Engelmann’s Crutch

2026-05-04

Walk into a Direct Instruction classroom and the first thing you notice is the rhythm. The teacher asks a question. Pauses. Says “get ready.” Snaps her fingers. Thirty kids answer in unison. Three seconds later, another question. Another snap. Another chorus. Ten responses per minute, per kid, for forty minutes straight.

It looks like a drill team. It works better than anything else we’ve ever measured. Project Follow Through — the largest education experiment in American history — tried nine teaching models on 100,000 disadvantaged kids. Direct Instruction was the only one that produced significant gains on basic skills, and on higher-order thinking, and on self-esteem. Third-graders with IQs under 71 hit the 70th percentile in reading. Nothing else came close.

The four mechanics that made it work are genuinely beautiful — and each one is something Hebrew classrooms quietly skip.

Strategies, not facts. A great DI program teaches the procedure, not the output. Ten letter-sounds plus a sounding-out rule lets a child read 25,000+ words. Memorizing 10 sight words lets them read 10. Hebrew is even more strategy-rich than English: a finite set of letters, a regular nikud system, three-letter shorashim, and a small number of binyan patterns generate the entire language. Teach the system, and a kid can decode and conjugate words they’ve never seen. But that is almost never how Hebrew is taught. The default play is to import a native speaker who doesn’t know how they decode — they just do — and let kids absorb by exposure. After twelve years, most can recognize a few hundred memorized words and conjugate nothing they weren’t drilled on. The strategy was never taught because the teacher never had to learn it.

Tracks, not units. Standard curriculum looks like a stack: Unit 1 for two weeks, Unit 2 for two weeks, Unit 3 for two weeks. The moment a unit ends, practice on it stops cold — and by the time it shows up again, the skill has decayed. A track runs sideways instead. A single skill (say, identifying a shoresh) appears in every lesson for thirty-plus lessons, mixed in with five other tracks at different stages of maturity. Skills don’t decay because they keep showing up. Hebrew curricula almost universally use units — colors this week, family next week, food the week after — because units are easy to plan and easy to test. The result is a student who could name fifteen colors in October and can’t anymore by March, because nothing brought “color” back into the lesson cycle.

Scripts, not improvisation. A DI script isn’t a teleprompter. It’s a ladder of five formats teaching the same skill at progressively less scaffolding — from heavy support (“Is there an e on the end? Will the vowel say its name?”) through to encountering the skill in natural reading with no cue at all. The teacher knows which rung each kid is on. No one advances until they’ve mastered the current rung. A native-speaker Hebrew teacher has none of this. She speaks Hebrew at the class and hopes osmosis works for the kids who happen to be wired for it. There is no rung 1, rung 2, rung 3 — there is only “more Hebrew” and “more frustration.” Mastery isn’t measured because mastery was never the design goal; exposure was.

Signals, not solo Q&A. In a normal classroom, a teacher who calls on individual students gives each kid about one oral response per minute. In a DI classroom using unison response on a snap, every kid gets ten. That’s a 10x increase in active practice per student per minute — and ragged responses instantly reveal who’s behind, so no one fakes their way through a year. Hebrew classrooms run the opposite extreme: the teacher talks, the class listens, maybe one kid gets called on. Most students go an entire period without producing a single word of Hebrew. They are spectators of someone else’s fluency.

But look closely at that last one. The snap. The “get ready.” The military-grade choreography of getting thirty mouths to move at the same instant.

That’s a crutch.

It’s a brilliant crutch — Engelmann earned every inch of it — but it exists because one human teacher cannot personally listen to thirty children answer one at a time. Unison response is the workaround. Most of DI’s elegance is workaround. The scripts are workarounds for inconsistent teachers. The fixed lesson sequence is a workaround for a class that has to move together. The whole apparatus is the best possible answer to: how do we teach a strategy, to mastery, with feedback every few seconds, to a group of thirty?

That constraint is gone.

A child sitting with an AI tutor doesn’t need to bark answers in unison. The microphone hears every response. The model corrects every error in real time. There’s no “ragged response” problem because there’s no group — there’s one student getting one immediate signal on every single attempt. The strategy stays. The tracks stay. The mastery rule stays. The crutches fall away.

This is what we’re building for Jewish Alpha School. Take Engelmann’s actual mechanics — the strategy design, the track structure, the format ladder, the response cadence — and run them through a tutor that never gets tired, never has to manage thirty kids, never has to compromise pace for the slowest learner or boredom for the fastest. The four mechanics get sharper, not softer, when you remove the crutch they were built to compensate for.

Hebrew is the perfect proving ground. The language is highly systematic, the classroom model has been failing it for fifty years, and there is a real community of parents who care deeply that their kids actually own the language — not just survive twelve years of trying.

If you’ve built scripted curriculum, designed mastery-based assessments, or worked with Direct Instruction in any form — email me. The crutch is coming off, and we need people who understand what was holding it up.