Why students forget vocabulary, and what the spacing research actually says to do about it.

Your Spanish 2 students could conjugate ser on Friday and could not remember it on Monday. The Quizlet sets are not the problem. The schedule is. Nicholas Cepeda and Henry Roediger spent twenty years showing exactly why.

The spacing effect is over a century old

Hermann Ebbinghaus ran the first experiments on himself in Berlin in the eighteen eighties (Über das Gedächtnis, 1885). He memorized lists of nonsense syllables, then tested himself at intervals. The curve he drew is one of the most replicated findings in psychology. We forget fast and then we forget slowly. About two thirds of new information is gone within twenty-four hours unless we revisit it.

Ebbinghaus also discovered that spreading practice over days was dramatically more efficient than cramming. He needed fewer total trials to reach the same retention if he split the study sessions. That finding sat in the literature for the next hundred years and was rediscovered by every generation. Vocabulary acquisition researchers (Bahrick 1979, Pimsleur 1967) replicated it in second-language contexts.

What Cepeda actually showed

Nicholas Cepeda and his colleagues ran the modern meta-analysis (Distributed Practice in Verbal Recall Tasks, Psychological Bulletin, 2006). They pulled 184 published studies and computed effect sizes for spacing. The average benefit of distributed practice over massed practice was large and consistent. Cramming worked for the next twenty minutes. It did not work for the next month.

The 2008 follow-up was the one that should change how you build a Spanish course. Cepeda and his team ran a clean experiment with 1,354 participants and varied two things: how long the gap between study sessions was, and how long after the second session the final test happened.

The result is now called the temporal ridgeline. There is an optimal gap, and it depends on how long you want to remember the material. If the final test is one week out, the best gap is about one day. If the final test is one month out, the best gap is roughly one week. The general rule that emerged was that the optimal gap is approximately ten to twenty percent of the retention interval. Want students to remember the preterite in May? Space the practice in November, January, and March.

That ratio is the part everyone misses. Spacing is not about spreading study out. It is about matching the gap to the goal. A vocabulary list practiced four days in a row and then dropped is functionally crammed.

Optimal gap was approximately ten to twenty percent of the retention interval, with shorter and longer gaps producing substantially worse retention.

The retrieval effect, separately

Spacing alone is half the story. Jeffrey Karpicke and Henry Roediger published a series of papers (The Critical Importance of Retrieval for Learning, Science, 2008) that showed a second, independent effect. Looking at material again is not the same as trying to remember it.

Their cleanest study had four groups all studying foreign vocabulary. One group studied and was tested repeatedly. One studied and was tested but had items drop out of practice once recalled. One studied repeatedly with no testing. One studied each item only once. A week later, the group that had been tested remembered roughly twice as much as the group that had been re-reading.

The follow-up paper (Karpicke and Blunt, Retrieval Practice Produces More Learning than Elaborative Studying with Concept Mapping, Science, 2011) compared retrieval practice against the most pedagogically respectable alternative the researchers could think of: concept mapping. Retrieval practice still won by a wide margin.

Robert Bjork has a useful frame for why. He calls these desirable difficulties (Bjork and Bjork, 2011). The harder you have to work to pull a word out of memory, the stronger the trace you leave for next time. Re-reading is easy and feels productive. It is the worst way to study.

Why Quizlet does not save you

Most Spanish students use Quizlet the way their parents used flashcards in 1984. They go through the deck a few times in one sitting the night before the quiz. They do well on the quiz. They forget the words by Tuesday.

The application supports better schedules. The Learn mode algorithm is roughly Leitner-style spaced repetition. Almost no student uses it that way. They click through the cards until the words feel familiar and stop. Familiarity is not retrieval. The studies are clear on this.

Anki, Memrise, and any properly built spaced repetition system (SRS) implement the Cepeda ratio explicitly. The algorithm pushes cards out further and further the more often you get them right and snaps them back if you miss. For vocabulary specifically, SRS is closer to optimal than anything a classroom teacher can implement by hand.

How to build a Spanish course around this

Cumulative review weeks

Build a four-week unit. In week four, instead of running new content, retest everything from weeks one through three. Not with the same activities. With retrieval. Quick-write the vocabulary from memory. Cloze paragraphs without word banks. Picture prompts where the only way to label is to remember.

Spiraled warm-ups

Start every class with three minutes of forced recall on material from earlier units. Not yesterday. Last month. Last unit. Make the gap big enough that the retrieval is hard. Bjork's desirable difficulties. If the warm-up feels easy, the gap is too short.

Final assessments that pull from the whole year

If a unit assessment only tests the unit, students cram for it and forget afterward. If the final pulls from the entire year, students have to maintain the material across months. That is the only path to durable acquisition. It is also unpopular with students until they realize their grades on the cumulative section are better than the unit-only section.

What this means for your Monday

Audit your last unit. How many minutes of class time did students spend retrieving vocabulary from previous units? For most of us, the honest answer is close to zero. We move on. We post a new Quizlet.

For the next unit, build in three retrieval moments per week, each pointing at vocabulary that is at least four weeks old. Five minutes each. Quick-writes, cloze, picture labeling, oral retell. Do not let students use the textbook or the Quizlet during the retrieval. The whole point is the effortful pull.

Then look at your final exam. If half of it cannot be cramped the week before, your students will not let themselves forget the material. That is the entire trick. Forty years of cognitive psychology agree.