The Good Robot

Louise Hickman on the Politics of Captions

April 25, 2023 Eleanor Drage
The Good Robot
Louise Hickman on the Politics of Captions
Show Notes Transcript

In this episode we chat to Louise Hickman, an activist and scholar based at the Minderoo Centre for Technology and Democracy at the University of Cambridge. Louise talks to us about stenography, the process of  transcribing speech into shorthand. You may be familiar with this from having seen court reporters write a transcription of a tribunal or case, but many stenographers also do crucial access work to create live captions of someone speaking. Stenographers create their own online dictionaries and then access words really quickly using keyboard shortcuts. We explore the political decision making process of captioning and why this matters. 


Kerry:

,I'm Dr. Kerry MacInerney from the University of Cambridge, Dr. Eleanor Drage and I are the hosts of The Good Robot Podcast. Join us as we ask the experts: what is good technology, is it even possible, and how can feminism help us work towards it? If you want to learn more about today's topic, head over to our website, www.thegoodrobot.co.uk, where we've got a full transcript of the episode and a especially curated reading list by every guest. We love hearing from listeners, so feel free to tweet or email us and we would also so appreciate it if you left us a review on the podcast app. Until then, sit back, relax, and enjoy the episode.

Eleanor:

In this episode, we chat to Louise Hickman, an activist and scholar based at the Minderoo Center for Technology and Democracy at the University of Cambridge Louise talks to us about stenography the process of transcribing speech into shorthand. Now, you might be familiar with this from having seen court reporters write transcriptions of tribunals or cases, but many stenographers also do crucial access work to create live captions of someone speaking. Stenographers create their own online dictionaries, and then access words really quickly using keyboard shortcuts. Having written all the transcriptions for this podcast for you guys, I'm fascinated by the process of captioning. So it was great for me to hear about Louise's work. I hope you enjoy the show.

Kerry:

Great. Well, thank you so much for being here with us. Just to kick us off, could you tell us a little bit about who you are, what you do, and what's brought you to thinking about feminism, disability, and technology?

Louise:

Thank you for having me., it's really great to be here. I was listening through the podcast last night and it was really exciting to see other speakers and guests that you've had on. So yeah, I think about disability and technology or feminist labor should back up a little bit. And. Start at the beginning with where I did my PhD. I did my PhD at UC San Diego, in the Communications Department where I did my dissertation on the processes of realtime writing which you might be more familiar through captioning of like live media content or social media content. And the reason I came to think about or think with realtime writing is I had many close relationships with stenographers in the classroom. It was noting here that I'm actually deaf and disabled. And so I used real time captioning first as a student and then I later used captioning had a Professor or an Adjunct teaching classes in the communication department. So through these relationships, I began to study the history's plural of captioning. So this took me in multiple directions. I went into the archive Gallagher University in Washington, dc which is a deaf university where I look at the kinda archive around Actioning as a kinda in through historical documents where in this instance I found letters dating back to the 1950s where there was exchanges between the university and the Library of Congress, and they were really starting to think about the possibility as using actioning as their kind of pedagogical tool for deaf people to view films that kind of was organized through lending libraries and all kind of quite patriotic. But yeah, we'll skip across that. But like most of my work, I would say it's quite ethnographic. So we can talk more about this later. Well, I think it's worth staying with the aspect of realtime writing for the moment. Like, you know, thinking about realtime writing or on demand work and thinking this through the lens of like disability studies or critical access studies as a way to really think about what did this mean for on demand labor in this context. Which is quite interesting because quite often when people hear about stenographers, they think about I assume courtroom workers, which is great, which is a part of this kind of history. But court workers quite often when they're transcribing the transcript of the spoken speech, they have the capacity to kind of correct the text after the fact, right? It's a kind of a lot of like tidying and maintenance work going on behind the scenes. But when we're thinking about access, which is a huge part of my work, access in this sense, real time writing is slightly more mechanical and I, I think we can expand on that later. And then what I mean by that is that there is this capacity where workers who are transcribing for deaf students or live like news coverage they're having to code real spoken speech in real time. So that's an entirely different work dynamic than correcting a transcript after the fact. So if we're thinking about real time labor, this has moved into other areas where I thought it doing work. As part of my first postdoc I was in the feminist labor lab directed by where we spent some time thinking about how to have fair wages for taxi drivers in San Diego. And as part of that work my addition to that project, we really grappled with the question of how taxi drivers have wheelchair accessible vehicles because they're much more expensive, the insurance is higher, the maintenance cost is higher, so there's no incentive for on demand workers to have these vehicle. So in, in the last few weeks, we've been doing a lot of organizing around this, where we've managed to kind of really pull a critical mass together to really kind of push this project forward. I could go on about that a while, but that is another dimension of real-time work, how I understand it. Since California, I've returned to the uk where I currently I am based at the Mindero Center for Technology and Democracy as research directive on work. And in this role I'm using a lot of the projects I've been working with, I'm also working in collaboration with Sarah Fox at Carnegie Mellon and Alexa Hagerty, of the University of Cambridge. And we are really thinking about the dimensions of access. We currently in a wider network of critical disability study scholars. We're working on a report at the moment, which had a quite a cheeky title laundry day from community alternative to access washing. So that is the current project that I am working on, but I hope that you can see that there is a thread there continually coming back to the conditions of realtime work.

Eleanor:

Amazing. Thank you. Well, you've given us a phenomenal introduction into all your incredible projects, can you tell us how they shape the way that you are thinking about what good technology is, whether it's even possible, and how feminism and disability studies can help us get there.

Louise:

Yeah. You know, I, I was expecting this question and I'm kind of dreading it after listening to the other podcasts, and it's really funny because I get this question a lot, but its slightly different, which is what is good captioning. Which again, it's like quite frustrating. Maybe I'm a rebel and I want to kind of move to the side of embracing human errors and like over automated ones. Maybe I wanna like embrace the politics of slowness rather than speed and efficiency. They're not always the mean to do this. I mean, I feel. Those kind of binary that is offered there, kind of set up a kinda conflict there. Attention you know, the access that is driven by the politic of social justice, but then there is the disability legislation that is that actually make this access happen. So the kind of distance between social justice and legislation is one that's quite tricky to navigate, right? And so, and of course, you know, this really changes from places to place quite often, but disabled newsers. So I guess if gets into the really messy area when a disabled person, for example, who might request. Captioning or access to a building or this is broadly understood at a kind of reasonable adjustment in the UK and a form of accommodation in the US. And so those kinda like parameters, when you're really starting to think about these kind of adjustments accommodations, it's like really moving away from the question of what good technology. And so again, it's like how can we bring them together? Like how do we bring them together in a way that we are pointing toward care that perhaps the captioner puts into their work. And I should say in this instance that most of Graphers I've worked. Really don't think of themselves as doing care work, which obviously kind of contradict my own position in thinking through what do they mean to do good captions, right? And so I guess we really are kind of thinking about the collective experience of what captioning is, right? And what it should be. But also we have to question like the real time conditions that stenographers are put under to produce the captions in the first place. Answering this question of what good technology is is really interesting because I approach my work through the lens of intimacy which you can think about this perhaps as a kind of partial account of what access means for disabled users. Access intimacy is a term coined by a disability justice activist, Mia Mingus from the Bay Area, where they really explore the contours of access intimacy: what did it mean for your colleagues or your allies to understand the conditions of what access looks like. So having that knowledge, and I can talk a bit more about it later, but Amy Hamraie talks about about access knowledge. So if I would like to offer kind of neater kind of definition of good technology, I'm always curious about the partial accounts. But also the mediation, like the mediation that we go through to account those technologies in the first place. I can give you a really quick example, which I find really fascinating, and it's like a moment of disruption is when at the beginning of a Zoom meeting, Sometimes you have to request captions, and some people are quite frustrated about it, but I actually think it's an interesting opportunity for people to check in and mediate what access looks like. Do we need captions? And then do, if we open up captions on Zoom, then the transcript is available for users in that meeting, do we want to share that information in the conversation? Private. So they opened up the like really kind of ethical questions about information sharing. Yeah, it's been a

Eleanor:

really interesting experience for Carrie and I doing this podcast as well, looking into what kind of transcripts we are after and what accessibility means to us, and we've been so inspired by your work over the years. I loved seeing what actually happens with stenography in the presentation that you gave at a conference we were at together a couple of weeks ago called AI An Keys in Berlin, and it's amazing to watch Stenographers in Action. It's an incredible thing. They're just so talented. It's amazing work. So I really want to know a little bit more about stenography and how it relates to real-time economies of delivery and instantaneous access. Right? And how access is always approximated. It's never fully accessed. Mm-hmm. And I was also wondering about the, the expense of this kind of access because, you know, stenographers are not cheap and they're socially rationed, and it's also a gendered form of labor. So you bring all these different things into the debate when you talk about stenography.

Louise:

Yeah, I, AI an anonymity was like a wonderful conference and I was so pleased when you asked the question about transcription. It was a really great observation for me to work with. Just to for listeners who are listening in today as part of this talk, I presented a film. Was commissioned by Luck Film Collective in London in 2022. I made this. Artist based in New York, Shannon Finnegan, and we worked together using Zoom, actually, which is kind of an interesting tool because it gave us this window into sharing our screens, which as the film showed kind of glimpses into the world of what a Grapher does when they're building. Job dictionaries. So in this film, I asked audience members to get up from their chairs and move closer to the kind of small monitor near the stages. And I did this in a way as an invite to think about our own embodiment when we're viewing films and like thinking about our embodied practices of access as well, like thinking about the people around us, et cetera. A part of this film, which which is quite central, is my relationship with a stenographer who I've worked with for a long time, and you kind of see the really nice banter between us and Jennifer is trying to pull up certain words that are in her dictionary. One moment where Jennifer trying to pull. Phone. And there's like this moment where it doesn't quite work. And then who finds it by Newton in asterisk. And so this is like this moment into an intimate moment of what it means again to do realtime work. You know, what did it mean to be under pressure to bring up, retrieve these words? Another aspect of this film, which I have never actually seen until this film is. Dictionary that the stenographer knew when working with myself. So there are encounters in there of readings graduate students, and I, I often tell this kind of Joke, which might be getting a bit tired some at this point if people have heard me speak before. But there was this great moment where I was a younger grad student and the readings was Mac Faber. But the dictionary called, it was relatively new, came up with dark vapor. So this is that kinda, I call it a phonetic mismatch or mishaps. And so it's really nice moment where you really are thinking about the social conditions of how real time speech is created. It's kinda like moving out a little bit of the intimacy. There's so much going on in this film and it's like, well, one of the central focus of this talk is thinking about how the shorthand, we actually come to mean multiple things. So, Access academic accreditations. They share the same shorthand brief on a keyboard, which is really kind of funny. If you think about the relationship access, academic and Accion, they're widely different. But what. stenographers rely on in this context is that they have to retrieve the right one. And so quite often they have to develop dictionary that have different social contexts. And so this interesting we, I think of how we think about ai, you know and how in the instances, you know, my kind. Joke about Matt Va is you can see the kind of like proclamation of the phonetic, like dark vapor, Matt Va. There's a kind of familiarity there. There's a kind of link there through the way it's sound. So that is what I would consider the kind of human error. So, That approximation of human errors is actually quite useful to thinking about captions. And so in that sense, cloud captioning is now using predictive text. So like in the sense it's gathering a huge amount of data to kind of predict the structure of speech. Which is hugely problematic because it doesn't account for local knowledge on the ground. It doesn't account for names or speakers and social context accents. And so in that sense, she really started to really understand the value. Human mistake and how they can be often corrected as well. You know, might be slowing down the generation of the tech or the realtime tech that's appearing on screen. But there is a kind of fence there that. The mistake makes sense. They have a kind of social context, if that makes sense. I, I'm might a real nerd when it comes to kind of thinking about realtime capturing and like, we haven't even stretched the surface of the transcription work you guys do yourself. I'm sure you've. The dilemmas where you've often thought about whether to remove the kind of stream of consciousness or somebody ums and, and whether that should be included. I've had interesting debate with my undergrads at various points where they've really kind of range, again, inclusions of the kind of questions of you know, I never

Eleanor:

know because I think, you know, when people are reading the script, I think some people are going to want to, to read something that reads really easily. But then on the other hand, if you read with the kind of ums and you know, maybe you also read for, for nervousness, you can read for emotion. You can see whether people are a bit awkward or trying to hedge their. Thoughts a little bit by saying kind of, or

Louise:

sort of,

Eleanor:

but it's, it's a kind of fascinating insight into the kinds of decisions that are being made when transcripts are produced.

Louise:

You know, it's really interesting if there is a kind of screenshot in the film where it says reading is not the same as listening. And that was like this really great moment that came out of the film which was like, I was able to identify that and show that in practice because I've had colleague I have a colleague of mine, Yona Glassman, who had also spent time looking at captioning and found that Captioners quite often clean up the spoken speech to make it readable. So there's an interesting relationship there because we're going in multiple directions. This is a text that's being read rather than listened to. But then we also have this AI on the horizon as well that really undermining this work that we do to deliver this text. And also a fornot, all this, like all the big lectures like and things like that, they all had a stenographer working, you know, transcribing those talks. So yeah, there's a huge history there. That's

Kerry:

so fascinating, and I have to confess, I don't actually end up making many of these political decisions because Eleanor very kindly does all the transcribing for the Good robot. So anyone who goes to our website, which is www. The good robot.co.uk and sees those lovely transcripts. That's all thanks to Eleanor. And one of the reasons why I don't do that is I'm dyslexic. I find it very, very hard to kind of translate word to sound and sound word and vice versa. And so it's interesting to kind of hear like, oh, actually that's a really conscious process in that Eleanor ghost in which he's trying to sort of best capture what our episodes sound like, but also produce something that's like really useful for people to be able to go back and refer to. I just wanted to ask you to kind of close one of the things which I think is really interesting about probably being in this space is the extent to which kind live experience and sort of autobiographical encounters shapes your academic work. And of course, sort of vice versa. And we were just recording yesterday with the wonderful Laura Fallo, who was sharing with us a series of vignettes. Stories or creative non-fiction that she written about her own experiences with technology. And so we were wondering how do you think we should and can like grapple with like lived experiences in our academic work while also resisting the kinds of commercialization I guess, of this knowledge, which is can also be really common.

Louise:

I've been thinking a lot about this and I expect this came up a lot in law. Conversation as well. It quite often, You know, it's a really tricky position to occupy the space as a somebody who's visibly disabled and how that visibility is incorporated into my work, or also bracketed off too. I, I talked about this earlier where I kind of referenced this earlier. Amy talked about access knowledge, and I think Nikki really fundamental in the way that I think about the work that I do that is epigraphic is like, and I think keep a struggle because there is this epigraphic quality, which I can draw attention to the. Everyday encountered that perhaps other people might. But to do the work that I do as an academic researcher, I'm always looking for ways to decenter myself, which seems quite paradoxical. You know, it's like, on the one hand I am like using my lived experience as a way to kind of think about these issues, but on the other hand, I'm also backing away and being like, okay, well, These issues apply to other concepts, and I'm always looking for way to open up the conversation for others to enter into and say, oh yeah, I have never thought of it in that way. And so it's a really pretty position to be in. And I'm not quite fully fleshed out that kinda position. And I, perhaps I make it more complicated cause I'm often thinking with others, you know, like captioners or personal assistance. So in that sense there's always a collective we, and so that's the way I think in that, again, de-centered my position as a researcher because I'm always thinking. The mediation of how we kind of conduct this work on co-producing knowledge, you know? So yeah, I think we'll do really tricky one. Thank you so much

Kerry:

and it's really helpful to hear you speak about that as well. It is something I really appreciated. She was a piece by Eugene Zuki who writes about being a multiracial woman in academia and she talks about sort of, Trying to bring her experience, not as someone who just happens to be multiracial or a woman of color, but saying, you know, as a mixed race person, she's really practiced in thinking about a lot of these issues, and it's a skill and it's a muscle that she's flexed and built up, which is why she's able to kind of do this kind of work and bring about these kinds of changes in these institutions. And so she's calling for that to be recognized as a skillset rather than saying, oh, well that's just something you are. Good at because you happen to fit in these particular identity categories. And yeah, that was just something I found really, you know, useful. So for our lovely listeners, we will put that on the reading list for today's episode and also happy to share. But yes, I mean largely we just wanna say thank you so much. It's been such a pleasure to be able to chat to you. It's certainly brightened what is otherwise a very gloomy and gray afternoon here in the uk. So thank you so much.

Louise:

Thank you for having me. This episode was made possible thanks to our generous funder, Christina Gold. It was written and produced

Eleanor:

by Dr. Ella Dre and Dr. Kerry McElrath, and

Louise:

edited by Laura's and.