Symbol PrimarySymbol Primary

3B: Pronunciation Training

Having a pleasant and easy-to-understand accent opens the door to quickly building strong connections with native speakers. Improving your accent reduces the friction in conversation, making it easier to make friends, impress a date, or get a job.

Pronunciation is largely a physical skill and requires physical training to master. Unfortunately, no matter how much input you get, input alone won’t guarantee good pronunciation.

Input gives you the ability to accurately perceive the sounds of your target language (TL). This enables you to self-correct your pronunciation.

Without this ability, achieving a native-like accent simply isn’t realistic. Native correction and accent elimination classes can help, but there is a limit to what you can accomplish if you can’t hear your accent.

Muscle Memory

The Refold logo is an origami crane because second language learning is like unfolding the crane and trying to use the same piece of paper to fold a new shape.

The challenge with folding a new piece of origami from used paper is that the paper retains the memory of the original shape and it is difficult to avoid following the same creases.

This metaphor applies perfectly to muscle memory in your mouth. A lot of the reason foreign accents occur is that your mouth is already trained to speak a certain way, and deviating from this is extremely difficult.

To speak without an accent, you need to build up the strength and precision of the patterns of mouth movements that native speakers use, and you need to relax the muscles that native speakers don't use.

Reprogramming Your Mouth

When you start speaking your TL, your mouth will hurt. You’ve never used your mouth like this before.

Treat your mouth like you would any other muscle in your body. This means warmups, stretching, and workouts. For the mouth, these are called "Articulation Exercises".


There are many types of articulation exercises. Some will be dependent on the language.

Mouth Relaxation Techniques

Your existing muscle memory will fight to take over when you try to speak. When you find this happening, try using mouth relaxation techniques that singers and public speakers use.

Mouth Stretching and Warmups

To speak your TL, your mouth needs to move completely differently than you are used to. Stretching your mouth can help loosen up tension in the muscles you use for your native language and allow them to move more naturally in your TL.

Also, when you find your mouth hurting from speaking your TL, stretching exercises can help the muscles recover more quickly.

The first half of the following video shows how to stretch your mouth. The second half is a set of articulation exercises specifically for English learners.

Tongue Twisters

Tongue Twisters are deliberately difficult expressions that sometimes generate humorous results when mispronounced. The difficulty with tongue twisters is the rapid alternation between similar (but distinct) sounds in the language.

Most languages have tongue twisters and they can be useful for training your mouth's ability to quickly move between different sounds.

Enunciation Exercises

These are like tongue twisters, but they usually aren't words or sentences. They are consonant and vowel combinations that alternate in sequence to help improve speaking precision and enunciation.

In the following video, you can see examples of these kinds of exercises for English:

Specific Sounds

Sometimes, you will struggle to produce a specific sound in your TL. The rolled-R sound is an example of a common issue for learners.

Learn the name of this sound and search Google for practice guides. Practice these problematic sounds daily until you can produce them.


The above exercises are helpful, but to develop natural pronunciation, you need to practice with real language.

Some people do this by recording themselves reading a book out loud and then listening back. The problem with this method is that you don't have a direct target to compare yourself against.

Instead, we recommend shadowing your parent.

What's Shadowing?

Shadowing is the practice of listening to your language parent, repeating back what they say, and then comparing your output to theirs. Because you have a direct target, you can accurately notice differences between their pronunciation and yours. You can then specifically practice these discrepancies, creating a feedback loop that aligns your pronunciation with theirs.


We don't recommend shadowing until you have level 5 comprehension of your parent. You should be able to clearly hear every word your parent says when they speak.

If you can't accurately hear your parent, you won’t be able to identify the differences between your speech and theirs. This can lead to building incorrect muscle memory and poor pronunciation which will be difficult to undo later.

Learn to Shadow in your NL

Shadowing is a skill that takes some practice to learn. Your brain isn't used to keeping exact wording in your short-term memory. Usually, it just grabs the meaning and discards the specific words.

To build this skill, practice shadowing in your NL first before trying with your TL.

Types of Shadowing

There are two different types of shadowing: continuous and perfect sentence. They each have different pros and cons, so we recommend doing a mix of both.

Continuous Shadowing

Continuous shadowing consists of listening to your parent and repeating everything they say out loud in real-time. If you make a mistake or fall behind your parent, you just keep going.

The non-stop aspect of continuous shadowing makes it conducive to entering a flow state, which is ideal for building muscle memory. The downside is that it can be a bit chaotic, making it harder to hone in on specific sounds you’re struggling with.

Check out the following video to see an example of shadowing in English:

In the beginning, you won’t be able to keep up with your parent. This is normal. The key is to just keep going without worrying about accuracy. Without pausing or rewinding, replicate your parent as best you can for the duration of the session.

It can be helpful to shadow audio you’ve already listened to once before. That said, we don’t recommend shadowing the same piece of audio multiple times.

Once you become comfortable with simply keeping up, shift your focus to how closely your pronunciation resembles theirs. Your attention should be split evenly between your parent and your voice. The goal of shadowing is to tune your pronunciation, so don’t worry about the actual content of what your parent is saying. Noticing discrepancies between your pronunciation and theirs will naturally make your pronunciation more accurate over time.

For an in-depth explanation of continuous shadowing check out the following video:

The simplest way to do continuous shadowing is to listen to audio of your parent speaking through either speakers or headphones, and repeat back what they say out loud. For this to work properly, you must be able to hear both your parent and yourself.

Finding the right volume level can be difficult. People often report that they either can’t clearly hear their voice over the audio or can’t clearly hear the audio over their voice.

You can get around this issue by using a microphone or audio interface that allows you to hear your voice through headphones. This also has the benefit of allowing you to hear your voice more objectively. You can either have your voice coming into one ear and your parent’s coming into the other or have both voices coming into both ears.

When using this technique, make sure you have an audio setup with minimal lag for your voice. If the lag is more than a few hundred milliseconds, it will completely interrupt your ability to speak. In fact, there is a device called a "Speech Jammer" that does exactly that.

Perfect Sentence Shadowing

Perfect Sentence shadowing is when you listen to an individual sentence spoken by your parent, record yourself imitating that sentence as closely as possible, and then compare your recording to the original to pick up on any discrepancies.

Perfect sentence shadowing is great for honing in on specific sounds or sound combinations that are giving you trouble. The downside is that the discontinuous nature makes it less effective for training overall rhythm and flow.

Use an app like Audacity to isolate a single sentence clip of your parent speaking. Listen to the sentence a few times to load it into your short term memory, then record yourself imitating the sentence as closely as possible. Listen back to both your recording and the original a few times, and notice whatever discrepancies you can. Then, re-record yourself pronouncing the sentence and attempt to reduce the discrepancies. Repeat this until you feel like you’ve plateaued, and then move on to a new sentence.

Tools like Audacity also allow you to overlay the two pieces of audio. This makes identifying discrepancies very easy. When you make a mistake, it will create dissonance which sounds bad to your ear. Check out this guide for an explanation on how to overlay audio.

The term “perfect sentence technique” was coined by the polyglot Marc Green to refer to a similar method. In Green’s technique, you sit with a native speaker and read a sentence out loud to them. They give you feedback on how you sound and then read the same sentence to you. You repeat this until you're happy with how you sound.

In Refold, by the time you practice perfect sentence shadowing, you’ve already developed your ability to accurately perceive the sounds of your TL. This allows you to function as your own partner and give yourself feedback. That said, if you ever feel that your pronunciation development is stuck, it can be helpful to find a native speaker and apply Green’s original method.


This content is funded by Patreon. If it helped you and you want to see us build more guides and tools, please support us!