Spaced repetition is often described as an app that shows a flashcard just before you forget it. That description is attractive—and incomplete.
The real learning mechanism is not the notification or the algorithm. It is the combination of spacing and successful, effortful retrieval. You meet information after a gap, try to recall it, receive feedback, and meet it again after another gap. The schedule creates an opportunity; retrieval does the memory work.
That distinction explains why some learners review thousands of cards yet freeze in conversation. They have optimised card recognition rather than building flexible access to meaning, sound and use.
What research actually supports
Research on second-language vocabulary generally supports spaced encounters and retrieval practice over massed repetition. Reviews also warn that there is no single scientifically perfect interval for every learner and every word.
Three findings matter for a practical system:
- Recalling beats rereading. Trying to produce an answer strengthens access more than repeatedly looking at the answer.
- A gap makes retrieval useful. If the item is still fully active in working memory, another repetition may feel fluent without creating durable learning.
- Difficulty must remain achievable. Retrieval that is challenging and successful is valuable. Repeated total failure means the card, cue or interval needs repair.
- frequent in the situations you care about;
- personally useful;
- repeatedly encountered but not retained;
- easy to confuse with another form;
- important for an exam or course goal;
- part of a reusable phrase or grammar pattern.
The goal is therefore not “make every review hard.” It is “make recall effortful enough to strengthen memory, with feedback before errors become stable.”
A starting schedule—not a law
For a new, useful word or sentence, this sequence is a reasonable starting point:
| Review | Approximate gap | What to do |
|---|---|---|
| First recall | 10 minutes | Recall meaning and say the item aloud |
| Second recall | 1 day | Produce it from a clear cue |
| Third recall | 3 days | Use it in a new sentence |
| Fourth recall | 7 days | Recall from audio or a situation |
| Fifth recall | 14 days | Mix it with similar items |
| Sixth recall | 30 days | Test delayed, independent use |
Treat these gaps as a starting scaffold. Easy cognates may move faster. Abstract words, unfamiliar sounds and easily confused forms may need shorter early gaps or better cues.
An adaptive SRS can manage intervals, but a simple calendar or spreadsheet can also work. The quality of each retrieval matters more than the brand of software.
Build four kinds of memory
A single “word = translation” card trains a narrow relationship. Useful language requires several access routes.
Meaning recognition
See or hear the target-language item and understand it.
> Front: escuchar > Back: to listen; example and audio
Form recall
See a meaning or situation and produce the target-language form.
> Front: “to listen” in Spanish > Back: escuchar
This direction is harder and closer to speaking or writing.
Context selection
Complete a sentence where meaning and grammar determine the answer.
> Quiero ___ música. > escuchar
Personal production
Answer a question using the item in a true sentence.
> ¿Qué música escuchas cuando estudias?
You do not need four cards for every word. High-value items deserve several retrieval routes; low-value words may need only recognition until they become relevant.
The 20-minute daily system
Minutes 1–10: due reviews
Answer before revealing. Say the response aloud when pronunciation matters. Mark honest failures rather than rewarding vague familiarity.
Minutes 11–14: repair difficult cards
Do not keep failing the same card unchanged. Add a clearer context, shorter answer, useful contrast or audio cue. If two items interfere, place them on a comparison card.
Minutes 15–17: add a small number of new items
Five well-chosen expressions are often more useful than twenty random words. Select language you encountered in a lesson, podcast, conversation or reading.
Minutes 18–20: free retrieval
Close the deck and speak or write using several reviewed items. This step checks whether knowledge survives outside the original card.
What deserves a card
Create a card when an item is:
Avoid turning every unknown word into a permanent obligation. Reading and listening will expose you to thousands of low-frequency items. Save deliberate review for language with a realistic future use.
Learn chunks, not only isolated words
An isolated word does not reveal its common partners, grammar or register. Compare memorising “decision” with learning:
- make a decision;
- reach a decision;
- a difficult decision;
- I haven't decided yet.
For languages with grammatical gender, case or particles, include the information needed to use the noun:
- German: die Entscheidung, plural die Entscheidungen;
- Japanese: a short sentence showing the natural particle;
- French: the article and a frequent verb or adjective;
- Korean: the word inside a complete ending pattern.
One compact, natural sentence often teaches more than several disconnected fields.
Use audio without making the answer obvious
Audio supports sound recognition and pronunciation, but choose the direction carefully.
- Audio on the front tests listening comprehension.
- Audio on the back gives feedback after recall.
- Recording yourself tests production and exposes hesitation.
Combine SRS with Langmitra's shadowing-method guide to move from word retrieval toward fluent phrases. Podcast lessons and graded input provide the encounters that make flashcards meaningful.
Prevent the review avalanche
Most SRS failures are workload failures. Every new card creates future reviews.
Use these controls:
- stop adding cards when the daily queue exceeds your available time;
- suspend obsolete, trivial or badly written cards;
- cap new items before reducing due reviews;
- keep one main deck per active language rather than many abandoned decks;
- treat repeated failure as a design problem, not a moral failure.
A learner adding 30 cards per day can create hundreds of weekly review decisions. A sustainable five-card habit produces slower-looking but more durable progress.
What to do with “leeches”
A leech is an item that repeatedly fails. Before forcing another repetition, diagnose it:
| Problem | Better repair |
|---|---|
| Cue is vague | Add a specific situation or sentence |
| Two forms compete | Make a contrast card explaining the choice |
| Answer is too long | Split it into smaller decisions |
| Word is abstract | Add a personal example or image |
| Sound is unfamiliar | Add slow and normal audio plus shadowing |
| Item is not useful | Suspend or delete it |
Forgetting is information about the learning setup. Use it to improve the card.
Spaced repetition cannot replace input
Flashcards intentionally narrow the task. Real language is variable, fast and contextual. You still need reading and listening to learn:
- which words commonly occur together;
- how meaning changes with context;
- what sounds natural rather than merely grammatical;
- how speakers reduce, connect and emphasise sounds;
- how discourse unfolds beyond one sentence.
A meta-analysis of incidental second-language vocabulary learning found meaningful gains from input while also indicating that incidental exposure alone usually teaches only a fraction of target words. The sensible conclusion is not “input or flashcards.” It is input to encounter language, deliberate retrieval to stabilise it, and output to make it accessible.
Use Langmitra's science-backed language-learning guide and podcast-based learning guide to build the rest of that system.
A weekly health check
Once per week, review four signals:
- Completion: Can you finish due reviews without rushing?
- Delayed recall: Do older items still appear after one to four weeks?
- Transfer: Can you use reviewed language in a fresh sentence?
- Encounter: Are the words appearing in your listening and reading?
- Nakata and Webb, review of adult second-language vocabulary training: https://www.cambridge.org/core/journals/studies-in-second-language-acquisition/article/review-of-laboratory-studies-of-adult-second-language-vocabulary-training/18F0A5D1FFC829CE05931B2EEE83124A
- Koval, lag effect in L2 vocabulary learning: https://www.cambridge.org/core/journals/applied-psycholinguistics/article/abs/testing-the-reminding-account-of-the-lag-effect-in-l2-vocabulary-learning/4D03D1F5169B719D90F14BBE474015F5
- Uchihara and colleagues, meta-analysis of incidental vocabulary learning: https://www.cambridge.org/core/journals/language-teaching/article/how-effective-is-second-language-incidental-vocabulary-learning-a-metaanalysis/E38E3468FD2090B1FA3051051DE8E70C
If card accuracy is high but speaking access is low, add production and context. If accuracy collapses, reduce new items and repair cues. If words never appear outside the deck, choose material closer to your actual goals.
Frequently asked questions
What is the best spaced-repetition interval?
There is no universal interval. Start with short early gaps and expand them after successful recall. Let difficulty, retention goal and performance adjust the schedule.
How many new words should I learn per day?
Choose the number whose future reviews you can sustain. Five to ten useful items is a sensible range for many busy learners, but consistency matters more than a fixed target.
Is recognition enough?
Recognition is useful for reading and listening, but speaking and writing require production practice. Include some meaning-to-language prompts and free output.
Should I use sentences or single words?
Use single words when the mapping is simple and context adds little. Use sentences or chunks for grammar, collocations, register and easily confused meanings.
Can I use spaced repetition for grammar?
Yes. Use minimal pairs and situation-based prompts that require a grammatical choice. Avoid cards that merely ask you to recite a long rule.
The durable system
The strongest spaced-repetition routine is small enough to survive busy weeks and connected to language you actually consume and use. Retrieve first, check feedback, repair weak cues, and test the knowledge outside the deck.