The Forgetting Curve: Where It All Began

In 1885, German psychologist Hermann Ebbinghaus published Uber das Gedachtnis, a monograph documenting the first systematic experiments on human memory. Working with nonsense syllables to eliminate prior associations, Ebbinghaus measured his own recall at various intervals and plotted what became known as the forgetting curve. His data showed that within 20 minutes of memorising new material, retention dropped to approximately 58%. After one hour, only 44% remained. By 24 hours, roughly 33% persisted. At the 31-day mark, retention had fallen to just 21%.

These figures have been replicated across dozens of subsequent studies. A 2015 meta-analysis by Murre and Dros at the University of Amsterdam confirmed that Ebbinghaus's original curve remains remarkably accurate, with minor variations depending on material complexity and individual differences. For adult language learners in Singapore grappling with Mandarin characters or Malay vocabulary, the practical implication is straightforward: without deliberate review scheduling, the majority of newly acquired words vanish within days.

Sebastian Leitner and the Cardbox System

The first widely adopted method for combating the forgetting curve came from German science journalist Sebastian Leitner, who published So lernt man lernen in 1972. Leitner proposed a physical flashcard system using five compartments. New cards begin in Box 1 and are reviewed daily. Correctly recalled cards advance to Box 2, which is reviewed every other day. Further correct recalls move cards through Box 3 (weekly), Box 4 (biweekly), and Box 5 (monthly). An incorrect answer at any stage sends the card back to Box 1.

The elegance of Leitner's system lies in its efficiency. Cards that are already well-known receive progressively less attention, while difficult items cycle frequently. A learner managing 500 flashcards might review only 60 to 80 per day, yet maintain strong recall across the entire set. This principle of adaptive spacing became the foundation for every digital spaced repetition system that followed.

The SM-2 Algorithm: Precision Scheduling

In 1987, Polish researcher Piotr Wozniak developed the SuperMemo algorithm, later refined into SM-2, which became the mathematical engine behind modern spaced repetition software. SM-2 assigns each card an easiness factor (EF), initially set at 2.5. After each review, the learner rates recall quality on a scale from 0 (complete blackout) to 5 (perfect recall with no hesitation).

The algorithm recalculates the EF after each review using the formula: EF' = EF + (0.1 - (5 - q) * (0.08 + (5 - q) * 0.02)), where q is the quality rating. The minimum EF is clamped at 1.3 to prevent intervals from collapsing too aggressively. Inter-repetition intervals follow a simple progression: the first review occurs after 1 day, the second after 6 days, and subsequent reviews multiply the previous interval by the current EF. A card with an EF of 2.5 reviewed correctly three times would have intervals of 1, 6, and 15 days.

Wozniak's research, documented extensively at supermemo.com, demonstrated that SM-2 could maintain 90–95% retention rates while reducing total review time by up to 70% compared to fixed-schedule revision. The algorithm has since been forked, modified, and embedded in dozens of applications, though its core logic remains remarkably unchanged after nearly four decades.

Three SRS Applications Compared

Anki: The Open-Source Standard

Anki, created by Damien Elmes in 2006, is free on desktop (Windows, macOS, Linux) and Android, with a one-time $24.99 charge for the iOS version. It uses a modified SM-2 algorithm and supports text, images, audio, and cloze deletions. The card library is entirely user-generated, with over 80 million shared decks available through AnkiWeb. Power users can install add-ons written in Python, adjusting everything from interval modifiers to card scheduling algorithms. The learning curve is steep: configuring optimal settings for, say, Mandarin character recognition requires adjusting new card limits, graduating intervals, and lapses thresholds. However, for users who invest 30 to 60 minutes in initial setup, Anki delivers the highest degree of control over review scheduling.

Memrise: Gamified Mnemonics

Memrise, founded in London in 2010, takes a different approach by combining SRS with mnemonic associations and short video clips of native speakers. The free tier includes basic courses, while the Pro subscription ($8.49/month or $59.99/year) unlocks advanced features including offline mode, difficult-word review, and grammar bots. Memrise pre-builds most courses, which reduces setup time but limits customisation. For adult learners in Singapore targeting Mandarin, Memrise offers HSK-aligned courses from beginner through HSK 6. Retention data from a 2023 study by University College London showed Memrise users recalled 47% of taught vocabulary after 90 days without review, compared to 31% for traditional study methods.

Migaku: Immersion-First Design

Migaku, developed by language learning content creator Yoga in 2021, targets intermediate and advanced learners who primarily acquire language through media consumption. Available as a Chrome extension and companion app at $9.99/month, Migaku integrates directly with Netflix, YouTube, and web pages. Users highlight unknown words in context, which are automatically converted into SRS flashcards with sentence examples, audio, and dictionary definitions. The approach aligns with Stephen Krashen's comprehensible input hypothesis by preserving the original context of each word. For Singapore-based learners consuming Japanese anime on Netflix or Korean drama on Viu, Migaku transforms passive viewing time into structured vocabulary acquisition.

A Practical Daily SRS Schedule

Based on retention data from Wozniak's research and practical testing by language learning communities, the following daily schedule optimises SRS effectiveness for working adults:

Morning review (15–20 minutes, 7:00–7:20 AM): Complete all due cards from previous sessions. Morning review exploits the testing effect when the brain is freshest. For a typical deck of 1,500 Mandarin characters, this translates to approximately 50 to 80 due reviews per day once the deck is mature.

Midday new cards (10–15 minutes, 12:30–12:45 PM): Introduce 10 to 20 new cards. Research from the University of California San Diego suggests that spacing new card introduction away from review sessions improves initial encoding by 12 to 18%. Adding new cards during a lunch break also creates a natural time boundary that prevents the common mistake of overloading with too many new items.

Pre-sleep review (10 minutes, 10:00–10:10 PM): A brief review of cards marked as "hard" or "again" during the morning session. A 2019 study published in Psychological Science found that information reviewed within 30 minutes before sleep showed a 17% improvement in next-day recall compared to identical reviews performed in the afternoon.

SRS for Mandarin Characters and Malay Vocabulary in Singapore

Singapore's bilingual education policy means most residents have at least partial exposure to a second language, but adult learners returning to or starting Mandarin face the specific challenge of character recognition. The Hanyu Shuiping Kaoshi (HSK) framework requires knowledge of 2,663 characters across six levels. Using SRS, learners typically add 10 characters per day, reaching HSK 4 (1,200 characters) in approximately four months and HSK 6 in eight to nine months.

For Malay vocabulary, which uses Latin script, the primary challenge shifts from character recognition to colloquial versus formal register differences. Bahasa Melayu as spoken at Geylang Serai markets differs substantially from the formal register used in government communications. SRS decks that include example sentences from both registers help learners navigate these differences. The National Library Board maintains a Malay-language collection at the Woodlands Regional Library that can serve as source material for building context-rich flashcard decks.

Regardless of the target language, consistent daily SRS sessions of 30 to 40 minutes outperform sporadic 2-hour study blocks. The spacing effect is cumulative: a learner who reviews 50 cards daily for 365 days will retain significantly more than one who reviews 350 cards in weekly 3-hour sessions over the same period.

References and Further Reading