Low-literacy / damaged-screen mode: no reading required; the entire first viewport is a single audio-start target; use actual photos, large visual cues, spoken language, and no precision tapping.
