Why our Mandarin quizzes test recall, not tapping
Open almost any language app's end-of-lesson quiz and you'll see the same thing: an English word on screen, four buttons, pick the translation. Easy. Satisfying. Almost useless.
That format tests recognition: matching a visible English prompt to a candidate. In real conversation nobody hands you the English first. You hear Chinese cold, and you either know it or you don't. Recognition feels like learning because it's fast; recall feels wobbly because it is — it's the actual act of pulling meaning out of memory, which is the only thing that transfers to a live conversation.
What our quiz stage does instead
The LearnTo Mandarin course inverts the flow. In the quiz, Chu plays the Chinese word — replay it as often as you want — and never says the English gloss. The character sits on screen; the meaning is nowhere in the question. The options below are meanings drawn from the unit's own vocabulary, so every wrong choice is a word you're also supposed to know.
The design constraint behind it: the meaning had to be learned earlier in the lesson, by listening to her say the word and reading the gloss. The quiz deliberately withholds what a recognition quiz would give away. If you can answer, you remembered. If you can't, you re-listen to Chu — which is the loop, not a failure state.
Small choices that keep it honest
- No English audio anywhere near the question. Early builds used a poster placeholder during the quiz until the spoken clips were re-recorded without glosses; the rule survived into the shipped clips: the quiz audio is the word, period.
- Listening doesn't grade you. The course tracks listening exposure separately from recall answers, and listening never scores pronunciation. Exposure is not mastery — pretending otherwise inflates streaks and deflates trust.
- Wrong answers stay quiet. The reaction engine pins mistakes to its lowest tier: a warm nod, no buzzer, no confetti. A shaky recall attempt is information, not a shame event.
- Units unlock in order. You can't quiz on Unit 9 words on day two, so recall failures always land on material you were actually taught.
Cognitive science isn't a secret here — testing effect, generation effect, desirable difficulty are decades old. What's unusual is a consumer app paying the UX cost of feeling harder in order to teach something that lasts.
Try the wobble
The whole experiment is free: 12 units, 226 words, 92 dialogue lines, recall quizzes included, at learnsto.com/mandarin. Take Unit 1's quiz and notice the difference between guessing A-or-B and actually remembering 谢谢. It feels worse. That's the tell.