The six Vietnamese tones, explained by what they do to your voice

Every Vietnamese syllable carries a tone, and changing it changes the word. The textbook example is honest but overused: ma ghost, má mother, mà but, mả grave, mã horse, mạ rice seedling. Six words, one set of consonants and vowels, six different pitch shapes.

What most explanations get wrong is describing tones as heights. They are not. They are movements — what your pitch does across the syllable — which is why a learner who has read six descriptions still cannot produce them and a learner who has copied fifty recordings can.

The six

Tone	Mark	Example	What your voice does
ngang	none	ma	Stays flat, mid-level. The default.
huyền	`	mà	Starts a little low and falls further. Breathy, relaxed.
sắc	´	má	Rises sharply from mid to high.
hỏi	̉	mả	Dips down, then comes back up. A question-mark shape.
ngã	~	mã	Rises, breaks in the middle with a catch in the throat, then rises again.
nặng	̣	mạ	Drops fast and stops short. The syllable is cut off.

Two of those need more than a row.

ngã is the one that has no English equivalent at all. The “break” is a glottal stop — the catch in the middle of uh-oh. Say uh-oh and notice what your throat does between the syllables; now do that inside a single syllable while your pitch is rising. That is ngã.

nặng is not just low. It is short. The vowel gets clipped and the syllable ends abruptly, as though you stopped it with your throat. Learners tend to produce a long, low tone instead, which sounds like a strange huyền.

The pairs that actually cause trouble

Not all six are equally confusable. In practice, learners collide on three pairs:

• sắc and ngã — both end high. The difference is the break in the middle of ngã. If your ngã has no catch, it is a sắc.
• hỏi and nặng — both start by going down. hỏi comes back up; nặng stops.
• huyền and ngang — the flat one and the falling one, when the fall is gentle. This is mostly a listening problem rather than a production one.

Spend your practice on those three pairs rather than on all fifteen possible combinations. The rest sort themselves out.

Southern speakers use five

Here is the fact that reconciles a lot of confusing advice: in Southern Vietnamese, hỏi and ngã have merged. Most speakers in Ho Chi Minh City produce them identically, as a dipping tone. The spelling still distinguishes them, and the merge is completely normal — nobody is speaking incorrectly.

So if you are learning Southern Vietnamese, you have six tones to read and five to say. If you are learning Northern, you have six of each, and ngã with its glottal break is a real target.

This matters for choosing material. A course that drills a Northern six-tone distinction is teaching a Southern learner a difference their own family does not make.

Why “listen and repeat” beats every explanation

Tone is carried by fundamental frequency — the pitch of your voice — and F0 is something your ear can copy long before your conscious mind can describe it. Every learner who ends up with good tones got there by imitation.

Two practical consequences:

Never learn tones from slowed-down audio. Stretching a recording to half speed smears the pitch contour, which is precisely the thing you are trying to hear. A slowed ngã often loses its break entirely, so the learner practises a tone the speaker never produced. HiAn deliberately ships one recording per line at natural speed for this reason, rather than a speed slider.

Learn tones inside words, not on isolated syllables. In connected speech a tone is shaped by the tones around it, and ma-má-mà drills teach a shape that does not survive a sentence. Learn cảm ơn as a unit, not as cảm plus ơn.

Seeing your own pitch

The gap between “I think I said it right” and “I said it right” is where most tone practice dies. You cannot hear your own pitch contour reliably while you are producing it.

HiAn’s shadowing mode records you and draws your pitch curve against the native speaker’s on the same axis, then says what differed in words — your pitch stayed level where it should fall. Deliberately there is no score. A percentage on pronunciation would be an invented number, and a learner who is told “72%” learns nothing they can act on, whereas “your voice did not come back up at the end” is a specific instruction.

A practice order that works

1. Two weeks on hearing, before producing. Play minimal pairs and choose which one you heard. Getting your ear ahead of your mouth makes everything after it faster.
2. Then the three confusable pairs, one pair at a time, in real words.
3. Then whole phrases, copying rhythm as well as pitch. Vietnamese is syllable-timed — every syllable takes about the same length — and getting that rhythm right does as much for intelligibility as the tones do.
4. Then conversations, where the tones have to survive being said at speed while you are thinking about something else.

That last step is the one people skip, and it is the only one that produces a person who can be understood in a noisy café.