Spaced Repetition: What 140 Years of Research Actually Says

In 1885, a German psychologist named Hermann Ebbinghaus decided to do something unusual. He wanted to study memory scientifically, which meant he needed to memorize material that had no prior associations. So he invented nonsense syllables, combinations like “DAX” and “BUP” and “ZOL,” and he memorized lists of them. Then he tested how quickly he forgot them.

What he discovered changed how we think about learning. Memory doesn’t just passively decay. It decays in a predictable pattern, and you can fight that decay if you know when to review.

What is the forgetting curve?

Ebbinghaus found that forgetting is steepest right after learning. Within an hour of studying something new, you’ve already lost about half of it. Within 24 hours, you’re down to maybe 30 percent. By the end of a week, you’re holding onto about a quarter of what you learned, if that.

This is the forgetting curve, and it’s exponential. Most of the loss happens early, then the rate slows down. Whatever survives the first few days is relatively stable, but it’s still decaying, just more gradually.

The important discovery was that reviewing information resets the curve, and each review makes the curve shallower. After one review, you might retain the information for a few days before it starts to fade. After two reviews, maybe a week. After three, maybe a month. The intervals keep expanding as the memory gets stronger.

But why does spacing reviews work better than cramming them together?

Why does spacing work better than cramming?

The “spacing effect” is one of the most replicated findings in all of psychology. If you spend an hour studying something, spreading that hour across multiple sessions produces better retention than cramming it into one session. Same total time, dramatically different results.

A 2006 meta-analysis by Cepeda and colleagues looked at 317 experiments from 184 different studies spanning over a century of research. The spacing effect showed up across different types of material, different age groups, different retention intervals, different testing conditions. It’s not a quirk of certain circumstances. It’s how human memory works.

Why does spacing help? Researchers have proposed several mechanisms:

When you review at different times, you encode the memory in different contexts, different moods, different environments. This creates more neural pathways to the information, more ways to retrieve it later.

Spacing also makes retrieval harder, and harder retrieval strengthens memory. This is what researchers call “desirable difficulty.” Recalling something easily doesn’t do much for retention. Recalling something that you almost forgot does.

And time between reviews allows for memory consolidation. Sleep plays a role here. The neural connections that encode your memory literally get reinforced while you’re not thinking about the material at all.

The science was clear by the early 20th century. The challenge was turning it into something practical.

How did spaced repetition become an algorithm?

Ebbinghaus’s findings were elegant, but they didn’t tell you exactly when to review. Paul Pimsleur, a language learning researcher, proposed specific intervals in 1967: five seconds, twenty-five seconds, two minutes, ten minutes, one hour, five hours, one day, five days, twenty-five days, four months, two years. The intervals roughly followed an exponential pattern.

Sebastian Leitner turned the spacing idea into a practical system in 1972. His flashcard method used physical boxes. New cards start in box one and are reviewed every day. When you get a card right, it moves to the next box, which you review less frequently. Get it wrong, and it goes back to box one. Simple, manual, effective.

The real breakthrough came from Piotr Wozniak, a Polish computer scientist who in 1987 created SuperMemo, the first spaced repetition software. His SM-2 algorithm didn’t just use fixed intervals. It adapted to your performance. When you reviewed an item, you rated how easy it was to recall. The algorithm used that rating to calculate the optimal interval until the next review. Hard items came back sooner. Easy items were spaced out further.

Modern algorithms like FSRS have refined this approach with more sophisticated models of memory, but the core insight remains Ebbinghaus’s: memory decays predictably, and well-timed reviews can counteract that decay with remarkable efficiency.

All of this research focused on memorizing facts and vocabulary. But pronunciation isn’t just memory, it’s a motor skill. Does spacing work the same way?

Does spaced repetition work for pronunciation?

Here’s where it gets interesting for language learners. Most spaced repetition research has focused on declarative memory: facts, vocabulary, definitions. But pronunciation is a motor skill. It involves procedural memory, which follows somewhat different rules.

The good news is that motor learning research shows similar benefits from spacing. Distributed practice beats massed practice for motor skills just as it does for verbal recall. Cramming pronunciation practice into one long session is less effective than spreading it across multiple shorter sessions.

But motor skills also have some unique properties. Motor memories are generally more resistant to forgetting than verbal memories. Once you can ride a bike, you don’t forget how, even after years without practice. Pronunciation patterns fall somewhere in between: they’re more durable than vocabulary, but they do decay without practice which is why people talk about fossilized accents that seem resistant to change.

What this means practically: pronunciation you’ve truly mastered, where the motor pattern has become automatic, will stick around longer than vocabulary you’ve merely memorized. But pronunciation you’re still working on, where you can produce it with effort but it’s not automatic, that needs regular review or it will degrade.

So spacing helps. But most people don’t practice pronunciation with any kind of spacing at all.

Why does most pronunciation practice fail?

Think about how you actually practice pronunciation. Maybe you drill a word a few times when you first learn it, then never practice it again. That’s not spaced repetition, that’s hoping the first exposure was enough. It usually wasn’t.

Maybe you avoid words you find difficult, substituting easier words when you can. That means zero practice, zero improvement.

Maybe you rely on random conversation, encountering words whenever they happen to come up. That’s unstructured exposure with no intentional spacing.

Or maybe you practice the same words every day whether you need to or not. That’s wasted effort. Once something is well-learned, daily review isn’t helping. You’d be better off extending the interval and spending that time on something that actually needs work.

Without deliberate spacing, you’re fighting against how memory works instead of using it. The question is what to do instead.

What does effective spaced pronunciation practice look like?

You need to track what you’ve practiced, you can’t space something if you don’t know when you last reviewed it. You need automated scheduling, because human intuition about when to review is notoriously bad. We think we’ve got something until the moment we need it and realize we don’t.

The intervals also need to adapt to your performance. Items you struggle with should come back sooner, while items you’ve mastered can wait longer. A good system adjusts based on how you actually perform, not just elapsed time, which requires immediate feedback on your production attempts.

And you need to trust the system. When the algorithm says review today, review today. Skipping a session because you “feel like you know it” defeats the purpose. The algorithm knows better than your intuition.

This is exactly what spaced repetition software does, and after 140 years of research, the evidence that it works is about as strong as evidence gets in psychology. The spacing effect isn’t a lifehack or a study tip. It’s a deep fact about how human memory functions.

You can ignore it and practice inefficiently. Or you can use it and get more retention from less time.

SpeechLoop is built on this research. Every pronunciation you practice gets scheduled by an algorithm that tracks how well you produced it and when you last reviewed it. Sounds you struggle with come back sooner. Sounds you’ve nailed get pushed out further. The algorithm decides what you practice, because it’s better at it than you are. Ebbinghaus and Wozniak did the hard work. We’re just applying it to pronunciation.

---

References

Ebbinghaus, H. (1885). Über das Gedächtnis: Untersuchungen zur experimentellen Psychologie [Memory: A contribution to experimental psychology]. Duncker & Humblot.

Pimsleur, P. (1967). A memory schedule. Modern Language Journal, 51(2), 73–75.

Cepeda, N. J., Pashler, H., Vul, E., Wixted, J. T., & Rohrer, D. (2006). Distributed practice in verbal recall tasks: A review and quantitative synthesis. Psychological Bulletin, 132(3), 354–380.

Dunlosky, J., Rawson, K. A., Marsh, E. J., Nathan, M. J., & Willingham, D. T. (2013). Improving students’ learning with effective learning techniques: Promising directions from cognitive and educational psychology. Psychological Science in the Public Interest, 14(1), 4–58.