The Science of Spaced Repetition: Why Forgetting Is Part of Learning

In 1885, the German psychologist Hermann Ebbinghaus ran a meticulous series of experiments on his own memory — memorising lists of nonsense syllables and then testing his recall at intervals over days and weeks. What he found has been replicated so consistently in the century since that it is now simply called the Forgetting Curve.

The curve is steep and fast. Without review, we forget approximately 70% of newly learned information within 24 hours. Within a week, the figure approaches 90%. The information isn't gone — it's in memory somewhere — but it is no longer readily retrievable.

This has obvious implications for language learning. Vocabulary memorised today and never reviewed again is, for practical purposes, lost within days.

Ebbinghaus also discovered the remedy. Reviewing material at spaced intervals — not immediately after learning it, but just before you would otherwise forget it — dramatically slows the forgetting curve. Each successful retrieval strengthens the memory trace and pushes the next review further into the future.

This is the spacing effect, and it is one of the most robustly replicated findings in cognitive psychology. Spaced practice outperforms massed practice (cramming) not just in short-term retention but in long-term retention — the kind of memory that is still there weeks and months later.

The elegant paradox at the heart of spaced repetition is that forgetting is not the enemy of memory. It is the mechanism through which memory becomes durable. The act of retrieving a memory — of pulling it back from near-forgetting — is what strengthens it. A word you always recall easily is not being meaningfully reinforced. A word you almost forget and then retrieve is becoming deeply encoded.

Spaced repetition systems (SRS) — most famously Anki — operationalise this principle through an algorithm that schedules reviews at mathematically optimal intervals based on your past performance.

When you review a flashcard and rate how difficult it was, the algorithm uses that information to calculate when you should see that card again. Easy cards get pushed far into the future — perhaps months. Hard cards come back tomorrow, or even later the same day. The system is continuously adjusting to your actual memory rather than following a fixed schedule.

The latest version of Anki uses an algorithm called FSRS — the Free Spaced Repetition Scheduler — that improves on the older SM-2 algorithm in ways that reduce review burden by 30 to 50% for the same retention rate. That's a significant efficiency gain: the same memory with fewer hours of review.

Spaced repetition is not a complete language learning method. It is an exceptionally efficient memory tool — but what you put into it matters as much as the system itself.

The classic failure mode is learning isolated vocabulary words out of context. You can successfully drill 5,000 Mandarin words with Anki and still be unable to construct a natural sentence in Mandarin, because recognition and production are different cognitive skills, and because words learned in isolation don't come with the pragmatic and collocational knowledge you need to actually use them.

The most effective use of spaced repetition is with sentence cards — full sentences from content you've actually encountered, with the target word in context. This connects vocabulary to meaning, grammar, and usage rather than just to a translation.

The most common mistake with SRS tools is creating cards faster than you review them. Reviews accumulate, the deck becomes overwhelming, and the whole system collapses into guilt and avoidance. This is a set-up problem, not a motivation problem.

A sustainable practice: add no more than ten to twenty new cards per day, and keep daily reviews to under twenty minutes. The goal is a habit you can maintain indefinitely, not a sprint that burns you out in a month.

• Use sentence cards, not isolated word cards
• Add cards from content you're already consuming — videos, articles, podcasts
• Cap new cards at 15–20 per day to keep reviews manageable
• Review every day — even five minutes beats skipping entirely
• Don't add cards you don't actually want to remember — quality beats quantity

The magic of spaced repetition is not visible in the short term. After a month, you will have a few hundred cards you reliably recall. After six months, a few thousand. After two years, vocabulary that would take a non-SRS learner a decade to acquire.

It is not dramatic. It does not feel like learning. It feels like a small, boring habit that one day produces the surprising experience of understanding something you didn't know you knew — a word appearing in a podcast, a phrase making sense in a film — and realising that it was waiting for you all along.