Unit: SIT216 User Centred Design | Deakin University
Grade: HD (88)
Team: Trung Kien Do, Insu Oh
MojiMoji is a mobile app concept for natural language learning through real-time AI conversation. Users pick a roleplay scenario, choose an AI character, and practice speaking in realistic situations — ordering at a café, renting a car in Melbourne, opening a bank account. The AI responds naturally, gives grammar feedback inline, and tracks learning progress over time.
This repo contains the full UX research and high-fidelity prototype work produced for the project: user research methodology, usability testing data, iterative design decisions, and future implementation proposals.
Target users: International students aged 18-24 living in Melbourne
Research methods: Mixed-methods usability testing, Google Form survey, in-person task observation
Participants: 5 international students (Vietnamese, Nepali, Sinhala, Korean, Mandarin speakers)
Testing conducted: May 2025, Deakin University workshop
The interface scored well across the board: 100% of testers rated it visually appealing, and all could find and use core features without confusion. The AI roleplay scenarios felt realistic enough that 80% of testers reported feeling more confident making real conversations after using the app. All testers said they would recommend it to other students.
The feedback and grammar correction system was the main area that didn't land as strongly, with most testers rating it neutral rather than useful. The translation and pronunciation tool was helpful for some and irrelevant for those already confident in those areas. Both point to the same underlying issue: the features weren't personalized enough to the individual's actual gaps.
The prototype went through three major iterations based on team discussions and early user feedback.
Iteration 1 — Scenario screen enhancement
The original prototype only showed scenario titles. Added category tags, difficulty indicators, descriptions, and filtering. Users needed enough context to pick a scenario appropriate for their level before starting.
Iteration 2 — Custom scenario creation
Added a full flow for users to build their own roleplay: icon, roles for both user and AI, context description, voice style, and a public/private toggle. Addresses the gap where predefined scenarios don't match what a user actually needs to practice.
Iteration 3 — AI avatar and scenario detail screen
Added a preview modal before starting any roleplay, showing the AI avatar, scenario description, difficulty, estimated time, and learning objectives. Users need to know what they're committing to before they start speaking.
Three priorities came out of usability testing:
1. Onboarding tutorial — One tester couldn't figure out the recording flow without help. A short interactive tutorial on first launch reduces drop-off before users even get to the core feature.
2. Multi-language AI conversation — The app supports 8 interface languages, but the AI could only converse in English at the time of testing. Multiple testers mentioned wanting to practice Mandarin or use the app in a non-English context. This is the most significant expansion for reaching an international market.
3. Progress Tracker — 80% of testers supported adding a dashboard showing day streaks, daily speech stats, a global XP score, and monthly activity. Designed and included in the prototype as a proposed next phase.
The high-fidelity prototype covers:
- Home screen with trending and new roleplay scenarios
- Scenario selection with category filtering and difficulty tags
- AI character selection (Marcus, Olivia, Ruby)
- Live conversation screen with inline grammar feedback and translation
- Post-conversation feedback report with fluency, vocabulary, and grammar scores
- Custom scenario creation flow (5 steps)
- Progress Tracker dashboard
- Language settings (English, Korean, Japanese, Chinese, Hindi, Vietnamese, German, and more)
View the interactive Figma prototype → (link to be added)
User Research Usability Testing Mixed-Methods Analysis High-Fidelity Prototyping HCI Principles Iterative Design Focus Groups Survey Design Data Analysis Figma