The hour after your lesson matters more than the lesson

Your tutor spent 11 minutes untangling por and para for you. She used an example about a bus ticket that finally made it click. You said "ohhh" out loud.

Nine days later you're in the same hole, and she's drawing the same diagram. Neither of you says anything about it.

The problem is the 6 days afterwards, when nothing asks you to produce any of it. That's a schedule problem, and you can fix it in about 15 minutes a week.

Start with your tutor's homework

Before any of this: if your tutor set you a task, do that task. Properly. It beats every generic routine on this page, including mine.

She watched you fail at something specific. She heard which verb form you dodge and which construction you route around every time. A task aimed at that is worth more than a schedule designed by someone who has never heard you speak.

So if you reliably do your tutor homework, you're already ahead of most people reading this, and I'd rather say that plainly than pretend you need a system. Use the routine below to cover the days the homework doesn't reach, or the weeks nothing gets set.

The routine, on a clock

Three touches. About 15 minutes total, against an hour you paid for.

One rule runs through all three: produce before you check. Every touch starts with you pulling the word out of nothing. The second you read first, you've swapped the exercise for a worse one.

Touch 1: within the first hour, 6 minutes

Close the notes. Open a blank page. For 4 minutes, write down every word and correction you can drag back out of the lesson.

It'll feel thin. You had an hour of material and you'll produce maybe 12 items, half of them fuzzy. Write the fuzzy ones anyway, with a question mark.

Then, and only then, 2 minutes with your notes open. Don't reread them start to finish. Scan for what you missed and mark those items. Six or 7 misses wouldn't surprise me for a dense hour.

Why blank-page first rather than notes first: rereading loses. Roediger and Karpicke ran the clean version of this in 2006 with prose passages. One group read a passage repeatedly, another read it once then took recall tests. With the final test 5 minutes later, the rereaders genuinely won. At 2 days and at a week, the recall group had them beaten. That's the testing effect, and it's why the order of those 6 minutes matters more than their length.

Now the part I should be straight about. That study justifies producing before you check. It says nothing about doing it inside an hour, and I don't have a paper that puts a clock on this touch the way Cepeda puts one on day 3. The hour is my number, not a finding. It's there because that's roughly how long you can still rebuild the lesson without help, while your tutor's phrasing and the order things came up in are still sitting there, and that window shuts fast. Ninety minutes is fine. The next morning is a different exercise, and a harder one.

A word list isn't always what you walk away with, though. If the hour was mostly conversation, write down the 3 or 4 sentences you tried to say and got corrected on, both versions, yours and hers. Those are the items.

And if it was a grammar untangle like the por/para one, don't copy her explanation. Write the rule in your own words, then invent one fresh example she never gave you. Generating your own material is remembered better than reading someone else's, which Slamecka and Graf pinned down in 1978, and the invented example is the thing you'll test yourself on later.

Touch 2: that night, 4 minutes

Before bed, take the list. Cover the target-language side. Go down the English column and say each one out loud. If it doesn't arrive in about 5 seconds, mark it and move on.

Four minutes is deliberately not enough time to study properly. It's a handoff: the material goes in front of a night's sleep while it's still active.

Gais and colleagues tested this in 2006 with high-school students learning English–German vocabulary, checking recall 48 hours after learning. Students who slept within a few hours of learning did better than students kept awake that first night, and the deprived group was still behind after a recovery night's sleep, which is what rules out simple tiredness as the explanation. Sleep does filing work on what you touched that day, covered properly in the sleep post.

Say it out loud rather than in your head. Producing the sound is closer to the thing you'll need to do in the next lesson, and it stops you from silently upgrading a half-remembered guess into a pass.

Touch 3: day 3, 5 minutes

Two days later, cold. No list in front of you until you've tried.

This is the touch people skip, and it's the one carrying the most weight. By day 3 the material has faded enough that pulling it back is genuinely hard.

You'll fail items here. Good. Kornell, Hays and Bjork ran 6 experiments in 2009 on questions built so the guess would fail, and a failed attempt followed by the answer still beat being handed the answer straight. That's the evidence the day-3 touch stands on.

There's a name for the category it belongs to. Elizabeth and Robert Bjork call conditions like this desirable difficulties, and their 2011 chapter names spacing and self-testing in the list, defining them as difficulties that earn the word by triggering "encoding and retrieval processes that support learning, comprehension, and remembering." Their caveat travels with it: a difficulty stops being desirable if you lack what you'd need to get through it. A cold test on a list you never wrote down is just a bad afternoon.

So why day 3 rather than the same evening or the following Sunday? Cepeda and colleagues taught facts to more than 1,350 people in 2008, varied the gap before a review, and tested up to a year later. The best gap gets longer as your deadline gets further away, while shrinking as a share of it: about 20 to 40% of a 1-week delay, falling to roughly 5 to 10% when the test is a year out.

That conditional result is the finding to hold onto, and it's the one their earlier review of 317 experiments on distributed practice landed on too. The gap that wins depends on how far away you need the material. Spreading study out generally beats piling it up, but the useful question was always how far apart, and that answer moves.

Their shortest delay was a week, and a lesson cycle is about that, so the top of the range points at day 2 or 3. I round to day 3 for 2 reasons: day 2 collides with the night touch for a lot of people, and a slightly-too-long gap costs less than a slightly-too-short one. Day 2 is defensible and I won't argue with anyone who prefers it. The honest version is that anywhere in that window beats the two things people actually do, which are the same evening and never.

Either way it's far enough from touch 2 that you've genuinely forgotten some of it, which is the difference between a review and a formality. Cram all three touches into one evening and you get the illusion of having studied and little else, which is the whole problem with cramming.

End by carrying your day-3 misses into the next lesson. That turns a review into an agenda, and your tutor starts from your actual gaps rather than from wherever the textbook happened to stop.

If you only ever do one touch

Most people won't run all three. Fine.

Doing one touch well matters more than the schedule. A single honest retrieval attempt on day 2 or 3, where you produce the words from nothing before checking, beats three sessions of scrolling your notes.

The size of that difference is the reason I'll defend this. Karpicke and Roediger had college students learn 40 Swahili-English pairs, then varied only what happened after the first correct answer. Items that kept being tested came back at about 80% a week later. Items that were repeatedly restudied but dropped from testing came back at 36%.

Karpicke & Roediger 2008 · Swahili vocabulary, one week later

0%

recalled when items kept being tested after the first correct answer

0%

recalled when items were restudied but dropped from testing

The students in that study couldn't feel the difference either. Every group predicted they'd recall about half the list, and those predictions didn't shift between conditions, which is worth sitting with if you've been judging your progress by how familiar the notes look.

So pick the day-3 touch if you're picking one. It's the one with the widest gap between how it feels and what it does.

Doing it on paper

All of this runs on a notebook and a covered column. No software involved, and it works.

Where it breaks is week 4, when 4 lessons' worth of items sit in one document and nothing tells you which of them are fading today. Manual review then defaults to reviewing everything or reviewing nothing, and both lose to reviewing the handful that are actually about to go. That's the one job software genuinely does better, and if your paper habit is holding, you don't need any.

The free handbook has this routine as a printable post-lesson page if you want it on paper. And if you want the longer version of why the material disappears in the first place, that's in the guide on remembering what your tutor teaches.

Questions people actually ask

---

Sources

Every load-bearing claim Verbamor makes is traced to its paper on the research page.