80%
Exercising a word beats rereading it.
Learners who kept retrieving new foreign words remembered about 80% a week later. Those who stopped once they knew them remembered about a third.
Karpicke & Roediger, Science, 2008
2×
Generating a word beats recognising it.
Practice that asks you to produce the answer had about twice the effect of practice that asks you to pick it, across studies of the testing effect.
Rowland, Psychological Bulletin, 2014
494%
Feedback after a miss matters most.
Seeing the right answer after a wrong one raised what learners kept a week later by 494%. Feedback after a right answer made little difference.
Pashler et al., 2005
839
Spacing is the biggest lever.
Across 839 assessments in 317 experiments, practice spread over days beat the same practice crammed together, and the best gap grows the longer you mean to remember.
Cepeda et al., 2006 and 2008
1 in 4
Learning a word isn’t keeping it.
Across 22 studies of deliberate vocabulary study, learners could produce only about a quarter of the words some time later. Spaced reviews are what turn a word learned into a word kept.
Webb, Yanagisawa & Uchihara, 2020
FSRS
Spaced by a model of your memory.
Reviews are timed by FSRS, an open-source memory model now built into Anki, which predicts when you are about to forget and outperforms older schedulers on a large public benchmark.
open-spaced-repetition, srs-benchmark