Welcome everyone to the Whose Voices? podcast. Here we’re reimagining the internet as a feminist, anti-colonial, anti-capitalist space that centers the people of the global majority. Together, we’ll explore how whose knowledge’s work is envisioning a digital world where our stories, our wisdom, and our futures take center stage. So, tune in and turn up the volume, because when we ask whose voices, the answer is clear. Ours!

Maari: Did you know that while there are over 7,000 languages spoken around the globe, a majority of the internet’s content exists in only under 10 languages. And most of which are Eurocolonial languages like English. This means that for the vast majority of the world, access to digital content and technologies in their native languages is severely limited.

Now let’s consider another layer of this challenge. The intersection of multilinguality and disability. In regions like the Asia-Pacific, where nearly half of the world’s population of people with disabilities reside and where over 2,300 languages are spoken, the impact of this digital language divide is even more pronounced.

Here, a lack of accessible technologies in local languages prevents people with disabilities from fully and meaningfully participating in the digital world. This isn’t just an inconvenience. It’s an amplification of existing inequalities with very real consequences.

Without the ability to access information, services, and opportunities online, people with disabilities face significant barriers in education, employment, and social inclusion. Without the ability to create and represent their knowledges online, they face a form of epistemic injustice.

Today, we’re coming to you from the vibrant city of Taipei at RightsCon 2025. We’ve just wrapped up an insightful dialogue session covering community-based approaches to accessible language technologies. We asked at our panel why language technologies are not made for us. and particularly not made for people with disabilities. Why are they always motivated and driven by profit seeking corporations? And what do communities do and have to say about these issues?

Before we get into the depth, we’d like to let you know that this conversation was recorded live at a very bustling conference at RightsCon 2025. You might hear during this podcast, some of that lively buzzing in the background. But we’ve done our best to make sure it does not obscure the conversation itself. With that said, we’re thrilled to continue the conversation here with our incredible panelists.

Joining us is Khansa Maria, a disability rights activist, researcher, and independent consultant whose expertise spans international development, disability justice, and inclusive education. Currently pursuing her PhD at University of Oxford’s Department of International Development, she brings her lived experience and knowledge to our discussion.

Khansa: Thank you, Maari. Really happy to be here.

We’re also honored to have Rahul Bajaj with us. A practicing lawyer with expertise in disability rights law in India. As the co-founder of Mission Accessibility, an organization championing the rights of the disabled and a legal consultant on disability rights issues, Rahul offers critical insights into the tech and legal landscape of accessibility.

Rahul: Hello Maari.

And completing our panel, we have Asmelash Teka Hadgu. As the co-founder and CTO of Lesan, he spearheaded the creation of cutting-edge machine translation systems for languages like Amharic and Tigrinya. His PhD research at Leibnitz University focused on applied machine learning for scholarly communication, crisis response, and NLP in low resource settings.

Additionally, as a part-time fellow at DAIR, distributed AI research, he’s building partnerships to develop language tools for the “low-resource languages”.

Asmelash: Yeah, thanks Maari.

Maari: And I’m Maari Maitreyi, the language justice program co-lead at Whose Knowledge. I’m a feminist, artist and scholar, exploring the intersections of technology, knowledge, and language. My work focuses on how these elements shape the experiences and realities of the people of the Global Majority. All right, let’s kick things off. My first question is for Khansa. Khansa, what’s your experience of navigating the internet, particularly as a person with visual impairment? What are the joys you’ve encountered?

Khansa: I think that’s a very interesting question. For me, the experiences that gave me joy are connecting with community. I found an entire disability community online, people that I did not know about, especially because I went to a mainstream school.

So you know, meeting those people, having that exposure, and I’ve made some really, really good friends. Like, I’ve met them in person now and it’s I think some of my best friends I’ve made online.

And I think it’s something very special about connecting with people who look like you or who may have the same experiences of marginalization. But you don’t necessarily get to meet them in real life and then you get to connect and find all these resources. And I think something else that gives me a lot of joy is like just audio books that I find online. Sometimes it’s like watching vlogs and stuff. So really just finding connection and accessible ways of engagement.

Maari: Thanks, Khansa. I think that makes a lot of sense. Community and access are key to the digital experience of people from marginalized locations. Rahul, let’s turn to you. How has your experience been, particularly accessing the internet in Hindi?

Rahul: Of course, I really enjoy using the internet in English as well, but there are some kinds of content which are better enjoyed in Hindi. Shayaris (poetry), you know, some kind of stories. I like reading short stories and also kind of use generated content where people put out podcasts and YouTube videos and such.

So I enjoy listening to that kind of content in Hindi as well. I’m told reels etc. are also quite popular, but those are not very accessible. So I don’t really - I’m not able to engage with them in Hindi or otherwise.

The other kind of Hindi content which I really do listen to a lot is OTT content. whether it be sports commentary or you know, just entertainment content, a lot of that is in Hindi. So I do mean where the platforms are accessible, I do use them, use that.

Maari: Some of this sounds fun, entertainment and sports, but of course there are challenges when things are not accessible. Khansa, can you share with us some of the challenges you face when you use the internet? What is some of the barriers that stand in the way of a more inclusive online experience for you?

Khansa: I think one of the biggest ones is when things are inaccessible. And I know that’s a very broad statement, but again, like Urdu language tech is inaccessible, so that’s a lot of things that I can’t, books that I can’t read, poetry that I can’t read.

Then if we talk about other things, so for instance, a lot of my friends like play Ludo Star and whatever online. All these games, and these are activities that I want to engage in, but they are just not accessible for me. Another thing that is a challenge that I don’t enjoy is simply, when I go through Instagram and stuff and I come across all these inaccessible posts with lack of image descriptions and all of these things and I’m like this looks interesting. I want to engage, but I just can’t.

Maari: Absolutely. It’s clear from what you’re saying that there’s several areas of inaccess, but it also seems like challenges multiply when you’re a non-English speaker in general. Let’s ask Asme about his experience of using the internet in his native language of Tigrinya.

Asmelash: The internet is a very, very different experience for me when I use it in English and in my native language, say Tigrinya.

Let’s start with the easy and pervasive tools such as search engines. Search engines don’t work well, knowledge repositories don’t have enough content, and social media is also full of harmful content in these languages.

So, in the age of - now we’re in 2025, so everybody’s talking about chatbots. Chatbots are actually broken in my native language of Tigrinya and similar languages. So yeah, it’s quite a divide.

Maari: And you’re actively working to address some of these challenges, particularly accessing the internet in Tigrinya. And you emphasize keeping your efforts community centered. Could you share more about the work that you’re doing? What drives your approach and how has community shaped the solutions you’re building?

Asmelash: I really like this question. So, we have at Lesan AI, we have built a couple of products now. We started with the machine translation product to begin with. And now like recently we have developed an automatic speech recognition system. And the community aspect of this manifests for example, when we source our data, we reach out to our community members to source text.

So we prompt them with different questions and we gather text and then we also reach out to our community members to kind of record you know, the audio representation of these things.

And this is how we start kind of creating you know, the audio and text pairs. That’s usually used to create speech-to-text systems.

I will mention this. This is in contrast to let’s say what big tech companies would do, which is basically scrape the web for whatever audio and transcripts they could find, you know, put these things together and create a technology and serve to the different language communities.

So our community centeredness or, we’re rooted around our community means that we are working with our community members in all the life cycle of creating these language technologies we build.

Maari: I think we all really love that you’re motivated to make tech because of community. Could you tell us what community means to you?

Asmelash: Sure, in many ways when I talk about working for our community, first and foremost, I’m thinking all the speakers of like a given language.

I usually talk about my community to speakers for example as a community. So, um, that’s basically how I define it and how that then translates into action is basically all speakers are in my mind when I design and implement solutions for out, for example.

It manifests for example in the different dialects I should consider to try to be representative as representative as possible.

Maari: One focus Asme is clear about is the centering of community. Let me turn to Khansa here. I know you’re a strong advocate for putting people first in your work on making the internet more accessible. Could you share what that means to you? And how does this philosophy guide your approach to some of the research you’re facilitating in Pakistan.

Khansa: I think putting people first is at the core of all the work that I do, whether that’s my advocacy or activism or anything and my research really, because I think everything needs to be community driven and people driven.

You can never treat people as passive objects. You need to treat them as active participants and value them as such.

Value their contributions, value their time, value their involvement within the project and the process itself.

Because the process does change and you need to sort of be guided by your participants and let them take a lead on how they want the process to work out. It’s ultimately collective and it’s ultimately impact driven, the work that we do. So I think for me it’s just about making them an equal part of the process, whether that’s for compensation, whether that’s valuing their contributions in the same way, or any of those things really.

Maari: Through your research with ALT - accessibility, languages and technologies team, I know you’ve been seeing a wealth of insights coming in. Could you share some of the key learnings that have stood out to you?

Khansa: And that’s a very good question. I think my biggest one was when my participants really told me when I pushed them like how do you use Urdu tech if you don’t know English? And they were like, actually, that’s not possible. If you don’t know English, you don’t get to use your tech, you know?

And I thought that was just appalling in this day and age. And when they told me that like to use Urdu tech, you need to use English. That also was very counter intuitive for me. And I just I think these are some of the biggest things that we need to remedy or work on as a as a community.

But I think one of the more surprising findings and you know, pleasantly so was people’s focus on content writing, content creation, whether that’s videos or audio and just the ways that they’re like it’s supposed to be a very visual mode of engagement, so to say. And they’re engaging in it and they’re doing really well and they’re really enjoying it.

Maari: Thanks, Khansa for those really revealing insights. I think the very fact that you just cannot have a holistic experience of Urdu on the internet without English is eye opening.

Rahul, you’ve been actively working on auditing online spaces, tackling accessibility and inclusivity in Hindi on websites, apps and other platforms head-on. Could you share more about this work?

Rahul: So, as part of the study we are doing with Whose Knowledge?, we had held a focus group discussion in Delhi in March 2024. And that focus group discussion revealed the kinds of platforms, it gave us insight into the kinds of platforms people normally like to use in Hindi, people with visual disabilities.

One important one was, you know, platforms where you can create content and also listen to poetry, shayaris as they are called. And for that purpose, we chose this platform called Pratilipi. Then the Civil Services exams are a very coveted means of obtaining secure income and livelihood in India.

So there’s an app called Sanskriti IAS. So like that news platforms, social media. So we chose a little bit of everything. And then we did a two-fold kind of audit on them, which is our normal approach also.

One is a technical audit where you go into the menu of under the web content accessibility guidelines, which are the global benchmarks for a website to be accessible. What specifically does this website or app not have that it needs to have for it to be accessible, which is the technical side.

Because ultimately if you are approaching a developer and if you can give them very specific feedback, they can get that implemented much more quickly in terms of fixing things. The second part is user inputs because the whole research process is user driven and that is core to how we work at mission accessibility also. So you know, we chose five users who had participated in that discussion last year and asked them to audit two platforms each.

And tell us very detailed feedback on a task wise basis. Okay. If you go to a platform, what do you do first? You access the homepage, then you access and you log in. So task by task, how much can you access it and how much can you not? And then we derived some findings from that.

Maari: I think your work is so critical because it can give a deeper picture of what it looks like to use the internet in a language like Hindi as a person with disabilities. I’m wondering in the data, what are some of the insights or trends you’ve been noticing?

Rahul: So we found that on the technical accessibility audit front, a lot of platform did not do very well. The failure rate was anywhere between 50 to 70% for most platform, which is to say that if you apply these accessibility criteria, which every platform needs to comply with 50 to 70% of them were not met.

You know, so for example, things like not having headers whereby you can quickly jump from one section to another on a platform if you’re visually disabled or having image descriptions for pictures or audio descriptions for videos. These are the kinds of things that often got reported and unlabeled buttons also where you where the button isn’t properly labeled. So you don’t know what it is for as a screen reader user.

Uh on the user front, we found that some people struggled to use a few of the platforms that we had asked them to audit. But actually speaking the user feedback was much more positive in the sense that they said that they were able to access the platform.

Now one immediate thought that occurs to me is that for a lot of people even if the platform is not fully accessible, they’re able to figure out hacks and ways of using it. For example, for Pratilipi, one person mentioned that they’ve been using it for five, seven years, so they know how to get around, but a new user may not be able to. Which is not obviously a desirable state of affairs.

So one clear finding is the need for targeted interventions to make these platforms more accessible. For the developers of these platforms to be open to feedback and course correction from persons with disabilities. You know, one clear example of this I can give you is of Pratilipi itself that one user mentioned that they weren’t able to correctly label the mislabeled buttons until 2022.

But now Google through the Playstore has disabled that option. So you can’t relabel buttons in an app. So even if the person himself wants to make it accessible for himself, he is not able to do so. So that’s something which is not desirable. So these are a few kind of key things that come to mind immediately.

Maari: It looks like technically people are still working through the challenges, even as there are language related ones. What challenges have you come across regarding Hindi language?

So far as the Hindi level aspect language aspect is concerned, two or three key findings that emerged were number one, some people mentioned that the platforms because they accessed all these platforms using their Hindi interface.

So a lot of instructions which are otherwise given in English get transliterated into Hindi quite literally. for example, if it says send to another mobile number, to it will say contact which is not very clear in terms of what is sought to be conveyed by the prompt which the platform conveys.

Another person also mentioned for a platform called Jagran, which is a news platform that the Hindi is quite pure and may not be always intelligible to somebody who’s not that conversant in the language. This person mentioned that since he had been studied in a Hindi medium school till class eighth, therefore he was able to follow it.

Then the other aspect was that some platforms even on their Hindi interface, a lot of content still remains in English. So therefore, it’s not really that you know, they are giving a lot of Hindi language content. These were a few important findings that emerged.

Maari: From what Rahul is saying, it’s clear that there are lots of issues that need working on. Let’s shift our focus to solutions and the path forward. Khansa, based on the insights you’ve gathered, what are some concrete steps that can be taken to improve online experience for people with disabilities and particularly those who use Urdu?

Khansa: I think there are two answers to that. The first is that we need desperately need Urdu tech in the first place. So, Urdu speech to text, and of course text to speech as well.

So that’s the first thing we need like an accessible version of it that people can actually use for all the tasks that they want to perform. How can we use the findings? I think the first is of course developing community driven resources for the people in terms of trainings, in terms of making sure that people know about the things that do exist.

And then of course, I think it’s so, so, so important to continue advocacy and continue engaging with these designers and these web you know, these creators. So anything that is created is accessible from the get go.

And anything that isn’t is made accessible. And I think that requires a lot of advocacy but also legal reform and legal push back because ultimately you have to have a mechanism of enforcement to make these things happen to make people realize that this is an important priority.

Maari: Rahul, you’ve been tackling some of these challenges within the legal realm as Khansa has brought up. Could you share what your experience has been like?

Rahul: So, in India, we are fortunate in that we have a very strong and forward-looking disability rights law, which is the Rights of Persons with Disabilities Act, which is based on the UN Convention on the Rights of Persons with Disabilities. So to that extent, it embodies the best of what the globe has to offer in a large measure.

Of course, it does have defects. For example, the penalty amounts for violation of the act are not that high. And you know, the enforcement mechanism has some weak areas. But in terms of my specific interventions. So after I came back to India after my post graduate study, I worked for Justice DY Chandrachur who was a judge of the Supreme Court, later went on to become Chief Justice of India.

And I worked on some really landmark disability rights cases with him, especially on the principle of reasonable accommodation where people with disabilities need to be given additional support, such as say a scribe to write an exam. And since then I have been focused on you know, building my own organization which focuses on disability rights.

And one important success we have achieved recently is that more than 150 entities in India which are not accessible, they have been penalized for failing to make their platforms accessible and they’ve been asked to do so in a time bound manner. So through that also a lot of good has happened in the sense that companies and government authorities are now beginning to realize that accessibility is a non-negotiable.

It’s not just a nice to have, but it is a must have.

Maari: When you say entities, you mean?

Rahul: So these are the basically government and private apps and website developers. That is what I mean, the law uses the term establishment. So government and private establishments that run digital platforms which are not accessible.

Maari: And how do you plan to take the learnings from your audits forward?

Rahul: So, we normally follow a five step process at Mission Accessibility, which which we would be keen to try and do here also with respect to these findings. You know, the first couple of steps involves trying to liaise with platforms and to tell them that look, these are some accessibility issues which your platform has. So if you could kindly look into this, take steps because not only is it a legal mandate that you must do it, and especially now in light of recent developments, but also it makes good commercial sense because you can expand your footprint and morally also it is the right thing to do.

So the first effort is always to engage and that’s not to say that they need to work with us or pay us to fix their platform. There are 100 other providers in the market, but they need to do it. Like it is not something which they can just ignore. That’s one part. The second part is, if that doesn’t work, then you try to use social media in a constructive way and to bring this to people’s attention that look these platforms are not accessible. So, if 10 people tag them, sometimes it can make a difference because companies are very sensitive about and government authorities also some of them are sensitive about their public image.

And if that also does not work, then obviously the legal advocacy route remains open where you try to liaise with the disability regulators and with the courts and you say that look, ultimately, this is a right that I have to access the platform. If I’m not able to access it, I won’t be able to read the news, I won’t be able to buy groceries, and it is a violation of my rights. And the platforms are then compelled to really take steps to make themselves accessible to an accessibility audit or otherwise.

Maari: That’s really very interesting to note the legal approach here, Rahul. Thank you for sharing your process. And finally, let’s go back to the context of Africa. My question to Asme is this. You’ve described yourself in the past as part of a pan-African AI movement. What are some of your visions for AI or language technologies in Africa?

Asmelash: Great. Well, I hope, it’s something that should kind of reflect our realities, the different communities and in Africa, we have upwards of 2,000 languages. And you can imagine at least I hope we have different builders distributed across the continent serving their kind of like language technologies, creating the best they can create.

And in a way, there is a small initiative that we have started called the Huniki Federation that aspires towards that, which is that different startups and organizations working on African languages, independent organizations, but they can come together sort of in a federation to to kind of serve African languages. That speaks to the sort of vision or what should we call it, like aspiration I have. Again, that would be contrasted to monopolies, like a company aspiring to serve it all. Yeah, so at least that’s where I hope we’re heading and that that’s what I wish for the African community.

Maari: I think the idea of a federation and a collective is so wonderful. In fact, a key message there is the fact that you’re trying to do something different from big tech, creating space for diversity in the interest of language and language resilience. Thank you for sharing that. I would love to know what are some of the organizations that are part of this alliance?

Asmelash: So, there is Lesan, the Ghana NLP, Lelapa AI and DAIR, which is these are the founding kind of members.

Maari: We look forward to learning more about these organizations, their work, and even the federation itself. I think we’ve learned so much from you, Asme, Khansa, and Rahul. Today on the various approaches to technology making in the service of people.

All right folks, that’s a wrap. As we close this episode, I want to thank our incredible panelists for sharing their valuable insights and experiences. From discussing the challenges of navigating the internet with disabilities to exploring community-driven solutions for language technologies, this has been a truly enlightening conversation.

Khansa: Thank you.

Asmelash: Awesome.

Rahul: Thank you, Maari. I really enjoyed the conversation.

Maari: This episode was meant to be a small taste of the panel we had at RightsCon this year and something that can promise the larger work and conversations we are working on in the coming months. We ask you to continue to tune into this space and join us on the journey to understand disability, community, language, and technology from a justice focus. Stay connected, stay inspired until next time.