Vietnamese for heritage speakers who understand but cannot speak

You understand your grandmother perfectly. You have understood her your whole life. And when she asks you something directly, the answer arrives in English and stops at your teeth.

This is one of the most common Vietnamese learning situations there is, and almost none of the available material is built for it.

Why beginner courses make it worse

Open any Vietnamese app as a heritage speaker and the first twenty lessons will ask you to pick the meaning of xin chào from four options.

You know what xin chào means. You knew before you opened the app. So the exercise teaches you nothing, and worse, it teaches you something false: it tells you the course thinks you are a beginner, which does not match your experience of yourself, and the mismatch is discouraging in a way that makes people quit within a fortnight.

The material is not wrong. It is aimed at someone else — a person building comprehension from zero. You already have comprehension. You have a production problem, and those are different enough to need a different order.

What is actually missing

Three things, in roughly this order of difficulty.

1. Retrieval speed

You know the word. It takes you four seconds to find it, and a conversation moves faster than that. Recognition is a much cheaper mental operation than recall, and years of passive exposure build the first without building the second.

The fix is timed production: being asked for the Vietnamese and having to produce it, quickly, without a multiple choice list to recognise it in. It feels harder because it is harder — that is the point.

2. The pronoun system

This is often the real blocker, and it is rarely named.

Growing up, you were a child. You called everyone by the terms a child uses and everyone called you con or cháu. Now you are thirty-four, and the woman who was cô when you were eight may be bác now, and your cousin who was anh is still anh but you are no longer em to everyone in the room.

Nobody explicitly taught you the system, because native speakers absorb it by living inside it. You absorbed the half you needed as a child and moved abroad before you needed the adult half. So the hesitation you feel is not about vocabulary at all — it is that the first word of the sentence requires a social judgment you were never taught to make, and getting it wrong in front of family is mortifying in a way that getting it wrong with a waiter is not.

3. Confidence, which is downstream of the other two

The freeze is not shyness. It is the accurate perception that you are about to make a mistake in front of people whose opinion matters more than anyone else’s. Fixing the first two makes it go away; being told to relax does not.

What to do instead

Start with production from day one. No recognition drills. If you are given a course that starts with choose the meaning, skip past it until the exercises ask you to say something.

Shadow long before you converse. Play a native line, say it immediately after, and compare. This trains the mouth without requiring a social judgement, so it is the cheapest possible practice and the fastest way to get your speed up.

Drill the specific people in your life. Not “how do you address an older woman” but “how do I address Cô Hằng, who is fifty-eight, on my mother’s side”. Six or seven real relationships covers almost everything you will say, and they are exactly the ones where the stakes feel high.

Accept that you will sound like a child at first. The Vietnamese you have is a nine-year-old’s Vietnamese, because that is when it stopped developing. This is not a failure; it is where you resume from. The vocabulary you need for adult conversation — work, money, plans, opinions — was simply never installed.

What HiAn does with this

There is a heritage mode, and it is not a cosmetic setting.

Turning it on inverts the exercise order for every lesson. Production types — produce the Vietnamese, fill the gap, timed response — come first, and the two purely recognition drills, choose the meaning and Vietnamese to English, are suppressed entirely. You are never shown a four-option list asking what a word you have known since childhood means.

The family tree feeds the same machinery: you enter the actual people, and the app resolves the pronoun pair for each one — what you call them, what you call yourself, whether the particle belongs — and then drills those inside real sentences with recordings.

That is the whole design. Not an easier course, and not a harder one. A different order, for someone who arrived with half the work already done.

One more thing

The reason this is worth doing is rarely the language.

Most heritage learners are not trying to reach a proficiency level. They are trying to have a conversation with a specific person, often an elderly one, and they are aware that the window is not infinite. That urgency is a real advantage — it makes the daily practice easy to sustain — but it also makes the freeze more painful.

So: aim narrow. Pick the five things you want to be able to say to that person. Learn those to the point of automaticity, in their dialect, with the right pronouns. Have the conversation. Then widen.

A course finishes. A conversation with your grandmother does not have a syllabus.