A helper says each word naturally, then models a tap for each syllable. The learner may tap, point or listen. Say the whole word naturally again afterward. The questions use stated models; different pronunciations are not mistakes. Do not count letters. No independent reading, timer or recording is needed. Questions check models, not speech or reading fluency.
Worked example
Sun has one beat, rabbit has two, and banana has three in our models. Ask your helper to demonstrate each before using the models.
Listen and speak first
Helper: say cat, then tiger, then potato. Let the learner try the beats. Model cat with one, tiger with two and potato with three. Say each word naturally again. Listen to the learner rather than judging pronunciation from a written answer. This spoken practice is not automatically scored.