How to Use Spaced Repetition for Language Learning: A Science-Backed Guide (2026)

1. Why Spaced Repetition Works for Vocabulary

Two mechanisms do the heavy lifting. The first is the spacing effect: information reviewed across distributed sessions is retained far better than the same total study time crammed together. The second is retrieval practice, sometimes called the testing effect. Every time you try to recall a word before flipping the card, you are not just checking your memory; you are strengthening it. Roediger and Karpicke (2006) showed that repeated retrieval produces substantially better long-term retention than repeated re-reading, even when the re-readers feel more confident.

• The forgetting curve resets with each review. A timely review near the point of near-forgetting flattens the curve and pushes the next safe review further out.
• Difficulty is the point. Reviews that feel slightly effortful (what Bjork calls "desirable difficulties") encode more durably than easy ones.
• Vocabulary is paired-associate learning, which is exactly the kind of arbitrary form-meaning mapping that spacing benefits most.

2. Setting Up Your Review Schedule

The core idea of a spaced repetition schedule is simple: review a new word after a short delay, and if you recall it correctly, lengthen the next interval. A classic manual progression looks like 1 day, 3 days, 7 days, 16 days, 35 days, and so on, with the interval shrinking again whenever you forget. You can run this by hand with paper cards (the Leitner box system), but it becomes unmanageable once you have more than a few hundred cards, because every card is on its own clock.

• New word: first review within a day, while the trace is still partly intact.
• Correct recall: roughly double the interval (1 to 3 days, then to about a week, then two weeks, then a month).
• Failed recall: reset the card to a short interval and rebuild from there.
• Consistency beats intensity: a 15-minute session every day outperforms a 2-hour session once a week, because the daily session keeps every card near its ideal review point.

3. How Many New Words Per Day Is Optimal

For most learners, 10 to 20 new words per day is the sustainable sweet spot, with 20 to 30 being realistic for full-time students or people in immersion. The number that matters is not how many cards you can add, but how many you can keep reviewing over the following weeks. Every new card you introduce today generates a tail of future reviews, so adding 50 new words a day quickly creates a review backlog that no one finishes.

4. Choosing the Right Cards: Sentence vs Isolated Word

A common debate among language learners is whether to study isolated word pairs ("perro = dog") or full sentence cards ("El perro está durmiendo"). The research leans toward context. Words learned in meaningful sentences benefit from richer encoding and are easier to actually produce in conversation, because you also absorb collocation, grammar, and register. Isolated pairs are faster to create and review, but they often produce words you recognize on a test yet cannot use in speech.

• Use sentence-context cards for most vocabulary. Seeing the target word in a natural sentence cues real usage, not just the dictionary gloss.
• Keep the minimum information principle. One target word per card. A sentence testing five new words at once schedules poorly and fails often.
• Add audio for any spoken language, and especially for tonal languages where the written form does not encode pronunciation.
• Test both directions sparingly. Recognition (target to native) builds reading; production (native to target) builds speaking. Production cards are harder but more useful.

5. Common Mistakes Language Learners Make With SRS

Spaced repetition is forgiving, but a handful of habits reliably wreck results. Most of them come from optimizing the wrong thing, chasing the number of cards added rather than the number of words retained and usable.

• Adding too many new cards. The most common failure mode. A 50-cards-per-day pace feels productive for a week, then collapses under review debt.
• Making cards too complex. Cards crammed with multiple facts, long definitions, or whole paragraphs review slowly and fail unpredictably. Keep one testable point per card.
• Treating recognition as knowing. Flipping a card and thinking "yeah, I knew that" without genuine retrieval defeats the testing effect. Commit to an answer before you flip.
• Skipping days, then cramming. A backlog of overdue cards is demoralizing and breaks the scheduling math. Even a 5-minute daily session keeps the system healthy.
• Studying pre-made decks of words you will never use. Generic frequency lists are fine for core vocabulary, but cards drawn from content you actually read, watch, and care about are remembered better.
• Ignoring the algorithm. Manually re-burying cards or overriding intervals because you "feel" you need them usually makes scheduling worse, not better.

6. Getting Started Today

You can start spaced repetition with nothing but index cards and the Leitner box, and plenty of polyglots did exactly that. But once your deck grows past a few hundred cards, manual scheduling becomes the bottleneck, and so does card creation. This is where software earns its place. Anki is the long-standing power-user choice and now supports FSRS, but it requires manual setup, add-on installation, and you still build every card by hand. That friction is why many learners install it and abandon it within a month.

• Pick a daily new-card cap (start at 10 to 15) and a fixed review time you can keep.
• Generate cards from content you care about, not just a generic list, so engagement stays high.
• Use sentence context with audio for words you want to actually speak.
• Review every day, even briefly, and let the algorithm handle the intervals.