SRS Language Learning: The Complete Guide to Spaced Repetition for Vocabulary

SRS language learning is, without exaggeration, the most efficient method for vocabulary acquisition ever validated by cognitive science. A spaced repetition system schedules each word for review at precisely the moment your brain is about to forget it — not too early, not too late — compounding retention across thousands of items without wasted effort. For language learners, that means reaching conversational fluency in a fraction of the time that random review or rereading would require. This makes SRS especially valuable for self-directed students relying on distance learning resources like MOOCs and async language courses, where the built-in retrieval structure of a classroom is missing. Beginner ESL decks typically start with concrete nouns and action verbs — the word classes most reliably retained when paired with imagery.

This guide is not a general overview of flashcard apps. If you want that, our best flashcard app for language learning guide covers the full competitive landscape. Here we go deeper: the algorithm trade-offs that actually matter for vocabulary, the sentence mining technique that dramatically improves card quality, honest vocabulary targets, a daily practice framework designed to last years rather than weeks, and a concrete strategy for avoiding the burnout that derails roughly 80% of SRS learners within their first month.

What Is SRS and Why Language Learners Need It

A spaced repetition system — SRS — is a study method that schedules review sessions based on a model of memory decay. Every time you review a flashcard and judge how well you recalled it, the algorithm updates its prediction of when you will forget that item and schedules the next review accordingly. Cards you recall easily get pushed weeks into the future. Cards you struggle with come back tomorrow.

The contrast with conventional study is stark. Most learners review vocabulary in fixed lists, cycling through the same 30 words regardless of which ones they already know. SRS eliminates that waste entirely. Every review minute is spent at the moment of maximum memory-strengthening value.

Language learning is uniquely well-suited to this approach. Vocabulary acquisition is at its core a memorization task — thousands of arbitrary sound-meaning pairings that must be stored, retrieved, and eventually automated. No amount of grammar study or conversation practice substitutes for the raw word knowledge that SRS builds systematically. Our guide to flashcards for memorizing words explains the card design principles that maximize retention for vocabulary specifically, covering everything from image pairing to context sentences.

The practical result: studies on spaced repetition for language learning consistently show that learners using SRS acquire vocabulary at two to four times the rate of those using massed practice (blocking all study into a single intensive session) — a gap that comes down to why repetition works when it is spaced. The difference is not marginal. It compounds dramatically over months and years.

A full introduction to the mechanics of spaced repetition, including the research on the spacing effect, is in our spaced repetition study techniques guide, and the retention rate explained piece covers why FSRS-tuned vocabulary decks aim for 90% recall rather than 100%. What follows here focuses specifically on applying those mechanics to language learning.

The Science: Spacing Effect, Forgetting Curve, and Vocabulary

The theoretical foundation of SRS dates to Hermann Ebbinghaus, the German psychologist who in 1885 mapped what is now called the forgetting curve: the exponential rate at which memories decay without reinforcement. Ebbinghaus found that roughly 50% of new information is forgotten within an hour, and 70% within 24 hours, without review. The decay then slows, but without intervention the memory approaches zero.

The critical insight from Ebbinghaus — and the one that SRS exploits — is that each successful retrieval resets and extends the forgetting curve. Review a word just before you forget it and the next forgetting curve is shallower and longer. Review it again at the right interval and it extends further still. After four or five well-timed reviews, a word enters long-term storage and may persist for years with only occasional maintenance reviews.

The spacing effect — the finding that distributed practice outperforms massed practice for long-term retention — is one of the most replicated results in cognitive psychology. Our explainer on spaced practice as a study method covers the broader research. For language learners, the practical implication is direct: 15 minutes of SRS vocabulary review per day produces dramatically better retention than a 90-minute session once a week covering the same material.

Vocabulary acquisition has a particular property that makes SRS especially effective: words exist in isolation until they connect to meaning, and that connection is initially fragile. The first few reviews of a new word are doing the work of consolidation — moving the item from short-term to long-term storage. SRS front-loads those early reviews at high frequency (tomorrow, then three days, then a week) precisely when consolidation is most at risk of failure. Later reviews are spaced wider because the memory is stable enough to survive longer gaps.

Research on active recall as a study method adds another layer. The act of attempting to retrieve a word before seeing the answer — even if you fail — produces stronger learning than passive re-exposure. This is the testing effect: retrieval attempts, successful or not, strengthen the neural pathways associated with the target memory. SRS flashcards force retrieval by design. You see the front of the card and must produce the answer before seeing the back. This is categorically different from reading a word list, even a list you review repeatedly.

FSRS vs SM-2: Which Algorithm Learns Your Memory?

All SRS tools are not equal. The scheduling algorithm determines how accurately the system models your individual memory, which directly affects how efficiently you learn. There are two dominant algorithms in consumer SRS tools today: SM-2 and FSRS.

SM-2 (SuperMemo 2) was developed by Piotr Wozniak in 1987 and published as part of the SuperMemo software. It was the first practical algorithm for automated spaced repetition scheduling and remains the most widely deployed SRS algorithm in history. Anki, the most popular free SRS tool, uses SM-2 as its default scheduler. SM-2 works by maintaining an "easiness factor" for each card that adjusts based on your performance ratings. Rate a card as "Again" and the easiness factor drops, meaning the card will be scheduled more frequently. Rate it as "Easy" and the factor increases, pushing the card further out. The next interval is calculated as the previous interval multiplied by the easiness factor.

SM-2 is robust and well-understood. Its two main limitations for language learning are that it treats all users identically (no per-learner memory modeling) and that it struggles with items where memory is genuinely unstable — words that you recognize one day and blank on the next. The algorithm interprets inconsistency as low easiness and schedules the card aggressively, which can create review debt on words that actually need less frequent review than SM-2 thinks.

FSRS (Free Spaced Repetition Scheduler) was published in September 2022 and takes a fundamentally different approach. Rather than using a fixed formula, FSRS uses a machine-learning model trained on large real-world review datasets to build a per-user memory model. It estimates the probability that you will correctly recall a given card on any given day, and schedules the review for the day when that probability drops below a target threshold (default 90%). Controlled studies comparing FSRS to SM-2 on real user data show FSRS achieves approximately 12.6% better scheduling efficiency — meaning you reach the same retention rate with fewer total reviews, or higher retention with the same number of reviews.

Aspect	SM-2	FSRS
Year / origin	1987, SuperMemo (Piotr Wozniak)	2022, open-source ML research
Scheduling approach	Fixed formula with per-card ease factor	ML model — estimates per-user recall probability
Efficiency at scale	Baseline (over-reviews stable vocab)	~12.6% fewer reviews for same retention
Tuning	Manual ease-factor adjustments	Auto-optimized from your own review history
Best for	Small decks (<500 cards); legacy apps	Large decks (2,000+); long-term vocabulary growth

For language learners, the practical difference matters most at scale. If you are maintaining a deck of 3,000–5,000 words, FSRS reduces daily review burden meaningfully by not over-scheduling cards you have consolidated and not under-scheduling cards at risk of dropping out. SM-2 tends to over-review stable vocabulary and under-review vocabulary that sits in an awkward middle zone of partial retention.

Which should you use? If you are starting fresh with a new tool, use FSRS where available. Flashcard Maker uses FSRS scheduling by default. Anki supports FSRS as an opt-in scheduler (replacing the default SM-2). Most other consumer SRS apps either use SM-2 or a proprietary variant. The algorithm matters most for large decks maintained over long periods — at 500 cards or fewer, the practical difference is small. Beyond 2,000 cards, FSRS produces a noticeably lighter review load for the same retention outcome.

Our broader comparison of language flashcard tools covers how each major app implements its scheduling, including audio support and deck structure options that matter for vocabulary work.

Sentence Mining: The i+1 Strategy That Makes SRS Stick

The single most common failure mode in SRS language learning is not algorithmic — it is card quality. Learners build decks of isolated word-translation pairs (“renard = fox”) and wonder why, despite consistent review, the words never feel natural in production. The problem is that word-translation pairs lack the semantic context that makes vocabulary retrievable in real communication.

Sentence mining is the practice of extracting cards from authentic native-language content you encounter during regular study or immersion — articles, novels, podcasts, subtitles, social media. Instead of a word-translation pair, a sentence-mined card looks like this:

Front: “Le renard _____ rapidement à travers les bois.” (The fox _____ quickly through the woods.)

Back: courait (was running) — courir, imperfect 3rd person singular

The sentence provides retrieval cues that the isolated word does not. You are not just trying to match a sound to a meaning — you are retrieving a word that fits a specific syntactic slot in a real sentence you have seen in context. This produces vocabulary knowledge that transfers to comprehension and production, not just recognition in isolation.

The i+1 principle (also called comprehensible input +1) was formalized by linguist Stephen Krashen. The idea is that the optimal input for language acquisition is material where you understand everything except exactly one new element. At i+1, the unknown element is inferrable from context, making it acquirable. Material with too many unknowns (i+5, i+10) becomes noise — you cannot infer meaning and acquisition stalls. Material with no unknowns (i+0) produces no new acquisition.

Applied to sentence mining, i+1 means: mine sentences that contain exactly one unknown word. The surrounding context should be fully comprehensible. This serves two purposes. First, it ensures the sentence provides enough context to make the unknown inferrable. Second, it keeps each card atomic — one unknown per card means each review tests one specific piece of knowledge.

In practice, finding i+1 sentences requires exposure to graded or frequency-ordered content. For absolute beginners, this might mean graded readers or high-frequency word lists. For intermediate learners, it means choosing reading and listening material just above your current level — where most words are known but one in every 20–50 is new. Native-level content like news articles and novels works well once you have a vocabulary base of around 2,000–3,000 words.

The workflow: encounter an unknown word in context, capture the full sentence plus a brief definition or translation, add it to your SRS deck. Review the sentence with the target word cloze-deleted (blanked out). This approach — sometimes called “sentence cards with cloze deletion” — is the format recommended by most serious language learners and supported by the research on contextual vocabulary acquisition.

Pairing sentence cards with imagery also helps. Studies on vocabulary with pictures consistently show that visual association roughly doubles retention for concrete nouns compared to text-translation pairs alone. For abstract vocabulary, example sentences outperform images — a practical reason to use both depending on the word type.

Realistic Vocabulary Targets: From 1,000 to 10,000 Words

One of the most confusing aspects of language learning is the gap between what different fluency claims actually mean in terms of vocabulary size. Here are evidence-based benchmarks:

~1,000 words: Survival level. At 1,000 high-frequency words in a language, you can navigate basic tourist and travel situations — ordering food, asking directions, buying things. You will miss most of what native speakers say in natural conversation. This is achievable in two to three months of consistent SRS practice starting from zero.

2,000–3,000 words: Conversational fluency threshold. Research on vocabulary and text comprehension consistently finds that knowing the 2,000 most frequent words of a language provides approximately 95% coverage of everyday spoken conversation. At 95% coverage, you can follow the gist of most conversations and fill gaps through inference. This is the first milestone where language use begins to feel natural rather than effortful. For most learners, reaching 2,000–3,000 words with solid recall takes 12–18 months of consistent SRS practice.

5,000–8,000 words: Comfortable reading and television comprehension. Written language uses a much wider vocabulary than spoken language. At 5,000 words you can read newspaper articles with occasional dictionary lookups. At 8,000, most popular fiction is comfortable. This range is where sentence mining from native content becomes the primary vocabulary acquisition strategy — you have enough foundation to extract i+1 sentences from authentic material.

8,000–10,000 words: Advanced / near-native comprehension. At 10,000 words, vocabulary is rarely a limiting factor in comprehension of non-specialized content. Native speakers of major languages typically have passive vocabularies of 50,000–100,000 words, so even advanced learners have room to grow. But at 10,000, vocabulary stops being the bottleneck — grammar, pragmatics, and exposure time become the limiting factors instead.

These targets apply to high-frequency vocabulary learned in context. A vocabulary of 2,000 high-frequency words is far more useful than 2,000 randomly selected words. Frequency lists based on large corpora of spoken and written language (like the Routledge Frequency Dictionaries series) give learners the highest return on their SRS investment by ensuring every word learned is actually encountered regularly.

For tools that support structured vocabulary work, our audio flashcard guide covers how to build decks that include native-speaker pronunciation alongside written forms — critical for avoiding the common failure mode of learning words you can read but not recognize when spoken.

Sustainable Daily Practice: Card Counts and Time Budgets

The most common miscalibration in SRS language learning is starting too aggressively. New learners, excited by the efficiency of spaced repetition, add 50–100 new cards per day in the first week. This creates a review debt that compounds rapidly: cards reviewed today come back in 1–3 days, so 100 new cards per day means 100–200 reviews due per day within a week. Daily review sessions balloon to 45–60 minutes. Burnout follows.

The research-backed sustainable range is 10–30 new cards per day. At 15 new cards per day — a figure that produces roughly 5,000 new words per year — the mature review load (cards introduced weeks or months ago that have reached wider spacing intervals) stabilizes at roughly 80–120 reviews per day for most learners. At 15–20 minutes of review time per session, that is a daily commitment most people can sustain indefinitely.

The math: 15 new cards/day × 365 days = 5,475 new words in a year. Combined with vocabulary acquired through immersion (reading, listening), a learner adding 15 cards/day consistently will cross the conversational fluency threshold (2,000–3,000 words with solid recall) in the first 6–12 months and reach intermediate reading level (5,000 words) within two years. These are realistic projections for consistent practice, not optimal-case estimates.

A practical daily structure that works:

• Reviews first. Complete all due reviews before adding new cards. Reviews maintain existing knowledge; new cards expand it. If you skip a day of reviews and only add new cards, debt compounds. If you skip new cards and only review, no debt accumulates. Reviews are non-negotiable; new cards are optional on any given day.
• Cap new cards per session. Most SRS apps let you set a daily new card limit. Set it at 10–20 and do not manually override it. The cap prevents enthusiasm from creating unsustainable debt.
• Track session length, not card count. On days when you have less time, prioritize reviews over new cards. A 10-minute session that clears your review queue is better than a 25-minute session that adds 30 new cards and leaves reviews half-done.
• Take breaks without guilt. Missing two or three days is normal. Returning after a break to a large review queue is not a crisis — increase your daily review limit temporarily to work down the backlog, but do not add new cards until the queue is manageable.

For context on how these principles apply across different subjects, our flashcard study techniques guide covers session structure, card design, and daily habit formation for academic subjects alongside language learning.

Why Most Learners Burn Out on SRS (and How to Avoid It)

Many learners who start an SRS vocabulary routine quit within the first few weeks. The hard truth about spaced repetition for language learning is that the algorithm is rarely the reason people stop — behavior is. Burnout patterns are visible in the review history of SRS apps and reflected in the language learning community's well-documented graveyard of abandoned Anki decks. Understanding why burnout happens is the first step to preventing it.

Card creation friction. The most common early quitter pattern: a learner starts by manually typing cards from a vocabulary list or textbook. The process is slow, tedious, and feels like homework before the actual studying even begins. After a week of spending 20 minutes creating cards and 10 minutes reviewing them, the ratio inverts and the learner quietly stops.

Fix: Minimize card creation time. Use pre-made frequency-list decks for the first 1,000–2,000 words (the AnkiWeb shared library has high-quality decks for every major language). When you transition to sentence mining, capture cards directly from content you are already reading rather than creating them from scratch. Flashcard Maker is built around this principle — you highlight a sentence on any webpage, right-click, and the card is captured in under two seconds without leaving the page. The friction of card creation is the single largest predictor of early dropout, and any reduction in it meaningfully extends practice streaks.

Review debt accumulation. Adding too many new cards early creates a review queue that grows faster than it can be cleared. Every missed day makes the backlog larger. Within two weeks, opening the app produces anxiety rather than satisfaction. The learner starts avoiding the app altogether.

Fix: Set a hard daily new-card limit of 15–20 and do not raise it for at least the first two months. If you miss three or more days, temporarily pause new cards and focus only on clearing the review backlog before resuming your normal pace.

Monotony and lack of connection to real language use. SRS review in isolation feels abstract. Reviewing 100 decontextualized flashcards per day with no immersion — no reading, no listening, no speaking — produces vocabulary that exists on cards but not in your active processing. The words never feel real, motivation drops, and the practice starts to feel pointless.

Fix: SRS should never be your only language contact. Pair it with 20–30 minutes of native content immersion (reading, listening, watching) in your target language daily. The immersion reinforces the words you are building in SRS and provides new i+1 sentences for card mining. Each reinforcement outside SRS makes the next SRS review faster and more satisfying.

Card quality debt. Poorly designed cards — too broad, ambiguous, or disconnected from real usage — produce inconsistent results. You get the card right when you remember it was a fox and wrong when you do not, but the recall isn't building genuine knowledge. Reviewing bad cards feels frustrating and unrewarding.

Fix: Delete or edit cards aggressively. If a card has been marked "Again" more than five times without improvement, rewrite it. Add context (a sentence, an image, a mnemonic). Or delete it and replace it with a better sentence-mined card. A small deck of well-crafted cards produces better outcomes and higher motivation than a large deck of mediocre ones. See our guide on what makes a good Anki deck for card design principles that apply across all SRS tools.

No clear milestone visibility. SRS can feel like an infinite treadmill with no signal of progress. You review cards every day, the queue never reaches zero, and you cannot tell whether you are actually learning the language.

Fix: Track vocabulary milestones rather than daily reviews. Note when you reach 500, 1,000, 2,000 words known. Cross-reference against comprehension: when you start to catch words in a TV show you could not follow three months ago, that is real progress. Milestone visibility makes the daily practice feel purposeful rather than mechanical.

Building an SRS Workflow With Flashcard Maker

Flashcard Maker is a Chrome desktop extension built around reducing the two friction points that kill most SRS habits: card creation time and the context switch away from your reading flow.

The core workflow is simple. While reading any article, news site, or online text in your target language, highlight the sentence or phrase containing the unfamiliar word. Right-click and choose “Create flashcard (as question)” or “Create flashcard (as answer)” from the context menu. The selected text is captured into a card without leaving the page. You can add a translation or note, assign it to a language-specific deck, and return to reading in under five seconds.

For review, Flashcard Maker uses FSRS scheduling — the same modern algorithm discussed earlier in this guide. After viewing each card, you rate your recall as Again, Hard, Good, or Easy. The scheduler updates the card's next review date based on those ratings and your personal memory curve, not a fixed formula. For learners maintaining large decks, this reduces unnecessary over-review of words you have solidly acquired.

All cards are stored locally in your browser via IndexedDB. No account is required, and nothing is sent to any server. For learners working with sensitive material — professional documents, specialized academic content — the local-first architecture means your study data stays entirely under your control.

Decks in Flashcard Maker can be organized however you prefer — by frequency tier, by topic, by source content, or by the date you started mining. Import existing vocabulary lists in Quizlet TSV or CSV format to seed your decks without starting from scratch. When you want to move your cards to another tool, export your deck as a Quizlet-ready TSV file for use in any compatible import workflow.

For learners using Anki on iPad or mobile devices for review on the go, Flashcard Maker serves as an efficient capture tool during desktop reading sessions. Capture cards while reading on your computer, then export the TSV for use in a complementary tool on mobile. The two tools cover different parts of the workflow without competing: Flashcard Maker handles frictionless web capture and FSRS-scheduled desktop review; a mobile SRS app handles review during commutes and spare moments.

A practical setup for sentence mining with Flashcard Maker:

1. Choose a reading source in your target language slightly above your current level (a news site, a graded reader site, a language-learning blog in the target language).
2. Read for comprehension. When you encounter a word you do not know, check if the sentence is otherwise fully comprehensible — if so, it is an i+1 candidate.
3. Highlight the sentence. Right-click → “Create flashcard (as question)” with the target word replaced by a blank, or capture the full sentence and edit it in the card panel.
4. Add the translation or definition as the answer. Optionally add a mnemonic or related image.
5. Continue reading. Do not interrupt your flow to review immediately — let the SRS scheduler surface the card at the right time.
6. Complete your daily review queue in the Chrome side panel each morning or evening. Rate each card honestly — the FSRS scheduler's accuracy depends on honest recall ratings, not optimistic ones.

At 15–20 new cards mined per day from authentic content, you will build a vocabulary base grounded in real usage rather than textbook abstractions. Combined with regular immersion in your target language, this workflow is the most efficient path from beginner to conversational fluency that current learning science supports.

For audio-heavy vocabulary work — tones in Mandarin, verb conjugations in French, pronunciation differences in Arabic — structuring cards with native audio playback matters enormously for phonology-dependent languages (covered in the audio flashcard deck-building guide linked earlier).

Frequently Asked Questions

How is SRS different from regular flashcards?

Regular flashcards show every card on a fixed cycle, so you waste time on words you already know. An SRS tracks how well you recall each card and schedules its next review for the moment you are about to forget it — pushing easy cards weeks out and bringing hard ones back quickly. Same cards, far less wasted effort.

How many new cards should I add per day for language learning?

10 to 30 new cards per day is the sustainable range; 15 is a good default. At 15 cards daily you add about 5,000 words a year, and the mature review load settles around 80 to 120 reviews per day — roughly 15 to 20 minutes. Adding 50 or more cards a day creates review debt that triggers burnout within weeks.

Should I use FSRS or SM-2?

Use FSRS where it is available. It models your individual memory and needs about 12.6% fewer reviews than SM-2 for the same retention, with the gap widening past 2,000 cards. SM-2 is fine for small decks under 500 cards or legacy apps. Flashcard Maker schedules with FSRS by default; Anki offers it as an opt-in scheduler.

How many words do I need to be fluent?

About 2,000 to 3,000 high-frequency words gives roughly 95% coverage of everyday spoken conversation — the conversational fluency threshold. Around 5,000 to 8,000 words covers comfortable reading and television, and 8,000 to 10,000 reaches near-native comprehension where vocabulary stops being the bottleneck.

Why do I keep forgetting words even with spaced repetition?

Usually card quality, not the algorithm. Isolated word-translation pairs lack the context that makes a word retrievable, so they feel unstable. Switch to sentence-mined cards with one unknown word in context, pair concrete nouns with images, and rewrite any card you have failed five or more times. Daily immersion outside your deck cements the words further.