Why do so many Indian children reach Class 3 without being able to read? What the evidence says schools should do differently

India’s national deadline for every child to read fluently by Class 3 falls this academic year. ASER data, the NCF’s own literacy framework, and two decades of reading research all point to the same gap: reading is still being taught as something children absorb, not something they must be explicitly taught.

Meenakshi Iyer | Specialist in education equity, school improvement, and teacher professional development

The National Initiative for Proficiency in Reading with Understanding and Numeracy, launched by the Ministry of Education on 5 July 2021, set a hard target: every child in India achieves foundational literacy and numeracy by the end of Class 3, by 2026-27. That is this academic year. ASER 2024, the most recent nationwide reading assessment, found that 23.4 percent of Class 3 children in government schools could read a Class 2-level text, the highest this figure has been since ASER began in 2005, up from 16.3 percent in 2022 and 20.9 percent in 2018. Read the other way round, roughly three in four Class 3 children in government schools still cannot. The National Achievement Survey 2021 put average Grade 3 language competency nationally at 59 percent, with a gap of nearly 20 points between the states in the top quartile and those at the bottom. The deadline is real. So is the distance still left to cover.

What does learning to read actually require?

In 1997, the US Congress asked the National Institute of Child Health and Human Development to convene a panel reviewing the research on how children learn to read. The National Reading Panel’s 2000 report, drawing on evidence from more than 100,000 studies, identified five components with consistently strong effects on reading outcomes: phonemic awareness, phonics, fluency, vocabulary, and comprehension. Its central finding runs against a common classroom assumption: reading is not acquired the way spoken language is, by exposure and immersion alone. Most children need direct, systematic instruction in how sounds map onto letters before they can decode words reliably, and the Panel found this held for children from kindergarten through Class 6, with the benefit strongest for children already struggling. A teacher who reads aloud to a class and hopes fluency follows is relying on a theory the evidence doesn’t support. A teacher who teaches the code explicitly, in a planned sequence, is following what the research actually shows.

Does India’s own curriculum framework already say this?

It does, on paper. The National Curriculum Framework for the Foundational Stage, released by NCERT in 2022, lays out a Four-Block Approach to literacy instruction that names phonological awareness, letter recognition, sound-symbol association and word recognition as core components, alongside read-aloud, shared reading and guided reading. This is substantively the same architecture the National Reading Panel described two decades earlier, adapted for Indian classrooms and Indian languages. There’s a reason to expect it might work even better here: several Indian languages, including Kannada, have more transparent, regular sound-to-letter relationships than English does, and research specifically on early Kannada readers has found the same strong link between phonological awareness and word-decoding skill that shows up in alphabetic languages worldwide. The gap, in other words, isn’t in what India’s policy says teachers should do. It’s between that policy text and what happens in the first period of a Monday morning in an actual classroom.

What does the national data say is actually happening?

Even after the strongest post-pandemic recovery ASER has recorded, roughly three in four Class 3 children in Indian government schools still could not read a Class 2-level text in 2024.

The Class 3 reading numbers tell a story of a system pulling itself back from a specific shock rather than solving a long-standing one: 20.9 percent could read at Class 2 level in 2018, that fell to 16.3 percent by 2022 after pandemic school closures, and recovered to 23.4 percent by 2024, the best the metric has been in the survey’s history, and still a minority. Class 5 reading levels show a similar pattern of recovery without resolution. NIPUN Bharat’s Foundational Learning Study, assessing roughly 86,000 Class 3 students across 10,000 schools, was designed specifically to set proficiency benchmarks against which this progress, or the lack of it, could be measured state by state. The mission itself was built on the NEP 2020’s own language, which called foundational literacy and numeracy the country’s “highest priority,” warning that the rest of the policy would be “largely irrelevant” for a large share of students if this basic step failed.

What actually works, proven at scale in India?

Pratham, the same organisation that runs ASER, pioneered an approach called Teaching at the Right Level in the early 2000s: children are assessed on their actual reading level regardless of age or grade, grouped accordingly for part of the school day, and taught with activities matched to where they actually are rather than where the syllabus assumes they should be. J-PAL, the MIT-based research lab co-founded by Nobel laureates Abhijit Banerjee and Esther Duflo, has run six randomised evaluations of TaRL across seven Indian states over more than fifteen years, and found some of the largest effect sizes recorded anywhere in education research. The approach has since scaled to reach more than 60 million children across India and Africa. It isn’t a guaranteed fix left unattended: one early evaluation found gains of 0.14 standard deviations in the first year and 0.28 in the second, fading to about 0.10 a year after the programme ended, evidence that matches the original essay’s point about consistency over crisis response better than any restated principle could. What TaRL demonstrates isn’t that a single intervention solves reading. It’s that diagnosing exactly where a child actually is, then teaching to that point specifically, outperforms generic grade-level instruction by a wide and repeatedly measured margin.

What to do

Assess every child’s actual reading level this week, not their grade placement. A simple one-on-one check, similar to the ASER reading tool or a TaRL-style diagnostic, will usually show a wider spread within one class than the timetable assumes.

Check whether your classroom’s reading instruction includes explicit phonological awareness and letter-sound teaching, per the NCF Foundational Stage’s own Four-Block Approach, rather than relying on read-aloud and exposure alone to build decoding skill.

Group children who are behind by their actual level for part of the day, even briefly, rather than only by grade, since this single design choice is the core mechanism behind TaRL’s measured gains.

Use NCERT’s Foundational Literacy and Numeracy materials or Pratham’s TaRL resources specifically, rather than building a reading intervention from scratch, since both already exist and are built for exactly this purpose.

Track each struggling reader’s specific level every month rather than only at the start and end of the year, since the clearest lesson from TaRL’s own evaluations is that gains fade quickly once sustained, close follow-through stops.

Sources: NIPUN Bharat Mission, Ministry of Education, Government of India (launched 5 July 2021) · ASER 2024 National Findings, ASER Centre/Pratham · National Achievement Survey 2021, Ministry of Education · National Reading Panel, “Report of the National Reading Panel: Teaching Children to Read” (2000) · National Curriculum Framework for the Foundational Stage, NCERT (2022) · Phonological awareness and word decoding research in early Kannada readers, peer-reviewed study · J-PAL (Abdul Latif Jameel Poverty Action Lab), evaluations of Pratham’s Teaching at the Right Level, multiple randomised controlled trials across Indian states