QuickChat
Most AAC apps produce children who only request.
The default pattern in AAC is: tap a word, hear it spoken, done. Communication ends after one utterance. Children learn to ask for things — but never to comment, joke, greet, question, or hold a conversation. They become fluent requesters and nothing else.
QuickChat breaks this pattern. After the child says something, the app shows what they could say next — spanning different communicative functions. The child doesn’t have to invent the next turn; they just pick one. Communication becomes a rhythm.
Three steps, then repeat
QuickChat activates after the child speaks. The sentence engine helps them build the utterance; QuickChat shows what to say next. No setup, no mode switch — it happens naturally as part of communication.
Speak
The child builds their sentence in the speech bubble. Tapping a compass word or scene object adds it — and the sentence engine instantly predicts what might come next. Dashed prediction tiles appear ranked by scene context. The child taps predictions to extend their sentence, or taps ▶ to speak. TTS fires, and the signing bubble shows the key word’s ASL sign.
Choose
Follow-up options appear in center stage, each a different communicative act — requesting more, commenting, or engaging socially. At Tier 2, three options keep choices simple and Switch Control–friendly. At Tier 3, a grid of six plus a scrollable strip gives faster communicators more to work with. The child doesn’t have to generate the next sentence; they just pick the one that matches what they want to say.
Loop
The child taps a follow-up. It speaks with signing. New follow-ups appear based on what was just said — generated by the sentence engine using grammar rules and context. The conversation continues as long as the child wants. There is no end screen, no score, no timer. The child can also tap any word on the compass edges at any time to start a new thread.
The loop models turn-taking without requiring a real conversation partner. But it’s designed to be better with a caregiver present — “You said pizza is yummy! I think so too!”
Left: the sentence engine builds an utterance word by word in the speech bubble (dashed tiles = predictions). Right: after speaking, the QuickChat tray rises from center stage with follow-up options. Toggle between Tier 2 and Tier 3 to see how the number of predictions and follow-ups adapts.
Build sentences word by word, with help
When a child taps a compass word or scene object, it appears in the speech bubble. The sentence engine instantly predicts what might come next — showing dashed suggestion tiles ranked by scene context, grammar rules, and usage patterns. The child taps a prediction to extend their sentence, or taps ▶ to speak.
Single tap = instant speech. No sentence building. The child is learning that touching makes words happen.
Max 2 words. The engine predicts scene-relevant nouns after a verb. Child taps a prediction to complete the phrase, then ▶ to speak “Want pizza.”
Multi-word sentences. Grammar-aware predictions update after each word — the engine knows “I want” should be followed by a noun, not another verb. More prediction tiles for children who can handle more choices.
Why dynamic prediction
Pre-authored sentence templates require a library for every scene, every word, every level — a combinatorial problem that doesn’t scale. The sentence engine uses grammar transition rules and scene context to generate predictions dynamically, so every scene gets equally rich language support without manual authoring. The grammar rules are clinically grounded in Fitzgerald Key category relationships: verbs predict nouns, nouns predict descriptors, and pragmatic words close sentences.
Carrier phrases with a gap
Adapted from Ms. Rachel’s cloze technique: the app presents a familiar phrase with one missing word. The child fills in the gap. Lower demand than open-ended production because the sentence structure is already primed — builds sentence-level communication from single-word users.
Child taps one word to complete the sentence. Speech fires: “I want pizza please!”
Why carrier phrases work
Cloze/carrier phrase techniques reduce cognitive load by providing the sentence frame. The child only needs to supply one word — the meaningful word. This is how parents naturally scaffold language: “You want... what?” The fill-in approach embeds this proven interaction pattern into the app.
Not just requesting — the full communicative repertoire
QuickChat follow-ups are engineered to span all eight communicative functions. Every set of follow-ups intentionally mixes functions, so the child is exposed to commenting, greeting, and questioning alongside requesting — whether they see three options at Tier 2 or six at Tier 3.
🤲 Requesting
🚫 Protesting
💬 Commenting
👋 Greeting
✅ Responding
❓ Questioning
💭 Expressing Feelings
📢 Directing Attention
Why all eight
ASHA’s position on AAC explicitly states that communication intervention should target the full range of communicative functions, not just requesting (ASHA, 2005). Children who learn only to request develop “learned helplessness” patterns where communication is transactional rather than social. QuickChat’s mixed-function follow-ups prevent this by naturally exposing the child to commenting, questioning, and social language alongside every request.
Built on validated communication science
Every element of QuickChat maps to established clinical practices and research in augmentative communication.
Turn-Taking Scaffold
The speak-choose-speak loop teaches conversational turn-taking without requiring a conversation partner. The child learns that communication is an exchange, not a one-shot event.
Tier-Adaptive Complexity
The sentence engine adapts prediction complexity to the child’s developmental tier — single words at Tier 1, two-word combinations at Tier 2, multi-word grammar-aware building at Tier 3. Each tier models language at and just above the child’s current level.
Fitzgerald Key Colors
Grammar category colors (yellow = who, green = doing, blue = describing, orange = what, purple = social) teach sentence structure through visual pattern, not explicit instruction.
Pragmatic Breadth
Follow-up options span eight communicative functions (requesting, protesting, commenting, greeting, responding, questioning, expressing feelings, directing attention) to prevent request-only communication patterns.
Carrier Phrase Scaffolding
Fill-in sentences reduce cognitive load by providing the frame. The child supplies only the meaningful word — the same technique SLPs use in every therapy session.
Clear boundaries
QuickChat is designed with specific clinical intent. It’s important to be clear about what it is not.
Not a chatbot
The engine uses clinical grammar rules and scene context — not conversational AI. Every prediction follows Fitzgerald Key relationships, not learned conversation patterns.
Not a quiz
No right answers, no scores. Every choice the child makes is valid communication.
Not a replacement
Models turn-taking, but real conversation with real people is always the goal.
Not a gate
The child can always tap any word on the compass. QuickChat suggests; it never constrains.
What it is: a conversation scaffold
QuickChat shows the child what communication can look like — that after saying one thing, there are more things to say. It teaches the rhythm of conversation (speak, listen, respond) through the most accessible interface possible: pick from a small set of follow-ups tailored to the child’s developmental tier. Over time, the child internalizes the pattern and brings it to real conversations with real people.
Every follow-up is reachable by every input method
At Tier 2, QuickChat shows three follow-up options — deliberate: three choices is the maximum that works well with Switch Control scanning while still offering meaningful variety. At Tier 3, the grid expands to six options plus a scroll strip, because children at this level communicate rapidly and benefit from more choices.
Why tier-gated options
Three follow-up options at Tier 2 is not arbitrary. Research on choice-making with young AAC users (Sigafoos et al., 2004) shows that three options balances variety with manageability for early communicators. Three also maps cleanly to Switch Control scan groups. At Tier 3, children have demonstrated the motor planning and decision-making capacity to handle more — the expanded grid gives them speed without sacrificing accessibility.