Set Up Shadowing Practice: Repeat a Line, Hide the Text, Record a Take
Set up one shadowing line: repeat with text, hide it to test listening, then record one no-text take as your baseline.
Use the transcript only long enough to make the wording and meaning clear, hide it while the audio still plays, then record one no-text take.
If your best shadowing only works while the sentence is glowing on-screen, you may be practising the sentence with your eyes.
Repeat: use the text as support, not as the task
Text support is allowed. British Council shadowing guidance explicitly recommends a copied text when learners need extra help, and its pronunciation guidance also recognises a listen-pause-imitate version of shadowing. The useful boundary is not “text or no text?” It is whether the text is helping you hear the line or quietly replacing the listening. See the British Council shadowing activity and British Council pronunciation guidance.
Choose one short line. Read it once for meaning. Then listen and repeat. If you need to pause after the model at first, pause. Your first job is to stop wondering what the sentence says so you can notice how it is said.
Original practice example
Could you let me know when you get there?
Let me know means “inform me” or “tell me.” “Tell me when you get there” is also grammatical and natural, but it is a little more direct. While the text is visible, make sure let me know is one meaningful chunk rather than four separate words.
Use this when you want someone to message or call after they arrive.
Listen, pause if needed, and repeat the line while looking at it. Stop once the wording and meaning are clear enough to test without the text.
Explore more language-learning guides in Media-Based Language Learning.
The transcript is allowed to help. It is just not allowed to move in permanently.
Hide: keep the audio and remove the visual support
Now hide the text but keep the same audio line. This is not a memory contest. The audio is still the cue.
Ask one question: What breaks when the text disappears?
If you lose one weak word, one short chunk or one connection, good—that is useful information. You have found the place where the screen was doing more work than your ears.
Original practice example
I didn't mean to interrupt.
Mean to + verb is the standard pattern for intention. “I didn't mean interrupt” is wrong in standard English. In natural speech, to can be weak, so it is exactly the kind of word a learner may keep seeing in print but stop hearing when the text disappears.
Use this to apologise for interrupting someone unintentionally.
Hide the text and replay the line. If to disappears from your perception, reveal only mean to interrupt, listen again, then hide it immediately.
Reveal only the part that actually broke
The easy mistake is to reveal the whole transcript again after one small failure. That turns a precise diagnosis back into general reading.
Instead, repair locally:
What broke when you hid the text?
I lost one word or short chunk.
Reveal only that chunk. Listen again, hide it, then retry the whole line.
I lost most of the sentence.
The line is not ready for the no-text pass. Return to one supported repeat, confirm the meaning, then try again. Blind failure is not a badge of honour.
I can follow the audio without text, but my mouth falls behind.
Keep the text hidden. Your next problem is timing or production, not visual support.
I can say it without text, but I think I memorised the printed sentence.
Finish this setup anyway. Keep the audio as the cue and make the recording. A fresh-clip check is the better way to test recitation later; do not turn this setup into a memory exam.
I can follow and reproduce the line.
Record it. You have reached the end of the setup.
This local reveal rule matters because text-presented shadowing is a legitimate scaffold, not something the learner has to ban completely. Hamada and Suzuki’s review describes written-script support as one option when learners struggle to keep up, while also treating shadowing as a family of techniques that can be adjusted to the learner and task. See the shadowing-technique review.
Record: capture one no-text take
Once the line survives with the text hidden, record one take.
Do not wait for a perfect take. Do not record twelve versions looking for the prettiest one. The first useful recording is simply evidence of what your ears and mouth can do once the visual support is gone.
Original practice example
I would've called you if I'd known.
Would've means would have, and I'd known means I had known. “I would called you if I knew” is wrong for the intended past counterfactual meaning. The contractions also make this a useful line for testing whether you can follow natural compressed speech without reading each word.
Use this when you are explaining what you would have done differently if you had known something earlier.
Hide the text, play the model, then record one attempt. Keep the take even if one transition is imperfect.
Original practice example
Would you mind sending it again?
Would you mind + -ing is the standard pattern. “Would you mind to send it again?” is wrong in standard English. The phrase is polite and useful, but the setup goal here is not grammar study forever—it is to get the exact line from supported repetition into one no-text take.
Use this when asking someone politely to resend a message, file or photo.
Record one no-text version, save it, and stop. Detailed review belongs after setup, not inside it.
Your first recording is a baseline, not a score
British Council guidance suggests recording and comparing pronunciation, but your first take does not need to answer every pronunciation question. Its job is smaller: show what survived after the transcript disappeared. See the recording guidance.
You can label the file simply: line, date, first no-text take. That is enough.
There is no magic repetition quota here. The point is to reach a useful test, not a magic number.
Make the text leave before the practice ends
A good setup does not prove that you have mastered the line. It proves something more modest and more useful: the transcript helped you build access to the audio, and you tested that access without looking.
Repeat. Hide. Record. Let the text be a scaffold, then let your ears take over.