What Is Spaced Repetition and Why It Works for Language Learning

Spaced repetition is a study method that schedules each review just before you would forget the material. Instead of drilling a word ten times today, you see it today, tomorrow, in four days, then in two weeks. Each successful recall pushes the next review further out, so old words cost you almost nothing to maintain.

In this guide, you will learn why forgetting is predictable, how SRS algorithms decide when a card comes back, which mistakes quietly break the system, and which tools implement it best.

The forgetting curve: why your brain leaks vocabulary

In 1885, German psychologist Hermann Ebbinghaus memorized lists of nonsense syllables and tested himself at intervals. He found that memory decays fast at first, then more slowly. Within a day, he had lost more than half of what he learned. This pattern, known as the forgetting curve, was successfully replicated in 2015, so it is not a dusty artifact.

For language learners, the curve explains a familiar frustration. You learn 20 words on Monday, feel confident, and blank on most of them by Friday. The effort was real. The timing was wrong.

However, the curve has a useful property: every review flattens it. After each successful recall, the memory decays more slowly than before. Review a word at the right moments and the interval between reviews can stretch from days to months. That is the entire idea behind spaced repetition. You are not fighting forgetting with volume. You are timing your reviews to reshape the curve.

There is a second ingredient: retrieval. A landmark 2008 study by Karpicke and Roediger showed that testing yourself builds far stronger retention than restudying. Consequently, rereading your vocabulary list does little. Pulling the word out of memory, with effort, is what makes it stick. Spaced repetition combines both principles: active recall, spaced over expanding intervals.

How spaced repetition works, step by step

In practice, the method runs on flashcards, paper or digital. Here is the loop:

1. Learn a new item. A word, a phrase, or a grammar pattern in a sentence.
2. Test yourself the next day. See the prompt, retrieve the answer before flipping the card.
3. Grade your recall. Correct and easy, correct but slow, or failed.
4. Let the schedule adjust. Success pushes the next review further out. Failure resets the card to short intervals.
5. Repeat daily. Only the cards due today appear, so daily workload stays manageable.

A typical schedule for a card you keep answering correctly looks like this:

Six short reviews over six months, and the word is close to permanent. Compare that with cramming the same word 15 times in one evening and losing it within a week. The spaced version costs less total time and produces a stronger memory.

Before software, learners ran this with the Leitner system: cardboard boxes where correct cards move to a less frequent box and failed cards move back to box one. It works, but it scales poorly. You need thousands of words to hold a conversation, and nobody wants to hand-sort thousands of cards. That is the problem algorithms solve.

SM-2 vs FSRS: how the algorithms schedule your reviews

Every SRS app answers one question: when should you see this card again? Two algorithms dominate, and understanding them helps you trust (or configure) your tool.

SM-2 dates from the late 1980s and still powers many apps. Each card carries an “ease” number. Answer correctly and the next interval is roughly the previous one multiplied by that ease, typically around 2.5. Fail, and the card restarts from one day. It is simple and effective, but rigid. It reacts to your last answer, not to your whole history, and repeated failures can trap a card in a punishing loop.

FSRS (Free Spaced Repetition Scheduler) is the modern alternative, and it ships as an option inside Anki. Instead of one ease number, it models each memory with three values: how likely you are to recall it now, how slowly it decays, and how hard the item is. Crucially, it trains those parameters on your own review history. In plain words: FSRS learns how your memory behaves and predicts your forgetting, while SM-2 applies the same fixed recipe to everyone. Users switching to FSRS typically see fewer daily reviews for the same retention. You can read how it works on the FSRS project page.

Big commercial apps run their own variants. For example, Duolingo published a half-life regression model that estimates how long each word survives in your memory and schedules practice around that estimate. Different math, same goal: predict forgetting, then intervene just before it happens.

You do not need to understand the formulas. You do need to know one rule: the algorithm only works if your grades are honest. Marking a card “easy” when you hesitated poisons the schedule.

Common mistakes that break spaced repetition

Spaced repetition fails for predictable reasons, and almost none of them are the algorithm’s fault.

Adding too many new cards. Every new card generates future reviews. Add 50 new words a day and within weeks you face 400 daily reviews, and you quit. Start with 10 new cards a day. Increase only when your review queue feels light.

Skipping days. The schedule assumes you show up. Miss a week and 300 overdue cards greet you at once, half of them forgotten. A 10-minute daily session beats a 70-minute Sunday marathon, both for memory and morale.

Vague, context-free cards. A card that says “take = prendre” is weak, because “take” has a dozen meanings. Put the word in a short phrase (“prendre une décision”) or a cloze sentence instead. Retrieval in context is what conversation actually demands.

Ignoring leeches. A leech is a card you keep failing, review after review. Do not keep grinding it. Rewrite it: shorten it, add context, or pair it with the word you confuse it with. If a card fails six times, the card is the problem, not your brain.

Grading dishonestly. Recognizing a word after flipping the card is not recall. If you could not produce the answer, mark it failed. Short-term pain, long-term memory.

Tools that use spaced repetition well

Most serious vocabulary apps implement some form of SRS. As a result, the real differences lie in algorithm quality, content, and how much control you get. Here are the strongest options, as of July 2026:

Anki is the reference implementation and the only mainstream tool that lets you run FSRS on your own material. Clozemaster applies SRS to sentences rather than isolated words, which fixes the context problem automatically. Memrise wraps its repetition in native-speaker video, ideal for beginners who want real audio from day one. Lingvist suits intermediate learners: it feeds you high-frequency words inside sentences, in the order you are most likely to need them. Quizlet and Busuu both embed SRS inside broader study platforms, with less control but zero setup.

For detailed reviews, pricing, and picks per learner profile, see our full guide to the best flashcard apps for language learning.

Make forgetting part of the plan

Spaced repetition works because it accepts a fact most study methods ignore: you will forget, on a predictable curve, unless you retrieve at the right moments. Schedule those retrievals and vocabulary that once leaked away becomes permanent for a few minutes of daily review.

Our recommendation is simple. Pick one tool today: Anki with a pre-made deck if you want power and FSRS, or Clozemaster if you want sentences with zero setup. Then commit to 10 minutes a day for 30 days before judging the method. Consistency, not the algorithm, is where most learners win or lose.

Spaced repetition also pairs well with real input like reading and listening, especially when progress stalls. If that is where you are, our guide on how to overcome a language learning plateau shows where review fits in a broader routine.