The Science of Spaced Repetition

In 1885, a German psychologist named Hermann Ebbinghaus sat down and began memorizing lists of nonsense syllables. He then measured how quickly he forgot them. The result was the forgetting curve — the first clear demonstration that memory decays predictably over time, and that the rate of decay slows with each successful recall.

That experiment is the origin of everything we now call spaced repetition. And more than a hundred years later, nobody has found a way to beat it.

What the forgetting curve actually shows

After you learn a new word, you forget most of it within a day. Within a week, you have forgotten nearly all of it — unless you see it again. Each time you successfully retrieve the word from memory, the curve flattens. The word sticks around longer before you need another review.

This is not a linear process. It is exponential. The interval between necessary reviews grows with each success. First you might need to see the word again after ten minutes. Then an hour. Then a day. Then three days. Then a week. Then a month. A single word, reviewed seven or eight times across three months, becomes effectively permanent.

Why this beats every other learning method

Most language apps ignore this entirely. They show you the same ten words every day for a week, even after you have clearly memorized them. That is not practice — that is rehearsal. Your brain stops paying attention because there is no effort required to recall the word.

Spaced repetition flips this. The algorithm waits until the moment you are about to forget the word, and only then shows it to you again. That retrieval effort — the near-failure of almost not remembering — is exactly what strengthens the memory trace.

There is a term for this in cognitive science: desirable difficulty. The harder it is to recall something (up to a point), the stronger the memory becomes when you succeed. Easy recalls build weak memories. Hard recalls build durable ones.

The algorithms behind the method

Modern spaced-repetition software uses one of two main families of algorithms. The first is SM-2, the algorithm behind SuperMemo and Anki. You grade each card from "again" (I forgot) to "easy" (I got it with no effort), and the algorithm schedules the next review accordingly.

The second family is FSRS (Free Spaced Repetition Scheduler), a newer approach that models your memory more precisely using probability. It predicts the exact moment you will hit 90% recall probability and schedules the review for then.

Both work. The important thing is that both are based on your performance, not on a fixed schedule. A word you find easy will drift to the back of the queue. A word that keeps tripping you up will reappear soon.

Why this matters for language learning

Language learners have a specific problem: the sheer volume of vocabulary. To read a newspaper, you need roughly 5,000 words. To read a novel, 10,000. To be conversationally fluent, 3,000 is a reasonable target.

You cannot rehearse 3,000 words daily. You physically do not have the time. The only way to make that volume tractable is with an algorithm that shows you only the words you need to see, on the day you need to see them. That is what spaced repetition does.

If you review 20 new cards per day and spend 10 minutes doing so, you will add roughly 7,000 words to your long-term memory over a year. There is no flashier technique that comes close.

How Lingualite uses spaced repetition

Every flashcard you create in Lingualite enters a spaced-repetition queue. The algorithm tracks which cards you remember easily, which ones you struggle with, and when each one is due for its next review. You never have to think about scheduling. You just open the app, review the cards it surfaces, and mark how confident you felt.

The science is boring. The result is not. Consistent short sessions, spaced correctly, build vocabulary faster than any gamified streak or social feature ever will.