August 9, 2026 · current

The Mixed Home Page — Words and Folders on One Grid

Vendor survey (WordPower, Proloquo2Go, TD Snap, LAMP, Grid 3) and evidence synthesis on mixing core words and folders on the home grid — no surveyed system ships a categories-only home.

Oso Voice research — 2026-08-09. Three parallel research threads (TouchChat WordPower home composition; Proloquo2Go / Proloquo / TD Snap Core First / LAMP Words for Life / Grid 3 Super Core home composition; the peer-reviewed evidence base) on one question, raised by Adam at the hub-and-spoke device checkpoint:

“In the end the main page can’t just be categories. It will end up being a mix.”

This synthesis is the input to a future brainstorm → spec cycle. It deliberately makes no design decisions — it establishes what the field ships, what the evidence supports, and which questions the spec will have to answer. Context: the shipped hub-and-spoke design (spec 2026-07-28-hub-and-spoke-navigation-design.md) renders Home as a pure grid of category folders and locked “a grid is never a mix of words and folders” as a v1 rule. The prior research synthesis (2026-07-28-aac-navigation-research.md) already noted WordPower’s home is grammar-ordered core words with nouns evicted behind a GROUPS button — this document goes deeper on exactly that axis.

Confidence labels: [measured] (vendor documents/screenshots directly inspected) [vendor] (vendor prose) [peer-reviewed] [secondary] [not found].

Headline

Adam’s instinct matches the entire field. No surveyed AAC system ships a categories-only home page. Every one of them fills the home grid’s center with core words and pushes navigation to the periphery (bottom row, right edge, top row) or out of the grid entirely — and the developmental evidence explains why: young children don’t organize vocabulary taxonomically, so a home page made only of taxonomic category folders is the one arrangement the literature would predict to be hardest for Oso Voice’s users. The open questions are not whether to mix, but which words earn home-grid cells, where the folder region sits, and how the mix stays position-stable across grid sizes. Nothing in the evidence requires abandoning the shipped architecture — the navigation tree, fixed positions, and folder mechanics all carry forward; what changes is the composition of one node (Home).

01 — What the field actually ships

System Home composition Folder/nav placement Folder visual distinction
TouchChat WordPower (PRC-Saltillo) Core words fill the center, Fitzgerald left→right (pronouns left, verbs middle, descriptors right); nouns evicted entirely behind GROUPS ~10 category buttons at the periphery — top row + edges; count stays ~constant from 20-cell to 140-cell grids (nav share shrinks from ~50% to ~6% as grids grow) [measured] Color-coded category buttons (lighter shades; orange GROUPS)
Proloquo2Go / Crescendo (AssistiveWare) Core words arranged by grammatical category, color-coded borders Fringe folders “at the bottom” of the home page [measured — manual p.15] Literally “folder-shaped buttons”
Proloquo (AssistiveWare, newer) Core in columns with tabs Fringe folders in a dedicated yellow right-side column [measured — support KB] “Shading at the top — similar to a filing cabinet’s real folders”
TD Snap Core First (Tobii Dynavox) Core words color-coded by column (Questions/Pronouns/Verbs/Little Words) Navigation lives in a persistent toolbar outside the grid; on-grid links only at the bottom of each column + pink right-edge cells; Word Forms lower-right corner at ≥5×5 [measured] Separate toolbar chrome; pink link cells
LAMP Words for Life (PRC) 84 fixed cells, ALL words — no folder cells at all; pressing a word IS the navigation (opens its word-family page). Deliberately removed Unity’s top-row Activity Row for violating one-motor-pattern-per-word [measured — vendor comparison doc] n/a — no folder region Background tint (blue) marks “this button opens a page”
Grid 3 Super Core (Smartbox) Chat strip top row; core words fill the body Extended-core jumps “along the bottom of the grid”; topic jumps in specific interior columns [measured — KB] MFK color coding — Smartbox’s own documentation attributes the scheme to Goossens, Crain & Elder 1992; note that docs/decisions/002-mfk-color-coding-as-accent.md treats Goossens’ 1992 scheme (pink verbs, yellow nouns) as a distinct convention it considered and rejected, so “same scheme Oso Voice uses” holds for the Modified Fitzgerald Key, not for Goossens’ own key

Cross-system invariants worth treating as strong priors:

  1. Core words own the center; navigation owns the periphery. Bottom row and right edge are the two documented folder regions (P2G, Proloquo, Super Core, Core First); top row is used for either category buttons (WordPower small grids) or chrome (Super Core’s chat strip) — never for the message-critical core words’ bulk.
  2. Folder count stays roughly constant as grids grow — WordPower keeps the same ~10 lettered category buttons from WP20 to WP140 [measured]. The grid growing means more words, not more folders. Labels/structure stay constant across sizes; absolute coordinates do not (only same-cell-count files, e.g. the 60↔60 bilingual pair built for shared keyguards, promise coordinate identity) [measured].
  3. Folders are visually marked as a class — folder shape, filing-cabinet tab shading, or a background tint distinct from part-of-speech coding. OpenAAC’s cross-vendor developer guidance: “usually a different border style or thickness, folder-shaped border, or a visual flag in the top corner” [secondary]. Oso Voice’s shipped folder cell (folder glyph + accent border) already follows this convention.
  4. Every vendor states the same rationale — motor planning — for keeping both word and folder placements fixed [vendor, unanimous].
  5. Nouns/fringe go behind folders; core never does. WordPower’s rule is the cleanest statement: nouns live only behind GROUPS; pronouns/verbs/descriptors live on the grid [measured].

02 — What the evidence base says (and doesn’t)

  • No study directly tests folder-cells-among-word-cells. [not found] The field’s own state-of-the-science review (Light, Wilkinson, Thiessen, Beukelman & Fager 2019, AAC 35(1)) explicitly names hybrid displays as understudied. Any confidence here is inference from adjacent literatures.
  • Young children don’t organize vocabulary taxonomically. [peer-reviewed] Fallon, Light & Achenbach (2003, AAC 19(2)): 4–5-year-olds free-sorting 42 symbols grouped by event schema, not category, and organized only ~40% of items at all. Drager et al. (2004, JSLHR 47(5)): 3-year-olds did better with contextual-scene than grid layouts after one session. Light et al. (2004): schematic organization easier than taxonomic for 4–5-year-olds. The often-cited “children shift to taxonomic competence around 6–7” is a clinical generalization, not a single traceable finding [secondary]. Implication: a home page that is only taxonomic category folders front-loads the hardest organizational scheme at the most-visited node.
  • Visual clustering helps; heterogeneous scatter hurts. [peer-reviewed] Wilkinson & McIlvane (2013, AJIDD 118(5), DS + ASD samples), Thistle & Wilkinson (2017), Wilkinson, Gilmore & Qian (2022, JSLHR 65(2)): clustered, visually-cued arrangements speed search and cut errors versus uncued arrays. Inference for the mix: folder cells should form a spatially contiguous, visually distinct region (edge row/column), not be interspersed cell-by-cell among words — which is also exactly what every vendor ships.
  • Once mixed, freeze it. [peer-reviewed] Thistle et al. (2018, AJSLP): consistent symbol location drove motor learning in preschoolers; reshuffled locations prevented improvement. Adding a folder region must not move existing word positions afterward — the same reserved-slots discipline the app already enforces. (LAMP’s stronger motor-planning claims are vendor/clinical-model evidence only — an 8-child pre/post study — [vendor].)
  • Navigation elements coexisting with content is not attention-ruinous. [secondary — not independently confirmed] An eye-tracking study reported in the 2019 review (O’Neill et al., 2018: children with ASD/DS/ID viewing scene displays with a navigation bar) found ~61–67% of fixation time on content vs ~30–33% on the navigation strip. Reassuring but a different paradigm; treat as weak support.
  • Population caveat, field-wide: nearly all the controlled studies use typically-developing children as stand-ins; Thistle et al. themselves flag that replication in AAC-using autistic children is missing.

03 — Implications specific to Oso Voice

What carries forward unchanged from the shipped design:

  • The navigation tree, NavNode model, folder cells, auto-split fallback, depth cap, and per-node position semantics all survive — a mixed Home is a change to one node’s composition, not to the architecture.
  • The folder-as-class visual convention is already shipped (glyph + accent border) and matches both vendor practice and the clustering evidence.
  • The reserved-slots / never-move discipline is exactly what the motor-planning evidence demands of a mixed page.

What the mixed-Home spec will have to decide (the brainstorm agenda):

  1. Which words earn Home cells? The field’s answer is core words arranged by grammatical function. Oso Voice already HAS an always-visible core-words toolbar (the retained co-presence mitigation) — so the design must resolve the redundancy: does the toolbar dissolve into the home grid (WordPower style)? Does Home carry a different word set (top personal/frequent words)? Carrying both invites duplication the vendors avoid.
  2. Where does the folder region sit? Bottom row and right column are the two documented conventions. A fixed folder region also has to compose with the Home cell rule (index 0 of non-home grids) and grid-size scaling.
  3. One GROUPS-style entry or per-category folders? WordPower centralizes nouns behind one orange GROUPS cell plus ~8 grammar-class buttons; Oso Voice currently shows 9 category folders. With a folder region of limited size, some consolidation (a single “Groups”/“Words” folder opening today’s category directory) may be forced at small grids — which today’s category-directory grid already implements as a ready-made target.
  4. Grid-size behavior. WordPower’s measured pattern — constant folder count, word count scales — is directly implementable with the existing buildNavigationTree capacity math, but it means the word portion of Home is what grows/shrinks, with the collapse/split logic Oso Voice already has.
  5. Interaction with the Tier-1 correlation mechanics still queued from the 2026-07-28 research (verb→category jumps, next-word strip): a home verb cell that jumps to its category (eat→food) is navigation-by-word, LAMP-style — potentially reducing how many folder cells Home even needs. These two threads should be brainstormed together, not sequentially.
  6. Position stability at the transition. Introducing the mix is itself a one-time layout change for existing users — the same class of change the grid-size warning covers. The spec should decide whether the mixed Home is a package-curation change (positions from the manifest, like everything else) or an app-computed composition.

04 — Confidence ladder

  • [measured] Vendor home-grid compositions and folder placements (WordPower QRG/manual annotated screenshots and per-size Category Links tables; P2G manual p.15; TD Snap page-set guide; Super Core KB; Unity/LAMP comparison docs).
  • [peer-reviewed] Taxonomic organization is developmentally hard for under-6s (Fallon 2003; Drager 2004; Light 2004); clustering/visual-cue benefits (Wilkinson line); position-consistency motor learning (Thistle 2018).
  • [vendor] All motor-planning rationales; LAMP’s efficacy claims.
  • [secondary] O’Neill nav-bar eye-tracking figures (secondhand via the 2019 review); the “age 6–7 taxonomic shift” heuristic; OpenAAC conventions page.
  • [not found] Any direct test of mixed word+folder grids; any study of folder-vs-word cell discrimination in children; longitudinal real-app data on any of these display variables.

Primary sources

  • PRC-Saltillo, WordPower Vocabulary Quick Reference (19784v1.02, 2022). PDF
  • PRC-Saltillo, WordPower Manual for TC-NC-CF (2023). PDF
  • AssistiveWare, Proloquo2Go 4.0 Manual, p.15. PDF
  • AssistiveWare, Proloquo buttons & folders support doc. assistiveware.com
  • AssistiveWare, Fringe words in Proloquo (E. Sheldon). assistiveware.com
  • Tobii Dynavox, Core First English Page Set Guide v1.1. PDF
  • Tobii Dynavox, Snap Core First User’s Manual v1.12, §7, §9. PDF
  • Liberator/PRC, Comparison of Unity and LAMP Words for Life (Halloran & Browning). PDF / Differences doc
  • Smartbox Hub, Using Super Core; Colour coding. hub.thinksmartbox.com
  • OpenAAC, developer considerations. openaac.org
  • Fallon, Light & Achenbach (2003). AAC 19(2), 74-85.
  • Drager, Light et al. (2004). JSLHR 47(5), 1133-1148.
  • Light, Drager, McCarthy et al. (2004). AAC 20(2), 63-88.
  • Wilkinson & Jagaroo (2004). AAC 20(3), 123-136.
  • Wilkinson & McIlvane (2013). AJIDD 118(5), 353-364.
  • Thistle & Wilkinson (2017), background color/arrangement cues (journal citation unconfirmed — see caveats).
  • Thistle, Holmes, Horn & Reum (2018). AJSLP 27. PubMed 29860450
  • Wilkinson, Gilmore & Qian (2022). JSLHR 65(2), 710-726.
  • Light, Wilkinson, Thiessen, Beukelman & Fager (2019). AAC 35(1), 42-55.

← All research