# The Missing Voice: What Clinical AAC Research Says About Building an App for Pre-Verbal Toddlers

**Deep Research Report — March 30, 2026**
**Domain: Healthcare (Speech-Language Pathology / AAC)**

---

## Ki: The Field Your Client Is Walking Into

### The Terminology Landscape

AAC (Augmentative and Alternative Communication) is the umbrella term. Within it:

- **SGD** (Speech-Generating Device) / **VOCA** (Voice Output Communication Aid) — interchangeable terms for any device that produces speech output. When Medicare began covering these in 2001, they standardized on "SGD."
- **High-tech AAC** — tablet-based apps like Proloquo2Go, TouchChat, LAMP Words for Life
- **Low-tech AAC** — picture boards, PECS cards, communication books
- **Aided language stimulation** (ALS) / **Aided language input** (ALI) — THE core teaching strategy. A communication partner speaks while simultaneously pointing to symbols on the child's AAC device. This models how to use the device in real conversation. This is not optional — it's the foundational clinical intervention.

### The "No Prerequisites" Principle — A Critical Blind Spot

**There are no prerequisites for AAC.** This is ASHA's (American Speech-Language-Hearing Association) official position, and it's the single most important clinical principle to internalize.

Historically, some SLPs believed children needed to demonstrate certain cognitive or motor prerequisites before receiving AAC. This has been definitively debunked. Research shows:

- AAC can be introduced before a child turns one
- Children do NOT need to "fail" at speech before getting AAC
- Waiting for speech to "catch up" is considered malpractice by modern standards
- SLPs with more assistive technology training are LESS likely to impose prerequisites

**Why this matters for the app:** The app cannot have "readiness gates" or levels that restrict vocabulary access. A 12-month-old should be able to use it alongside a 4-year-old. The design must accommodate the full range from day one.

### AAC Does Not Prevent Speech — It Accelerates It

This is the #1 concern parents have, and the research is unequivocal:

> A widely cited meta-analysis by Millar, Light, and Schlosser (2006) reviewed 27 studies and found that the vast majority of individuals showed an **increase** in speech after beginning AAC. **No participants showed a decrease.**

Research shows toddlers with developmental delays who receive AAC support build larger vocabularies than those receiving speech therapy alone. AAC gives children a way to practice language before they can produce the words vocally. This directly supports the client's positioning: the app is a bridge, not a crutch.

**Sources:** [ASHA Practice Portal](https://www.asha.org/practice-portal/professional-issues/augmentative-and-alternative-communication/) | [HealthyChildren.org AAC Guide](https://www.healthychildren.org/English/health-issues/conditions/developmental-disabilities/Pages/augmentative-and-alternative-communication-for-children.aspx) | [Communication Community - Does AAC Prevent Speech?](https://www.communicationcommunity.com/does-aac-prevent-speech-development/) | [PMC - SLP Practices in AAC During Early Intervention](https://pmc.ncbi.nlm.nih.gov/articles/PMC9549491/)

---

## Sho: The Clinical Foundation — What the Research Actually Says

### 1. Core Vocabulary: The ~50 Words That Matter Most

Core vocabulary research is the bedrock of modern AAC practice. The landmark study by **Banajee, DiCarlo, and Stricklin (2003)** collected naturally occurring vocabulary from 50 toddlers (24–36 months) across five preschools during play and snack activities.

**Key finding:** All 50 children used 9 identical words across both routines. These words were **pronouns, verbs, prepositions, and demonstratives. Nouns were absent from the core list.** This is counterintuitive but critical: the most important words for communication are NOT the picture-friendly nouns (pizza, dog, ball) — they're the structural words that make sentences work.

#### The 100 High-Frequency Core Words (AAC Language Lab / PRC-Saltillo)

Research across multiple studies converges on these categories:

| Category | Words | Fitzgerald Color |
|----------|-------|-----------------|
| **Social/Interjections** | yes, no, hi, bye, please, thank you, sorry, excuse me, uh oh, wow, yay | Pink/Purple |
| **Pronouns** | I, me, my, mine, you, your, he, she, it, we, they, that, this | Yellow |
| **Verbs** | want, go, get, make, do, put, turn, look, see, eat, drink, play, help, stop, come, give, take, open, close, like, love, know, think, tell, say, read, write, sit, stand, walk, run, push, pull, throw, catch, fall, wait, try, need, feel, hurt | Green |
| **Descriptors** | big, little, small, more, all, some, good, bad, hot, cold, new, old, same, different, happy, sad, mad, scared, tired, sick, yummy, yucky, pretty, ugly, fast, slow, hard, easy, loud, quiet | Blue |
| **Prepositions/Misc** | in, on, out, up, down, off, under, over, here, there, where, what, who, when, how, why, not, and, or, but, with, for, to, the, a, an | Pink (prepositions) / Purple (questions) / Black (misc) |
| **Time/Sequence** | now, later, again, done, all done, first, then, next, before, after | Brown/Black |
| **Nouns (few)** | mom, dad, home, school, bathroom | Orange |

**Critical insight:** ~80% of daily communication uses only 200-400 words, and these are overwhelmingly NOT nouns. The app's core word bar should reflect this research — it should be verb-heavy and pronoun-heavy, not noun-heavy.

#### How Toddlers Actually Use Core vs. Fringe Words

A 2024 study in the American Journal of Speech-Language Pathology found that toddlers use core words as the structural backbone of every utterance, with fringe (topic-specific) words slotted in. The pattern is:

```
CORE + CORE + FRINGE
"I"  + "want" + "pizza"
"more" + "play" + "blocks"
```

This directly validates the sentence template engine design — templates built from core words with fringe word slots.

**Sources:** [Banajee et al. on ResearchGate](https://www.researchgate.net/publication/232041093_Core_Vocabulary_Determination_for_Toddlers) | [AJSLP - How Toddlers Use Core and Fringe Vocabulary](https://pubs.asha.org/doi/10.1044/2024_AJSLP-23-00366) | [AAC Language Lab - 100 Core Words](https://aaclanguagelab.com/resources/100-high-frequency-core-word-list) | [AssistiveWare - Teaching Core Words](https://www.assistiveware.com/blog/teaching-core-words-building-blocks-communication-and-curriculum)

---

### 2. Sentence Complexity by Developmental Stage

Mean Length of Utterance (MLU) is the standard clinical measure. This directly informs what sentence templates the app should present.

| Age | MLU | Typical Utterance Length | Sentence Template Complexity |
|-----|-----|------------------------|------------------------------|
| 12–18 months | 1.0 | Single words | "more" / "no" / "mama" |
| 18–24 months | 1.0–2.0 | 1–2 words | "want milk" / "go play" |
| 24–30 months | 2.0–2.5 | 2–3 words | "I want milk" / "more juice please" |
| 30–36 months | 2.5–3.0 | 3–4 words | "I want to play" / "mommy is eating" |
| 36–48 months | 3.0–3.75 | 3–5 words | "I want to eat pizza" / "can we go outside" |
| 48–60 months | 3.75–5.0 | 4–6 words | "I don't want to go to sleep" |

**Design implication:** The app should present sentences at MULTIPLE complexity levels simultaneously. When a child taps "pizza," show:

- **1-word:** "Pizza!" (for emergent communicators)
- **2-word:** "Want pizza" / "More pizza"
- **3-word:** "I want pizza" / "Pizza is yummy"
- **4-word:** "I want more pizza" / "Let's eat some pizza"

The child naturally gravitates to their developmental level. The longer sentences serve as language modeling — the child sees and hears sentence structures slightly above their current level, which is exactly how language acquisition works.

**Sources:** [iCommunicate - Milestones 18-24 months](https://www.icommunicatetherapy.com/child-speech-language/child-speech-language-development/child-speech-language-milestones-stages-of-development/speech-and-language-developmental-milestones-18-24-months-1-5-2-years/) | [iCommunicate - Milestones 36-48 months](https://www.icommunicatetherapy.com/child-speech-language/child-speech-language-development/child-speech-language-milestones-stages-of-development/speech-language-developmental-milestones-36-48-months-3-4-years/) | [Number Analytics - MLU Guide](https://www.numberanalytics.com/blog/unlocking-child-language-development-mean-length-utterance)

---

### 3. The Modified Fitzgerald Key — Color Coding Standard

Two color coding systems dominate AAC. The most widely used is the **Modified Fitzgerald Key** (adapted from Edith Fitzgerald's 1929 work for deaf education). The alternative is the **Goossens', Crain, and Elder** system (1992).

#### Modified Fitzgerald Key (Most Common)

| Color | Category | Examples |
|-------|----------|----------|
| **Yellow** | People / Pronouns | I, you, he, she, we, mom, dad |
| **Green** | Verbs / Actions | want, go, eat, play, help, stop |
| **Blue** | Descriptors (adj/adv) | big, little, hot, cold, happy, sad |
| **Orange** | Nouns | pizza, dog, ball, car, house |
| **Pink/Purple** | Social / Questions | please, hi, bye, what, where, why |
| **Red** | Important / Negation / Emergency | no, stop, don't, help (emergency), not |
| **White/Black** | Miscellaneous | the, a, and, in, on, to |

#### Goossens', Crain, and Elder (Alternative)

| Color | Category |
|-------|----------|
| **Pink/Red** | Social/Pragmatic |
| **Yellow** | People/Pronouns |
| **Green** | Actions/Verbs |
| **Blue** | Descriptions |
| **Orange** | Nouns |
| **White** | Miscellaneous |

#### Thames Valley Children's Centre (Simplified, Common in Practice)

| Color | Category |
|-------|----------|
| **Pink/Purple** | Social words |
| **Yellow** | People (including pronouns) |
| **Green** | Verbs (action words) |
| **Blue** | Descriptive (adjectives, adverbs) |
| **Orange** | Nouns (places, things) |
| **Black** | Miscellaneous (articles, conjunctions, prepositions, time, numbers) |

**The critical rule:** Consistency matters more than which system you pick. Whichever colors you choose, they must be consistent across the ENTIRE app — every screen, every mode, every template. SLPs will immediately notice inconsistency.

**Design implication:** The simplified Thames Valley system (6 colors) is the best fit for the POC. It's clean enough for children to internalize, and the reduced color palette creates a more visually cohesive design.

**Sources:** [Thames Valley Children's Centre - Fitzgerald Key](https://www.tvcc.on.ca/resource/fitzgerald-key-colour-coding) | [Communication Community - Fitzgerald Key](https://www.communicationcommunity.com/fitzgerald-key-for-aac/) | [Smarty Symbols - AAC Color Conventions](https://smartysymbols.com/aac-color-conventions/) | [PrAACtical AAC - Colorful Considerations](https://praacticalaac.org/strategy/communication-boards-colorful-considerations/)

---

### 4. Visual Scene Displays (VSDs) — The Research Is On Your Side

This is where the research directly validates the app's scene-based approach.

**Key finding:** Typically developing 2½-year-olds performed significantly better at locating language concepts with Visual Scene Displays than with traditional grid displays (Wilkinson & Light, cited in PMC review).

Additional evidence:
- Infants with complex communication needs used VSDs to participate in social interactions before they were one year old
- VSDs are now considered an **evidence-based practice** for communication intervention in young children with ASD/IDD
- Video VSDs (with motion) show emerging evidence of capturing visual attention even more effectively
- Toddlers relate to VSDs more quickly than traditional grid systems

#### What Makes a VSD Effective (Research-Based Design Rules)

1. **Use personalized, meaningful content** — photos/scenes of the child's actual environment outperform generic content. For a POC, illustrated scenes are acceptable, but MVP should support parent-uploaded photos.
2. **Include people** — children attend more to scenes with human figures, especially faces. Scenes should include characters (mom at the stove, child at the table).
3. **Navigation thumbnails** — research recommends thumbnail navigation bars adjacent to the main scene, not hidden behind menus.
4. **Limit hotspots per scene** — too many tappable areas overwhelm. Research supports careful selection of the most relevant vocabulary per scene.
5. **Consistent core word access** — core words should be accessible from any scene, not buried in navigation.

**Design implication:** The app's scene-first approach is strongly supported by research. The hybrid model (scenes for exploration + core word bar) aligns perfectly with evidence-based practice. One refinement: research emphasizes personalized content (real photos of the child's kitchen, not generic kitchens). For POC, illustrated scenes work. For MVP, allowing parents to upload photos of their actual environment would be clinically powerful.

**Sources:** [PMC - Designing Effective AAC Displays](https://pmc.ncbi.nlm.nih.gov/articles/PMC6436972/) | [ASHA - Visual Scene Displays](https://pubs.asha.org/doi/10.1044/2022_PERSP-22-00162) | [PMC - Using VSDs with Young Children: Evidence-Based Practice](https://pmc.ncbi.nlm.nih.gov/articles/PMC12645487/) | [PMC - Visual Attention to Human Figures](https://pmc.ncbi.nlm.nih.gov/articles/PMC3462357/)

---

### 5. Motor Planning — The Design Constraint You Can't Ignore

**Words must ALWAYS be in the same location.** This is non-negotiable in clinical AAC.

Research by Thistle and Wilkinson (2015) demonstrated that preschoolers with consistent symbol locations showed significantly faster response times by the fifth session compared to those with variable arrangements.

The principle: communication through AAC works like typing on a keyboard. You don't look at each key — you develop muscle memory. For a child using an AAC device, the motor path to "want" must always be the same sequence of taps, regardless of context or screen.

**LAMP (Language Acquisition through Motor Planning)** is an entire methodology built on this principle:
- One consistent motor pattern per word
- Words never move as vocabulary grows
- Repetition builds automaticity
- The motor plan stays consistent across time, environments, and developmental levels

**Design implications for the app:**

1. **Core words must be in exactly the same position on every screen** — the core word bar can't rearrange, reflow, or adapt. "Want" is always in the same spot.
2. **Scene-specific vocabulary should maintain position within its scene** — if milk is top-left in the fridge, it's ALWAYS top-left in the fridge.
3. **Don't "simplify" by hiding words** — a common mistake. Reducing the grid to show fewer options makes it "easier" but destroys motor memory. Better to show all words but highlight the ones being taught.
4. **No adaptive rearrangement** — don't move frequently-used items to the top. This breaks motor planning.

> **⚠️ This conflicts with the "Favorites" feature in the current spec.** If "favorites" means pinning items to the top of scenes, it changes the position of other items. A better approach: favorites as a separate quick-access area that doesn't disturb scene layouts.

**Sources:** [AAC Community - Motor Planning](https://aaccommunity.net/ccc/motor-planning/) | [PubMed - Consistent Symbol Location](https://pubmed.ncbi.nlm.nih.gov/29860450/) | [AAC Language Lab - Motor Planning](https://aaclanguagelab.com/assets/uploads/AAC_and_Motor_Planning_for_Readtopia.pdf) | [Avaz - Motor Planning and AAC](https://avazapp.com/blog/benefits-of-aac-implementation-in-autistic-individuals-emphasis-on-motor-planning/)

---

### 6. Communicator Profiles — Design for the Full Continuum

The Bridge School defines three communicator profiles, and the app's target audience spans the first two:

| Profile | Description | AAC Needs | App Behavior |
|---------|-------------|-----------|-------------|
| **Emergent** | No reliable symbolic communication. Uses gestures, vocalizations, facial expressions. Communicates about "here and now." | Simple cause-effect interactions. Single symbols. Aided language stimulation by partner. | Single-tap words with immediate speech output. Sentence suggestions as modeling (not yet for independent use). |
| **Context-Dependent** | Uses symbolic communication reliably, but limited to certain partners, topics, or contexts. | Expanding vocabulary, building sentence construction, broadening communication partners. | Sentence suggestions for independent use. Sentence builder. Core + fringe combinations. |
| **Independent** | Full symbolic communication across partners and contexts. | Robust vocabulary (2000+), morphology, literacy. | Beyond POC scope — this is where Proloquo2Go lives. |

**Design implication:** The app should work for both emergent and context-dependent communicators WITHOUT requiring mode-switching. An emergent communicator taps "pizza" and hears "pizza" — that's their whole interaction. A context-dependent communicator taps "pizza," sees the sentence suggestions, and taps "I want pizza." Same screen, same flow, different levels of engagement.

**Sources:** [Bridge School - Communicator Profiles](https://communication.bridgeschool.org/what-is-communicative-competence/communicator-profiles/) | [NWACS - Communication Ability Levels](https://nwacs.info/communication-ability-levels) | [Bridge School - AAC System Design Principles](https://communication.bridgeschool.org/intervention/aac-system-design-guiding-principles/aac-system-design-principles-for-communicator-profiles/)

---

### 7. Pragmatic Functions — Don't Just Build a Requesting Machine

This is one of the biggest criticisms of AAC apps and the biggest mistake in AAC implementation. Most AAC interactions become purely instrumental: "I want X." But communication serves many functions:

| Function | Example | Template Pattern |
|----------|---------|-----------------|
| **Requesting** | "I want pizza" | [person] want [item] |
| **Protesting/Rejecting** | "No pizza" / "I don't want that" | no [item] / [person] don't want [item] |
| **Commenting** | "Pizza is yummy" / "Look, a dog!" | [item] is [descriptor] / look [item] |
| **Greeting** | "Hi mom" / "Bye dad" | hi [person] / bye [person] |
| **Responding** | "Yes" / "No" / "I don't know" | yes / no / [person] don't know |
| **Asking questions** | "Where pizza?" / "What's that?" | where [item] / what that |
| **Expressing feelings** | "I'm happy" / "That scared me" | [person] feel [emotion] / that [descriptor] |
| **Social closeness** | "I love you" / "Mom is funny" | [person] love [person] / [person] is [descriptor] |

**Design implication:** Sentence templates must cover ALL pragmatic functions, not just requesting. When a child taps "pizza," the suggestions should include requests ("I want pizza"), comments ("pizza is yummy"), social references ("mom likes pizza"), and rejections ("no pizza"). This is what clinicians will specifically evaluate.

**Sources:** [Communication Community - 6 Communication Functions and AAC](https://www.communicationcommunity.com/communication-functions-and-aac/) | [AbilityPath - Pragmatic Functions](https://abilitypath.org/ap-resources/pragmatic-functions/) | [Avaz - Building Language for All Communicative Functions](https://avazapp.com/blog/building-language-for-all-communicative-functions/)

---

### 8. How SLPs Evaluate AAC Apps — The Checklist

SLPs use **feature matching** to evaluate whether an AAC system fits a child. The key evaluation dimensions:

#### Feature Match Categories

| Feature | What SLPs Check | POC Implication |
|---------|----------------|-----------------|
| **Symbol system** | Iconicity (how recognizable), consistency, cultural appropriateness | ARASAAC has good iconicity research; document the choice |
| **Vocabulary organization** | Taxonomic (category-based) vs. activity-based vs. semantic-syntactic | Our hybrid scene + core words covers this |
| **Vocabulary size & growth path** | Does the system start simple and scale? Can vocabulary be added? | Spec already addresses this |
| **Motor access** | Touch target size, consistency of location, switch access | 60pt targets + consistent positioning |
| **Language representation** | Single words, phrases, sentence construction, morphology | Template engine handles this |
| **Voice output** | Quality, age-appropriateness, language options | iOS TTS — document voice selection |
| **Customizability** | Can parents/SLPs add vocabulary, modify layouts? | Parent settings cover basics |
| **Portability** | Can it go everywhere the child goes? | iPad — yes |
| **Aided language stimulation support** | Does the app support modeling by communication partners? | This is a gap — see Ten section |

#### Key Assessment Frameworks

- **SETT Framework** (Student, Environment, Tasks, Tools) — holistic assessment considering the child, their settings, what they need to communicate about, and which tools fit
- **MOSAIC** — observation-based assessment across conversation, partner interaction, capability, environment, and behavior
- **Dynamic Assessment** — trial-based evaluation where the SLP tests different systems with the child to see which produces the best outcomes

**Sources:** [AAC Community - Evaluation Tools](https://aaccommunity.net/caac_slp/evaluation-tools/) | [SLP Toolkit - AAC Assessment](https://www.slptoolkit.com/blog/aac-assessment-in-schools-a-team-approach/) | [ASHA - AAC Practice Portal](https://www.asha.org/practice-portal/professional-issues/augmentative-and-alternative-communication/) | [Speech Pathology - Pediatric AAC Assessment](https://medslpcollective.com/wp-content/uploads/2022/04/Katie-Threlkeld.pdf)

---

## Ten: The Blind Spots — What the Current Design Gets Wrong (or Doesn't Address)

### 1. Aided Language Stimulation Support — The Missing Feature

The single most important AAC intervention strategy is **aided language stimulation** — where a parent/caregiver speaks while pointing to symbols on the child's device, modeling how to use it. Every SLP will ask: "How does the app support modeling?"

The current spec has NO feature for this. This is the biggest gap.

**Recommendation for POC:** Add a "modeling mode" or "partner mode" — when activated (from parent settings or a quick toggle), the app highlights symbols as the parent taps them while talking. This doesn't need to be complex. Even a simple visual emphasis (glow, scale-up) when a parent taps symbols while speaking to the child would demonstrate awareness of this practice.

At minimum, the app should include a brief onboarding guide explaining aided language stimulation to parents — most parents don't know what it is, and it's the #1 factor in AAC success.

### 2. The "Favorites" Feature Breaks Motor Planning

As flagged in Section 5: if "favorites" reorders items within scenes, it destroys motor memory. The child learns that milk is in position 3 on shelf 2 of the fridge. If a parent "favorites" milk and it moves to position 1, the child's learned motor path is broken.

**Fix:** Favorites should be a separate quick-access overlay or bar — a shortcut that doesn't disturb the spatial layout of any scene. Think of it as a "bookmarks" strip, not a reordering.

### 3. Sentence Suggestions Should Show Multiple Complexity Levels

The current spec shows sentences at roughly the same complexity (3-4 words). Research on MLU progression says the app should present sentences at 1-word, 2-word, 3-word, and 4+ word levels simultaneously. The child naturally engages at their level, and seeing higher-complexity sentences serves as language modeling.

### 4. The App Needs to Support More Than Requesting

Current sentence examples lean heavily toward requesting ("I want pizza," "let's eat pizza," "more pizza please"). The template library needs to cover all pragmatic functions: commenting, protesting, social, emotional, questioning. See Section 7 for the full breakdown.

### 5. Personalized Content Is Clinically Superior

Research consistently shows personalized VSDs (photos of the child's actual environment, actual family members) produce better outcomes than generic content. For POC, illustrated scenes are fine. But the MVP roadmap should prioritize parent-uploaded photos — a clinician will ask about this.

### 6. Caregiver Training Is the #1 Barrier

Research identifies caregiver buy-in and training as the single biggest barrier to AAC success. The app is only as effective as the adults using it WITH the child. For POC, this could be as simple as:
- A first-launch tutorial explaining how to use the app with the child (not just handing it to them)
- A brief "what is aided language stimulation" guide
- Tips that appear contextually ("Try modeling this sentence by tapping the symbols as you say them aloud")

### 7. The Competitive Landscape Is Weaker Than It Looks

The existing AAC apps have significant weaknesses the client can exploit:

| App | Price | Strengths | Weaknesses |
|-----|-------|-----------|------------|
| **Proloquo2Go** | $250 | Gold standard, huge vocabulary, motor planning | Overwhelming UI, 80% of users don't follow recommended settings, vocabulary navigation confusing for toddlers |
| **TouchChat** | $300 | Clear hierarchical structure, SLP-friendly | Limited academic vocabulary, requires extensive custom programming, anxiety-inducing multi-activation levels |
| **LAMP Words for Life** | $300 | Excellent motor planning, consistent location | Abstract icons, steep learning curve, buttons require good memory |
| **Avaz AAC** | $10/mo | Kid-friendly design, supports real photos | Less vocabulary depth, subscription model |
| **CoughDrop** | Free–$35/mo | Cloud-based, multi-device sync, open ecosystem | Less polished, smaller user base |
| **My First AAC** | $5 | Designed specifically for toddlers, affordable | Very limited vocabulary and features |

**The gap:** No existing app combines (a) modern, child-friendly visual design, (b) scene-based navigation validated by VSD research, (c) contextual sentence modeling, AND (d) a clinical vocabulary foundation. They all have some of these. None have all four.

**Sources:** [Speech and Language Kids - AAC Apps Review 2025](https://www.speechandlanguagekids.com/aac-apps-review/) | [Spoken AAC - Best AAC Apps 2026](https://spokenaac.com/best-aac-apps/) | [Goally - Compare AAC Apps](https://getgoally.com/compare-aac-apps/) | [Smiles Therapy - Proloquo2Go Pros and Cons](https://www.smilestherapy.com/pros-and-cons-of-the-proloquo2go-aac-app/)

---

## Ketsu: What This Means for the App

### Spec Changes Required

Based on this research, the following changes to the design spec are recommended:

#### Critical (Must Have for POC)

1. **Add aided language stimulation awareness** — at minimum, a parent onboarding guide explaining ALS. Ideally, a "modeling mode" toggle.
2. **Fix the Favorites feature** — separate quick-access bar, not scene reordering.
3. **Multi-complexity sentence suggestions** — show 1-word through 4-word options, not just 3-word sentences.
4. **Expand pragmatic functions in templates** — ensure templates cover requesting, protesting, commenting, greeting, questioning, expressing feelings, and social closeness.
5. **Lock symbol positions** — add explicit requirement that scene layouts are spatially fixed. Items never move. Core words never rearrange.
6. **Use the Modified Fitzgerald Key consistently** — adopt the simplified 6-color Thames Valley system across all screens.

#### Important (Should Have for POC)

7. **First-launch caregiver tutorial** — brief, visual guide on how to use the app WITH the child.
8. **Include characters/people in scenes** — research shows children attend more to scenes with human figures.
9. **Don't gate vocabulary by "level"** — all vocabulary should be accessible from day one. Don't hide words to make things "simpler."

#### MVP Roadmap Items

10. **Parent-uploaded photos for custom VSDs** — clinically superior to generic illustrations.
11. **Communication partner training module** — in-app guidance for parents and SLPs.
12. **Data logging for SLP reporting** — which words used, frequency, communication functions attempted.
13. **Switch access support** — iOS has this built in, but the app needs to be tested with it.

### The Core Word List for POC (~50 words)

Based on the research, here are the recommended ~50 core words for the persistent bar, organized by Fitzgerald Key color:

**Yellow (People/Pronouns):** I, me, my, you, he, she, we, it
**Green (Verbs):** want, go, get, make, eat, drink, play, help, stop, come, give, look, like, love, feel, put, open, turn
**Blue (Descriptors):** more, big, little, good, bad, hot, cold, happy, sad, yummy, yucky
**Pink/Purple (Social/Questions):** yes, no, hi, bye, please, thank you, what, where
**Red (Important/Negation):** no, stop, don't, all done
**Orange (Nouns — minimal):** mom, dad
**Black (Misc):** in, on, up, not, and, the

**Total: ~52 words** — these cover the most frequently used words in toddler communication across multiple research studies.

### The SLP Skill Foundation

This research should form the basis of the SLP skill the user plans to create. Key rules the skill should enforce:

1. All vocabulary uses consistent Fitzgerald Key color coding
2. Symbol positions never change (motor planning)
3. Sentence templates cover all pragmatic functions, not just requesting
4. Multiple sentence complexity levels are always presented
5. No vocabulary gating or prerequisite requirements
6. Core words are always accessible from any screen
7. Grammar is handled correctly (conjugation, pluralization, articles)
8. Aided language stimulation is supported or at least acknowledged

---

## Comprehensive Source List

### Research & Academic
- [Banajee, DiCarlo, Stricklin (2003) - Core Vocabulary Determination for Toddlers](https://www.researchgate.net/publication/232041093_Core_Vocabulary_Determination_for_Toddlers)
- [AJSLP (2024) - How Toddlers Use Core and Fringe Vocabulary](https://pubs.asha.org/doi/10.1044/2024_AJSLP-23-00366)
- [PMC - Designing Effective AAC Displays](https://pmc.ncbi.nlm.nih.gov/articles/PMC6436972/)
- [PMC - Using VSDs with Young Children: Evidence-Based Practice](https://pmc.ncbi.nlm.nih.gov/articles/PMC12645487/)
- [PMC - Visual Attention to Human Figures in AAC](https://pmc.ncbi.nlm.nih.gov/articles/PMC3462357/)
- [PMC - SLP Practices in AAC During Early Intervention](https://pmc.ncbi.nlm.nih.gov/articles/PMC9549491/)
- [PMC - AAC for Children with IDD: Mega-Review](https://pmc.ncbi.nlm.nih.gov/articles/PMC8009928/)
- [PubMed - Consistent Symbol Location and Motor Learning](https://pubmed.ncbi.nlm.nih.gov/29860450/)
- [PMC - ARASAAC Norms Study](https://pmc.ncbi.nlm.nih.gov/articles/PMC6305113/)

### Professional Organizations & Practice Guidelines
- [ASHA - AAC Practice Portal](https://www.asha.org/practice-portal/professional-issues/augmentative-and-alternative-communication/)
- [AAP Pediatrics - Prescribing Assistive Technology](https://publications.aap.org/pediatrics/article/156/1/e2025072216/202154/Prescribing-Assistive-Technology-Focus-on-Children)
- [HealthyChildren.org - AAC for Kids](https://www.healthychildren.org/English/health-issues/conditions/developmental-disabilities/Pages/augmentative-and-alternative-communication-for-children.aspx)

### Clinical Resources & Practitioner Guides
- [AAC Language Lab - 100 High Frequency Core Words](https://aaclanguagelab.com/resources/100-high-frequency-core-word-list)
- [AssistiveWare - Teaching Core Words](https://www.assistiveware.com/blog/teaching-core-words-building-blocks-communication-and-curriculum)
- [AssistiveWare - Aided Language Stimulation](https://www.assistiveware.com/learn-aac/aided-language-stimulation)
- [Communication Community - Fitzgerald Key](https://www.communicationcommunity.com/fitzgerald-key-for-aac/)
- [Thames Valley Children's Centre - Fitzgerald Key](https://www.tvcc.on.ca/resource/fitzgerald-key-colour-coding)
- [Bridge School - Communicator Profiles](https://communication.bridgeschool.org/what-is-communicative-competence/communicator-profiles/)
- [NWACS - Communication Ability Levels](https://nwacs.info/communication-ability-levels)
- [AAC Community - Motor Planning](https://aaccommunity.net/ccc/motor-planning/)
- [PrAACtical AAC - Vocabulary Lists](https://praacticalaac.org/praactical/aac-vocabulary-lists/)
- [Speech and Language Kids - AAC Apps Review 2025](https://www.speechandlanguagekids.com/aac-apps-review/)
- [Communication Community - Does AAC Prevent Speech?](https://www.communicationcommunity.com/does-aac-prevent-speech-development/)

### AAC App Resources
- [Spoken AAC - Best AAC Apps 2026](https://spokenaac.com/best-aac-apps/)
- [Goally - Compare AAC Apps](https://getgoally.com/compare-aac-apps/)
- [Smiles Therapy - Proloquo2Go Review](https://www.smilestherapy.com/pros-and-cons-of-the-proloquo2go-aac-app/)
- [ARASAAC - Symbol Database](https://arasaac.org/)