Show the same sentence in writing and you understand it immediately. Have someone say it and you hear a blurred stream of sound, catch two words and lose the rest. This is the complaint language learners voice most often, and it is almost always blamed on the wrong cause.
The problem isn't speed
Speed gets accused first: "they talk too fast". Yet slow the same recording down and parts of it still make no sense. Because the real issue is this: you know the word, but you don't know its sound.
When you learn a word with your eyes, you invent a pronunciation for it in your head. If that invented sound differs from the real one, you won't recognise the word when it reaches your ear. Then you open the transcript and get a surprise: the word you missed is one you know perfectly well. That is the heart of the listening problem.
Second cause: words stick together
In real speech, words don't stand apart the way they do in a dictionary — they run into each other. "What are you going to do" becomes "whatcha gonna do". You are waiting for four dictionary words and a single block of sound arrives instead; you heard something that isn't in any dictionary, so you start searching for it.
This happens in every language and is routinely left untaught. Textbooks present words separately, clearly and slowly; life speaks by linking, swallowing and accelerating. The only way to close that gap is to listen to real speech rather than textbook audio.
Third cause: where you stall, you miss the train
When an unfamiliar word arrives, your mind stops to decode it. During that one-second pause the speaker has produced three more words, and now you are behind. Shortly after, you understand nothing at all — even though you actually knew the rest of the sentence.
The fix here is not learning more words but learning to let go. Don't stop on the word you missed; stay in the flow. The goal in listening is not to catch every word but to follow the speech.
What actually works
1. Not ten minutes once — one minute ten times. Replaying a short passage does far more than playing a long recording once. Each pass dissolves a little more of it, and the sounds settle into place.
2. The three-pass method. Listen once with no subtitles and tell yourself what you understood. Listen again with the transcript. Then listen once more without it. What you hear on the third pass is the set of sounds you acquired on the second.
3. Find the word you missed. When you open the transcript, mark the spot you failed and ask one question: did I already know this word? The answer is usually yes. Say that word aloud a few times — from now on your ear will recognise it.
4. Shadow the speaker. Play the recording and repeat along half a second behind. This single exercise repairs listening and pronunciation together, because you don't merely hear the sound, you produce it.
5. Pick material at the right difficulty. A recording you understand about ninety percent of is the right material. One you understand half of doesn't teach, it exhausts. Slightly above your level moves you forward; far above it puts you off.
6. Use slowdown as a temporary tool. 0.75 speed helps at the start, but staying there never prepares you for real speech. After a few passes, go back to normal.
The passive listening myth
"Let it play in the background, my ear will get used to it" is a widespread belief and largely wrong. Sound heard without attention is indistinguishable from noise. Twenty minutes a day of active listening — pausing, replaying, checking the transcript — gives you more than two hours of background audio.
How to start today
Choose a short recording: one or two minutes, with a transcript. Run the three-pass method, say the words you missed out loud, and play the same clip once more the next day. A week later, open the first day's recording and you will hear the difference yourself.
If you don't know where to find material at your level, take the free placement test first; the result comes on the CEFR scale. Then work through the dialogues in the free lessons with this method, repeating every line aloud.