Does that sound right?
On the hidden potential of sign language data
Sign language also fails because of technology – when supply and demand don’t find each other. How do we fix that?
When you evaluate sign language services per region, you hit a data barrier: whether sign language is offered is usually not tied to a place in the data, but to specific events at specific times. In museums or at the theatre, for example.
Event platforms often flag this information too, for example with a small icon showing a sign made with two hands in a pinch grip.
But: places offer sign language today and not tomorrow. Individual events come and go. And unfortunately, the service often comes and goes with them. How do you count that?
On top of that, the information is usually not machine-readable yet. That is why the sign language situation cannot be captured consistently per place and region. Can AI help here? Only to a limited extent – because the information also has to be correct. The lack of machine readability makes verification harder, for example when “clearing out” outdated data, or as a reporting chain when users spot errors in the information.
For other helpful place features, such as wheelchair accessibility, a coherent statistical picture is already emerging. This makes helpful statistics possible across countries. There is no way around machine readability for this.
With sign language, we are only just getting started.
Apps can find and share the information automatically.
Yet there are already suitable standards: anyone who organises an event and promotes tickets on a platform like Eventbrite is already taking part. So is anyone who offers a hotel or a holiday apartment.
The platforms for this already pass on data – about events, places and services. The machines are already talking to each other. The exchange happens by itself. This is how the date of an event, the address, special discounts and more end up in search engines today. Other apps find the information automatically through the corresponding open standards. That increases reach and revenue.
One of the methods behind this is widely used around the world and is called “Schema.org”. It is used by all common search engines and LLMs. It makes events, services, places and offers visible to Google, Gemini, ChatGPT and Claude.
There is no reason not to try this for sign language too.
However: a lot of accessibility information, such as sign language, is usually not included there yet (or only rudimentarily). Event platforms, for example, “simply” lack the corresponding form field when entering an event. People who are affected are still too rarely involved in designing these apps.
We are sure: these technical standards can be extended. And their use can be pushed forward. Why do we believe that? In many other data formats, information has recently become a legal requirement – for example on comprehensive accessibility features in public transport. Across Europe, no less. In some cases, data is already exchanged in real time, e.g. for elevators.
That would be one of many good reasons to anchor sign language more firmly in technical standards too – and ultimately to make sharing the information a legal requirement once feasibility has been proven. Funding could also be tied to the condition of sharing the information.
Just one answer is missing: how do we adapt the existing standards?
A machine-readable standard could make local sign language services more visible and easier to capture. It would prove feasibility – and thus make a law possible that makes this information mandatory.
With the a11y-Score, you would then see more clearly in which regions sign language has been taken into account. And where there is still a need to talk.
Interested? Feel free to write to us.