The Evidence
None Of This Is A Guess.
Aphasia interrupts retrieval, not intelligence. The evidence below is why words are organized by job and color, why the board offers rather than demands, and why the family guide is part of the product.
Back to the Stroke & Aphasia pageHow To Read This Page
Every finding below names its sources. Where the field has no evidence, we say so instead of implying otherwise. Where Grand Voice OS goes beyond the literature, we mark it as our extension, not as a study result. Ask us for any source in full: admin@titaniumdigitalholdings.com.
What The Evidence Says
The Fitzgerald Key, in its Goossens and Crain modification, has organized AAC vocabulary by grammatical role and color for decades. GVOS uses it on every board, so a returning user is home on day one.
Fitzgerald Key; Goossens & Crain modified color coding (public domain)
Word prediction has measured communication-rate gains of up to 59 percent for motor-impaired typists. GVOS builds prediction into Word Builder's suggestion strip and the suggest board's five-word offers.
Trnka et al., University of Delaware
Communication-partner training is one of the most cited unmet needs in AAC, and untrained partners are a leading driver of abandonment. Every kit ships with How to Talk with Me because listening is a skill the family deserves to be taught.
Communication partner instruction literature; AAC abandonment studies
Consistent symbol location is the single best-replicated learnability finding in AAC research. Motor plans build around position, and moving a word costs the user the practice they paid for. Every GVOS board obeys this as law.
Thistle 2018; Dukhovny & Gahl 2014
Typical aided communication runs 8 to 10 words per minute. Phrase-based prestored speech measured 36 to 64 words per minute. That is why banked phrases, pinned folders, and one-tap libraries sit at the center of the system.
Todman 2000
Still Being Compiled
The dossier grows. These are the source sets currently being compiled for this page, named so the gap is visible instead of hidden: