opposition from the public.
      183
             Lai 1999.
      184
             Zeng 2007.
      185
             Wang 2019.




100                                                                              LOUD AND CLEAR: Effective Language of Instruction Policies for Learning
186
      Evans and Morrison 2017.
187
      Evans and Morrison 2017.
188
      EDB 2012.
189
      Education Commission 1990; Evans and Morrison 2017.
190
      Evans and Morrison 2017.
191
      Tsui et al. 1999.
192
      Tsui et al. 1999.
193
      The survey response rate was 50%
194
      Yip, Tsang, and Cheung 2003; EDB 2002.
195
      Ng 2007.
196
      Tsang 2008.
197
      Evans and Morrison 2017.
198
   Algeria, Bahrain, Chad, Comoros, Djibouti, Arab Republic of Egypt, Eritrea, Iraq, Jordan, Kuwait Lebanon, Libya, Mauritania, Morocco,
Oman, Qatar, Saudi Arabia, Somalia, Sudan, Syrian Arab Republic, Tanzania, Tunisia, United Arab Emirates, West Bank and Gaza, and the
Republic of Yemen.
199
      Versteegh 1997.
200
      Al-Huri 2015.
201
      UNESCO 2019.
202
      Feitelson, Goldstein, Iraqi, and Share 1993; Abu-Rabia 2000.
203
      Taha-Thomure 2019.
204
      Taha-Thomure 2008.
205
      Taha-Thomure 2019.
206
      MINEDU 2018.
207
      MINEDU 2018.
208
      DIGEIBIR 2013; Hynsjö & Damon 2015.
209
      Garcia 2010.
210
      Buhman & Trudell 2008.
211
      World Bank 2002.
212
      Aprendo en Casa.
213
      Andina 2019.
214
      Montoya Rojas 2001; Garcia 2010.
215
      Kudó 2004.
216
      Rodriguez Lozano 2012.
217
      Buhman & Trudell 2008.
218
      Lopez & Kuper 2000.
219
      Hynsjö & Damon 2015.
220
      ASER 2018 Feitelson, Goldstein, Iraqi, and Share 1993.
221
      80 percent of primary schools are funded and managed by central and state governments and local bodies
 Based on School Education in India, Flash Statistics published by NUEPA in 2016. http://dise.in/Downloads/Publications/Docu-
222


ments/U-DISE-SchoolEducationInIndia-2015-16.pdf
223
      Jhingran, D. (2005). Language disadvantage: The learning challenge in primary education. Delhi. APH Publishing.




REFERENCES                                                                                                                                 101
