Champion · IID Policy Hackathon 2026

ঝরে পড়াদের বাংলা পড়া

Bangla reading for working teens in Dhaka

We find each teen's reading level, and teach from there.

Rafi's story

  1. Rafi, 15, at the tea stall, looks up at shop signs he cannot read. Caption: রাফি, বয়স ১৫। ছোট থেকেই পড়া ছেড়ে কাজ করে। Thought: এই লেখাগুলো কী বলে?
  2. A crowded room seated by age; the teacher tells Rafi to sit with children his size. Caption: ১ জন শিক্ষক, ৩০ জন ছেলেমেয়ে। বসানো হয় বয়স দেখে। Teacher: রাফি, তোমার বয়সীদের সাথে বসো।
  3. Rafi holds a children's primer with a big অ while two teens laugh. Caption: হাতে ছয় বছরের শিশুর বই। Teens: হা হা! বাচ্চাদের বই পড়ে!
  4. A fluent girl reads aloud while Rafi struggles over his slate. Caption: কে কতটা পড়তে পারে, কেউ দেখে না। Girl: ...তারপর সে বাজারে গেল। Rafi thinks: আমি এখনো পড়তে পারছি না!
  5. The teacher asks where Rafi is; his place on the bench is empty. Caption: আসা কমিয়ে দেয়। তারপর আবার ঝরে পড়ে। Teacher: রাফি কোথায়? Learner: ও আর আসে না।
  6. Back at the stall, Rafi asks an older man to read his phone for him. Caption: দোকানে ফিরে, এখনো অন্যকে দিয়ে পড়িয়ে নিতে হয়। Rafi: চাচা, এখানে কী লেখা?

Rafi is an illustrative character drawn from practitioner interviews.

The problem

ঝরে পড়াদের বাংলা পড়া আর হয়ে ওঠে না...

Older beginners are taught like 6-year-olds, and drop out again.

2 in 3
of the poorest children (7⁠–⁠14) can't read with understanding.*MICS 2025
12 of 14
practitioners place learners without a reading check.*Our survey, n = 14: an early signal
Our proposed solution

Start learning based on the reading level, not by age.

Goal: each learner reaches level ৫, বুঝে পড়া (reading with understanding), in 3 to 5 months.

অক্ষর থেকে অর্থOne Bangla reading scale: 7 levelsBuilt on NAPE, ASER, EGRA and TaRL.See the full scale
০ধ্বনি
১বর্ণ
২শব্দ
৩যুক্তবর্ণ ও বাক্য
৪সাবলীল পড়া
৫বুঝে পড়া
৬গভীরভাবে বোঝা
  1. Check in 5 minutes

    Web app. Voice-guided.

  2. Find your level

    and continue learning.

  3. Teach with everyday Bangla

    Printed: signs, prices, bKash.

The future we want

Anyone who wants to READ Bangla, should be able to read Bangla.

In the room

  1. 1Each learner knows their level after a reading check.
  2. 2Small groups by level, not age.
  3. 3A ready level-wise sheet every week.
  4. 4Learners read a story and say what it means.
Rafi uses the reading app on his phone and reads Bangla by himself. Bubbles: এই অ্যাপটা তো দারুণ! বাংলা পড়া এত সহজ! / ওয়াও! এত সহজভাবে পড়া শেখা যায়! / আমি এখন সহজেই বাংলা পড়তে পারি!
Rafi reads Bangla by himself. Later: for teens with their own phone.

Underneath: the systemBNFE and the Ministry build the check in. Progress is tracked by level, not attendance.

Deeper still: the beliefEvery learner can progress from where they are.

  • Pebble

    Now · our team

    Apps ready to test; open release; scorer-agreement test first.

  • Rock

    0–6 months · NGO partners

    Pilot in 3–5 Dhaka centres. Track scorer agreement, attendance and level gains. Nothing signed yet.

  • Boulder

    BNFE, MoPME · via NFE Act 2014 s.12

    BNFE circular under the NFE Act 2014 (s.12): record a reading level at intake in the NFE MIS for SKILFO and ALO. PEDP-5, approved 4 Nov 2026, has no reported out-of-school line.

Dignity: results are never ranked or shown to the class. Could be left out: girls kept at home, learners with disabilities, non-Bangla speakers, teens without phones.

Our ask

A BNFE circular (NFE Act 2014 s.12): a reading-level check at NFE, SKILFO and ALO intake, mapped to EMDC's levels. Plus an open teen primer (s.9(4)).

Questions

Common questions

Who is this for?

Teens in Dhaka who left school early to work and never learned to read Bangla well. Most programmes seat them by age and hand them children's books, so many drop out again.

Does a learner need to read to take the check?

No. The check is voice-guided and starts at sounds (level ০), so a learner who can't read yet can still take it.

Try the reading check
Does every learner need a phone?

No. One shared phone per centre runs the check. Teaching uses printed level sheets built on everyday Bangla: signs, prices, bKash.

Does AI decide a learner's level?

No. Fixed rules score the check. AI only drafts level sheets and transcribes the optional read-aloud, and a person checks every level.

What happens to a learner's data?

Results are never ranked or shown to the class, and learners use a nickname. Guardians give consent for anyone under 18, as the Personal Data Protection Act 2026 requires. The read-aloud is optional: the audio goes to a transcription service and we keep only the score.

Isn't the government already doing this?

Partly. UNICEF's EMDC places 7 to 14 year-olds by level. SKILFO and ALO now reach older teens, but neither publishes a reading placement. We plug a check into their intake instead of building a new programme.

What would a pilot cost?

Our estimate for 100 learners in 4 centres over 3 months is BDT 80,000, about BDT 800 per learner, mostly people and phones.

Has it been tested with learners?

Not yet. The apps are ready to test. Next is a pilot in 3 to 5 Dhaka centres, starting with a test of whether two scorers agree. Nothing is signed yet.

I run a centre or programme. How can I take part?

Write to hello@shadiqbal.com. We are looking for centres for the pilot and for teachers to review the Bangla content.

Email Team NASS
Sources and evidence

What our numbers stand on

  • Problem

    MICS 2025

    BBS and UNICEF, 2025

    Among children aged 7 to 14 from the poorest 20% of homes, only 33.6% read with understanding: 2 in 3 can't.

  • Problem

    Annual Primary School Statistics 2024

    Directorate of Primary Education

    16.25% leave before finishing primary. Dropout peaks in Grades III and IV, where reading is weakest.

  • Problem

    Team NASS practitioner survey

    Team NASS, Oct 2026 (n = 14)

    12 of 14 practitioners place learners without a reading check. An early signal from a small sample.

  • Gap

    EMDC, Stage 1 report

    UNICEF Innocenti, Dec 2025

    Level placement already works here for 7 to 14 year-olds. Not yet for teens.

  • Gap

    SKILFO

    Government of Bangladesh and UNICEF, Jun 2025

    BNFE's literacy programme for ages 14 to 17, scaling to 100,000+ learners. No published reading placement.

  • Policy

    Non-Formal Education Act 2014

    Government of Bangladesh

    s.12 lets BNFE set what providers report, so a reading-level field needs a circular, not a new law.

  • Policy

    Personal Data Protection Act 2026

    Government of Bangladesh

    Guardian consent for anyone under 18; a voiceprint is sensitive data. It shapes our safeguards.

  • Evidence

    Teaching at the Right Level

    J-PAL

    Randomised evaluations: grouping children by level, not grade, improves basic reading.

  • Scale

    অক্ষর থেকে অর্থ

    Team NASS reading scale

    Built on NAPE, ASER, EGRA and TaRL. Every level and the full reference list.

The team

Team NASS

Champion · IID Policy Hackathon 2026

Team NASS, champions of IID Policy Hackathon 2026
  • Shad Iqbal

    Team lead, product

    Framed the problem, set the scope, and built the first testable version: the level check and the reading scale.

  • Niloy Biswas

    Data and research

    Worked on the evidence: the practitioner survey, the sources and the numbers.

  • Adil Mahmoud Rion

    Backend and app engineering

    Built the backend and the সহজ বাংলা learner app.

  • Md. Kamrul Hasan Shovon

    Teacher, Bangla content and pedagogy

    Shaped the Bangla reading content and how each level is taught.

From 10 Minute School, in a personal capacity.