The Forgetting Curve: Why You Forget a Word You Learned Yesterday

On Tuesday you looked the word up. You studied it properly, wrote it out, even noticed the trick of it: "accommodate" has room for two c's and two m's. On Thursday you went to type it in an email, and your fingers produced "accomodate" before you could stop them. A small voice said what it always says. You literally just learned this.

That experience is so common that it has a name and a shape you can draw on the back of an envelope. The forgetting curve was first measured in the 1880s, and it has survived nearly a century and a half of people trying to knock it down. It explains why a spelling crammed on Thursday night holds up for Friday's test and has slipped away by Monday.

More usefully, it tells you what to do about it. The answer turns out to be less work than most people are doing now, spent at better moments. This piece covers what the curve is, what modern research has added, and how one stubborn word behaves over two weeks when you treat it the right way.

A man, a stack of nonsense, and a stopwatch

Hermann Ebbinghaus wanted to measure memory the way a physicist measures a falling object, and he ran straight into a problem. Real words arrive loaded with meaning and association, and those make some things far easier to remember than others. So he invented his own material. He built a large stock of meaningless syllables, consonant, vowel, consonant, things like "dax" and "bup" and "zol", and learned lists of them at a steady pace until he could recite each list without a mistake.

Then he waited. After 20 minutes, or an hour, or a day, or a month, he sat down and learned the same list again.

His measure was the clever part. Instead of asking how much he could recall, which is crude, he timed how long the relearning took. If a list first took 10 minutes and relearning it took four, he had saved six minutes, or 60 percent. That saving told him how much of the original work was still in his head, including the part he could no longer consciously reach.

The results, published in 1885, have a shape everyone who has ever studied will recognize. After 20 minutes, his savings had already fallen to a little under 60 percent. After an hour, to under half. After a day, to about a third. And then the fall almost stopped. Six days later he still saved about a quarter of the effort, and a month later about a fifth.

The curve is a cliff followed by a long, gentle slope. Most of what you are going to lose, you lose in the first day.

The curve has held up

One man testing himself on nonsense syllables in the 1880s is a slender basis for a law of nature, so in 2015 two researchers at the University of Amsterdam, Jaap Murre and Joeri Dros, did it again. One of them spent many hours learning and relearning lists of syllables, following Ebbinghaus's method as closely as a modern lab could. The new curve lay almost on top of the old one.

It did have one small wrinkle. At the 24-hour mark, the savings were slightly higher than the smooth curve predicted, as if something had happened overnight to protect the memory. The authors suggested sleep as a likely reason, which fits a large body of later work on how the sleeping brain consolidates what it learned during the day. We go into that in why sleep is part of learning to spell.

Real words fade more slowly, but the same way

You are not learning nonsense. "Accommodate" has a meaning you will never forget, a sound you hear every week, and a story you can attach to it. All of that slows the fall. You will not lose a real spelling as fast as Ebbinghaus lost "zol".

But notice what you actually forget. Nobody forgets what "accommodate" means. What slips is one small, arbitrary detail: whether the m is doubled. The meaning is anchored in a hundred other things you know, while that second m is anchored in almost nothing, so it behaves much more like a nonsense syllable than the rest of the word does. The hard part of a hard word sits on the steep part of the curve.

There is a second trap. When you see "accomodate" written down, it may look faintly wrong, and that feeling fools you into thinking you know the word. Recognizing a spelling and producing one are different acts. Recognition survives far longer than recall, which is why you can be sure you know a word and still type it wrong.

Why cramming wins on Friday and loses by Monday

Cramming is the art of arranging to be at the top of the curve at the moment of the test. It works. That is the problem with it.

A spelling test on Friday morning measures how high the curve is at nine o'clock. It does not measure how steep it is, so 20 words studied hard on Thursday night can all come out right on Friday and start to slide by lunchtime. By Monday, the words that have not been used since are on the long slope, and they show up misspelled in the next piece of writing. Parents see this constantly, and we cover it from their side in why children spell words right on the test and wrong in writing.

The research here is unusually solid. In 2006, Nicholas Cepeda, Harold Pashler and their colleagues pulled together hundreds of experiments that compared the same amount of study packed into one session with the same amount spread across several. Spreading won, across ages, materials and decades of lab work. Crucially, the comparison was always the same total time. Spacing is not a way of doing more work. It is a way of getting more from the work you do.

What bends the curve

Here is the finding that changes how you should study. Every time you successfully pull a memory back out, the curve that follows is flatter than the one before. The first time, a new spelling might start to blur within a day. After one good recall, it may last several days. After a few more, weeks and then months.

The key word is recall, which means producing the word from memory rather than looking at it again. In a well-known 2006 study, Henry Roediger and Jeffrey Karpicke had students either reread a short text or test themselves on it. Five minutes later, the rereaders did slightly better. A week later, the students who had tested themselves remembered far more. The students themselves predicted the opposite result, because rereading feels fluent and testing feels like failing.

The psychologist Robert Bjork has a name for this kind of useful struggle: a desirable difficulty. Recall that takes some effort does more for a memory than recall that is easy, and that is why timing matters so much. Test yourself two minutes after studying and the word is still sitting in front of your mind; you learn almost nothing. Wait a month and it may be gone, so you are simply learning it again from scratch. The sweet spot is the moment when the word has started to fade but you can still reach it with a bit of effort.

How long the gaps should be

If each recall flattens the curve, the gaps between recalls should grow. That is the idea behind every expanding schedule, from index-card systems to modern software.

The best evidence on how long to wait comes from a 2008 study by Cepeda and colleagues. More than a thousand people learned obscure facts, reviewed them after a gap of anything from minutes to several months, and were tested up to a year later. The ideal gap depended on how long the memory had to last. For a test a week away, a review after a day or so worked best. For a test a year away, a gap of several weeks did better. Put simply, the longer you want to remember something, the longer you can afford to wait between reviews.

For spelling, where you want the word for life, a rough rhythm of one day, then about three, then a week, two weeks, a month, and then every few months is a sensible starting point. A word you get wrong goes back to the start. Modern scheduling programs refine this by tracking every word you practice and estimating the moment you are about to forget each one, which is exactly what Spelling.School does behind the scenes. If you are curious about the machinery, see how an algorithm decides when you see a word again.

One word, two weeks

Here is what the curve looks like when you work with it. Take "accommodate" and give it the treatment.

Two things stand out. The whole thing costs you less than three minutes spread over two weeks, far less than writing the word out 20 times on the first night. And the mistake on day eight is not a setback. A recall that fails, followed by the correct answer, tells your memory exactly where the weak point is, and that word will now be checked sooner until it holds.

Multiply that by 10 or 20 words at a time and you have a routine that fits in a few minutes a day. The daily habit is what matters, because it means each word gets checked on the day it is due rather than on the day you happen to feel like studying.

Forgetting is not a character flaw

People who struggle with spelling often decide they have a bad memory. Almost always, they have an ordinary memory being used in a way that guarantees forgetting: learn once, never retrieve, and wonder where it went.

Forgetting is not a failure of the system. It is the system working. A brain that kept every detail it met at full strength would drown in them, so it lets go of whatever it has not needed lately. Robert and Elizabeth Bjork describe memory as having two separate strengths: how well something is stored, and how easily you can get at it right now. A word can be well stored and hard to reach, which is why a spelling you "lost" often comes back quickly, just as Ebbinghaus found when his relearning went faster than his learning.

That should be encouraging. The words you keep misspelling are not proof of a defect. They are words that were never retrieved at the right moments, and that can be fixed.

What to do next

Choose five words you have misspelled in the past week, the real ones from your own emails and messages rather than a list from the internet. Look at each one closely today and find the exact letter that trips you. Tomorrow, before you look at them again, have someone read them to you, or record yourself saying them, and write them from memory. Do it again in three days and again in a week. Every miss goes back to tomorrow.

That is the forgetting curve put to work. If you would rather have the words read aloud and the schedule kept for you, you can try five words free on the Spelling.School home page, no account needed. For more of the science behind spacing, read how spaced repetition boosts memory and why it works.

Questions people ask

How fast do we forget new spellings?

Most forgetting happens within the first day. In Ebbinghaus's classic experiments, he had lost roughly two thirds of the benefit of learning a list after one day, and forgetting slowed sharply after that. Real words fade more slowly than the nonsense syllables he used, but the pattern is the same, so the first day or two after learning a spelling is when it is most at risk.

Why does cramming for a spelling test not last?

Cramming puts a word at the peak of the forgetting curve at the moment of the test, but it does nothing to make the curve less steep afterward. The word was fresh, not learned. Spreading the same study time over several days, with the word recalled from memory at each session, produces spellings that last for weeks and months instead of hours.

Does forgetting mean I have a bad memory?

No. Forgetting is how normal memory works: the brain lets go of details it has not needed recently, and the letters inside a word are exactly that kind of detail. People who think they have a bad memory for spelling have usually learned words once and never practiced recalling them. Spaced recall fixes that for almost everyone.

How many times do you need to see a word to remember its spelling?

Seeing a word is not what makes it stick; recalling it from memory is. There is no fixed number, but most people find that a word they have spelled correctly from memory four to six times, with the gaps growing from a day to a few weeks, stays spelled correctly for months. A word that needs more tries simply gets more, and a miss is useful because it shows which letter is weak.

What is the best gap between practice sessions?

The best gap depends on how long you need to remember the word. A large 2008 study by Cepeda and colleagues found that the ideal gap grows with the length of time you want to remember something. For spelling you want to keep for life, a good pattern is to review after one day, then about three days, then a week, two weeks and a month, and to start again whenever you get a word wrong.

Fix the words you actually get wrong

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