21 topics
Spaced repetition, set up so you do not abandon it
A review scheduled the day before you would have forgotten a word is worth far more than one scheduled a day after, and almost all of spaced repetition is that single idea implemented in software. Each successful review pushes the next one further out — three days, then a week, then three weeks — because the memory is measurably more durable each time. Everything that goes wrong with these systems goes wrong in the same three places: how many cards you add, how the cards are written, and what happens when you fail one.
Where students get stuck
I have eight hundred reviews due and I have stopped opening it
Review load is not something you choose daily, it is set weeks earlier by how many new cards you added. As a rough rule your steady daily reviews land around ten times your daily new cards, so twenty new a day becomes two hundred reviews a day once the deck matures. Digging out works in one order: set new cards to zero, do not delete anything, and clear the backlog oldest first over a week or two. Then restart new cards at a number you would still do on a bad day. Five a day forever beats fifty a day for a fortnight.
There are cards I have failed twenty times
Those are leeches, and the card is at fault rather than your memory. Almost always it is one of three faults: two meanings crammed onto one card, a word with no context so nothing cues it, or a near-synonym pair that cannot be told apart from the prompt alone. Delete or suspend it without guilt and rewrite it as a cloze sentence you actually encountered, with the target word blanked. A card that has failed twenty times will fail a twenty-first, and each attempt costs you review time that a working card could have used.
I know the card but the word never comes to me in conversation
Recognition and production are different memories, and a word-to-translation card only trains one of them. Worse, the card itself becomes the cue — you know the answer because you recognise the card, not the word. The repair is cloze deletion on a sentence you met in real reading, with the word removed and the rest of the sentence as context. That forces retrieval from meaning rather than from a familiar prompt, and it carries the grammar and collocation along with it instead of stranding the word alone.
Why did the interval jump from three days to three weeks?
Each card carries a multiplier that grows when you pass and shrinks when you fail, and the next interval is roughly the last one times that multiplier — commonly around two and a half. So four days becomes ten, ten becomes twenty-five, and the growth looks alarming while being exactly the point. Answering easy rather than good on a card you actually found effortful inflates that multiplier and the card disappears for months. Use the grading honestly: good means you retrieved it with effort, which is the response that builds memory.
Should I aim to remember everything I have added?
No, and chasing that is the most common way people bury themselves. Retention is a dial: pushing your target from around ninety percent up to ninety-eight shortens every interval and can multiply your daily reviews several times over for a few extra words remembered. Ninety percent means one card in ten fails, and those failures are the system working — a lapsed card resets and comes back sooner. If you are failing almost nothing, your intervals are too short and you are paying for reviews you did not need.
What's covered
Spaced Repetition topics you can work through with a tutor, generate practice on, or turn into flashcards and a study plan.
Setting up a deck
- Choosing a daily new-card limit you can sustain
- Premade decks versus cards you made yourself
- One deck per language, and what to keep out of it
- Deciding what deserves a card at all
Card design
- One fact per card and why two answers break it
- Cloze deletion from a sentence you actually met
- Recognition cards versus production cards
- Adding audio, images and example context
- Handling near-synonyms and word families
Scheduling mechanics
- Intervals, ease multipliers and interval growth
- Learning steps before a card graduates
- Lapses, relearning steps and leech thresholds
- Choosing a retention target and what it costs
Feeding the deck from real input
- Sentence mining from reading and subtitles
- The one-unknown-word rule for mined sentences
- Adding words from a conversation you had
- Pruning cards you no longer need
Staying with it
- Recovering from a backlog without deleting the deck
- Reviewing in short sessions across the day
- Pausing new cards during exam weeks
- Knowing when a word has graduated out of the system
Spaced Repetition questions
Do flashcards actually teach you a language?
They maintain vocabulary you have already met, efficiently, and that is all. Comprehension and speaking come from reading and conversation. Treated as the whole method they produce someone who knows a lot of words and cannot use them.
Should I use a premade deck or make my own?
Cards you made from something you actually read are remembered far better, because the context is yours. A premade deck is reasonable for a fixed list like exam vocabulary, and painful as a general approach.
Can it make the cards for me?
It generates flashcards from your own coursework, and you can turn a passage you were reading into cards with the sentence context kept. Reviewing them is still the part that does the work, and you should cut the ones that do not earn their place.
Is twenty minutes a day enough?
Yes, if your new-card rate is set so that twenty minutes actually clears the queue. The failure mode is adding at a rate that needs forty and then missing days, which compounds into a backlog you avoid.
Stuck on spaced repetition right now?
Talk it through out loud, share your screen, and watch it worked out step by step on a whiteboard.
Start free — no card