בס״ד

Why you forget Gemara words, and how spaced review fixes it

You learn a daf on Sunday. You look up a word, then look it up again a few lines later. By Thursday you're looking it up a third time. That isn't a sign you're bad at languages. It's what memory does, and it has been measured for more than a century.

What Ebbinghaus measured

In the 1880s the German psychologist Hermann Ebbinghaus ran an experiment on himself. He learned lists of nonsense syllables until he could recite each list without a mistake. Then, after a set delay, he learned the same list again and timed how long it took.

He called the difference "savings." If relearning took less time than learning it the first time, something had stayed in memory, and the time saved showed how much.

His numbers, as reported by Murre and Dros (2015), went like this. After 20 minutes he saved 58% of the original learning time. After an hour, 44%. After 9 hours, 36%. After a day, 34%. After 2 days, 28%. After 6 days, 25%. After 31 days, 21%.

Two things stand out. First, most of the loss happens early: savings fell from 58% to 34% within a day, and only about 13 points more over the next month. Second, something was still there after a month. Relearning was faster than learning from nothing.

The same curve, 130 years later

In 2015 Jaap Murre and Joeri Dros at the University of Amsterdam repeated the experiment. One subject, Dros himself, spent 70 hours learning 70 lists of nonsense syllables and relearning them after 20 minutes, 1 hour, 9 hours, 1 day, 2 days, 6 days or 31 days.

His savings were 47% at 20 minutes, 37% at an hour, 28% at 9 hours, 32% at a day, 23% at 2 days, 17% at 6 days and 4% at 31 days. The 31-day lists were all learned early in the experiment, and learning took a little longer as the weeks went on. Correcting for that, the authors put the 31-day figure at about 14%, still lower than Ebbinghaus's.

Their conclusion: "the Ebbinghaus forgetting curve has indeed been replicated." They also found the curve is not completely smooth. There seems to be a small jump up at the one-day point, where memory looks a little better than the trend predicts. They suggest sleep may play a part, and say plainly that this needs more experiments to confirm.

What this means for a word on the daf

Nonsense syllables are harder to keep than real words, and a word you met inside a sugya you followed has more to hold on to. So you'll keep more than Ebbinghaus did. But the shape of the curve is the useful part:

Again, later, beats again, now

If you can only spend ten minutes on a list of words, is it better to go over them ten times tonight, or a few times over several days? Researchers have tested this question many times.

Cepeda and colleagues (2006) pulled together 839 measurements from 317 experiments in 184 published articles on spaced practice of verbal material. Their analysis found that the gap between study sessions matters, and that the best gap grows with how long you want to remember. If you want a word for next week, review it over a few days. If you want it for the whole masechta, keep coming back to it over weeks.

Cramming works for tomorrow morning. Spacing works for the rest of the masechta.

How Sugya.app is built around this

We built the app so that this happens without you having to plan it.

Every word is also shown in the real line where it appears. That part isn't from these studies. It's our own view that a word tied to a sugya you know is easier to hold than a word on its own.

A simple routine

  1. Before the shiur, learn the new words of the perek.
  2. The next day, do a short review before anything new.
  3. Once a week, go through your weakest words.

That's enough to keep most of what you learn. Start learning · How review works · Using it with a class

Sources

More