(Note: this turned out to be a long intro, so, like, don’t feel like you have to read it. It’s a free internet.)
Hello! My name is Nina Markl and I am an academic researching and teaching at the intersection of language, technology and power. As it says on the title, this website is an “unordered” collection of my thoughts on language, technology and power. In his excellent book “The Mechanic and the Luddite: A ruthless criticism of technology and capitalism”, Jathan Sadowski discusses the figure of the “mechanic”: a person who tinkers, who takes things apart and puts them back together, who has agency and control in their engagement with technology, and important knowledge about its workings. This is not so much contrasted than complemented by the figure of the “luddite”, who is deeply concerned with the way power is wielded through technology – and often shifted ever further from workers to bosses.
Unlike many other people working on computing (broadly construed), I was never particularly interested in technology as a child. I don’t have stories of tinkering with computers, radios, or even bikes (which I love now). I was only ever interested in people and their relationships and spent most of my time reading novels, and as I grew older, feminist and leftist political writing. I started out studying linguistics as an undergraduate, and realised quickly that what I was actually interested in wasn’t language as such (though there is real joy to be found in the study of linguistic systems) but rather what people do with language. I was immediately drawn to the study of language change (how did we end up here? where are we going?) and language variation (why do different people talk so differently? why do we see such seemingly endless creativity?). Fundamentally, many of these central questions in this area of “sociolinguistics” come back to: how is power constructed through language? In my very first sociolinguistics lecture, in my first year of undergraduate study, the wonderful Lauren Hall-Lew -- who would, some years later become my PhD supervisor, mentor and pal – said (paraphrasing slightly): “statements about language are never just about language. statements about language are always statements about people”. This particular foundational insight has stayed with me and guided my research, study and teaching ever since. As much as I love language (and linguists!), the reason I study it is the way it exposes and, in fact, upholds social structures and relationships that are otherwise difficult to see. Language mediates power relations at all levels – from interpersonal dynamics between friends, partners, colleagues, family members, to relations between citizens and their governments. How we use language is central to how we express who we are and where we sit in the social hierarchy – and it is central to how we are perceived by others, where we are and are not allowed to enter (literally and metaphorically), who we are and are not allowed to challenge.
It was a genuine coincidence I ended up working on “language technologies” during my PhD. I wanted to pursue a PhD, I wanted to stay in the same city where my partner was doing his PhD – and as luck would have it, a new PhD programme on natural language processing was soliciting applications the same week my undergraduate thesis was due. Having gotten the PhD and gainful employment, I can now admit that I didn’t even know what “natural language processing” was before I sat down to write an application which proposed, essentially, a computational approach to the work I was already doing (studying language variation in Edinburgh). I remember that many of the seminars I attended that first semester of my PhD in 2019, were about variations of BERT (the OG LLM). I didn’t understand the technical details, but more importantly I often found myself wondering why certain tools were developed and implemented. At the time, I thought that I was simply missing important context and didn’t have enough technical knowledge to be allowed an opinion on the practical and scientific utility of different models and research programmes. It wasn’t until years later that I realised that the central question of why really is left as an exercise to the reader in much of the computer science literature. The other thing that stood out to me immediately, was a wildly oversimplified notion of “language” that generally didn’t consider how language is actually used, therefore failing miserable when applied by real people to real conversations and texts. Again, I assumed I was missing a trick. Surely these things are only meant as simplified models and we are all aware that language is too complex to reduce to a dictionary. Surely we are all aware that language is a fundamentally social activity that is about so much more than information (which, in many ways, computers are already extremely good at in storing, exchanging and manipulating). In some ways, things have changed as language technologies have become more popular as both research objects and products and the conversation has become significantly more nuanced. In others, we still see wild claims equating coherent production of plausible sentences as one step before production of a (terrifying or wonderful, take your pick) God. All of this is to say that what I eventually realised is that I still didn’t find the technology as such all that interesting (sorry!), but I was fascinated by what the design and deployment of language technologies does socially and politically. This led me first to research on “algorithmic bias” and then to a broader study of the political economy, history and impact of computing.
As Jathan Sadowski puts it in “The Mechanic and the Luddite”, “technologies articulate broader dynamics—political, economic, social, cultural, moral—and give them material form in the world.” This succinct statement (unintentionally, I assume) echoes one of my favourite statements about language by Susan Gal and Judith T. Irvine (2000): “[Language users’] ideologies about language locate linguistic phenomena as part of, and evidence for, what they believe to be systematic behavioral, aesthetic, affective, and moral contrasts among the social groups indexed.” I still have to think through more deeply how and why these two framings fit together, but I think it’s something like this: there are broader material and ideological structures which force people into different social positions within a hierarchy. The argument that Irvine and Gal make is that when we see differences in other people’s linguistic behaviour (using “ain’t” instead of “is not”, using a minoritised language on public transport), we connect these differences (these aberrations from the norm) to these social positions and use them to justify and explain why different people are where they are (and why we are where we are). Crucially, we already believe in these “systematic behavioral, aesthetic, affective, and moral contrasts” – statements about language are never about language. Some ways of talking are evidence of education, others of lack thereof, both an outcome of and justification for inequality. In the same way that language expresses or voices the these broader material and ideological structures, technologies, as Sadowksi puts it “give them material form”.
All of this is a very long-winded way to explain why I’m interested in the mechanics of both technology and language. And I haven’t even touched on my interest on workers who act as such mechanics.