Why Your TH Sound Still Sounds Wrong (And How to Fix It)

September 15, 2026

The TH sound causes more trouble for second language speakers than almost any other sound in English. It shows up constantly, you can barely finish a sentence without it, but it barely exists in most other languages. That combination, extremely common and extremely unfamiliar, is exactly what makes it so hard to fix on your own.

There are three specific things that have to work together to make this sound accurately. Most people are only getting one or two of them right, which is why the sound keeps coming out close, but not quite there.

1. The tongue actually has to clear the teeth

There is no way around this one. The tongue has to bite down, just slightly, between the teeth. You do not need to stick your tongue far out to do this, a small amount of clearance is enough, but that clearance has to actually happen. When it does not, the sound collapses into something else entirely, often closer to an S or a T, which is the single most common mistake second language speakers make with this sound.

A simple way to build awareness here is to practice in front of a mirror or record yourself on video. Watch specifically for whether the tongue is clearing the teeth, or whether it is staying behind them, pointing up, or dropping down. You cannot fix a placement problem you cannot see.

2. Control the airflow like a wind instrument

Even with the tongue in the right place, the sound will not land if the airflow is not controlled. Think of it the way a saxophone player thinks about a held note, there has to be steady, controlled air moving past the tongue for the sound to actually sustain.

A useful drill: hold the sound out for as long as you can, well past how it would ever actually be used in a sentence. Extending it artificially, almost like sustaining a musical note, builds the airflow control you need. Once you can hold it steadily, you can bring it back down to a natural length and the sound holds together.

3. There are two different TH sounds, not one

This is the one most people never realize. English actually has a voiced TH and a voiceless TH, and mixing them up is immediately noticeable to a native listener, even though the physical difference is small.

The clearest way to hear the difference is through a minimal pair like close and close, where the only difference between the two words is whether that final sound is voiced or not. The exact same swap happens with TH. A word like with uses the voiceless version, but many learners default to the voiced version instead, which native speakers register as subtly, but clearly, wrong.

How to actually practice this

Awareness comes before accuracy. Most learners go through three stages: first you cannot hear the mistake at all, then you start noticing it after the fact, and eventually you catch yourself mid sentence and self correct in real time. That self correcting stage feels awkward, but it is a genuinely good sign, it means your ear is catching what your mouth used to miss.

A practical way to build practice material: ask an AI tool to generate a short paragraph at your level, and ask it to bold every TH sound and mark whether each one is voiced or voiceless. Read that paragraph out loud repeatedly. Once individual words feel steady, move up to full sentences, then paragraphs, then real conversation, which is always the hardest stage since there is no script to fall back on.

This last stage, real conversational use, is usually where a second pair of ears makes the biggest difference. Self correction gets you a long way, but catching your own patterns in live conversation is exactly the kind of thing coaching exists to speed up.

If the TH sound is something you have been quietly working around for years, there is a good chance it is not the only pattern worth a closer look. Join the Accent Forge community to work through this and the rest of the sound system properly, with feedback from an actual coach along the way.