Skip to content
EyeSayAAC

Research and evidence

Evidence states, stated literally.

EyeSay is an early access product. We separate established AAC evidence from EyeSay-specific product behaviour, hypotheses, internal evaluation and external validation so those categories cannot quietly blur together.

Current evidence status

What each kind of statement means

External evidence

Established AAC evidence

Evidence about AAC practice generally is cited as external evidence. It is never presented as proof that EyeSay produces the same outcome.

Implemented and testable

EyeSay product behaviour

Statements about stable motor positions, communicator authorship and partner-model isolation describe implemented product behaviour that can be checked in the software and regression tests.

Hypothesis — not demonstrated

EyeSay hypotheses

Questions about whether EyeSay's architecture changes communication gaps, search effort or long-term use remain hypotheses until appropriate EyeSay-specific evidence exists.

Internal only

EyeSay internal evaluation

Internal usability, reliability and instrumentation checks help us improve the product. They are not independent validation and are not clinical evidence.

Not yet completed

External validation

We will describe external validation as planned, underway, completed or independently replicated only when that exact state is true. No clinical-validation claim is currently made for EyeSay.

Established AAC evidence we draw on

Broader AAC literature informs design questions such as aided language modelling, access to a broad range of communicative functions and the importance of ongoing, person-centred system support. Those sources do not demonstrate that EyeSay itself improves language, participation or clinical outcomes.

EyeSay hypotheses we plan to test

The following are hypotheses, not demonstrated outcomes. We publish them so future evaluation can falsify them rather than allowing product claims to outrun evidence.

  • Whether a stable motor-language plane plus a separate Context Halo reduces unresolved word-search events over time.
  • Whether keeping refusal and self-advocacy permanently reachable changes the range of communicative functions expressed in ordinary use.
  • Whether same-topology partner modelling is easier for partners to use consistently without contaminating communicator authorship.
  • Whether promotion from contextual suggestions into permanent vocabulary can close observed communication gaps without destabilising learned positions.

Internal evaluation

Internal evaluation currently covers product behaviour, regression safety, accessibility checks, reliability, performance and privacy boundaries. It can tell us whether EyeSay behaves as designed. It cannot establish clinical effectiveness, comparative superiority or independent validation.

External validation status

External validation is not yet completed. When external work begins, we will distinguish planned, underway, completed and independently replicated work rather than collapsing those states into a single "validated" label. Until appropriate EyeSay-specific evidence exists, we make no claim of clinical validation, clinical proof or clinical superiority.

Planned evaluation questions

  • Can people reach a broad range of permanent vocabulary while the 48 stable motor positions remain unchanged?
  • Which searches end without a communicator selection, and do those gaps change after deliberate vocabulary promotion?
  • How often do partners model language, and can modelling opportunities be measured without scoring the communicator for imitation or compliance?
  • Do the technical authorship boundaries reliably prevent Halo suggestions and partner-model events from becoming communicator speech or history?
  • Which access barriers appear across touch, keyboard and future specialist access methods without inferring intent, competence, mood or correctness?

What we will not claim

  • That EyeSay diagnoses, treats or cures autism or any other condition.
  • That EyeSay has demonstrated clinical validation or clinical superiority before appropriate EyeSay-specific evidence exists.
  • That Context Halo or analytics can determine what a communicator feels, means or intends.
  • That aided language modelling establishes any particular language or communication outcome for EyeSay.
  • Outcomes that have not been measured in real use with an appropriate study design.

External references

These references support statements about AAC practice generally. They are dated and sourced here precisely so readers can distinguish them from EyeSay-specific evidence.

  1. American Speech-Language-Hearing Association (ASHA)

    Augmentative and Alternative Communication (AAC) — Practice Portal

    Living practice portal; citation date not specified · Accessed 26 August 2026

    General AAC practice source, including aided language modelling / augmented input. This is not EyeSay-specific evidence.

    Open source: Augmentative and Alternative Communication (AAC) — Practice Portal (opens in a new tab)

  2. Biggs, Carter & Gilson

    Systematic Review of Interventions Involving Aided AAC Modeling for Children With Complex Communication Needs

    Augmentative and Alternative Communication, 2018 · Accessed 26 August 2026

    Systematic review of aided AAC modelling interventions. This supports discussion of the broader practice, not a claim about EyeSay outcomes.

    Open source: Systematic Review of Interventions Involving Aided AAC Modeling for Children With Complex Communication Needs (opens in a new tab)

  3. Wallace, Rispoli, Eiser-Hess & Gonçalves

    Peer-mediated aided AAC modeling: a systematic review

    Augmentative and Alternative Communication, 12 May 2026 (online ahead of print) · Accessed 26 August 2026

    Recent systematic review describing the evidence base and its limitations in school-based peer-mediated aided AAC modelling.

    Open source: Peer-mediated aided AAC modeling: a systematic review (opens in a new tab)

Working with us

We are looking for families, schools and speech and language therapists willing to evaluate the system in ordinary daily use and report where it fails. Measurement should use privacy-safe derived metrics wherever possible; we do not need raw child communication transcripts for marketing analytics.