Lingua
If you have ever written a vocabulary list, reviewed it ten times in one evening and found it gone by Friday, you have run the experiment that led to spaced repetition. The problem was never the number of reviews. It was their timing.
The principle: review just before you forget
Memory for a new word decays quickly at first and more slowly after each successful recall. A review right after learning does little, because the word is still fresh. A review after the word is entirely gone does little too, because you are learning it from scratch. The review that helps most comes when the word is hard to recall but still recallable. Each such review makes the memory more durable, so the next review can wait longer: a day, then three, then a week, then a month.
That growing schedule is spaced repetition. The spacing effect behind it is one of the best-documented results in memory research: the Cepeda meta-analysis of over three hundred experiments found spaced study reliably beat massed study, with the ideal gap growing as the desired retention grew.
The algorithm, in plain terms
Modern trainers descend from an algorithm called SM-2, the one the SuperMemo and Anki programs made famous. For each card you say how well you remembered it. The card's interval grows by a factor if you remembered easily, grows less if it was hard, and resets to the beginning if you failed. Words you know well drift out to reviews months apart; words you keep missing stay close.
The practical consequence is efficiency. A deck of a thousand words, maintained daily, takes perhaps ten minutes a day, because on any given day only the handful due for review appear. Rereading the thousand-word list would take an hour and work worse.
Why it beats rereading
Two mechanisms are at work, and both favour the flashcard over the list.
- Retrieval, not recognition. A card asks you to produce the answer before showing it. Looking at a list only asks you to recognise words that are already in front of you, which feels like knowing and is not. The testing effect research by Roediger and Karpicke showed that retrieving beats rereading even when it feels less productive.
- Effort at the right moment. The algorithm places each retrieval where it costs some effort, and effortful recall is the kind that strengthens.
How to use it without drowning
Spaced repetition fails in one predictable way: the deck grows faster than you review it, the backlog becomes a wall, and you stop. A few rules keep it alive.
- Add fewer words than you think. Five to ten new words a day is plenty. Every new word is a commitment to several future reviews.
- Review every day, briefly. The schedule assumes you show up; a week away turns thirty due cards into two hundred.
- Add words with context. A card that shows the sentence the word came from is easier to recall and far more useful than a bare translation.
- Delete without guilt. A word you failed five times and never needed is not a word you need. Remove it.
- Say the answer aloud. Production, not silent recognition, and a little pronunciation practice as a bonus.
The trainer in LinguaPair
The vocabulary section is a spaced-repetition trainer built on the same SM-2 family of scheduling, with two things that make the deck less work to build.
- Words arrive from real use. Words from your translator conversations with your partner are picked out automatically, and words you meet in reading or films go in with one tap on "Learn". You rarely type a card by hand.
- Every card comes with its sentence and its sound. The example, the translation and the pronunciation are attached when the card is made, so the context rule takes care of itself.
- Sessions fit the day. Due words appear in a short session; a daily XP goal and a streak keep the "review every day" rule visible.
A few minutes every day, a few new words at a time, and the words stop leaving.