My Guide — The Companion
The child’s avatar is their voice. Who teaches them how to use it?
Without a companion, the app has two choices for modeling words and demonstrating signs: the avatar does it (confusing — the child’s representation is doing something the child isn’t doing) or the UI does it (cold — a clinical tool, not a world).
The companion solves this. It’s a separate character — a friendly guide that models, teaches, and celebrates. The child’s avatar stays purely their voice. The companion is someone helping them find it.
Bottom-right corner. Always visible. Always friendly.
The companion sits in the bottom-right corner of the compass frame, small and ambient. It breathes gently — a living presence, not a static icon. When idle, it’s just there — the child knows a friend is available. One tap activates the engagement mode.
The companion is the app’s warmth. Without it, Meadow is a speech device with buttons. With it, Meadow is a place with someone in it. That’s the difference between “time for your AAC app” and “I want to go to Meadow.”
Consistent Position
Bottom-right corner, every scene. Predictability builds trust.
Gentle Breathing
Subtle scale animation (1.0× → 1.04×) signals “I’m here.” Never pushy.
Child-Discoverable
No parent gate. The child can find and tap their guide independently.
Signing Anchor
The signing bubble overlaps the companion. Signs come from the guide.
Persistent · Gentle pulse
The companion reaches out when the child doesn’t
The youngest children don’t yet know that touching the screen makes words happen. One tap on the companion starts a brief, warm engagement cycle — modeling words, inviting interaction, celebrating any response. The companion does what a good SLP does: reaches out first.
Spotlight
The scene gently dims. One word glows with a golden ring, enlarged 15% from its natural position. Everything else stays visible but muted. The word stays in context — milk glows in the kitchen where milk lives, not on an isolated flashcard.
Present — Triple Modality
Three outputs fire simultaneously: the word’s picture stays spotlighted in-scene, the companion speaks the word with warmth and emphasis (parentese TTS), and the companion demonstrates the ASL sign in the signing bubble. The child gets the word through eyes, ears, and hands all at once.
Wait — Expectant Pause
A “Your turn!” indicator appears. The companion waits — 7 seconds for First Words, 5 seconds for Word Combinations and Sentences. This is the heart of the technique: the silence after modeling invites the child to act. The spotlight ring continues its gentle pulse, holding attention.
Celebrate
Every response gets celebrated — the level matches the child’s effort. The companion delivers the celebration, not the UI. “You said milk!” comes from the guide’s warm voice, making it feel like praise from a friend, not feedback from a machine.
Advance
The spotlight releases. A progress dot fills. The next word lights up. After the final word, the companion delivers a wrap-up: “Great talking!” with a gentle wave. The scene un-dims and returns to reactive communication mode. Tap the companion mid-cycle to end early — no punishment, ever.
No wrong answers. Just a gradient of celebration.
The child can respond in three ways. Each gets a different level of celebration from the companion.
Big Celebration
Companion says: “You said milk!” with enthusiastic voice. Gentle burst animation. Progress dot sparkles.
Small Celebration
Brief sparkle near touch point. Soft chime. Progress dot fills normally.
Gentle Advance
Companion softly repeats: “That’s milk! Let’s try another one.” The loop never stalls.
One engine, three developmental tiers
The engagement cycle automatically adjusts based on the child’s developmental tier setting. Younger children get shorter cycles, longer pauses, and bigger celebrations. All parameters are configurable by the parent or SLP behind the parent gate.
| Parameter | First Words | Word Combinations | Sentences |
|---|---|---|---|
| Words per cycle | 3 | 4 | 5 |
| Expectant pause | 7 seconds | 5 seconds | 5 seconds |
| Celebration intensity | Maximum — biggest celebrations | Standard | Brief — more mature feedback |
| Word selection pool | Core words + scene words | Full scene vocabulary | Full scene + sentence scaffolding |
Smart word selection
Words are randomized from the scene’s vocabulary, but the selection is weighted: words the child has never engaged with are always prioritized first (ensuring full coverage). Among previously-engaged words, those with the oldest engagement timestamp are preferred — naturally cycling through the entire vocabulary over repeated sessions.
Signs come from the guide, not from thin air
When any word is spoken in Meadow — by tapping a word tile, by QuickChat, or during engagement — the companion demonstrates the ASL sign in the signing bubble. The bubble overlaps the companion’s position, making it visually clear that the guide is showing the child the sign.
This solves a common AAC problem: sign animations that appear in a disconnected corner of the screen feel like UI decoration, not communication. When signs come from a character the child has a relationship with, they become something to imitate, not just something to see.
Why signs need a source
Children learn signs through imitation of people, not by watching abstract animations. The companion gives the sign a social source — “my guide is showing me how to say milk with my hands.” This maps to the natural way sign language is acquired: by watching someone you trust do it first.
A voice the child wants to hear
The companion has its own voice — warm, playful, and distinct from the child’s speech output voice. When the child taps a word, their voice speaks. When the companion models, teaches, or celebrates, the companion’s voice speaks. Two characters, two voices, zero confusion.
Warm & Playful
Pre-recorded via ElevenLabs Voice Design. Consistent across sessions. Feels like a friend, not a teacher.
Parentese Prosody
During engagement mode, the companion speaks in parentese — slightly higher pitch, slower tempo, exaggerated vowels. The speech pattern most effective for early language acquisition.
Expressive Celebrations
Celebrations use the expressive tone: genuine enthusiasm. “You said milk!” with warmth, not robotic feedback.
Two voices, not one
The child’s speech output voice (male or female child voice, selected during setup) is the child’s identity. The companion voice is the guide’s identity. Mixing them would blur whose voice is whose. A child who hears “their” voice say “You said milk! Good job!” gets confused — did I say that? Am I praising myself? The companion’s distinct voice keeps the roles clean.
The companion knows when to help and when to be quiet
The companion is not always-on. It has clear rules for when it’s active and when it stays silent. The child’s voice is never competed with.
Active — companion is engaged
- Engagement mode (child or caregiver tapped companion)
- Signing demonstrations (any word spoken)
- Celebrations during engagement cycles
- Onboarding greeting (“Welcome to Meadow!”)
Silent — companion is ambient
- Direct word taps (the child’s voice speaks)
- QuickChat conversation loop
- Feelings tray expression
- Routine and scene navigation
Why silence matters
If the companion celebrated every single word tap, it would become the voice of the app. The child would associate communication with the companion’s response, not with their own expression. Silence during pure communication ensures the child experiences: “I tapped milk → I said milk.” Not: “I tapped milk → the bear said something about milk.”
Designed for partnerships from day one
The companion is an architectural slot, not a hardcoded character. The default companion ships with Meadow. A licensed character can swap in with zero engine changes — same behaviors, same voice contract, same signing demonstrations, new face.
Default Companion
Ships with Meadow. Warm, gender-neutral, designed to be universally appealing to the 12–48 month range.
Licensed Character
A Bluey, an Elmo, a Daniel Tiger — any character a child already has a relationship with. The parasocial bond transfers into the app.
Same Engine
Character swap changes visuals and voice pack. Behavior rules, signing, engagement loop, and position are all unchanged. One interface, infinite faces.
Parasocial relationships as clinical lever
Young children with developmental delays often form strong parasocial relationships with screen characters. Research shows these relationships can increase engagement, reduce anxiety, and improve learning outcomes (Calvert, 2017). A child who already trusts Bluey will engage more readily when Bluey is their Meadow companion. The character slot lets Meadow tap into existing trust without rebuilding it from scratch.
Built on what SLPs already know works
The companion’s design maps directly to validated clinical techniques. Every behavior has a research basis.
Modeling vs. Self-Expression
The separation between the child’s voice (avatar) and the teaching voice (companion) mirrors how SLPs work: the therapist models, the child produces. Mixing the two confuses whose communication act is happening.
Expectant Pausing
The structured wait after the companion models a word is the single most effective technique for eliciting production in pre-verbal children. Built into every engagement cycle.
Triple-Modality Output
Picture + companion speech + companion sign, simultaneously. Multiple channels increase encoding strength and serve different learning styles.
Differential Reinforcement
Bigger celebrations from the companion for more intentional responses. The child naturally drifts from random touches toward purposeful communication.
Warm Celebration
Praise comes from the companion — a character, not a UI element. Human-feeling, enthusiastic but genuine, modeled on Ms. Rachel’s approach: brief, warm, and focused on the child’s communicative act.
Clear boundaries
The companion is designed with specific clinical intent. It’s important to be clear about what it is not.
Not a conversation partner
The child communicates with real people. The companion models and guides — it never pretends to be a peer.
Not AI-driven
All companion behaviors are pre-authored and clinically vetted. No generative AI, no unpredictable responses.
Not a game character
No leveling, no unlocking, no collectibles. The companion is a guide, not a gamification mechanic.
Not passive content
The companion doesn’t play videos or monologue. Every engagement is interactive with structured pauses for the child’s turn.
What it is: a bridge between silence and speech
The companion bridges the gap between a child who doesn’t yet know that touching a screen makes words happen and a child who uses Meadow as their voice. It reaches out first, models the rhythm of communication, celebrates every attempt, and then gets out of the way. The companion’s success is measured by how quickly the child stops needing it.
Every child can reach their guide
The companion and engagement mode are fully accessible from day one.
Timer fallback in engagement
If a child uses Switch Control and no switch activation occurs within the pause window, the cycle still advances. The companion softly repeats the word and moves on. The child is never stuck. This makes the engagement mode usable across the full spectrum of motor ability — from children who can accurately tap a target, to those who can only make gross motor contact, to those who cannot yet touch the screen at all.