Learning English in your sleep: the podcast won't stick, but sleep still helps

Every night, Marta falls asleep with her phone by the pillow and a YouTube video promising "English while you sleep": eight hours of phrases in her ear. In the morning she can't recall a single one. It isn't a lack of discipline. The method fails by design.

Audio playing while you sleep doesn't leave you with words you can use when you wake up. A 2023 historical review of 51 papers concludes that the evidence for new knowledge learned in sleep carrying over to waking "remains inconclusive." What does help is the reverse order: practice actively in the evening, then go to sleep. Learning is your job while you're awake. Storing it is sleep's job.

What the 1950s EEG studies found#

Sleep learning got its first serious test in the 1950s, with electroencephalography (EEG). In the 1956 study by William Emmons and Charles Simon, researchers played words during the night and cut the audio as soon as the EEG showed alpha waves, the sign that the sleeper was drifting toward waking. The next day, participants recalled no more than a control group. They only retained what they heard while drowsy or close to waking.

The 1956 study has weaknesses that the 2023 review points out, but its conclusion holds: what plays while you're truly asleep doesn't reach the memory you use when you're awake. Ken Paller and Delphine Oudiette sum it up in Scientific American: "participants in these experiments appeared to learn only if they were not really sleeping when information was presented to them."

What the 2019 Bern study really showed#

The most-cited experiment in favor of sleep learning shows far less than its headlines. At the University of Bern, Marc Alain Züst, Simon Ruch, Roland Wiest and Katharina Henke had 41 German speakers nap and, during deep (slow-wave) sleep, played them pairs of an invented word and a German word, like tofer–Haus ("house"). On waking, nobody consciously remembered anything, so they were only asked to guess whether each invented word meant something bigger or smaller than a shoebox.

The result landed barely above chance. When the German word lined up with 2 to 4 slow-wave peaks, participants guessed right 59.9% of the time; with 0 or 1 peaks, they stayed at chance, around 50%. Across all pairs, accuracy was 52.7%, against a coin flip's 50%. That's a faint hunch about words that don't exist, not vocabulary you can use.

Whatever gets learned this way stays out of conscious reach. In their 2020 review, Simon Ruch and Katharina Henke wrote that sleep learning "invariably produces memory traces that are consciously inaccessible in the awake state."

Sound in deep sleep can even get in the way of learning later. Thomas Andrillon and his team played noise patterns to 20 sleepers. The ones played in deep NREM sleep didn't help on waking and "impaired their ability to learn this exact same information when awake." The study used sounds, not words, so it doesn't prove a podcast erases vocabulary. It does show the sleeping brain isn't a passive sponge.

What sleep does do: store what you practiced awake#

Sleep isn't for loading in new information. It's for locking in what already got in. Björn Rasch and Jan Born describe, in Physiological Reviews, "sleep as a brain state optimizing memory consolidation, in opposition to the waking brain being optimized for encoding of memories."

New words change status overnight. Nicolas Dumay and Gareth Gaskell taught new spoken words to two groups. Words learned at 8 p.m. and followed by sleep acted like part of the vocabulary 12 hours later, competing with similar real words. Words learned at 8 a.m. and followed by 12 waking hours showed no such effect. In the authors' words, "sleep provides an opportunity for hippocampal information to be fed into long-term neocortical memory."

The sooner you sleep after studying, the better. Steffen Gais, Brian Lucas and Jan Born taught American high school students lists of 24 English–German word pairs. Sleep helped most when it came within a few hours of studying. A night awake after learning hurt recall 48 hours later, even after a night of recovery sleep.

Sleeping between two study sessions works even better. In the study by Stéphanie Mazza and her team, 40 French speakers learned 16 Swahili words over two sessions, some sleeping in between and some not. The sleep group needed half as much practice to relearn the words and, a week later, recalled 15.20 out of 16 words versus 11.25. Six months on, the gap held: 8.67 versus 3.35.

The lab exception: reactivating what you already know#

Nighttime audio does help in one case, and it requires learning first. Thomas Schreiner and Björn Rasch taught Dutch words to German speakers and, during NREM sleep, replayed some of those already-learned words. Their recall improved by almost 10 percentage points; the same replay with participants awake did nothing. Testing came that same night, after about 3 hours of sleep, with two closely related languages. Schreiner said it plainly: "You can only successfully activate words that you have learned before you go to sleep. Playing back words you don't know while you're asleep has no effect."

Nighttime reactivation is fragile. When Schreiner, Lehmann and Rasch added a second sound 200 milliseconds after each word (the correct translation, a wrong one, or a plain tone), the benefit vanished; pushed back to 1,500 milliseconds, it returned. One brief cue, then silence: the exact opposite of a podcast that never stops.

The effect of this reactivation is real but modest. A meta-analysis of 91 experiments with more than 2,000 participants put the effect size at g = 0.29. Xiaoqing Hu and his co-authors conclude that it "can improve memory performance with a small to moderate effect." It works in N2 and slow-wave sleep, not in REM sleep or while awake: a lab finding with EEG, not a recipe for home.

What to do tonight instead of pressing play#

The routine that follows from these studies is simple: practice actively in the evening, then go to sleep. No study has tested it exactly as-is with Spanish speakers or with speaking practice; it applies what the research on vocabulary and sleep shows. Say the phrases you want to keep out loud, recall them without looking, and let the night do its part.

Don't cut sleep to study more. Sleep loss before learning costs you too: in the study by Seung-Schik Yoo and Matthew Walker covered by the Harvard Gazette, people who went about 35 hours without sleep did 19% worse on average at memorizing new material. The authors conclude that "sleep before learning is critical in preparing the human brain for next-day memory formation."

Split your practice across two days with a night in between. Review tonight, sleep, and say the same phrases again tomorrow before looking at them. Stéphanie Mazza's team put it this way: "Sleeping after learning is definitely a good strategy, but sleeping between two learning sessions is a better strategy."

The evening session doesn't have to be long. Tabla is built for 5- to 20-minute sessions where you say English phrases out loud and the app checks your pronunciation: do one before dinner, sleep, and repeat it the next evening. Its personalized podcast of the phrases you practiced is for your commute or a walk, not for falling asleep to.

Tonight you don't need earbuds. You need ten minutes of out-loud practice before bed and a full night of sleep.

Try this today: before bed, say five English phrases you want to keep out loud and get into bed with no audio playing; tomorrow evening, try saying them again before you look.