Ask this in a language forum and you start a fight. One camp says delete Anki, read books, and let vocabulary grow as a child’s does, through repeated meetings in real sentences. The other says reading without review is a sieve: you recognize a word, forget it three pages later, and only a scheduled card fixes that.
Each camp argues against a cartoon of the other: joyless robots drilling word lists, or readers drowning in books they barely understand. The useful question is narrower: if you already read a lot, does a spaced repetition habit repay its time?
That is a trade-off, not ideology. Spaced repetition is one of the best-studied findings in memory science and does something reading does not do reliably. Reading does several things spaced repetition cannot. Below: what each does, where each leaves holes, who should keep a deck, who can close it, and a middle path.
What Spaced Repetition Genuinely Does
In 1885 the German psychologist Hermann Ebbinghaus published the first experimental study of memory. He memorized nonsense syllables and tracked his forgetting, producing the forgetting curve: fast loss at first, then slower, roughly exponential decay. He also found that spreading study over time gave far better retention than cramming the same effort. This spacing effect has held across nearly 140 years of replication on free recall, recognition, cued recall and more. Robert Bjork and others formalized it into the modern theory of spaced practice in the 1960s and 1970s.
Spaced repetition software turns that into a schedule:
- Anki is free and open source on desktop and Android, and shows each card just before you would forget it. A correct answer lengthens the next interval; a miss shortens it.
- Schedulers: Anki long used a variant of SM-2, designed for SuperMemo in the late 1980s. In 2023 it added the option of FSRS, tuned on hundreds of millions of real reviews to hit a target retention with fewer cards shown.
- Cost: AnkiMobile for iPhone and iPad is a paid one-time purchase; the desktop apps and AnkiDroid are free.
What you buy is durable recall of items you choose. For eighty words you must know cold (tax vocabulary for an exam, irregular verbs that keep tripping you, medical terms for your job), it is close to the most efficient method ever invented. You also control the deck, so nothing important slips through because you happened not to meet it.
Paul Nation’s four strands of a balanced language course, each getting roughly equal time, are meaning-focused input (reading and listening for content), meaning-focused output (speaking and writing to communicate), language-focused learning (deliberate study of vocabulary, grammar and pronunciation) and fluency development (getting faster with what you know). Spaced repetition serves one strand, language-focused learning, and serves it well. The mistake is expecting it to cover the other three.
What Reading Does That Flashcards Cannot
- Volume. An hour of comfortable reading puts thousands of words in front of you, mostly familiar, a few new, all in living sentences. A heavy Anki session rehearses perhaps a hundred cards. The several thousand words needed to read freely come from exposure counts only reading and listening supply; see how I learned 3,000 words through reading.
- Collocation, register and shades of meaning. You learn that you make a decision but take a photo, whether a word is street slang or courtroom formal, and the shades a two-sided card cannot hold, averaged from twenty contexts. A card says what a word means; a book shows what it does.
- Automaticity. Understanding without translating comes only from fast, meaningful reading, Nation’s fluency strand. A card trains retrieval on prompt in isolation; reading trains instant recognition inside a sentence you care about. The broader case is in why reading is the best way to learn a language.
- Breadth. You cannot card a word you have not met. Reading supplies the candidates, including rare, specific words no premade frequency deck includes. Decks are downstream of input; reading is the input.
The Real Tension: Incidental Versus Deliberate
Learning a word while reading for the story is incidental learning. It works, slowly and unevenly. Estimates vary, but a word commonly takes a dozen or more meaningful encounters to become reliably yours. High-frequency words get there fast; mid- and low-frequency words may take very long, since you rarely meet them. The encounter math is in how to remember the words you read: pure reading retains common words well and lets many less common ones slip.
Spaced repetition is deliberate learning: you schedule encounters with a word you chose instead of waiting for chance. That fixes the needed-but-rare word. It also trains recall under pressure, because a card forces active retrieval while reading mostly trains recognition, and producing a word is harder than recognizing it.
Each leaves a gap:
- Reading alone: strong passive recognition of common words, weak active command of rarer ones, especially for production. You understand far more than you can say.
- Spaced repetition alone: words you recite on cue but cannot use fluently, no help with reading fluency, and a chore that tends to collapse before it teaches enough words to read freely.
The question is which gap matters most for you now.
Who Should Keep Spaced Repetition, and Who Can Drop It
Keep it if:
- You are a beginner. Reading teaches efficiently only once you understand most of the page. A deck of the first thousand or two high-frequency words gets you to that threshold faster. Decks are a ladder to reading, not a permanent home.
- You face an exam with a fixed list or deadline. Deliberate scheduling beats hoping the right words appear in your books in time.
- Your goal is production. Reading is weak for retrieval. Pulling words from your head on demand needs the active recall cards force, especially for words you meet too rarely to overlearn.
You can drop or shrink it if you are past beginner, read a lot at a comfortable difficulty, and want to understand rather than pass a test or speak under pressure. Reading then supplies the volume, context and most of the repetition, and a daily deck of a few hundred cards may redo it less pleasantly. Many advanced learners quietly stop here and lose nothing: their books became their spaced repetition. The logic that makes Duolingo frustrating for advanced learners applies to a giant deck: past a point, isolated drilling stops helping and real input does.
Most people live in the middle, and the answer there is a small, sharp deck fed by reading.
The Practical Middle Path: Read First, Mine Sparingly
- Read first. Reading supplies words, context and comprehensible input, and it is the part you enjoy, which keeps you going for years. In an hour of study, the great majority should be reading. Choose books where you understand most words unaided, the comfortable band in how to read without a dictionary, so only a thin stream of new words needs capturing.
- Mine sparingly. Carding every unknown word is the fastest way to a deck you dread. Capture words that recur, block a sentence you care about, or are clearly common in this kind of writing. Rule of thumb: a word that stopped you twice earns a card. Met once with the sentence still clear, let it go; if reading never brings it back, it did not matter.
- Review only high-value items. The deck catches what reading would lose: rarer words, words you need to produce, words that keep slipping. A lean deck reviewed a few minutes a day gives nearly all the retention of a giant one with a fraction of the misery and little risk of quitting.
- Do not let card-making eat reading time. The deck is the supplement, reading the meal. If the supplement takes longer, cut the supplement. The reading targets in how many words to read in a foreign language are too large to spend half your time on cards.
This approximates Nation’s four strands for one person: reading covers meaning-focused input and, as it speeds up, fluency development; the small deck covers language-focused learning. Output remains, which neither supplies, so you still need to speak and write.
Sentence Cards Versus Word Cards
- Word cards: bare word on one side, meaning on the other. Fast to make and review, and exactly what the “decontextualized” critics mean. Fine for a small set of high-frequency basics with one clear meaning, and efficient for raw coverage, as in a beginner deck.
- Sentence cards: the word inside the sentence where you met it, highlighted or blanked, tested in context. They keep collocation and register, show what the word does, and carry a hook from the story you were reading. They take a little longer to make and review.
For words mined from reading, sentence cards usually win, because they keep the context that makes book learning valuable. A common compromise: word cards for a small high-frequency starter set while racing to the reading threshold, then sentence cards mined from real reading. A sentence card from a novel you love is not dry; it is a piece of that novel, scheduled.
The Burnout Failure Mode
Anki usually fails because people quit, not because it does not work. The shape is predictable: you add cards faster than you review, the daily count climbs, one morning 400 cards are due and you feel dread. You skip a day: 600. You skip a week: a wall of stale reviews that feels like debt collection. You stop, and often blame spaced repetition itself and swear off all review, when the real culprit was a deck that grew faster than you could sustain.
The causes are self-inflicted:
- Adding too many cards, especially every unknown word from reading.
- Cards too hard or ambiguous.
- Retention targets set too high, making intervals short and reviews relentless.
- Treating a missed day as failure instead of a blip.
The defenses:
- Cap new cards per day well below what feels generous; today’s new cards are next month’s reviews.
- Mine sparingly, so the deck stays curated.
- Keep cards clear and answerable.
- After a break, grind the backlog down gradually instead of declaring bankruptcy.
- If the deck makes you want to quit the language, cut the deck, not the language.
A tiny deck you keep beats a huge one abandoned in three weeks, the same lesson that separates extensive from intensive reading: the lighter, sustainable habit wins.
Mining Words From Reading With Lingo7
The friction in read-first, mine-sparingly is the hopping: reading on one device, looking words up, copying them into Anki, building cards, finding context. Each hop risks your reading flow or the deck. Lingo7 puts it in one place.
- Tap an unknown word for an instant translation in context.
- Saved words go into built-in spaced repetition, so mining and review live inside the reading app. No tool switching, no hand-copied sentences.
- Each word arrives attached to the sentence you met it in, the sentence-card advantage, captured automatically.
- Synchronized audio adds the listening strand and reinforces the same words through a second channel.
Lingo7 supports 50 languages, runs on iOS and Android, and is free to start. Per-language guides like learning Spanish show how to start reading from your current level. The aim is to make mining cheap enough that you actually do it.
The Bottom Line
Do you still need Anki if you read a lot? Probably less than you think, and rarely zero.
Spaced repetition drives chosen words into durable recall, exactly what reading does least reliably for rare and production vocabulary. Reading gives volume, context, collocation, breadth and the automaticity of meeting words at speed in sentences you care about. Pure reading leaves gaps in active recall and low-frequency words; pure flashcarding leaves words you can recite but not use, in a habit that tends to collapse early.
Keep a deck if you are a beginner climbing to the reading threshold, have an exam with a fixed list, or aim for production. Shrink it toward nothing if you are past beginner, read a lot and mostly want to understand. Everyone else: read first, mine only high-value words, prefer sentence cards, keep the deck small enough never to burn out, and never let card-making crowd out reading. Anki patches the gaps the engine leaves; reading is the engine.