We meet a lot of students who arrive slightly angry. They have done a year of daily app practice, they have a long streak, and they still froze when a waiter in Florence asked them something unscripted. Nothing went wrong. They ran into the ceiling.
What apps do well
- Daily contact. Turning up every day is most of language learning, and apps are engineered to make you turn up.
- Vocabulary volume. Spaced repetition genuinely works, and is boring for a human to run.
- Recognition. Reading and hearing common words until they are automatic.
- Zero friction. No scheduling, no cost, no embarrassment.
Where the ceiling is
Around A2. You can recognise a lot, produce set phrases, and read simple text. Then progress stops, for three structural reasons:
- Nothing hears you. Speech recognition checks whether you said the target sentence. It cannot tell you that your sentence was grammatical but no Italian would phrase it that way.
- There is no unpredictability. Every prompt is expected. Real conversation is a stream of things you did not prepare for, and handling that is a separate skill that only practice with a person builds.
- Errors are never diagnosed. An app marks an answer wrong. A teacher notices that you have avoided the subjunctive for three weeks and works out why.
An app can teach you the words. Only a person can teach you to be understood while getting them slightly wrong.
The honest comparison
Per hour, a private lesson is far more expensive and produces disproportionately more speaking ability. Per month, an app is close to free and produces vocabulary you cannot yet deploy. They are not competitors. The efficient combination is lessons for production and correction, an app for the daily twenty minutes of drilling between them.
How to tell which you need right now
If you cannot yet read a menu, an app will move you forward quickly and cheaply. If you can read comfortably but freeze when spoken to, more app time will not fix it — that is precisely the gap conversation classes exist to close.