If your character tests can be beaten by pointing a phone at the screen, the problem is not your students’ honesty, it is the test design. Phone OCR and instant translation have made recognition-based assessment trivially cheatable. The good news for teachers and tutors is that the fix is the same change that makes the test pedagogically better: assess production, not recognition.

Why recognition tests are now defeated

OCR reads characters off a screen or page instantly, and translation apps supply meanings on demand. So any task where the prompt and the answer can be scanned, match this character to its meaning, pick the right reading, identify the word, is vulnerable. The technology answers it faster than the student can.

You cannot out-design a camera at the recognition level. The only durable move is to test something a camera cannot do for the student.

Test production, not recognition

A character a student must produce from memory leaves nothing to scan. There is no prompt-and-answer pair for OCR to bridge, because the answer has to come from the student’s own recall. Concretely:

  • In-person handwriting. Have students write characters or sentences by hand, observed, with phones away. Producing 汉字 from a meaning or pinyin prompt cannot be OCR’d.
  • Dictation to handwriting. Read a word; students write the character. The input is audio, the output is their hand, neither is scannable.
  • From-memory composition. Short written answers produced on paper in class.

This is the recognition-versus-recall distinction we cover in the case for a dedicated Hanzi writing app: recall is the harder, real skill, and conveniently, it is also the cheat-resistant one.

The deeper issue: OCR erodes the skill itself

There is a reason beyond cheating to push toward production. Habitual OCR use erodes students’ own recall, the character-amnesia effect we cover in relied too much on Pleco OCR and is OCR making character amnesia worse. So designing assessments around production does not just stop cheating; it pushes students toward the practice that actually builds the skill.

How to help students get there

If the test rewards recall, students need a way to build it:

  • Assign from-memory writing practice, not recognition flashcards.
  • Emphasise stroke order so characters are legible and recall is automatic, see Hanzi stroke order practice.
  • Keep it spaced and small, so it is sustainable.

Where Hanzi Write Practice fits, honestly

Hanzi Write Practice is a student practice tool, not a proctoring or anti-cheat system, and it would be wrong to claim otherwise. What it does is help students build the production skill your better tests measure: drawing characters from memory on a grid, with stroke feedback and spaced repetition. When students can actually write, the incentive and the ability to OCR-cheat both shrink.

Design tests around producing characters, point students at real recall practice, and the camera loses its job.

Join early access and point your students at real recall practice.