Drafted with AI assistance and reviewed, fact-checked and edited by hand before publishing - see our AI transparency page for how we use AI across the app.
The idea for this feature came from Faizel Manjoo, Founder of Education Beyond and Founding Educational Partner of Spelling.Live.
Faizel has worked in education for more than 24 years, with experience spanning classroom teaching, school leadership and working directly with private families. His work includes educational assessment, academic development and helping children prepare for competitive school entry.
As part of his involvement in the educational development of Spelling.Live, Faizel proposed introducing a three-week checkpoint to establish whether children have retained what they have learned and can apply spelling patterns to unfamiliar words. He also helped shape the educational principles behind the feature, including the distinction between remembering previously practised words and applying learning independently to new words.
What follows is how these principles have been incorporated into Spelling.Live.
The question spaced repetition can't answer
Spelling.Live already brings words back on a spaced-repetition schedule: a word your child gets right comes back after longer and longer gaps, and a word they get wrong comes back tomorrow. See exactly how words come back, and the settings that change it.
But every one of those reviews asks about a word your child has already practised. If they spell "rain" correctly on review, we know they remember "rain". We don't know whether they can spell "paint", "snail" or "explain". Those words use the same pattern, ai as in rain, but the child has never been given them.
That gap matters. A child can remember a list without having learned the pattern behind it, and ten words held in memory for a Friday test fade. Knowing the pattern is what lets a child spell words they meet in their own writing.
So the checkpoint asks both questions: do they remember it, and can they use it?
What the checkpoint does
Every three weeks, starting three weeks after your child's first practice that counts, the app offers a short checkpoint. It never blocks anything, and if now isn't a good time you can tap "Remind me tomorrow".
It looks at the spelling patterns your child has practised in the last three weeks with at least two different words, along with any patterns that are due a recheck. It picks up to 3 patterns for children in Reception to Year 2, and up to 5 for older children.
For each pattern, your child spells up to four words:
- 2 words they have practised. This is retention: can they still spell them?
- 2 words they have never been given that use the same pattern. This is transfer: can they apply the pattern to new words?
That comes to up to 12 words for younger children and up to 20 for older ones. If we don't yet have enough checked new words for a pattern, your child spells only its two practised words, so the total can be lower. The number shown before the checkpoint starts is always the number of words your child will actually be asked.
The new words are always hand-chosen, never generated by AI. They come from our curated phonics lists, the National Curriculum's own example words and worked examples we have checked by hand. A wrong or made-up word in a check like this would give you a wrong answer about your child, so we don't take the risk.
The app says each word and reads it in a sentence. Your child gets one try per word, with no hints, no flash of the word and no letter tiles. They type or write on screen, whichever they normally use. This is the one part of the app where we deliberately hold help back, because the point is to see what they can do on their own.
Only the right mistakes count
A mistake only counts against a pattern if the mistake is in that pattern. If your child writes "rane" for "rain", that counts against ai, because the ai is what went wrong. If they write "paimt" for "paint", the ai is right and the slip is in the n, so it doesn't count against ai.
This stops one slip being blamed on the wrong rule.
What the outcomes mean
Each pattern gets one of four outcomes:
- Secure. All four words right, both the practised words and the new ones. The pattern is checked again in about two to three months.
- Remembered, new words not yet checked. Both practised words right, but we didn't have checked new words to test whether your child can apply the pattern. Retention is confirmed; application isn't yet, so we don't call it Secure. It is checked again in about two to three months.
- Needs more practice. One mistake, or both practised words remembered but both new words missed. The pattern comes back into practice automatically, about two words per session for the next three weeks, and is rechecked at the next checkpoint.
- Not secure. A practised word wrong plus another mistake, or three or more mistakes. The pattern comes back with about four words per session and goes first. It is rechecked once your child has spelled three words with that pattern correctly in practice.
You don't have to plan any of this.
What you see, and what your child sees
Parents and teachers see retention and transfer separately, because they tell you different things. A line might read:
ai: Remembered 2 of 2 words, Applied 0 of 2 new words. Needs more practice.
That child has learned their words but not yet the pattern. A child who missed a practised word but spelled both new ones is in a different place, and the report shows it. If we don't have enough checked new words for a pattern, the report says "New-word check not available" rather than guessing.
Your child sees only a celebration of what they got right. They never see a list of what they got wrong. The checkpoint is for the grown-ups to act on, not for a child to feel tested by.
A smaller change that is already live
While building this we also changed how one kind of answer counts. A word your child only got right after help, or after missing it first, used to count the same as a word they knew straight away. It now counts as practice, not proof that they remember it, and the word comes back sooner. This follows another of Education Beyond's principles: a correct answer reached with help is practice, not evidence that the child can recall the word.
Thank you to Education Beyond
The three-week checkpoint was developed from the educational principles proposed by Faizel Manjoo, Founder of Education Beyond and Founding Educational Partner of Spelling.Live.
Faizel's contribution focused on how to assess retention and transfer, distinguish independent recall from assisted performance, and use checkpoint results to inform subsequent learning and reinforcement.
Spelling.Live has translated these principles into the app's checkpoint functionality.
You can find out more about Faizel's wider educational work at educationbeyond.co.uk.
Join the parent community
Faizel also runs the Education Beyond parent community on WhatsApp, for parents who would value guidance and discussion about their children's education beyond spelling. Join the Education Beyond parent community on WhatsApp.
The community is run by Education Beyond, not by Spelling.Live, so WhatsApp's and Education Beyond's own terms and privacy policies apply.