The shadowing technique: what it is, and how to practice it
Copy a voice you can hear, out loud, in real time. It is the oldest trick in pronunciation practice, and it still works because it forces your mouth to keep up with your ear.
Ookie · Updated 9 October 2026 · 6 min read
- Shadowing means speaking along with a model, not waiting for it.
- Short sessions work better than long ones.
- One correction per take is the whole point.
Five steps that keep it short
01Pick one sentence you actually care about
A line from something you were reading. If you would not say it out loud in a conversation, skip it.
02Hear the model twice before you speak
Once to understand it, once listening only to the rhythm. Where does the voice rise, and where does it land?
03Speak along, not after
Start slightly behind the model and stay with it. This is the part that feels strange for a minute and useful by the third.
04Record yourself saying it alone
No model this time. This is the take you compare, and the one Ookie scores.
05Fix one sound, then move on
Not four things. One. Tomorrow, the same loop with a new sentence.
Try it on this sentence
Read it twice, shadow it once, then record yourself. Listen to appreciate and it join into one word.
Thank you so much, I really appreciate it.
Ookie does the boring parts: it plays the sentence in your own voice with a native accent, records your take, and tells you the one thing worth fixing. You still have to say it out loud.
Questions
How long should a shadowing session be?
Five to ten minutes is plenty. Short and daily beats long and occasional, and your attention is what runs out first.
Do I need a native speaker to listen to me?
No. You need one honest comparison: your take, then the model sentence. Hearing the two back to back is what shows you the difference.
Is shadowing useful for beginners?
Yes, as long as the sentence is short enough that you can finish it. Start with five to eight words.
Sources
- Whitworth 2024 — pronunciation feedback and noticing
- Probst 2002 — shadowing in second-language listening
- Ding 2019 — shadowing and speech production