Spaced Repetition: The Science Behind Never Forgetting Vocabulary

Spaced repetition is the most evidence-backed vocabulary learning technique that exists. It works by exploiting the brain's memory consolidation process — reviewing information at precisely the moment before you're about to forget it, which forces deep encoding rather than surface recall. Done correctly, it's possible to permanently retain thousands of words in a fraction of the study time that rote repetition requires.

Most language learners have heard of spaced repetition. Far fewer understand the actual science behind why it works — or why their current implementation of it might be producing far less than the technique is capable of delivering.

The Ebbinghaus Forgetting Curve

In 1885, German psychologist Hermann Ebbinghaus published the results of years of self-experimentation on memory. By memorizing lists of nonsense syllables and testing his recall at intervals, he produced what is now called the forgetting curve: a mathematical description of how memory decays over time.

The curve has a specific shape: steep and fast. After learning something new, you retain roughly:

• ~58% after 20 minutes
• ~44% after 1 hour
• ~36% after 9 hours
• ~26% after 1 day
• ~23% after 2 days
• ~21% after 6 days
• ~18% after 1 month

If you study vocabulary once and never review it, you will have forgotten roughly three-quarters of it within a week. This is not a failure of intelligence or effort. It is how human memory is designed.

How Spaced Repetition Defeats the Forgetting Curve

Ebbinghaus also discovered something else: each time you successfully recall information just before forgetting it, two things happen. First, you re-encode the information — the memory trace is refreshed. Second, the new forgetting curve is shallower than the previous one. Each successful recall increases the stability of the memory and extends the time before the next review is needed.

This is the core mechanism of spaced repetition. Instead of reviewing vocabulary at fixed intervals (every day, every week), you review each item at the precise moment its memory stability is about to drop below the recall threshold — typically defined as an 80–90% probability of remembering it.

The practical result: after enough reviews, words require review intervals measured not in days but in months, then years. Words you have reviewed six or seven times with a good spaced repetition system will remain accessible essentially indefinitely with minimal maintenance.

The Algorithms Behind Modern SRS

The first systematic implementation of spaced repetition as a learning tool was the SM-2 algorithm, developed by Piotr Woźniak in 1987. SM-2 calculates each item's next review date based on two variables: the interval since the last review and the ease of the previous recall (how hard or easy the item was to remember).

Modern systems use more sophisticated descendants of SM-2. The FSRS algorithm, now used in Anki and other platforms, incorporates research from cognitive psychology to predict forgetting more accurately by also modeling individual learner memory profiles, not just item difficulty.

Fluentera's flashcard system builds on these principles — tracking your performance on each vocabulary item and surfacing it for review at the optimal moment, integrated into story-based lessons rather than isolated drill sessions.

Optimal Review Intervals: What the Research Shows

For new vocabulary, research suggests initial review intervals of roughly:

• First review: Same day or 1 day after initial learning
• Second review: 3–5 days after
• Third review: 10–14 days after
• Fourth review: 3–4 weeks after
• Fifth review: 2–3 months after

After five well-timed reviews, a vocabulary item typically enters long-term storage with review intervals of 6 months to a year. This is what "permanent retention" looks like in practice — not truly permanent, but stable enough to require only occasional maintenance.

Why Most Learners' SRS Practice Underperforms

Reviewing too soon (over-reviewing)

Many learners review their flashcards every day, even items they got right easily. This feels productive but violates the core principle: you want to review at the edge of forgetting, not comfortably within the zone of easy recall. Reviewing too soon wastes time and, critically, produces shallower encoding than a correctly-timed review does.

Using recognition instead of recall

Seeing a word and knowing what it means (recognition) is far easier than producing it from scratch when needed (recall). Flashcards that only test in one direction — foreign word to native meaning — build recognition vocabulary. For active speaking ability, you need recall practice: native meaning to foreign word. Both directions are valuable, but recall is harder and more useful.

Learning isolated words without context

Vocabulary learned in context — in a sentence, in a story, in a situation — is retained significantly better than vocabulary learned as a bare translation pair. The context creates additional retrieval cues: if you can't remember the word, you might remember the story or sentence it came from. Fluentera always introduces vocabulary in context precisely because isolated word-pair learning is the weakest form of vocabulary acquisition.

Not being honest on ratings

SRS systems ask you to rate how difficult a recall was. Learners who consistently over-rate their recall (rating "easy" items as "hard" to get more repetitions, or rating hard items as "good" to avoid revisiting them) corrupt their own algorithm. The system can only optimize intervals for items it has accurate data on.

Practical Setup: Getting Started With SRS

The best SRS system is the one you actually use consistently. Research comparing systems suggests that daily usage matters far more than which algorithm the software uses. Pick a platform and commit to a daily session, even a short one — 10–15 minutes of focused review is enough to maintain 200–400 active vocabulary items.

For new vocabulary items, aim to add no more than 10–20 new words per day to prevent your review queue from growing faster than you can manage it. The most common SRS mistake is adding too many new cards at once, generating a review backlog that becomes discouraging and then abandoned.

Frequently Asked Questions

How many words can I realistically retain using spaced repetition?

With consistent 10–15 minute daily sessions, most learners can maintain between 2,000 and 5,000 active vocabulary items depending on their review efficiency and how many new words they add per day. 2,000 words represents roughly the vocabulary of a lower-intermediate learner; 5,000 covers most common situations. Near-native fluency typically requires passive recognition of 10,000+ words, but active use of 5,000 gets you very far.

Is spaced repetition useful for grammar, or only vocabulary?

Primarily vocabulary, but grammar points can be reviewed as SRS cards if they are specific and testable. For example: "What is the past tense conjugation of 'aller' for 'je'?" works well as an SRS card. Abstract grammar rules ("how does the subjunctive work?") are better learned through exposure and practice than through flashcard review.

Does spaced repetition work for characters (Japanese kanji, Chinese hanzi, Arabic script)?

Yes — and it is particularly well-suited to character learning because the volume is high and the forgetting curve is steep. Learning 2,000 kanji for basic Japanese literacy is the kind of task that benefits enormously from algorithmic spacing; the difference in retention between SRS and rote review for this task has been measured and is substantial.

Should I build my own card decks or use pre-built ones?

Both have value. Pre-built decks (Anki has thousands for major languages) get you started immediately and are useful for core vocabulary. Building your own cards from vocabulary you encounter while reading, listening, or speaking is more effective for retention because the personal context of where you found the word becomes an additional memory cue. Ideal practice combines both.