Cavendish Cards is a deck of pictures a person points at to say something they could not otherwise say. So is an AAC board. The resemblance is real, it is the first thing most people notice, and it is the most consequential misunderstanding this project has.
The two things are not the same, and the difference is not a technicality. It decides what belongs in the deck, what the deck is allowed to grow into, and — most importantly — what a person is still owed after the cards are on the table.
The standing rule: this deck sits alongside a person's real communication tools, never in place of them. If a person needs a way to communicate, they need AAC. "They have the cards" is never a reason to under-provide it. This holds at any age, and especially for children.
Why this page exists
The mistake is easy to make and expensive to hold.
It is easy because the form is nearly identical. Pictures. A grid or a spread. A person who is not speaking, pointing. If you have ever seen an AAC board, a set of Cavendish Cards on a table looks like a small, pretty one.
It is expensive because of what follows. When a room decides the cards are the communication system, three things happen: nobody chases the device, the trial, the funding, or the training; the person's actual range of expression is quietly capped at whatever this deck happens to name; and everyone feels good about it, because something visible is being done. The deck becomes a reason to stop, which is the opposite of what it is for.
The cards are cheap, free, printable, CC0, and available this afternoon. Real AAC is often expensive, gate-kept, waitlisted, and fought for. That asymmetry is the whole mechanism. A free thing that resembles the expensive thing will be offered as a substitute for it, every time, by people with no bad intent and no budget. Naming it is the only defense the deck has.
What AAC is
Augmentative and alternative communication is everything a person uses to communicate other than speech: gestures and signs, a letterboard, a symbol book, a speech-generating device, a phone with text. Unaided (the body) and aided (something outside it). Low-tech and high-tech, usually both at once, because batteries die and pools are wet.
What matters here is not the technology. It is the job. AAC exists so a person can say anything they might need to say — including the things nobody planned for, the things that are inconvenient, and the things that keep them safe. Robust AAC is built around core vocabulary: the small set of high-frequency words that make up most of what anyone says, combinable into sentences that were never on anybody's list. It is a language system, and it belongs to the person.
Four commitments come with it, and this project holds all four:
- No prerequisites. Access is not earned by demonstrating readiness, attention, or "communicative intent." Candidacy models have been rejected by the field for decades, and the least dangerous assumption is always competence.
- Available all the time. Not at the table, not in the session, not when the aide has it. A communication system that can be taken away is not a communication system.
- Modeled, not tested. Communication partners use the system themselves, out loud, without demanding a response. Nobody learns a language they only ever hear as a quiz.
- Never a threat to speech. Giving a person AAC does not suppress speech, and the evidence points the other way. "Let's wait and see if words come" is a decision to leave someone without a voice in the meantime.
Anyone might need it, and need it intermittently. A person who speaks most days may lose speech in a meltdown, in shutdown, in a loud room, or with a stranger — and part-time AAC users are among the most under-served people in the field. The deck's no words right now names exactly that weather. It does not answer it.
What Cavendish Cards is
A lens, not a language.
The deck makes one hard-to-voice thing sayable: sensory, regulatory, and emotional weather, and the conditions that help. It names states and needs that are difficult to put into words even for fluent speakers — and then it stops. Its vocabulary is small and opinionated on purpose. Completeness is not the goal; chasing it is how the deck drifts.
A spread is a design brief for the environment, not a message. cave, buzzy, headphones is a to-do list for the room.
Six differences that matter
Curated, not comprehensive. AAC has to let a person say anything — including "stop," "that hurts," "not him," "I want to go home." This deck names a specific inner world and refuses to expand past it. A vocabulary that is deliberately incomplete cannot be anybody's communication system, and a deck that grew until it could would have become a bad AAC board by accident.
A reframe, not a referent. An AAC symbol is a neutral pointer: "juice" means the drink, and the symbol has no opinion. Every card here carries a worldview in its Notes — what the state is, why it is not misbehavior, what to do. An AAC symbol that argued with you would be a broken AAC symbol. A card that is only a label doesn't belong in this deck.
Precision is required there, and not here. If an AAC user says "juice" and gets water, the system failed. If a person lays foggy and the facilitator isn't sure what it means, nothing failed — the card opened something. There is no right card and no wrong reading. A tool with no accuracy requirement cannot carry the weight of a person's actual words.
The work lands in different places. AAC is the person's expressive channel; using it well means the person said what they meant. A spread is a relational object: it obligates the facilitator and changes the room.
Privacy is the default here, and transmission is the default there. Communication is addressed to someone. A spread is not: cards lie face-down, a person can lay one for themselves and never turn it up, and turning a card up is the act of consent. A system whose default is "this stays private" is not a way to talk to people.
Optional here, non-negotiable there. Nobody has to play. Declining a spread costs nothing, and "I don't want to" is a complete answer. AAC is not an invitation and not an activity — it is access, and it is a right.
The deck deliberately moves labor off the person and onto the environment. AAC deliberately gives the person their own power to speak. Both are right. They are not the same move.
The inclusion test, and why the line is where it is
From CARD.md, the rule every proposed card has to pass:
A card earns its place if it names a hard-to-voice inner state or a niche-construction need — and carries a reframe in its Notes. It is not for wants, objects, or actions a person could request.
The quick test: "can't tell" belongs. "hungry" does not.
"Can't tell" names interoceptive uncertainty — a real, hard-to-voice experience that most people have no words for, and that gets misread as not caring or not trying. "Hungry" is a request. It is ordinary communication, a person can make it through any channel they have, and it belongs to their AAC.
This looks fussy from the outside. It is the single rule holding the project's shape. Every request-shaped card added is a step toward a board with ninety words on it that somebody will call adequate. The failure mode is not one bad card — it is a slow slide toward comprehensiveness, ending in a deck that has become a communication system nobody designed, evaluated, or resourced. Growing toward AAC is a failure mode, not a roadmap.
If you want the deck to be able to say more things: that is a real need, and the answer is AAC, not more cards.
"They have the cards"
The sentence to listen for. It shows up in IEP meetings, in handover notes, in "we've tried everything."
Whenever the cards are being counted as communication provision, something has gone wrong. The correct response is to put it in writing the other way around: the spread is evidence of what this person is carrying, and evidence that they need a real way to say things. A deck of prompt cards is never a reason to close a referral, delay an assessment, deny a device, or decide someone is "communicating fine now."
Say it plainly in the room: these are not their words. They still need their words.
Is it PECS?
No, and the difference is the ethos.
Picture exchange protocols are behaviorist by construction: phases, prompting hierarchies, reinforcers, an exchange trained until it is reliable. Even on their own terms they teach requesting first and most, which is why so many people leave them able to ask for things and unable to say anything else. Much of the Autistic community and a good many speech-language pathologists argue for skipping straight to robust, language-based AAC.
This deck carries no behaviorism at all. No card is a target, a reward, a step, or a compliance check. There is no prompting, no correct response, nothing to fade, and nothing to reinforce. Turning a card up is communication, not performance. Nobody hands a card to anybody to get something.
Pictures are not the common thread. The paradigm is what differs, and it differs completely.
How the two actually work together
Alongside is not a disclaimer. It is a working relationship.
- A spread can ask for AAC.
another way to talknames the need for a second channel — pointing, typing, writing, signing, showing. It is not the channel. When it comes up, the answer is to hand over the device, the board, the whiteboard, and the time. - A spread can be a design brief for AAC access. Device charged and within reach. Not left in the bag on trips. Partners who model. Nobody talking over it. Time to compose, without the question being reworded.
- AAC users play the deck through their AAC. Pointing is one access method among many — eye gaze, partner-assisted scanning, directing someone else to move the cards, using the web deck with a switch or screen reader. Nothing in the deck requires hands, speech, or literacy.
- Model the deck like you model AAC. Lay your own spread. Say what your own weather is. A facilitator who only ever asks other people to play is running an assessment, which this is not.
- The deck is not a stepping stone to AAC, and not a trial of it. Playing well says nothing about whether somebody needs a device. Not playing says nothing either.
What we owe AAC users
Standing commitments, not aspirations:
- The deck is never a prerequisite, a gate, or a readiness check for getting AAC.
- The deck is never offered in place of an evaluation, a trial, funding, or training.
- AAC is never withdrawn, withheld, or made contingent — not for behavior, not for a session, not for a spread.
- Nothing in a spread is used to argue someone communicates well enough to need less.
- The deck never asks a person to abandon their own channel to use this one.
If you remember one thing
The deck is a lens, not a language. It makes one hard-to-voice thing sayable and hands the room a to-do list. It does not give anyone their words, and it was never built to.
If a person needs a way to communicate, they need AAC. Get them AAC.
Lineage and further reading
- National Joint Committee for the Communication Needs of Persons with Severe Disabilities — the Communication Bill of Rights, and the position against eligibility prerequisites for communication services
- Anne Donnellan — "The criterion of the least dangerous assumption" (1984); presume competence
- Millar, Light & Schlosser — meta-analysis on AAC and natural speech development (2006): AAC does not impede speech
- Janice Light & David McNaughton — communicative competence in AAC
- Stimpunks Foundation — Communication Bandwidth · Situational mutism
- Cavendish Cards — the inclusion test in
CARD.md· guidebook · why this exists · example spreads
Dedicated to the public domain under CC0 1.0. Home: https://stimpunks.org/projects/cavendish-space-project/