Conversational Shadowing: Practicing Reactions

Shadowing means listening to speech and saying it aloud at nearly the same time. The standard version and its variants are covered in /what-is-english-shadowing.

Most shadowing practice is built for monologues. One voice speaks. You shadow that voice. That helps with long turns, but it skips the most common English in a conversation: the listener's side.

Conversation is turn-taking. Little words like "mm-hm", "no way", "right", "wait, really?", and "that makes sense" are the turns listeners take while someone else speaks. They carry the conversation. They show attention. They also shape how a native speaker responds to you.

This page goes deep on one method: shadowing only the listener in a two-voice recording.

• "15-Minute English Podcast" — conversational rhythm
• "No Partner? No Problem!" — three ways to practice alone

The gap: sentence practice does not build reactions

Monologue shadowing trains sentences. You hear a full sentence and you say a full sentence. Real conversation does not run that way. Speak to any native speaker and you will hear short sounds between turns: "uh-huh", "yeah", "hm", "exactly". These are not filler. They tell the speaker whether to continue, slow down, explain again, or stop.

By B1 or B2, you already know what these words mean. The problem is not meaning. The problem is producing them fast, with the right pitch, while the other person is still talking. Textbooks do not drill that. Most shadowing exercises do not drill that.

I have not seen a standard course treat "mm-hm" and "wait, really?" as core material. That is odd, because they do more social work than many full sentences.

Here is a normal exchange with the listener's turns in strong:

Speaker: I got the job.
Listener: No way.
Speaker: They called this morning.
Listener: That's great.
Speaker: I start Monday.
Listener: Wow.

The listener's words are tiny, but they keep the speaker going. Without them, the speaker feels like they are talking into a wall.

Picking a recording

Not every video works. A podcast with two clear voices works best. One main speaker, one listener who reacts. Avoid panel discussions at first: too many voices, too much overlap. Avoid single-voice narration because there is no listener to shadow.

Interviews where the host says "right", laughs, and pushes back are good material. The main speaker should talk most of the time, but the listener must be audible.

Shadow only the listener

Pick an interview or podcast with two voices. Do not shadow the main speaker. Shadow the interviewer or guest who reacts. Every "right", "exactly", "I see what you mean" is your line. The main speaker talks. You produce only the listener's turns, at the same time as the listener.

Example: the main speaker says "So I moved to London without a job and without a place to stay." The listener says "No way." You say "No way" with the listener, matching the speed and pitch. Then the speaker continues. The listener says "Right." You say "Right." Your mouth never forms long sentences.

Do one full interview this way. Then swap roles. On the second pass, shadow the main speaker while the listener reacts over you. That teaches you to hold a sentence while someone says "yeah" or "mm-hm" into your ear.

Timing matters more than words. A late "really?" is worse than none. If the video has already moved to the next fact and you say "really?", you break the conversation. The reaction belongs in the pause, not after it. This is why you copy the recording exactly at first. You are learning when the sound lands, not just what it means.

Pitch changes the meaning

A flat "mm-hm" can sound bored. A rising "mm-hm" can sound interested. Same letters, different message. If you only learn the word, you miss the function. This is why reading a reaction list is not enough. You need to copy the sound from a real voice in a real conversation.

Backchannels like "uh-huh", "hm", and "mm" are often written as if they are all the same. They are not. In an interview, a listener will use them to mean "go on" or "I need a second." The context and the pitch carry the meaning.

Build a reaction stock

A small set of 10 to 15 reactions covers a lot of daily conversation. You want them automatic. When the other person says something surprising, you should not be searching your memory for "no way." It should come out the same way your first-language reactions come out.

Do not read the table once and stop. Pick five. Drill them with a real conversation. Listen for one of these situations and produce the reaction in the pause before the recorded listener does. If you are too slow, loop the line and try again.

Some reactions are not words. "Mm-hm" and "uh-huh" matter because they let you respond without interrupting. They are small, but they are not empty.

Use series dialogues for messier speech

Podcast interviews are cleaner. Series scenes are closer to real talk. Two characters cut each other off and drop words. That is good material for this skill. See /learn-english-with-movies for how to mine scenes.

Pick a short two-person scene. First pass, shadow the character who is mostly listening. Second pass, shadow the character who drives the scene. Do not shadow both at once. The overlap is too thick until you know the lines.

A 30-second scene is enough. Run it five or six times. The words become less important than the rhythm. That rhythm is what you are practicing.

Solo role-play with pause and compare

This is the harder method. Play a conversation. Stop right before the listener replies. Say your own reply out loud. Then play the listener's actual reply and compare.

If the speaker says "The rent doubled, so I left," pause. You might say "That's awful." The video might say "Oh no." Both work. The point is not exact matching. The point is producing a reaction in under a second. If you pause for three seconds, that is the problem to fix.

This combines prediction and production. You predict what kind of turn comes next, produce it, then check it against a native speaker. It shows gaps that understanding alone does not show. Many learners can understand "The rent doubled, so I left" but cannot produce a quick natural reaction to it.

Practice inside the app

English Shadowing runs in the phone browser. You paste a YouTube link or search inside the app. For this practice, pick a podcast-style video with two clear voices. The app highlights the currently spoken word and shows a translation under each word, but the main job here is the listener's short turns.

When a reaction flies by too fast, tap the sentence to jump back to it. Use the repeat button to loop one sentence. Slow the video to 0.75x or 0.5x if the timing feels tight, then move back to full speed.

Tap any word to save it to a vocabulary list with spaced-repetition flashcards. For reactions, the flashcard is secondary. The stronger memory comes from saying the word in rhythm, but the list helps you keep a personal set of reactions you have actually heard.

The app is at https://englishshadowing.app/. It is free and there is nothing to install.

Where this shows up

Reaction shadowing pays off fastest in /talk-with-native-speakers. Native speakers read the small sounds you make while they talk. A well-timed "right" or "oh no" makes them feel heard. A silent listener can seem confused or cold, even if you understand everything.

It also changes what you notice in your own speaking. You stop treating conversation as a list of full sentences and start hearing the gaps between them. That is part of /speak-english-fluently too.

Start with four minutes of an interview. Shadow only the listener today. Tomorrow, do the same four minutes and swap roles. After two weeks, add one series scene and the pause-compare exercise. That is enough.