What shadowing is
Shadowing is the near-simultaneous vocal imitation of speech: you hear a phrase and repeat it with only a slight delay, word for word, in the same language — described in the research literature as "immediate vocalization of presented auditory stimuli … word-for-word repetition, in the same language, parrot-style." It comes from the training of simultaneous interpreters, who must listen and speak at once, and it has since become a mainstream language-learning method. The point is not translation or comprehension in the moment — it is copying the sound: the rhythm, stress and intonation of a fluent speaker.
What the research actually shows
Shadowing is unusual among popular techniques in having a real evidence base. A 2025 systematic review that gathered dozens of studies concluded that shadowing training "can help improve learners' comprehensibility, intelligibility, and accentedness, as well as certain aspects of suprasegmental pronunciation control, such as fluency and prosody." In plain terms: done consistently, shadowing tends to make learners easier to understand and more fluent in rhythm and flow.
That is encouraging, not a promise. The gains in the research come from sustained, correct practice, and studies vary in size and design. Treat shadowing as a well-supported method to practise, not a guarantee of a particular result.
How to shadow, step by step
Choose audio slightly below or at your level, ideally with a transcript, and keep clips short. Then run this loop:
- Listen once. Play a short segment and just listen — get the overall sound and meaning.
- Shadow with the text. Play it again and speak along, a beat behind, reading the transcript. Focus on matching rhythm and stress, not perfect words.
- Shadow without the text. Play it once more and shadow from listening alone. This is the core rep — copy the sound in real time.
- Record and compare. Record yourself shadowing, play it against the original, and pick the single biggest difference to target next time.
A realistic weekly dose
Consistency and correct repetition matter more than length. In one published study, learners shadowed for roughly 10 minutes at least four times a week across eight weeks and showed measurable gains in comprehensibility and fluency. That is a good template to copy:
| Variable | Realistic target |
|---|---|
| Session length | ~10 minutes — long enough to warm up and get reps, short enough to sustain daily. |
| Frequency | At least 4 times a week; daily is fine if sessions stay short. |
| Duration | Give it several weeks (the study ran 8) before judging progress — pronunciation change is gradual. |
| Material | Short clips at or just below your level, reused until they feel automatic before moving on. |
Shadowing with an AI partner
Shadowing's weak point when you practise alone is feedback: it is hard to hear your own gaps. This is where an AI speaking partner helps. FluentDojo gives you per-sentence pronunciation scoring, so after shadowing a model you can say the same lines to the AI and see which sounds and stress patterns still differ, then drill that one fix. You can also move from imitation into real-time conversation with the same coach — turning a solo copy-drill into responsive practice. If speaking feels hard at first, type to the AI and build up to voice.
Free users get 3 voice sessions per day and 20 per month after sign-up (members: 15/day, 200/month) — a 10-minute shadowing-plus-feedback session fits in the free tier. Quotas describe available sessions, not a result.
What it can and can't do
FluentDojo is a browser-based AI English speaking and pronunciation partner — no native app to download. Shadowing plus per-sentence feedback is a well-supported way to work on pronunciation and fluency, but it is not a teacher and does not promise fluency, a full accent change, or an exam score. For official exam rules, use the testing body's resources.
Pair shadowing with solo speaking practice and targeted pronunciation practice, or fold it into the 30-day routine. Browse all speaking guides.