Most learners have experienced the same frustrating pattern: a new word feels clear today, familiar tomorrow, and strangely unreachable a week later.
This is not a personal failure. It is one of the most normal features of human memory.
When we learn something new, our ability to recall it often declines over time. In many situations, the decline is steep at first and then becomes slower. This general pattern is commonly known as the forgetting curve.
The idea goes back to Hermann Ebbinghaus, one of the pioneers of experimental memory research. His classic work showed that newly learned material can fade quickly without review. Modern research has replicated the broad pattern of his forgetting curve, although the exact shape is not perfectly smooth and should not be treated as a universal law for every learner or every type of material.
That distinction is important. The forgetting curve is not a single fixed formula that predicts every memory. It is a useful model for understanding a real learning problem: if we do not return to what we learn, access to that knowledge often becomes weaker.
Forgetting is not the opposite of learning
Forgetting can feel like learning has failed, but that is not the right way to think about it.
Memory is not simply “saved” or “deleted.” A word may be easy to recognize but hard to produce. It may be available in one context but not another. It may come back immediately when we see the first letter, hear it in a sentence, or meet it in a story.
In language learning, this matters a lot. Knowing a word is not a single state. A learner may gradually become able to:
- recognize the word while reading;
- understand it while listening;
- recall its meaning without help;
- pronounce it;
- use it in a sentence;
- recognize its forms, collocations, and common contexts.
Because of this, vocabulary knowledge usually needs repeated encounters. One exposure is rarely enough for durable, flexible use.
Spaced repetition slows forgetting
The practical response to forgetting is not to panic and not to cram. It is to review at the right time.
Spaced repetition means that reviews are spread out over time instead of being packed into a single session. Early reviews happen sooner, when the memory is still fragile. Later reviews can be placed farther apart, because successful recall tends to make the memory more stable.
This is supported by the broader spacing-effect literature. A large review and quantitative synthesis of distributed practice looked at hundreds of assessments across many experiments and found that spacing study sessions over time generally improves later recall compared with massed practice.
A simple schedule might look like this:
Day 0 → Day 1 → Day 3 → Day 7 → Day 21 → later
The exact days are not sacred. The principle is what matters: review shortly before the memory becomes too weak, then increase the interval after successful recall.
Active recall matters
Not all reviews are equally useful.
Simply seeing the answer again is weaker than trying to retrieve it. When learners actively recall a word, choose an answer, translate from memory, complete a sentence, pronounce the item, or use it in context, the review becomes more powerful.
This is often called retrieval practice or the testing effect. Research by Roediger and Karpicke showed that taking a memory test does not only measure what someone knows; it can also improve later retention.
For language learning, this means that a good review system should not only show old words again. It should ask the learner to bring them back.
The interactive model
The interactive chart below shows a simplified memory model inspired by the forgetting curve and spaced repetition research.
The orange line shows the learning path when reviews are completed successfully. Each review brings recall back up and makes the next forgetting curve flatter.
The red dashed line shows what may happen after the first learning session if there is no review.
The gray dashed lines show what may happen if the learner stops reviewing after one of the later review points. These lines are important because they show the key idea of the model: spaced repetition does not remove forgetting. It slows it down.
The two parameters in the model
The chart uses two simple parameters.
Initial half-life
This represents how stable the memory is immediately after first learning the material. In this model, the half-life is the time it takes for recall probability to fall to 50%.
A low initial half-life means the memory fades quickly. This could happen with a difficult, unfamiliar, weakly encoded word. A higher initial half-life means the memory is more stable from the beginning. This could happen when the word is meaningful, emotionally relevant, similar to something the learner already knows, or learned in a strong context.
Half-life multiplier after each successful review
This represents how much more stable the memory becomes after a successful review. If the multiplier is high, each successful review makes the next forgetting curve much flatter. If it is lower, the memory still improves, but more gradually.
In the model, recall is calculated with a simple half-life formula:
Recall = 100 × 0.5 ^ (elapsed time / half-life)
After each successful review, the half-life increases:
New half-life = previous half-life × multiplier
This is not meant to claim that the brain literally stores every word with a precise mathematical timer. The formula is an educational model. Its purpose is to make the core memory principle visible: successful reviews can increase stability, so forgetting becomes slower over time.
Half-life models are also used in real language-learning systems. For example, Duolingo’s Half-Life Regression model estimates memory strength and predicts recall probability using a half-life-based approach for spaced repetition practice.
Why the curve still declines after several reviews
One surprising detail in the chart is that forgetting still happens even after several successful reviews.
This is realistic.
Spaced repetition does not mean that four reviews make a word permanent forever. It means that the memory becomes more stable. The decline becomes slower, but it does not necessarily disappear.
In the default version of the chart, the reviews happen on days 1, 3, 7, and 21. After the fourth review, the next forgetting curve is much flatter than the first one. But if the learner never meets the word again — not in reading, not in listening, not in conversation, not in exercises — recall can still weaken over the following weeks or months.
This is especially true for active vocabulary. A word may remain recognizable for a long time, while still becoming hard to produce in speech or writing.
So the message is not:
Four reviews are enough forever.
The better message is:
Each successful, well-timed review makes forgetting slower.
Real learning is richer than the chart
The chart shows an isolated memory item. Real language learning is richer.
A learner may review a word in an app, then later hear it in a song, read it in a story, see it in subtitles, use it in a sentence, or meet it again in a different topic. Each meaningful encounter can act as another form of review.
That is why spaced repetition works best when it is combined with rich input and active use.
Flashcards can help keep important words alive. But stories, listening, conversation, pronunciation, examples, and real contexts help turn those words into usable language.
The best learning system does not treat memory as a mechanical checklist. It creates repeated, meaningful encounters at the right time.
What the graph should — and should not — be used for
The chart should be read as a simplified explanation, not as a universal prediction.
It does not say that every learner forgets at the same speed. It does not say that every word has the same curve. It does not say that review always restores perfect knowledge. And it does not replace the many factors that influence memory: attention, sleep, prior knowledge, difficulty, interference, motivation, emotional relevance, and the quality of practice.
But the main idea is well supported:
Without review, recall often fades. With well-timed successful reviews, forgetting slows down. With active recall, the review becomes stronger.
For learners, this should be encouraging rather than discouraging.
Forgetting is normal. Review is not a sign that learning failed. Review is how fragile knowledge becomes durable knowledge.
The practical takeaway
A good vocabulary system should do three things:
- introduce new material clearly;
- bring it back before it disappears;
- increase the interval after successful recall.
That is the logic behind spaced repetition.
Not endless repetition.
Not cramming.
Not passive rereading.
Just well-timed remembering.
