Listening & speaking · 6 min read
Shadowing Works Better When You Know What You Are Copying
Practice rhythm and phrase boundaries with a short clip, then check whether you can understand and adapt what you repeated.
Published Updated

Shadowing asks you to speak along with, or just behind, a recording. It can feel satisfying to keep up with the sound. But keeping up does not show that you understood the sentence or could use it without the speaker.
Give the exercise a narrower aim. You might copy where a speaker pauses, the stress in a question, or the connection between two familiar words. Then finish with a small change to the sentence. That final step shows you what belongs to your own speech and what still depends on the recording.
A model you can use now
Use this supplied request as your first model. Its meaning is a polite request for water; the changed request asks for tea. Listen, repeat after it, then turn the sound off and request a different drink. Keep “a glass of water” together rather than pausing after every word. Device speech illustrates a possible reading, not the only appropriate rhythm.
Model transcript
Could I have a glass of water, please? Could I have a cup of tea, please? I am cold.
The transcript makes this a usable reading task when audio is unavailable. A synthetic model is a practice aid, not evidence of real-world listening or pronunciation ability.
Pick a phrase you can explain
Start with one or two sentences of clear audio and an accurate transcript. Choose a speaker whose language variety fits your current goal. For a first attempt, avoid jokes, background noise, and several people talking at once. Those can be useful later, but they make it hard to identify what you are copying.
Read the phrase and explain its meaning in plain language. You can use a translation for this check. If several words remain unknown, study them first or choose an easier phrase. Trying to match sounds while also guessing meaning gives you too many tasks at once.
A question such as 'Could I have a glass of water, please?' is useful because you can later change the requested item. You should know what the question is asking before trying to reproduce its rhythm.
Mark one feature on the transcript
Listen without speaking and mark the pause before 'please,' if your recording contains one. Then identify the words that receive more emphasis. Do not assume every recording will stress the same words: a speaker correcting an order might emphasize 'water' differently from someone making a routine request.
Choose one feature to follow. For example, keep 'a glass of water' together instead of pausing between every word. Speak after the recording first. Once that feels manageable, try speaking just behind it. If you repeatedly lose your place, shorten the phrase rather than pushing through a longer clip.
Lower the recording volume enough to hear yourself. When the source audio overwhelms your voice, it can conceal missing syllables and extra pauses. You can also alternate one repetition with audio and one without it.
Compare one detail instead of your whole accent
Record a short attempt if you want to inspect it. Compare the phrase boundary you marked, not whether your voice sounds like the speaker's. Different voices have different pitch ranges and qualities. Those differences do not by themselves tell you whether the question is understandable.
Write a precise observation: 'I pause after glass,' or 'I rush the last word and lose please.' Then repeat with attention to that detail. A note saying 'sounds bad' is difficult to use, even if it describes how you feel listening back.
A pronunciation tool or listener can provide another perspective, but automated scores should not be your sole measure. A score may respond to recording quality or the system's recognition limits. If the score and a knowledgeable listener disagree, investigate the sentence and recording before assuming your accent needs to change.
Move through supports without calling every stage independent speech
Label four attempts with the same understood request. First, listen while reading without speaking. Second, repeat after the audio with the transcript visible. Third, speak just behind the audio without text. Fourth, turn off the audio and request a different item. Each stage supplies different support. A strong third attempt can still depend on the speaker's timing; the fourth asks you to choose your own words.
For 'Could I have a glass of water, please?', change the request to 'Could I have a cup of tea, please?' Then add a reason you understand, such as 'I am cold.' If only the added reason stalls, keep the request intact and prepare that small addition. If the request itself disappears without the model, step back to repeating after audio and then retrieve it from a situation cue. Do not speed up an unsupported sentence to imitate progress.
Ask a listener what you requested after the no-model attempt. If the item is understood, check the phrase feature you were practicing rather than demanding a complete accent judgment. Without a listener, compare the recording to your intended item and phrase boundary, and mark listener confirmation as not available. A practice record should distinguish what you heard yourself from what another person actually understood.
Leave the recording and make your own request
Turn the audio off and ask for a cup of tea instead of a glass of water. Keep the polite request structure while adjusting the phrase. If the new sentence falls apart, return to speaking it slowly with meaning, then try it in context. Shadowing gave you a model; adapting it reveals a different skill to practice.
You can also change the situation. Say the request to a friend, then to a server. Consider how direct or formal you want to be, rather than carrying one recording's delivery into every exchange. A teacher or course can help with the differences between those situations.
Use shadowing as one part of practice. It does not replace listening for new information, answering unexpected questions, or receiving feedback. Stop if matching a speaker becomes physically uncomfortable, and avoid forcing your voice into an unsuitable pitch.
Apply what you read · no account needed
Check your understanding
Separate supported imitation from independent adaptation and choose an appropriate next step when a phrase breaks down.
Use it in a different situation
Record the same request with a model and without it using a different item. Compare the requested meaning and one phrase feature; note which supports were present.
This is a self-check, not a language-level assessment. Your answers stay on this page and are not sent to our server.




