AI-POWERED LEarning App
Hora da Libras
ROLE
UX Designer
EXPERTISE
UX/UI Design
YEAR
2025
Teaching Brazilian Sign Language through play and evidence.
Hora da Libras started as an inter-university research project to help children learn Libras as a first language. The first pilot showed communicability breakdowns and uneven task success. I redesigned the product using Semiotic Engineering, treating the interface as a message from designer to user. I validated the changes through MAC (Communicability Evaluation Method) and SUS (System Usability Scale). This case explains how we clarified that message, reduced cognitive load with chunking, and improved usability across core exercises.
I led the full product design process, including discovery, UX strategy, information architecture, UI, prototyping, and research operations. I also set up a shared Notion workspace to manage scope, hypotheses, evidence, releases, and rituals.
Team: 1 designer (me), 2 developers, programming instructors, and a speech‑language pathology specialist team.
Note: This project was done in Portuguese and later translated to English by me.
Challenges
Teach sign language to deaf children with late or limited exposure to Libras, especially those born to hearing families.
Wide variation in language proficiency and attention span; many users are pre-literate.
Limited caregiver insight into progress, reducing at-home support.
Ethical and scheduling constraints for research sessions with children.
High Level Goals
Make the app usable without reading so children learn by watching and doing.
Keep kids oriented at all times with a clear next step and visible progress.
Use short, fun activities with instant feedback to sustain engagement.
Impact
reduction in communicability breakdowns after redesign, making the interface significantly clearer for young deaf users.
SUS scores improved from 72.5 to 84.16 after the redesign, reaching an excellent level.
task completion in redesigned exercises, achieved by all 6 deaf children during usability testing.
Adopted by 2 federal universities (UFRJ & UFF) in their speech therapy programs, confirming real-world educational impact.
Process Overview
We ran a fast, user-centered loop. Semiotic Engineering guided decisions so the interface clearly “speaks” to the child through visuals. I was the sole designer, aligning developers, programming teachers, and speech-language specialists in a shared Notion.
Research & Planning
Desk research and short field conversations with educators and families to understand context and constraints. Success defined as fewer points of confusion, higher task completion, and better SUS.
Design & Prototyping
I started with a simplified UI and navigation flow aligned with WCAG 2.0 (Web Content Accessibility Guidelines), each screen focused on a single action. I also built interactive prototypes and a lightweight design system to support rapid iterations.
Implementation
Delivered specs, motion, and feedback patterns. Optimized media for low-end devices and prepared reusable templates for sign videos.
Testing & Optimization
Usability sessions and SUS informed iterations on progress visibility, return paths, and immediate feedback until flows felt predictable and self-explanatory.
User Persona
To design for the real context of deaf children and their families, I created a primary persona that anchored every decision from IA to microcopy. Bringing Maria to life kept us focused on visual-first interaction, shared devices, and the need for fast, clear feedback. The persona informed the Lean Canvas, guided user stories, and tied research insights to UI choices.
Maria Gomes | 10
Public School Student struggling with math learning
Location
Rio de Janeiro,
Brazil
Education
2th grade student
Challenges
Limited early exposure to Libras, gets lost with text-heavy apps
Goals
Communicate with family and classmates using everyday signs
Learn through short, playful tasks and see her progress
Feel independent and proud of small wins
Needs
Visual-first instructions with one clear action per screen
Predictable navigation and large tap targets
Instant feedback and a safe way to try again
Light, well-framed sign videos that work on low-end phones and weak internet
Pain Points
Tries to learn Libras using apps that require reading Portuguese to move forward
Ambiguous icons and hidden actions that create detours
Heavy videos that stutter or fail to load
No guidance on what to practice next or how caregivers can help
User Journey
Maria tries a traditional Libras app made for Portuguese readers. Text onboarding and menu labels force a parent to translate, so she feels it isn’t for her. With Hora da Libras, visual tiles and a short sign video invite action and she starts on Animals.
Conflicting
Traditional tasks mix icons and give no hint after errors. Hora da Libras offers help cues and Sinalário to guide a retry.
Thoughts: I cannot read this. I need my parent to explain. This app is not for me.
Actions: Hands the phone to a parent, waits for translation, taps randomly, feels reluctant to continue.
Thoughts: These prompts are long. I am not sure what the task is or if I did it right.
Actions: Skims or skips text, guesses answers, backtracks often, spends extra time with little sense of progress.
Thoughts: Icons are confusing and after an error I get no clue. I must ask for help.
Actions: Repeats taps without understanding, opens menus looking for hints, calls a parent, leaves the exercise unfinished.
Thoughts: I am tired and I do not see what I achieved. Better stop.
Actions: Closes the app, switches to another activity, does not return later because progress feels lost or unclear.
Trying
Traditional lessons lean on text and long prompts. Hora da Libras uses quick watch, choose, confirm loops with instant feedback.
Starting
Traditional app expects reading. Hora da Libras greets with visual tiles and a short sign video.
Quitting
Traditional app ends in frustration and no clear progress. Hora da Libras saves progress and suggests the next step.
Method and Toolkit
The project was guided by Semiotic Engineering, which views the interface as a message from the designer to the child. If children struggle to understand what is happening or what to do next, the message is failing. To detect these breakdowns, I applied the Communicability Evaluation Method (MAC) and tagged moments of confusion with labels like “What now?”, “Where am I?”, or “What is this?”.
To measure overall usability, I used the System Usability Scale (SUS), a standardized 10-item questionnaire. Combined with desk research on early sign-language learning, this toolkit shaped the hypotheses and a lightweight testing plan with educators and families.
Round 1 Findings
The first evaluation with deaf children showed clear patterns. Even though the interface followed WCAG guidelines, problems still appeared. MAC tags clustered around progression, the sign dictionary, and the old ranking area. These screens showed unclear next steps, weak return paths, and ambiguous icons. Caregivers often had to mediate at these points.
To capture evidence, we recorded each session with two synchronized cameras. One focused on the screen and the other on facial expressions and body language. We tagged and analyzed the videos, then used them as the basis for MAC and SUS scoring.
The baseline SUS averaged 72.5. This placed the app in a good but improvable range. Observations also showed detours after finishing a unit. These results confirmed our assumptions and pointed directly to which screens and messages required redesign.
From Tags to Fixes
I reviewed the synchronized video recordings and tagged breakdowns using the Communicability Evaluation Method and logged each one with its fix. The redesign added a persistent primary action, clearer progress, and a simple Home → Unit → Exercise flow with a reliable return path.
For harder content like points of articulation, I introduced preparatory steps and split complex activities into smaller blocks (chunking method). I also applied verb-first labels, consistent icons, and immediate feedback. The sign dictionary gained a guided review and an easy way back to the current task. Each change addressed a specific breakdown observed in testing.
Visual Literacy as Design Strategy
The redesign of Hora da Libras used Donis A. Dondis’ A Primer of Visual Literacy to guide how the interface communicates without written Portuguese. Each button type received its own color so learners could identify categories at a glance. Orange marked Handshape (CM), yellow marked Place of Articulation (PA), green marked animals and thematic signs, and light green marked general signs.
Progress bars and consistent exercise icons reinforced this system and showed learners what to do next. Combined with Semiotic Engineering and MAC analysis, these changes made instructions clear, reduced confusion, and helped children move through the units with more independence.
Competitor analysis
I conducted a competitor analysis to understand the strengths and weaknesses of existing math education platforms, focusing on how they engage students, personalize learning experiences, and align with educational goals. This helped identify opportunities for differentiation and innovation within the MathNex product.
LibrasLab
A fantasy-based math RPG where students complete math challenges through battles and story-driven quests.
✅ Pros:
• Video of a signer on every task, clear framing, large tap targets
• Speed control and a “translate” option help hearing users check understanding
❌ Cons:
• Prompts and options depend on reading Portuguese, which excludes pre-literate children
• Tasks map sign → Portuguese word rather than building Libras as a first language
SENAI Libras
Strong "sign dictionary" tool, a technical reference for hearing learners or vocational contexts. Misaligned with first-language acquisition needs,
✅ Pros:
• Large Libras glossary with search, A–Z index, and filters by domain
• Step-by-step videos for each entry support dictionary lookups
❌ Cons:
• Dictionary, not a learning path; no scaffolded progression or short practice tasks
• 3D characters may miss subtle sign parameters (handshape, movement, orientation, location)
Librário
An adaptive math platform powered by AI, focusing on personalized learning paths for elementary students.
✅ Pros:
• Simple, familiar game loop that motivates play with turn taking and quick wins
• Encourages repetition and visual recall of signs across rounds
sign for a Portuguese word or sentence
❌ Cons:
• UI and cards use Portuguese words and buttons, so pre-literate children need mediation
• Still photos reduce key parameters of the sign; no slow/zoom or movement cues
VLibras
A visual and interactive platform that teaches math through engaging mini-games and virtual objects for younger students.
✅ Pros:
• Fast PT-BR to Libras translation with a large sign library
• Useful 3D avatars for hearing users to preview a sign for a Portuguese word or sentence
❌ Cons:
• Built as a translator, not a learning path
• Feedback is absent beyond play or pause, so errors do not teach what to notice
Hora da Libras
By analyzing competitors, we saw that most lacked key features like adaptive learning (TRI), financial literacy content, AI-driven mentoring, and AR. MathNex combines all of these in one platform—offering a unique, personalized, and gamified experience that no other app at our price point provides.
LibrasLab
Senai Libras
Librário
Built for L1 (first language) acquisition by deaf children
Short, scaffolded tasks (chunking)
Hints and parameter cues after errors
Progress visibility and auto save
Real signer videos (not 3D avatar)
Solution
Hora da Libras is a visual-first app for deaf children learning Libras as a first language. The interface explains itself without reading, guides one action at a time, and rewards small wins. Content is organized by child-friendly semantic fields, with progress saved so families know what to practice next. The work was guided by Semiotic Engineering and iterated through usability sessions using MAC to locate confusion and SUS to confirm overall ease.
Image Association Onboarding
Unit 1 starts with image-to-image matches so learners grasp the mechanic before any signs. A progress bar tracks each small, usability-tested step, and augmentative visuals clarify Portuguese labels, showing progress and exercise type on the selection screen (single choice or matching).
Handshape Foundations
At the start of Unit 2, the concept of Handshape (Configuração de Mão, CM) is introduced. Activities correlate animals from Unit 1 with their respective handshapes, helping learners notice form before location, and the unit advances through a short, well-scaffolded series of exercises.
Place of Articulation and Chunking
Only after CM do we bring in Place of Articulation (Ponto de Articulação, PA). We apply the chunking method described in the thesis, starting with simpler selections and then moving to matching. This reduces cognitive load while the progress bar continues to signal advancement.
Each completed exercise adds points to the learner’s total. At the end of the video, those points appear in the user’s profile.
Next Steps
We plan to invest in video rotoscoping to create a new game mechanic focused on the phonological parameter of Movement (M). This expands the content beyond handshape and place of articulation. We will also add new units covering diverse semantic fields to broaden learners’ vocabulary.
To sustain growth, a premium plan will be introduced. It will give access to these new units and advanced features while keeping the core experience free and accessible.
Case studies
© 2025 – Daniel Monterazo
height
88
height
88