Sign16

The person

The challenge came from Bob’s gay Deaf friends in Singapore in the mid-1990s. The complaint behind it was partially fair: most accessibility research at the time went to blind and low-vision users, or to mobility-impaired users; the Deaf and Hard-of-Hearing community got very little. Bob was challenged to come up with something. The agreed price was a beer.

The constraint

Mid-1990s Singapore had unusually early 3G cellphone adoption. The Deaf community used 3G video calling to sign peer-to-peer in real time, which solved synchronous communication beautifully. Asynchronous text on the same phones still required typing English, because no sign-writing system existed for cellphones. The era’s SMS input ran on a 16-key keypad. Whatever a sign-writing system was going to be, it was going to have to fit there.

The use case extended beyond chat. A Deaf student would watch an English lecture through a sign-language translator, see English slides, then be asked to take notes in English — a perpetual game of Telephone from English to Sign and back to English again. “It’s mad, just mad.”

The insight

Bob couldn’t sign — still can’t,“only slightly encumbered by the fact that I can’t sign.”The non-fluency turned out to be the key. From Bob’s telling:“If you don’t know what the signer is saying, the speech becomes performance, it becomes dance. So, maybe if I simply described what I saw, that would be a form of sign-writing?”

Encoding the form of the sign rather than its meaningsidestepped the requirement to know any specific sign language. A signer would recognise their own actions in form-descriptions; the writer didn’t need to know what the signs meant. The writer also didn’t need to be a member of the signing community to build the tool, as long as the tool encoded what was visibly happening rather than what the visible thing meant.

The artefact

Sign16: 16 keys mapped to body parts and abstract movement primitives — away, moving together, and so on — with a custom font installed on the phone display so that the descriptions rendered as compact symbols rather than verbose English text. Some feature phones of the era allowed font installation even when they didn’t allow app installation; that opening was what made the system shippable in the first place. A later extension for laptop keyboards used a richer symbol set and the same encoding principle.

The teaching

Two lessons from Sign16, both of which Bob carried forward.

The first is that constraint-driven design produces solutions that freedom won’t. Sixteen keys forced a productive simplification — body-part plus movement-primitive — that a richer input device would never have arrived at. The simplification was not a compromise on what the system could do; it was the structural insight that let it do anything at all.

The second is that the writer of an accessibility tool does not have to be a member of the community the tool is for, as long as the tool encodes form rather than meaning and the community can read its own behaviour back from it. That is a stricter design discipline than it sounds: it rules out making implicit assumptions about what the community already understands, and it forces the encoding to be behavioural rather than semantic.

The coda

Bob won the beer. “The guys continued to sign to each other on their phones like it never happened.” Worth keeping verbatim — it is a self-aware ending to a story about a tool that solved a problem its users didn’t quite have.

A side observation that earns its place: Tantric on a Friday evening was loud and crowded, and ordering drinks at a packed bar is hard. For the Deaf friends, less so — one of them worked behind the bar, so they would simply sign their order over the heads of the people in front, and by the time they pressed through to the front, the drinks were ready. Accessibility research rarely surfaces the moments where the disabled approach is the advantageous one. “Sign is the greatest gift to a young person who likes to go clubbing.”