Spaced repetition for vocabulary, explained

Spaced repetition is one idea: review each thing just before you'd forget it, and the forgetting slows down every time. It is the most reliably effective technique in all of learning research — and vocabulary is the single best use case it has. Here's why it works, in plain words, and why so many people who believe in it still quit.

You don't have a memory problem — you have a scheduling problem

In the 1880s, Hermann Ebbinghaus memorized nonsense syllables and tested himself at intervals, producing the first picture of the forgetting curve: memory doesn't fade linearly, it falls off a cliff. Most of what you learn today is gone within days — unless you meet it again.

That's the entire diagnosis for vocabulary. You look a word up, you understand it, and by next week it's gone — not because you're bad at memorizing, but because nothing brought the word back before the curve hit zero.

Each well-timed review flattens the curve

Two findings, replicated for over a century, explain why the fix works:

• The spacing effect. The same number of reviews produces far stronger memory when they're spread out than when they're crammed together. Cramming feels productive precisely because the word is fresh — and freshness is the sign the review taught you nothing.
• Retrieval practice. Struggling to recall a word strengthens it far more than rereading it. The slight difficulty of "wait, I know this…" is not a failure state; it's the exercise itself.

Put together: recall each word at the moment recall is just barely possible. Every successful recall flattens that word's forgetting curve, so the next review can wait longer — days become weeks become months. A word reviewed six or seven times on that schedule is close to permanent, for a total time cost of a couple of minutes.

Why vocabulary is the perfect use case

Spaced repetition isn't the right tool for everything — you can't flashcard your way to good pronunciation or graceful writing. It shines when the material is thousands of small, independent items, each needing its own schedule. That is a description of vocabulary. Word number 3,041 doesn't care that you know word 3,040; each one fades on its own curve, and only software can track thousands of curves at once.

From SM-2 to FSRS, in plain words

For decades, most flashcard apps ran on SM-2, a 1980s-era formula: answer well and the interval multiplies, answer badly and the card resets. It works, but it's a blunt heuristic — it doesn't model you.

FSRS (Free Spaced Repetition Scheduler) is the modern replacement. It maintains a small memory model of every card — how hard it is for you, how stable it's become, how likely you are to recall it right now — and fits that model to your actual review history. The practical result: fewer reviews for the same retention, because words you know well drift far out instead of returning on a fixed multiplier. It's now the default in serious tools — Anki ships it, and Capecho schedules every review with it, computed server-side so your Mac and iPhone always agree on what's due.

Why people quit — and it's not the reviews

Here's the uncomfortable finding hiding in every abandoned flashcard deck: the reviews were never the problem. A due queue takes minutes a day. People quit spaced repetition for reasons that live outside the algorithm:

• The authoring tax. Every card had to be built by hand — look up, copy, paste, format — so the pipeline of new cards dried up, and the habit starved.
• Deck hoarding. Downloading a 5,000-word shared deck feels like progress, but drilling words you've never met in the wild is the least effective — and least pleasant — spaced repetition there is.
• Review debt. Miss a week and the queue balloons; the guilt does the rest. (A good scheduler handles this gracefully — the words most at risk come first — but the dread is real.)

The fix for all three is the same: make the input side nearly free, and only feed the system words you actually met.

Context makes every review cheaper

One more multiplier. A word drilled as a bare definition gives your memory a single, fragile hook. A word reviewed inside the sentence you met it in carries the topic, the tone, the grammar, and your memory of the moment — richer encoding going in, easier retrieval coming out, which means longer intervals and fewer reviews. Context isn't decoration on top of spaced repetition; it's what makes the schedule efficient.

How to actually stick with it

• Capture at the moment of meeting. If saving a word costs more than a second or two, you'll defer it, and deferred words are lost. This is the step worth automating completely.
• Only your own words. Your reading is the syllabus. Skip the shared mega-decks.
• Small daily sessions. A few cards while the kettle boils beats a Sunday study block; consistency is the whole game, and short sessions are what make consistency survivable.
• Trust the schedule. Don't re-cram words you got wrong; the algorithm already moved them closer. Rate honestly and close the app.

Spaced repetition works. It has worked in every study for a hundred years. The craft is arranging your tools so the system keeps getting fed — capture the words you meet, and let the schedule do the remembering.