Shadowing technique: how to practice it for listening and speaking

Shadowing is a practice technique where you listen to audio in the language you are learning and repeat it out loud almost simultaneously, like a shadow half a second behind the speaker. It trains listening, rhythm and intonation. It does not replace real conversation or feedback from a teacher.

The difference from repeating sentence by sentence is time pressure. When you pause and repeat, you can rebuild the sentence from what you read. When you follow the audio without stopping it, you have to process sound while you speak, and you quickly notice what you are not hearing: words that run together, weak syllables, and endings that almost vanish in fast speech.

Hamada (2016) studied shadowing with Japanese university students learning English as a foreign language. The title of the paper asks the useful question: who benefits, and how? Gains in listening comprehension were concentrated in the lower-proficiency group, while the intermediate group did not improve in the same way. It is a small study in one language and setting, not a guarantee for every learner.

To decide whether it fits you, start from your goal. The Common European Framework of Reference (CEFR) descriptors, published by the Council of Europe, describe what listening and spoken production look like at each level. If your aim is to follow conversations at natural speed, shadowing makes sense as a support activity. If vocabulary or grammar is what holds you back, other activities will give you more for the same time.

The material decides half of the exercise. Audio that is too hard leaves you mumbling without understanding, and audio that is too easy teaches you nothing new. Look for clips of 30 to 60 seconds that you understand almost completely when you read the transcript, even if some parts slip past you without it.

Spend 10 to 15 minutes on a single clip. Repeating it several times is part of the method: the first pass shows you the problem, and the following passes let you fix it before you move to new audio.

Take this sentence from a dialogue written for this guide: “I was going to call you, but my phone ran out of battery.” On paper it looks simple. At normal speed, “was” will probably sound weak, “going to” may come close to “gonna” in casual speech, and “ran out of” arrives as one block, with no gaps between the words.

On your first pass you might manage “I was going to call you” and lose the ending. Mark “ran out of,” repeat it five times copying how the sounds link, then go back to the full sentence. When you play your recording, check only two things: whether you link those three words, and whether you stress “phone” and the first syllable of “battery.” Fixing one point per session works better than trying to sound identical to the recording.

Shadowing is tiring, which makes it easy to slide into automatic mumbling that fills time without teaching you anything. These signs tell you to change the clip, shorten it, or go back to the transcript step:

Lumia AI does not listen to your recordings or score how you sound, so that comparison is yours to make, or a teacher's. Where it can help is the vocabulary in the clip: you can create flashcards by hand with phrases from the transcript, such as “run out of” with its meaning and your own example sentence, and add an optional hint.

If the audio is on YouTube, Lumia can generate flashcards or a quiz from the link when the service can access the video, and in the tutor chat you can ask about an expression at a specific moment. Check the draft against the transcript before you study it. During review, rate each card Again, Hard, Good or Easy based on what you actually recalled.