Existing children’s self-reporting tools like surveys and diaries often feel restrictive, leading to disengagement and low-quality responses. LLM-powered chatbots can adapt with simplified wording or empathetic tone, but such adaptations remain insufficient: responses may be adult-centered, complex, or formulaic, undermining engagement and response quality. We explore rhyme as a child-centered conversational style. In a co-design workshop with 35 children, participants envisioned dialogue that was short, playful, and soothing. Building on these insights, we designed a voice-based sleep diary in rhyming style and conducted a within-subjects study (rhyming vs. prose) with 42 children. Rhyming prompts improved response quality across question types, while maintaining high engagement even among children who preferred prose. We contribute empirical evidence and design insights showing how rhyme can exemplify broader child-centered strategies beyond capability adaptation. Although limited to short-term lab sessions, this work provides a first step toward conversational style as a design lever for children’s self-reporting.
ACM CHI Conference on Human Factors in Computing Systems