This project aggregates a lot of Auslan and other sign language data in easy machine readable formats. The data itself is often subject to copyright and other rules, so if you're building something commercial make sure to check each and every underlying source and secure appropriate licenses.
Copyright is a commercial right, and doesn't generally apply to non-commercial activities so individuals, hobbiests, and non-profit community groups should not be affected by that requirement when nobody's making money.
In addition to copyright, data available through this interface is also bound by:
- Commercial activities must have Deaf and/or Sumain leadership with veto power over projects and direction (see reasoning 1)
- Projects must not be related to policing, millitary, surveilance, auto-moderation, or censorship. (see reasoning 2)
- Projects must not attempt to build or support the building of automatic machine translation between a nation's recognised spoken/written language, and a grammatically distinct sign language. (i.e. Auslan to English machine translation is not allowed, but Auslan to ASL machine translation is acceptable) (see reasoning 3)
- Any project interacting with sign language data that is to be widely available (i.e. an app anyone can download or a website) must be made in consultation with Deaf and/or Sumain people, and must be removed from access if there appears to be a widespread community consensus that the app/website/service is harmful or unethical. (see reasoning 4)
- Deaf: a person who identifies as Deaf (with a capital D), who uses an established sign language as their primary or preferred form of communication, and is active and immersed in Deaf Culture.
- Sumain: a person who identifies as Sumain, who has a strong relationship with visual communication and generally prefers to communicate using a sign language, and is fluent and culturally immersed in that sign language.
- Too often, services that aim to support Deaf and/or disabled people are run by abled people with little understanding of the needs and risks associated with the services they provide. Deaf people have fought hard, for centuries, to protect and nurture the world's sign languages in to the beautiful, poetic, efficient, expressive, deep languages they have become today. Only Deaf and/or Sumain people are qualified to assess the risks associated with any technology that impacts language and culture, and commercial activities are likely to get tied up in government funding and other oppressive systems that put it at greater risk of harm and coersion. Deaf community in australia also has a growing consensus that hearing people profiting off Deaf languages is cultural appropriation and a kind of wealth theft.
- Sign Languages enjoy a position of privacy in the modern world, immune from being understood by outsiders. This is important, because sign languages generally exist within their own deep and distinct cultural groups, which have very different norms around respect, fairness, and mutual aid. Surface level access to signed content might appear rude, discriminatory, overly direct, or even offensive to outsiders who do not have sufficient cultural understanding to accurately interpret the communication in context. Building systems to surveil or moderate Deaf communications carries a high risk of misunderstanding, overpolicing, real discrimination, and cultural oppression.
- Fluent Auslan signers generally agree that visual-spatial sign mode natural languages like Auslan have such distinct grammar when compared to written mode languages, that high quality machine translation is not plausible. With the current political situation where Auslan interpreters are paid for by businesses and government services that barely understand the sign languages they're funding, there is too great a risk that low quality machine translation services would undercut the pay rate of qualified expert human interpreters. Under systems like australia's NDIS (National Disability Insurance Scheme) it is a requirement that people use the cheapest available disability supports, so the existance of low quality machine translation would be an existential risk to human interpreting services, and carries a high risk of forcing Deaf people to do intense advocacy work to argue why they need appropriate and accurate interpreting services instead of a computer system that government workers would not be able to see the flaws in. Translation between same-mode languages like ASL and Auslan is a more plausible task to do well, and in consultation with Deaf interpreters who do this sort of work, most agreed that experiments in machine translation between sign languages is unlikely to be harmful and has much better potential for high quality results.
- Some projects are non-commercial in nature, but carry a high risk of linguistic or cultural impact. That impact can be very worthwhile and positive, but only Deaf and/or Sumain people are qualified to assess if any piece of technology is having a positive impact. The magic of software is it's ability to amplify work. Build something once and then everyone can use it. We need to ensure that impact is positive, and doing so requires consultation with a variety of Deaf and/or Sumain people, listening to and being guided by feedback, and in the worst case, pulling down and destroying any service or technology which is widely understood to be harmful by Deaf/Sumain communities.