There's no AI tutor. No gamification. No magic. Spellimus is retrieval practice, spaced review, mastery pacing, and audio-first input — four ideas with 40+ years of cognitive science behind them — stitched together into a weekly rhythm.
Every word is spoken, not shown. Reading the word first would defeat the test — spelling is sound-to-symbol.
Recall is harder than recognition, and it's what sticks. The child types every letter — no multiple choice, no drag-and-drop.
Missed words return after 1 day, 3 days and 7 days. We use the research, not a fixed schedule.
A word is "mastered" when the child gets it right three times in a row, unaided, across at least two sessions.
The single best-evidenced thing you can do to move a word from short-term to long-term memory is to test yourself on it. Not re-read it. Not copy it three times. Actually try to spell it, from audio alone, and see what happens.
This is the Karpicke-Roediger result, replicated so many times it's uncontroversial: students who self-test between reading sessions remember 50–80% more a week later than students who re-read for the same amount of time.
"Retrieval is not a neutral event. It changes memory. Every act of retrieving a word strengthens the trace of that word and its context."
If a child spells "mischievous" correctly on Monday and never sees it again, there's a ~70% chance they'll miss it by Friday. This is Ebbinghaus's forgetting curve, documented in 1885 and still the single most predictive graph in memory research.
The fix is counter-intuitive: less practice, but spread out. A missed word returns on day 1, day 3 and day 7. Each successful retrieval flattens the curve further. By Friday's test, the word is on solid ground.
"Distributing study episodes over time produces more durable learning than massing them, even when the total time spent is held constant. The effect is robust, large, and extremely well-replicated."
English spelling is an imperfect phonetic system, but it is a phonetic system. A child who can hear "mischievous" said clearly — with stress on the second syllable, not the third — has a much better chance of mapping it to letters than a child who only sees it on paper.
We use UK and US neural voices for the base library, and we let parents and teachers record their own voices. For EAL children, hearing a parent's familiar voice say a new word is a measurable advantage: it lowers cognitive load enough to actually attend to the graphemes.
"Phonological awareness is the single strongest pre-literacy predictor of later spelling ability. Any intervention that puts the spoken word before the written one is operating on the right cognitive pathway."
The right amount of practice for a word isn't 10 minutes. It's the amount it takes to get that child's specific hand-typing-that-word into long-term memory. For some words that's 30 seconds, for others it's ten sessions.
Our default threshold is three consecutive correct retrievals, unaided, spread across at least two sessions. Tutors can loosen it for younger learners (two consecutive, for Year 1), or tighten it for exam prep (four, for 11+). A word is never "mastered" after one good day — that's when revision is still fragile.
"The primary benefit of criterion-based learning is that it calibrates effort to the individual child. Time-based practice reliably over-trains some words and under-trains others."
We shipped streaks for two months in 2023 and killed them. Streaks work wonderfully for DAU metrics and awfully for children: a missed day becomes catastrophic, a family holiday becomes a crisis, and — worst — kids start gaming the system by doing the minimum to keep the streak alive.
We replaced streaks with weekly rhythm. There's a list each week. A 6-minute session, three times. That's it. Completing the rhythm feels good because it's achievable, not because a digital flame shames you.
"Reward structures that punish gaps in practice inadvertently suppress intrinsic motivation. Children begin to engage with the streak, not the task."
Panic, guilt, fake taps to keep it alive. Bad weekends. Sad bedtimes.
Weekly, forgiving, real. Sessions missed roll into weekend if useful, vanish if not.
Spellimus is a structured weekly practice tool for spelling. It's not a phonics programme — we assume the child has some phonics instruction elsewhere. It's not a handwriting tool. It's not an AI tutor that will answer questions about what a word means. It is, very deliberately, a narrow thing done well.
If your child struggles with decoding, not spelling, we'd point you to Read Write Inc. or Little Wandle first. If you want a language-learning app, we'd point you to Duolingo. We're at our best as a weekly thing that runs for five years straight next to the other things you're already doing.
A small reading list. Everything open-access or available through most library systems.
Trends in Cognitive Sciences, 15(1), 20–27. The canonical review.
Applied Cognitive Psychology. Shows the effect persists in real classrooms.
Reading Research Quarterly. On why spelling practice drives writing fluency downstream.
Education Endowment Foundation. Moderate evidence, low cost, widely adopted.
National Literacy Panel. Audio-first for phonological awareness.
Contemporary Educational Psychology, 61. On why streaks backfire at scale.
For families, teachers, and tuition centres. No card, no setup call — just your first weekly list.
No card required · up to 6 children per family · cancel anytime