What a term of spelling data reveals about how children spell

Where these numbers come from

Everything below is drawn from a sample of anonymised, word-level practice over a single term (April to July 2026): more than 100,000 spelling attempts by children aged roughly 5 to 11, working through the statutory UK primary spelling curriculum. Most of it is children practising at home rather than one class working from one teacher's list, so no single classroom's choices skew the picture — but it's a sample, and we've treated it as one.

We have been careful with it. There are no pupil names or per-child figures anywhere here — every number is an aggregate for a word or a year group, with its sample size shown. We stripped out demo and test accounts, and we have flagged the few findings that rest on thinner data. Where the data can't honestly support a claim, we say so rather than dressing it up. If you'd like the short version of the memory science behind all this, our guide to the science of spelling covers it.

The five hardest words in the primary curriculum

Ranked by how often children spelled them correctly, the five hardest words were all Year 3/4 "tricky" words with a double letter, an awkward unstressed vowel, or both.

"Accidentally" was spelled correctly barely a third of the time. All five share the same trap: children can hear that something doubles or that there's a vowel they can't quite place, but they can't tell which. More on that pattern below.

And, for balance, the words children found easiest — all short, and spelled close to how they sound:

The longer the word, the harder it gets — up to a point

Sort every word by its length and the accuracy falls in an almost perfect staircase: each extra few letters costs roughly another ten points.

But length isn't destiny. Some short words are much harder than they look, and some long ones much easier. "Reign" (five letters) was spelled correctly only 61.8% of the time — worse than "important" (nine letters) at 83.3%. The difference is transparency: "important" is spelled almost exactly as it sounds, while "reign" hides a silent letter and a homophone. Length is a rough proxy for difficulty; how regularly a word maps sound to spelling matters more.

A hall of fame of misspellings

This is the part most teachers will recognise instantly. For every word with enough wrong answers to be reliable, we found the single most common misspelling and how dominant it was. The striking thing is how concentrated the errors are: for many words, one specific wrong spelling accounts for the majority of all mistakes. These aren't scattered guesses — they're the same error, made the same way, by child after child.

A few worth pausing on. "Woman" became "women" 82% of the time — that's a grammar slip (singular versus plural) wearing a spelling costume, not a phonics problem. "Friend" became "freind" in a third of errors: the textbook "i before e" violation, live in the wild. And "February" lost its first R to become "febuary" — children spell it the way it's usually said. If you want to help a child fix these, the trick is almost always to make the tricky part visible; our parents' guide on helping children remember spellings works through how.

Children write real words, not gibberish

Here's a finding with a genuinely practical edge. When children got a word wrong, they very often didn't produce a jumble of letters — they wrote a different, correctly-spelled English word that sounds the same.

This matters because a spelling test marks these wrong, but re-drilling the letters won't fix them. A child who writes "hole" for "whole" hasn't failed at phonics — they've spelled a real word perfectly; they've just picked the wrong member of a homophone pair. The fix is meaning, not mechanics: teach the pair together ("where and were", "the whole thing has a w"), rather than assigning more of the same spelling practice. It's a different problem, and it needs a different lesson.

Some words never really click — even by Year 6

We took a handful of words taught in Year 1 and 2 and tracked how well children spelled them at each actual age, from Year 2 up to Year 6. Most words improve as children get older. A few simply don't.

"Because" does what you'd hope: a clean climb from 78% to nearly perfect by Year 6. "Beautiful" doesn't. A Year 5 child was actually more likely to misspell "beautiful" than a Year 4 child, and it only really lands at Year 6. "Whole" just drifts in the low-to-mid 70s for a child's entire time at primary school. These are common exception words — words that break the usual patterns — and the data says plainly that they don't get fixed by children simply growing older. They need to be taught, and re-taught, deliberately.

What kind of mistake is it?

Reading through the most common errors, they fall into a handful of recognisable families. This grouping is our own judgement rather than a machine-counted statistic, so treat the shares as a rough read of the landscape, not a precise measurement — but the families themselves are clear:

• Homophones — the wrong real word (roughly a third): where/were, whole/hole, break/brake, would/wood, steak/stake.
• The doubled letter in the wrong place (about a quarter): "dissapear" for disappear, "adress" for address, "grammer" for grammar.
• The vanishing unstressed vowel (roughly a fifth): "seperate" for separate, "calender" for calendar — the schwa sound that could be almost any vowel.
• The dropped syllable (about a sixth): "libary", "intrest", "febuary" — words spelled the way they're said quickly.
• The "i before e" tangle (a small but stubborn group): "freind", "belive", "hight".

The useful takeaway is that most spelling errors aren't mysterious. They cluster into a few predictable mechanisms, and each mechanism has its own fix.

Some words have one fix; others are pure chaos

Not every word behaves the same way when children get it wrong. For some, nearly all the errors converge on a single alternative spelling. For others, almost every wrong answer is unique — no two children make the same mistake.

Predictable words — one dominant error — include centre (72% of errors are "center"), breathe (75% "breath"), believe (67% "belive") and library (63% "libary"). Chaotic words — where the errors scatter with no single culprit — include "various", "material", "enough", "occasionally" and "ordinary", each with roughly two-thirds of their wrong answers being one-offs.

This is genuinely useful for planning. The predictable words are the efficient ones to target: teach the one specific confusion ("it's centre, not center") and you'll clear most of the class's errors on that word in a single move. The chaotic words are a different signal — when children misspell "various" fifteen different ways, they're not confusing it with one thing, they simply don't have a firm image of the word yet. Those need varied exposure and use in context, not a single corrective rule.

Does practice actually work? Yes — but it has limits

We split the curriculum into three difficulty tiers and tracked accuracy by how many times a child had attempted each word. Practice helps everywhere, and it helps hardest words the most.

The hardest words climb more than 33 points from the first try to the sixth — repetition works, and it works most where it's needed most. But notice the ceiling: even after six or more goes, the hardest words (79%) still hadn't caught up to where the easiest words started on their very first attempt (87%). For genuinely difficult words, practice narrows the gap but doesn't close it inside a term. That's the honest case for combining retrieval practice with explicit teaching of the tricky part, rather than relying on repetition alone. (One thing this data can't prove, despite our hoping it would: that spacing practice across days beats cramming. Within this dataset, recently-practised words were spelled best — ordinary forgetting — so we're not going to claim a spacing effect we can't show. The wider research still supports little-and-often; our term of data just isn't the right instrument to demonstrate it.)

What we're not claiming

A quick, honest word on the limits, because data like this is easy to over-read. This is one term from one app, weighted toward children whose families chose to practise, so it isn't a representative national sample. The thinnest findings here — some of the Year 5/6 words — rest on a few dozen attempts, and we've flagged those rather than headlining them. We excluded teacher-authored custom word lists, because one class's unusual list can masquerade as a national trend. And, as above, we're deliberately not claiming a spacing effect the data can't support. What's left is, we think, a fair and genuinely useful picture — but it's a read on this term's practice, not the last word on how children spell.

If you teach spelling, three things to take away

1. The hardest words are predictable — front-load them. "Accidentally", "business", "separate", "occasionally" and the double-letter family will eat a disproportionate amount of your class's errors. They reward being taught explicitly and early.
2. Sort errors by type, not just right or wrong. A homophone slip ("hole" for "whole") needs a meaning lesson; a schwa error ("seperate") needs the vowel made visible; a chaotic word needs more exposure. The mistake tells you the fix.
3. Some words won't yield to age or repetition alone. "Beautiful" and "whole" sit stubbornly still across the years. Those are the ones to keep circling back to, deliberately, rather than assuming they'll come good on their own.

If any of this is useful, the free statutory word lists and spelling resources on the site are organised by year group and built around exactly these patterns.

Method: a sample of more than 100,000 anonymised, non-practice spelling attempts across the statutory UK primary curriculum, April–July 2026. Demo and test accounts excluded; teacher-authored custom lists excluded; per-word sample sizes shown throughout. No pupil, class, or school is identified. Figures are word-level or year-group aggregates.