Spaced Repetition and Retrieval Practice: What the Research Actually Says

What the evidence shows

Cramming before an exam is widespread and largely ineffective for long-term retention. Rowland (2014) synthesised 47 studies (N=6,801): retrieval practice (actively recalling information from memory) outperforms passive re-reading with an effect size of d=0.87. Using the calibration scale for second language acquisition research developed by Plonsky & Oswald (2014, N>604,000), this places retrieval practice firmly in the large-effect category (d≈1.0 threshold). Spacing practice across time adds further gains: Kim & Webb (2022) analysed 98 effects across 48 experiments and 3,411 participants and found spaced practice beats massed practice with a medium-to-large effect on delayed tests.

The testing effect

Every time you read a word again, your brain registers familiarity. Every time you close the book and try to recall it, your brain constructs and reinforces a retrieval pathway: the same pathway you will need when speaking. This is the testing effect, and it is robust across age groups, subjects, and test formats (Adesope et al., 2017, Review of Educational Research, DOI 10.3102/0034654316689306).

Format matters. Free recall (producing the target word with no cues) produces stronger retention than cued recall (flashcards, gap-fill). Both, however, substantially outperform passive study. Oral production in a real dialogue, as in Lexy's Stage 3 (Lesson), engages procedural memory alongside declarative memory, which is the mechanism underlying spoken fluency.

Spaced vs. massed practice

Massed practice (cramming) creates an illusion of knowledge: material feels easy to access because it is fresh. Learners misread this fluency as long-term mastery. Kim & Webb (2022) tested this across 48 language-learning experiments: on delayed tests, spaced practice produced a medium-to-large advantage over massed. On immediate tests, short and long intervals performed similarly, with the divergence emerging when retention is measured after a gap. Kim & Webb (2023) replicated this with 150 Korean learners studying 48 words; a two-week delayed test showed spaced groups outperforming massed groups on both gap-fill and flashcard tasks.

One widely repeated claim, that expanding intervals (e.g., 1, 3, 7, 14 days) are superior to equal intervals, is not supported by the Kim & Webb (2022) data: expanding and equal schedules were statistically indistinguishable. What matters is distributing practice across time, not the precise geometry of intervals.

Optimal gaps: Cepeda et al.

Cepeda et al. (2008) tracked learners over one year and proposed a practical rule: the optimal inter-study interval is 10–20% of the desired retention interval (DOI 10.1111/j.1467-9280.2008.02209.x). Want to remember a word one month from now? Study it every 3–6 days. Want to remember in a year? Study every 5–10 weeks. A schedule optimised for tomorrow's test is not optimised for next year's conversation.

Sleep as a free consolidation tool

Memory consolidation (the transfer of traces from the hippocampus to the cortex) happens primarily during sleep. Schimke et al. (2021) meta-analysed 25 studies (N=1,396): learning followed by sleep outperforms learning followed by wakefulness with Hedge's g=0.50 overall; g=0.57 for free recall; g=0.52 for recognition; g=0.39 for lexical integration (DOI 10.3758/s13423-021-01980-3). Studying vocabulary in the evening before a normal night's sleep is measurably more effective than studying in the morning with no subsequent sleep. This requires no extra time, only a schedule adjustment.

How Lexy uses these principles

Lexy structures each lesson across 49 languages into seven sequential stages, integrating retrieval practice and spaced intervals throughout:

1. Vocabulary: interactive flashcards and a matching quiz require active retrieval rather than passive reading.

2. Warm-up: a brief written chat with Lexy to prime communicative schemas before speaking.

3. Lesson: a 25-minute live voice conversation with the AI avatar without typing, practicing spoken retrieval without error interruption in lessons 1–15 to protect speech flow.

4. Practice: keyboard-based phrase exercises, listening tasks, and tests for grammar consolidation without vocal fatigue.

5. Dialogue: a two-voice listening conversation exposing learners to natural dialogue pacing.

6. Story: a single-narrator reading story with sentence-by-sentence translation and transliteration.

7. Memory Flashcards: spaced-repetition flip-cards reviewing vocabulary from the previous lesson starting from lesson 2 onward.

The 16-lesson course establishes phonetic foundations and core vocabulary, reinforced by spaced sessions and active recall.

Limitations

Most studies in the evidence base used controlled laboratory conditions with isolated word lists and fixed intervals. Li (2010) showed that lab effects consistently exceed classroom effects in SLA; real-world gains are likely smaller. Kim & Webb (2022) drew primarily from East Asian learners of English; generalisability to other language pairs needs separate investigation. Cepeda et al. tested native English speakers learning English words; optimal gaps for L2 learners may differ. Finally, most effect sizes reflect explicit knowledge (translation tests); data on implicit acquisition in natural communication contexts remain sparse.

Four takeaways

1. Retrieve, don't re-read. d=0.87 (Rowland, 2014) is one of the largest effects in educational psychology. Any format that forces recall beats passive review.
2. Distribute your sessions. 98 effects confirmed spaced > massed on delayed tests (Kim & Webb, 2022). Cramming prepares you for tomorrow's exam, not next year's conversation.
3. Use the 10–20% rule. For one-month retention, study every 3–6 days. For one-year retention, every 5–10 weeks (Cepeda et al., 2008).
4. Study before sleep. g=0.50 advantage for sleep-after-learning (Schimke et al., 2021). It costs no extra time, only a schedule change.

🎯 90-Second Quiz 1 / 5

Test Your Knowledge: Article Quiz

90-second quiz: check whether you understood the key themes of this article. 5 quick questions. If you make a mistake, we explain it right away.

1. Who first experimentally documented the "forgetting curve" in 1885?

Hermann Ebbinghaus Ivan Pavlov Jean Piaget