Most “learn English faster” advice sounds confident. It also gets the science backwards.

A YouTube video told me to repeat after native speakers in real time, drill my top 500 sentences every day, and listen to English I understood about 70% of — and that the strain, plus eight or nine hours of sleep, would rewire my brain. The advice was specific. The confidence was high. Almost none of it works the way the video says.

So I went back and read the research. Four claims, four corrections, then what I actually do in my classroom here in Fujieda. If you want a worked example of how I place a new student, there’s the diagnostic session.

In this guide: four popular shortcuts, what the research found, and what to do instead. Read time: ~6 minutes.

Table of Contents

Myth 1: Copying native speakers will fix your pronunciation

The claim: put on headphones, listen to a native speaker, repeat what they say in real time. Your pronunciation will improve fast. This is usually called shadowing.

Here is the problem. What changes your pronunciation is not the repeating. It is someone telling you which sound came out wrong, and how to fix it.

Researchers tested this on Japanese learners working on the English “r” — one of the hardest sounds to add if you grew up speaking Japanese. One group practised in ordinary speaking activities with nobody correcting them. They did not improve at all. The other group got corrected: the teacher said the word back the right way, right after the mistake. Only that group improved.

A 2024 review gathered fifteen separate studies of pronunciation software and found the same pattern. Software that showed people exactly which sound was wrong worked well. Software that only marked them right or wrong — which is about what copying a speaker gives you — worked much less well. Practising alone barely helped. Practising with another person worked as well as the best software did.

Two things worth knowing before you pay for a year of it.

Copying is usually sold as the way to fix your rhythm and intonation — the music of English, where your voice rises and falls. That is the part the research supports least. The gains turn up on single sounds, like the “r” in right or the “th” in think. Not on the music.

And short bursts do nothing you could measure. One to four weeks of practice showed almost no change at all. Improvements only started showing up after about five weeks.

Copying can be useful. It is not enough on its own. The repetition only starts doing real work when somebody tells you what you missed.

Myth 2: Drill 500 sentences a day and the grammar will stick

The claim: write out your 500 most useful sentences. Drill them daily. Depth over breadth, the way an athlete drills the basics.

There is nothing new about this. It is how English was taught from the 1940s through the 1960s. It had a name — the Audiolingual Method — and teachers stopped using it decades ago. Here is what happened when someone tested it.

In 1983 a study followed learners in Quebec who were drilling grammar patterns exactly this way. Inside the drill, they were accurate. Outside the drill, they could not use the same forms at all. When the drill stopped, or the sentence pattern changed, what they had drilled fell apart completely. They went back to speaking the way they had before the drills started.

The research on vocabulary points the other way. What builds a language is meeting thousands of words, in lots of different situations, spread out over time. Not 500 sentences on a loop.

Grammar does not get hammered in. It settles once you have met the same pattern in enough different places that you stop having to think about it. The repetition that helps is meeting real language again and again — not reciting the same sentences.

Myth 3: Understand about 70% and your brain adapts

The claim: find podcasts or videos where you understand about 70%. Your brain adapts at the edge of what it can handle.

The number is backwards, and not by a little.

Researchers have measured how many unknown words a reader can carry before a text stops working. To follow what you are reading and pick up new words from it, you need to already know about 98% of the words on the page. That is one unknown word in every fifty.

(That figure comes from research on reading. I have not found a comparable one for listening, so treat it as the reading number rather than a universal one. Either way, nothing in the research sits anywhere near 70%.)

At 70%, you spend all your attention working out the words you do not know. Nothing is left over to notice how the sentence is built, or to pick up anything new. The thread of meaning breaks. You get tense, and then you stop listening.

There is a famous idea behind the video’s advice, and it is half right. Stephen Krashen argued that we learn from language a little above our current level. The trouble is that he never said what “a little above” means, and nobody since has found a way to measure it. Researchers have been making that criticism since 1978: you cannot test an idea you cannot measure.

Vocabulary researchers have offered a working stand-in. Counted in actual words, “a little above your level” comes out at somewhere between two and five unknown words per hundred. Not thirty.

A simpler rule: if you can follow the story, you are in the right place. If you cannot, drop down a level. The edge is not a percentage you calculate. It is the feeling that the meaning is still coming through.

Myth 4: If it feels bad, your brain is growing

The claim: learning a language feels hard because your brain is being forced to grow. Sleep eight or nine hours and the breakthroughs arrive.

The first half is wrong, in a specific way. Being anxious does not grow your brain. It gets in the way of your brain.

There is a middle ground, and it is worth being fair about. A bit of challenge does keep people alert and stops them drifting off. Psychologists have mapped this: performance climbs as pressure rises, up to a point, and then it drops. Being completely comfortable is not the target either. The danger is on the far side of the peak.

Past that peak, several findings line up. Anxiety gets in the way at every stage — taking English in, sorting it out, and getting it back out again. An anxious brain is splitting its attention between the language and the feeling.

One study tested this directly. Learners who felt relaxed got much more out of being corrected than learners who felt anxious. Same correction, different result, depending on how safe the person felt.

Work on how the brain handles threat explains why. When your brain decides a situation is a threat, it switches into defence. Thinking flexibly gets harder, and so does forming memories that last. The defensive reaction gets in the way of the learning you are consciously trying to do.

Effort does help memory — spacing practice out over time, meeting the same thing in different situations. But that only works when the person is not anxious. Effort and anxiety are not the same thing, and treating them as the same is where this myth comes from.

Then there is the sleep. Sleeping well is good advice, and things really do settle in your memory over time. But I could not find a study that ties any particular number of hours to learning a language. Sleep because it helps you think clearly. The eight or nine hours is the video’s number, not a research finding.

A teacher who makes the room feel safe is doing more for your English than one who makes it feel like a test.

What actually works

Pronunciation practice with someone correcting you. A teacher saying it back the right way. Drills on pairs of words that differ by one sound — right and light, think and sink. Software that scores each sound separately and tells you which one you got wrong. The correction is the active ingredient, not the repetition. AI voice chat is not this — it is built to understand you, not to judge how you sound, so a wrong sound usually just passes. I wrote more about that in using an AI chatbot wisely.

Lots of words, not a fixed set of sentences. Thousands of words met in many different situations, with the timing spread out so you meet each one again just before you would forget it. The Leitner box guide and the Eiken retrieval practice post are worked examples.

Reading and listening you mostly understand — around 95 to 98% of the words already familiar. Stories, podcasts and graded readers where the meaning stays clear. Just listening will not give you accurate grammar on its own; you need someone pointing at the parts that are not sticking. The cooking and English post shows how a relaxed, hands-on task takes the pressure off and lets the listening do its work.

A classroom where mistakes are expected. Practice pays off in a room that feels safe. You get more out of the same exercise with a teacher who lowers the tension than with one who raises it. That is the kind of learning environment I try to create at Starfish English.

Book a Free Trial Lesson

The bigger picture

All four myths share one thing. They promise a shortcut through the discomfort.

The research points somewhere less exciting. English you can mostly follow. Practice with someone who tells you what you missed. Enough time, spread out, for it to settle. And a teacher or a partner who makes the room safe enough that you keep turning up. People move through that in different orders and at different speeds, but the direction holds.

Nobody is going to build a video around “understand almost every word, and keep going for five weeks”. It is still closer to the truth than what the video was selling.

Learn About the Free Trial Lesson