The World Bank publishes learning poverty by country in its World Development Indicators (series SE.LPV.PRIM). The figures below are the most recent value for each country in the series as last updated on 13 July 2026. The data years differ widely, because each estimate rests on whichever learning assessment the country last ran that meets the World Bank's standard.

The EGRA Toolkit (RTI International for USAID, second edition 2016) builds its assessment on the US National Reading Panel report of 2000 and later reviews. It states that five components are generally accepted as necessary to master reading: phonological awareness, phonics, vocabulary, fluency and comprehension. Higher-order skills build on lower-order ones.

Fluency and comprehension are measured separately but are connected in a simple way. A reader holds a limited amount in working memory. The EGRA Toolkit notes that oral reading fluency has been shown to be predictive of reading comprehension, which is why it sits at the centre of most early grade assessments.

Put together, the components form a pathway. Children do not finish one stage before starting the next, and good classrooms work on several at once, but a child stuck at an early stage cannot benefit from instruction aimed at a later one. Diagnosis means finding the earliest stage where the child is stuck.

Most reading research was done in English, an alphabetic script. Most South Asian children learn to read first in a different kind of script, and many learn two or three scripts by the end of primary school. The kind of script shapes what is hard to learn and in what order.

No single sequence has been agreed across South Asian languages, and the research base is far thinner than for English. The moves below follow from the properties of the script and from Nag's findings. Adapt them to the language and test them.

NIPUN Bharat lists the major aspects and components of early mathematics. Arithmetic is one of them, but the list starts before counting and runs to shapes, measurement and data. A numeracy programme that teaches only sums leaves most of it out.

NIPUN Bharat sets learning goals, which it calls lakshya, from Balvatika (the pre-school year before Grade 1) to Grade 3. The numeracy goals below are as printed in the 2021 mission document; the Ministry has since moved the end point to Grade 2. They describe what a child should do by the end of each year, so they double as a checklist for a classroom diagnostic.

The Annual Status of Education Report is a household survey of rural India run with Pratham. In each district a local organisation or institution conducts it, and children are tested at home, so children who are out of school or absent are counted. Reading and arithmetic tasks are given one-on-one to every sampled child aged 5 to 16, and the method has remained the same since 2006, which allows comparison over time.

ASER 2024 describes basic arithmetic as improving substantially in both government and private schools, reaching the highest level in over a decade. The gains were largest in government schools between 2022 and 2024. The school observations in the same report show FLN directives, training and materials reaching most of them.

National averages hide very different state systems. The table shows the share of government school children in Std III who could read a Std II level text, from the ASER 2024 state report cards (Table 5). The national findings name Himachal Pradesh, Uttarakhand, Kerala, Uttar Pradesh, Haryana, Odisha and Maharashtra as states with gains of more than 10 percentage points since 2022.

The Ministry of Education and NCERT ran the Foundational Learning Study in March and April 2022 and published national, state and district reports on 6 September 2022 (PIB, 30 December 2022). It tested Grade 3 children one-on-one in 20 languages and placed them on four levels of the Global Proficiency Framework: below partially meets, partially meets, meets and exceeds global minimum proficiency.

The Early Grade Reading Assessment (EGRA) and the Early Grade Mathematics Assessment (EGMA) are oral, one-on-one instruments developed by RTI International with USAID funding and adapted to each language. The EGRA Toolkit second edition dates from 2016; the EGMA toolkit by Platas, Ketterlin-Geller, Brombacher and Sitabkhan from 2014. Nepal used a nationally representative EGRA in 2014 to set initial reading benchmarks.

Chapter 2 of the National Education Policy 2020 is titled "Foundational Literacy and Numeracy: An Urgent & Necessary Prerequisite to Learning". Paragraph 2.1 estimates that over 5 crore children in elementary school had not attained foundational literacy and numeracy. Paragraph 2.2 sets the priority.

NEP 2020 replaces the old 10+2 description of schooling with a 5+3+3+4 curricular structure covering ages 3 to 18 (para 4.2). The first five years form the Foundational Stage: three years of anganwadi or pre-school plus Grades 1 and 2, covering ages 3 to 8, taught through flexible, multilevel, play and activity-based learning.

The Department of School Education and Literacy launched the National Initiative for Proficiency in Reading with Understanding and Numeracy (NIPUN Bharat) on 5 July 2021, under the centrally sponsored scheme Samagra Shiksha (PIB, July 2023). The NEP target year of 2025 became 2026-27 in the mission.

NIPUN Bharat expresses its targets as lakshya, learning goals for each year from Balvatika to Grade 3. The literacy goals below are as printed in the mission document. The Ministry now says the lakshya run from Balvatika up to Grade 2 (PIB, 29 July 2026); the revised list was not available to check, so the 2021 goals are shown.

NEP 2020 para 2.5 notes that many children fall behind within the first weeks of Grade 1 because they had no quality pre-school. NCERT's response is Vidya Pravesh, guidelines for a three-month play-based school preparation module (July 2021), described as an essential component of NIPUN Bharat. It runs for about 12 weeks at the start of Grade 1, four hours a day, and covers pre-literacy, pre-numeracy, cognitive and social skills.

The Right of Children to Free and Compulsory Education Act 2009 gives content to Article 21A. It does not use the phrase foundational literacy, but several sections speak directly to how the early grades are taught and assessed.

The Global Education Evidence Advisory Panel (GEEAP), convened by the FCDO, the World Bank, UNICEF and USAID, published its 2023 update on cost-effective approaches to improve learning in low- and middle-income countries. It searched over 13,000 additional studies and draws on over 550 evaluations in all. Two of its three Great Buys are about how teaching is organised.

The approach regroups children by what they can do, measured with a simple ASER-style oral test, for a dedicated part of the day. Each group works on the next skill it needs using activities, games and short texts, and children move up as soon as they are ready. J-PAL counts six randomised evaluations in seven Indian states and reports that the approach has reached over 60 million children in India and Africa.

Abhijit Banerjee, Shawn Cole, Esther Duflo and Leigh Linden evaluated two programmes in urban Indian schools ("Remedying education: evidence from two randomized experiments in India", Quarterly Journal of Economics 122(3), 2007). The first hired young women from the community, balsakhis, to teach children lagging behind in basic literacy and numeracy. The second gave children computer-assisted mathematics.

Pratham's model worked with volunteers. The question for policy was whether it could work inside government schools with government teachers. Banerjee, Banerji, Berry, Duflo, Kannan, Mukerji, Shotland and Walton report a series of trials in four states.

Muralidharan, Singh and Ganimian evaluated Mindspark, an adaptive computer programme used in after-school centres in Delhi, through a lottery that gave some applicants a voucher ("Disrupting education? Experimental evidence on technology-aided instruction in India", American Economic Review 109(4), 2019). The software tests each child and serves content at her actual level, which in their sample was often several grades below the one she was enrolled in.

Nepal ran a national version for reading. Its National Early Grade Reading Program, a five-year, USD 71 million programme from 2014/15 to 2019/20, aimed to improve reading in Nepali for children in Grades 1 to 3 of all community (government-funded) schools in all 77 districts. USAID supported it through a USD 53.8 million contract with RTI International that worked in 16 target districts (CAMRIS International, Early Grade Reading Program Performance Evaluation 2019, February 2020).

India's neighbours face the same problem at different levels. Pakistan runs its own ASER, led by Idara-e-Taleem-o-Aagahi (ITA), on the same household model. The table brings together what could be checked, with the year each figure refers to.

Many South Asian children start school in a language they do not speak at home: a tribal child in Odisha taught in Odia, a Bhojpuri-speaking child taught in standard Hindi, a child in Pakistan taught in Urdu or English when the home language is Pashto or Saraiki. Section 2 showed why this matters: a child can decode a language she does not understand, and comprehension, the point of reading, stays out of reach.

Policy, statute, the Constitution and the Supreme Court each say something different, and they do not carry the same force. A practitioner designing an FLN programme in a minority language needs to know which is which.

Karnataka had made the mother tongue or Kannada the compulsory medium of instruction in primary classes as a condition of recognising schools. On 6 May 2014 a Constitution Bench of the Supreme Court, in a judgment delivered by Justice A.K. Patnaik, struck down that requirement for unaided and minority schools.

A teacher in a Delhi resettlement colony, a Jharkhand village or a tea garden in Assam may have children speaking four or five home languages. Mother tongue instruction for each is impossible. NEP 2020 and the government's own description of Vidya Pravesh point to a multilingual approach instead.

A practical decision table for an NGO or district team. It assumes a government school system; for private schools, the 2014 judgment means the choice rests with the school and parents.

Matthew Kraft, David Blazar and Dylan Hogan meta-analysed 60 studies of teacher coaching with causal designs ("The effect of teacher coaching on instruction and achievement", Review of Educational Research 88(4), 2018, pages 547 to 588). Much of the evidence comes from literacy coaching for pre-kindergarten and elementary teachers in the United States.

FLN materials are often designed for one teacher with one grade. ASER 2024's school visits show that is not the typical classroom in rural government schools. The RTE Act's Schedule allows two teachers for a primary school of up to sixty children, so a small school is by law a multigrade school.

Every FLN method assumes that a teacher and the children are in the room, on task, for a set period each day. ASER 2024 records attendance on the day of its school visit, and the RTE Act sets rules on staffing and on taking teachers away from teaching.

NIPUN Bharat gives panchayats and School Management Committees roles: baseline analysis to identify struggling learners, ensuring enrolment, connecting volunteer parents to schools. The Jaunpur study is the caution. Giving communities information about learning did not, by itself, change what happened in schools.

A teacher or programme worker can place every child on a reading and an arithmetic ladder in about five minutes each, using the same logic as ASER. Make one card in the school language. Test children one at a time, away from the others, and mark the highest level each child reaches comfortably.

Illustrative example. A Grade 3 government school class in a Hindi-medium block has 40 children. The teacher runs the diagnostic in the first week of the session. Her results, which roughly follow the ASER 2024 national spread for Std III, are below.

The most common measurement mistake in FLN work is using a tool built for one question to answer another. Start from the decision the data will inform, then pick the instrument.

A decision table for programme design, built from the evidence in Section 7. Each row names a common field situation and the option with the strongest support. Treat it as a starting point for a theory of change, then test it.

Illustrative example, invented to show the steps; the figures describe no real block. A block in eastern Uttar Pradesh has 120 government primary schools and 6,000 children in Grades 3 to 5. A baseline using the ASER card finds that 30% of them can read a Grade 2 story.

Continuing the Illustrative block example. All figures are invented for the exercise and should be replaced with local prices. The point is the structure: cost every component, divide by children reached, and decide in advance what will be checked.

Some questions in foundational learning remain open, and practitioners will meet strong views on each. The table summarises the debate and what the evidence covered in this course can and cannot say.

Foundational learning numbers are quoted widely in proposals, news reports and government documents, and they are often misread. Each mistake below appears regularly.

A short reference list of the terms and acronyms that recur in Indian and South Asian foundational learning work, with the slide section where each is explained.