Spaced Repetition for Languages: Flashcards + Reading

Editorial illustration showing the TortoLingua turtle discovering meaning through context for the article "How Spaced Repetition Works for Language Learning".

What spaced repetition means for language learning

Spaced repetition means returning to material after a gap rather than repeating it only in one sitting. A flashcard system can schedule those returns. Reading can also bring you back to a word, but it does not give you the same control over when that word appears.

The useful distinction is between remembering a particular answer and learning how a word works in different sentences. You can use cards, reading or both. Neither a card marked “easy” nor a familiar word in a story proves that you know every use of it.

The forgetting curve is not your personal countdown

Hermann Ebbinghaus’s work with lists of nonsense syllables helped establish the study of forgetting. The familiar curve is often presented as a fixed percentage of words everyone loses after an hour or a day. That is not a sound way to predict your vocabulary.

Murre and Dros’s replication (2015) explains the savings method: measuring how much effort is saved when learning the material again. Savings are not the same as the percentage of foreign words a reader can recall. The material, person, test and delay matter.

This is why an app cannot infer the exact moment you will forget a word simply from a universal curve.

From expanding intervals to a box of cards

In “A Memory Schedule” (1967), Pimsleur proposed graduated interval recall for language learning. The idea of expanding gaps is one historical approach, not a mandatory sequence of seconds, days and years for every word.

The Leitner approach makes the scheduling idea tangible with groups or boxes of cards. In a simple example:

  1. Put a new card in the group you review more often.
  2. Try to recall the answer before turning it over.
  3. After a correct answer, move it to a group reviewed less often.
  4. If you cannot recall it, return it to a group that receives more attention.

You choose whether to use this method. This example explains how the groups work; it does not prescribe daily reviews or a weekly target.

Spaced repetition software automates scheduling. Depending on the system, answers, ratings or other information influence later reviews. A scheduled review is an estimate, not knowledge of the exact moment you will forget. These card controls should not be confused with TortoLingua’s reading flow.

What the research supports—and where it stops

Cepeda and colleagues (2006) synthesized distributed-practice research in verbal recall tasks. Their review included 317 experiments. The interval between study episodes and the interval until the final test jointly affected retention. A gap that suits one retention goal need not suit another.

That supports taking the timing of practice seriously. It does not establish a universal “five repetitions on five days” rule, show that longer gaps are always better, or make one review schedule a guarantee of language proficiency.

Webb (2007) studied 121 Japanese learners of English with 1, 3, 7 or 10 encounters. Ten tests examined aspects such as spelling, association, grammatical function, syntax, meaning and form. Increasing encounters produced gains in at least one aspect at each increase. This does not make 10 encounters—or a range such as 7–16—a universal mastery threshold.

Nakata and Elgort (2021) compared spaced and massed encounters with new words in English sentences, followed by meaning feedback. Spacing helped on explicit meaning-recall and meaning–form matching tests. It did not produce the same advantage on the semantic-priming measure used to examine tacit knowledge. The outcome depended on what was tested, not just whether the word reappeared.

Cards and reading: compare the task, not the label

Imagine a card with perro on one side and “dog” on the other. Recalling the translation answers that card’s question. It does not itself test whether you can spell the word, recognize it in speech or choose an appropriate phrase. A different card can ask a different question; not all cards are isolated word–translation pairs.

A sentence gives you an example of use. Seeing a word in another sentence may show a different grammatical role, companion word or meaning. It does not follow that every encounter teaches a new layer or that any reading task outperforms any card task.

ActivityWhat it lets you doWhat it does not establish
Recall a word from a cardCheck an answer to a chosen prompt.Mastery of every meaning or use.
Read a word in a sentenceSee how it contributes to that message.A fixed review interval or guaranteed recall later.
Combine a chosen review with readingRevisit an item and examine examples of use.One best balance for every learner.

Word knowledge includes more than a translation, a distinction developed in Nation’s work. Avoid using a vocabulary count such as 2,000 or 3,000 word families as permission to abandon cards or as proof that all original books are now understandable. The coverage guide explains why the text and the reader matter.

If you want to try it

Choose an item you want to remember and a question worth answering: its meaning in a sentence, its spelling or a useful phrase. Try recalling it before checking. If the prompt is ambiguous or the answer keeps failing, you can change the prompt, look for a clearer example or stop using the card.

For reading, choose a passage you can follow. You may notice words again naturally, but a book does not promise to repeat each chosen word on a schedule. You decide how much to read and whether to add deliberate review; there is no required daily quota.

Extensive reading and learning a language by reading offer ways to explore this alongside comprehensible input. The background on extensive reading also distinguishes it from close study.

Where TortoLingua fits

In TortoLingua, mark words you do not know while reading. The app uses those marks to choose the next text automatically; if the current text is too difficult, it offers a simpler one. This does not mean you manually choose the level, rate flashcards or control a review queue.

That reading support is not evidence of a particular spaced-repetition algorithm or a guarantee that a marked word will return at an optimal interval. Speaking, listening and writing remain separate activities when those are your goals.

Further reading and sources

For a specific next question, continue with:

The primary research linked above supports the bounded claims in its own tasks and samples. Historical background includes Ebbinghaus’s Über das Gedächtnis (1885), Pimsleur’s “A Memory Schedule” (1967), and Leitner’s So lernt man lernen. Nation’s Learning Vocabulary in Another Language (2001) and Webb and Nation’s How Vocabulary Is Learned (2017) provide broader vocabulary discussions; they are not used here to prescribe a fixed number of encounters or rank all reading above all flashcards.

Last updated

TortoLingua app

Read in the browser or install the mobile app.

Open the web app or find the App Store and Google Play download links.

Get the app