Why Spelling.School Starts With a Word, Not a Lesson

Most of us learned spelling the same way. A rule on Monday, a list of words that follow it, some copying in between, and a test on Friday. So it can be surprising to open Spelling.School and find no lesson waiting. You hear a word, and you are asked to spell it.

Every part of that was decided on purpose, and most of it was decided because of what learning research says about how spelling is actually learned. The easiest way to show the reasoning is to follow one word through the app. We will use receive, which an enormous number of people write as "recieve".

It started with my son

I built Spelling.School for my son. He always struggled with spelling, and I could not find anything that helped. When I watched his first sessions in the app, the misses were "theif", "swiming" and "runing", and they were not random. The same few patterns went wrong again and again: the order of i and e, a consonant that should double before -ing. What he needed was not another lesson about those patterns. It was practice on the particular words where they kept tripping him up.

What I asked for then is still the heart of the app: "Whatever people are messing up, bring them a bunch of that until they get it." Everything below is how we tried to do that well.

First, you try it

When "receive" comes up for the first time, you are not shown it. You hear it said, you hear it in a sentence ("I was happy to receive a letter from my friend"), you see that it is a verb meaning to get or be given something, and you type your best attempt.

Hear it, see what it means, type it from memory. Nothing on the screen gives the spelling away.

Trying before being taught feels backwards, but it helps. In experiments published in 2009, Nate Kornell, Matthew Hays and Robert Bjork had people attempt answers they could not yet know. Getting it wrong first made them learn the answer better, once they saw it, than people who had simply studied it from the start. Lindsey Richland and colleagues found the same with reading passages that year and called it the pretesting effect. The failed attempt gives the right answer somewhere to land.

It also saves your time. Most words in any level are words you already spell. Trying first lets the app find that out in a second, so your minutes go on the words you miss. If you type "receive" correctly and confidently, it is marked as known and set aside for a while. If you get it right only after a lot of backspacing, the app treats it as a guess, because heavy correction while typing is one of the clearest signs that a person was unsure.

Look at what is not on that screen. There are no answer choices, and the count of letters is hidden, because it gives away a dropped or doubled letter. On a phone the app draws its own keyboard, because your phone's keyboard suggests words as you type, and a suggestion is the answer.

Nothing to choose from

There are no multiple-choice questions anywhere in Spelling.School. You never pick "receive" out of a list of four spellings.

Two findings decided that. The first is the testing effect: producing something from memory strengthens it far more than looking at it again. In 2006 Henry Roediger and Jeffrey Karpicke showed that students who practiced recall remembered much more a week later than students who reread the same material, even though the rereaders felt more confident. Recognizing a spelling is a much weaker kind of recall than producing it. (The testing effect goes into this in detail.)

The second is more alarming. In 1988 Alan Brown found that simply seeing misspelled words makes people more likely to misspell them later, even words they knew beforehand. Larry Jacoby and Ann Hollingshead found the same two years later, in a paper called "Reading student essays may be hazardous to your spelling." A quiz that shows you "recieve", "receeve" and "receve" next to the right answer is teaching you three wrong spellings. So the app never asks you to pick the right spelling out of a lineup of wrong ones. What you study is your own attempt, marked letter by letter beside the right one.

The moment you miss

Say you typed "recieve". This is what comes back.

The exact letters that went wrong, and the rule they broke, at the moment it matters.

The two letters you swapped are marked. The pattern they belong to is named, and the rule is shown: i before e, except after c. This is where the teaching happens in Spelling.School, and it is the part people most often ask about.

Explicit teaching of spelling patterns works, and nothing here argues otherwise. A 2014 meta-analysis by Steve Graham and Tanya Santangelo found that teaching spelling explicitly improved spelling, and reading and writing too, at every age they looked at. English is also far more orderly than its reputation. A large study of about 17,000 words in the 1960s found that roughly half are spelled predictably from their sounds, and most of the rest become predictable once you know a word's history and meaning. (Is English spelling really irregular? has the numbers.)

The real question is when to teach a rule. A classroom teaches i before e to thirty children on the same Tuesday, because that is the only way a classroom can work. One learner has no such constraint. If you already spell "believe" and "ceiling" without thinking, a lesson on i before e is time you will not get back. If you write "recieve", the best moment to meet that rule is the moment you write "recieve". Every word in the app is tagged with the spelling patterns it contains, about sixty in all, down to the exact letters, so the rule can be the one your mistake actually broke. When an attempt is too far off to pin on one pattern, the app shows the word's own patterns instead.

The rule has to be precise to help. "This vowel sound can be spelled several ways" is true and not much use. Where our wording is too broad, that is ours to improve, and we are improving it.

Then you type it right

You cannot move on by looking. You type "receive" correctly first.

The correction becomes one more act of recall, and the word returns later in the same session for an unprompted try.

This comes from research on feedback. In 2005 Harold Pashler and colleagues found that, for people learning new words, being shown the right answer after an error made a large difference to what they remembered, while feedback after a correct answer added little. The correction is where the learning is, so the app makes you produce it rather than glance at it. A few cards later "receive" comes back, and you spell it again with nothing to copy.

Linnea Ehri's work on orthographic mapping explains why producing the letters matters so much for spelling. A word becomes automatic when its sounds are bound to its letters in memory, and that binding is built by attending to each letter as you say the word, not by recognizing its outline. (How a word becomes automatic explains the idea for parents.)

When the same mistake keeps happening

One miss on "receive" is a slip. Three misses on the same pattern, across different words, is a habit, and the app treats it that way. It switches on what we call a weak spot for i before e. Your next sessions bring in a couple of extra words that share it, "ceiling", "deceive", "believe", "field", mixed in among your other words, until you have spelled five of them correctly on the first try in a row.

If you want the whole rule laid out, it has its own page, with the exceptions that have to be learned by heart and a two-minute drill.

Each of the patterns has a page like this. The lesson is there; it arrives when your own mistakes say you need it.

Notice that the extra words come mixed into your session rather than as a page of ie and ei drills. Doug Rohrer and Kelli Taylor showed in 2007 that mixing different kinds of problems produces better learning than practicing one kind in a block, even though the block feels easier. Ten "-ie-" words in a row teach you to type "ie". A mix teaches you to decide between ie and ei, which is the skill a real sentence asks for. The same thinking shuffles every session: reviews and new words together, never a list in alphabetical order. (Interleaving explains the research.)

So the school is there. It just does not start with a lecture.

Then it comes back

A word you spelled correctly once is not yet a word you can spell. Most of what anyone learns fades within days unless it is revisited. A 2006 review by Nicholas Cepeda, Harold Pashler and colleagues, covering hundreds of experiments, found that spacing those revisits out over time builds far stronger memory than the same practice crammed together.

Spelling.School schedules "receive", and every other word, with FSRS, an open-source model that estimates how likely you are to still remember each word. Missed this morning, it comes back soon. Right three times, it waits weeks, then months. You never make a list or decide what to review. (The forgetting schedule shows how the timing works.)

That is also why sessions are a few minutes long and daily. Phillippa Lally's research on habits found that a small daily action becomes automatic in roughly two months, and that missing a single day does not undo the progress. So the app asks when you would like to practice and reminds you then, spreads out the reviews after a week away rather than greeting you with hundreds, and does not punish a missed day.

Memory and understanding

A fair objection to any tool like this is that it is just memorization with an algorithm. The word carries a lot of baggage, so it is worth being plain about it.

Spelling is a memory skill. Nobody can work out "receive" from first principles in the middle of a sentence; every good speller is drawing on thousands of words held in memory. Knowing the rule makes that memory easier to build and harder to lose, which is why the rule appears at the moment of the miss. But knowing a rule and having the word ready when you write are different things, and only practice turns one into the other. Anyone can hear "i before e" a hundred times. What changes spelling is meeting the rule on the word you just got wrong, every time you get it wrong, until you stop.

Who this is for

If you want a course that walks through every spelling rule in order, like a school year, good ones exist, and for a young child just learning to read, that order matters a great deal. Our post on how spelling should be taught describes what a good one looks like.

Spelling.School is for anyone who already reads and writes English and keeps losing marks or confidence to the same words: the adult who was never properly taught, the student sitting IELTS, the nurse writing charts, the child who aces the Friday list and misspells it on Monday. For all of them the fastest route is the one my son needed. Find the words that are actually wrong, show exactly why, and keep at them until they are right without thinking.

To see this with your own words, try the Spelling Fingerprint, a free four-minute check that finds the patterns behind your mistakes, with your own answers as the evidence. It is the closest thing we have to a first lesson, because it is built entirely from you.

Questions people ask

Does Spelling.School teach spelling rules?

Yes. Every word is tagged with the spelling patterns it contains, about sixty in all, and when you miss a word the app marks the wrong letters and shows the rule behind them. Patterns you keep missing become weak spots, and you get extra practice words for them until you spell them reliably. Each pattern also has its own page that explains the rule, its exceptions and the words people most often miss.

Why does Spelling.School ask me to spell a word before teaching it?

Because trying first helps you learn, even when the attempt is wrong. Research by Kornell, Hays and Bjork in 2009 found that an unsuccessful attempt makes the correct answer easier to remember once it is shown. Trying first also lets the app skip the words you already know, so your practice time goes on the words you actually miss.

Why are there no multiple-choice spelling questions?

Two reasons. Producing a word from memory strengthens it much more than recognizing it in a list, which is the testing effect described by Roediger and Karpicke. And studies by Alan Brown and by Jacoby and Hollingshead found that seeing misspelled words makes people more likely to misspell them later, so a quiz full of wrong spellings can do harm.

Is spaced repetition just memorization?

Spaced repetition is a way of timing practice so that each word returns as it starts to fade, which builds memory that lasts far longer than cramming. Spelling does depend on memory, since nobody can stop to work out "receive" from rules in the middle of a sentence. Good practice combines the two: the reason for the spelling when you miss it, and repeated recall until the word is automatic.

Why do I have to retype a word after I get it wrong?

Retyping the correct spelling turns the correction into active practice rather than a glance at the answer. Research on learning words, including a 2005 study by Pashler and colleagues, found that feedback after an error is where most of the learning happens. The word also returns a few cards later in the same session, so you get a second, unprompted try.

Is Spelling.School good for a young child just starting to read and spell?

It is best for learners who can already read simple English and want to fix the words they get wrong. A child at the very start of reading benefits most from systematic teaching of letter sounds and patterns in order, usually led by a teacher or parent. Spelling.School works well alongside that, once a child is writing words on their own.

Fix the words you actually get wrong

A few minutes a day, spaced repetition, built only for spelling. Try five words without signing up.