Navigating a Digital Humanities PhD: Some Reflections

This post is inspired by a discussion held at the m3d conference as part of Biblissim-ia 2025 Cluster 3, where I shared feedback on my experience as a Digital Humanities PhD student. Some of these thoughts were sparked by questions asked during the round table, and I thought it could be somewhat useful for future students and supervisors to put them down in writing.

Disclaimer I: Everything I share here comes purely from my own perspective and personal experience as a third-year PhD student, and is not intended to be aphoristic in nature. I am a traditionally trained philologist/classicist turned Digital Palaeographer via a Digital Humanities Master’s, and now am co-advised by both a Palaeographer and a Computer Vision professor. I am also a (rather well-integrated) immigrant in France, having done most of my higher education away from home. I am also an inherently optimistic person, which might affect how I read situations.

Disclaimer II: The relevance of this advice will depend greatly on the technical level required by the thesis project. The DH field has a wide spectrum of computational profiles, where prerequisites and expectations on the technical side vary. Some projects adapt existing tools to traditional questions, while others develop sur mesure tools to tackle new questions, and they are all important for the scholarship. My experience is in an experimental, hybrid project in computer vision for palaeographical analysis with high technical requirements, so these thoughts and suggestions may apply better to those on a similar path.

I will try to refine and update this post as I approach the PhD defense, but here’s a first draft.

Terminology note: In this post, I refer to “the Humanities” to describe Liberal Arts more broadly, but due to my background, I’m biased toward historical and primarily philological disciplines (such as classical studies and palaeography). Similarly, I use “CS” as shorthand for computer science fields, though in my case, this mainly refers to computer vision and AI.

1. Technical Skills, Training, Research Mindset

Manage expectations

Coming from a Humanities background, many of us implicitly learn (and feel) that knowledge is cumulative—and thus linear: you read more, you understand more, you filter, you synthesise, you produce scholarship. In CS, this process is much less linear, though cumulative nonetheless. You often don’t know how to feel progress, because you may spend weeks troubleshooting something very basic or confronting problems no one fully understands, especially when you’re new to the field.

It took me a long time—and a lot of conversations with people around me—to realise that in CS, everyone is constantly learning and failing, including supervisors. A lot of progress is trial and error. If I could go back to year one, I would tell myself what others kept telling me but I wouldn’t believe: that not knowing or retaining everything is normal, and failing is often the path to understanding. Switching from a purely Humanities approach to Digital Humanities means you are constantly in a learning space. This space is precious and should be your new comfort zone, guilt-free, at least for the first 1–2 years.

Build foundational knowledge

Math and AI are exciting and scary at the same time, and most Humanities-background students feel lost and mostly like imposters. One of the most helpful things in my PhD was my advisor Mathieu’s one-on-one classes (after following - for the third time - his Introduction to Deep Learning course) on basic notations and operations for deep learning networks. I appreciated that he tried not to feed my imposter syndrome, answered my seemingly “stupid” questions, and trusted that I was capable of eventually understanding, even though math was not a language I had learned to reason with.

From my side, I spent countless hours watching YouTube videos and tutorials about networks, optimisation, image manipulation. The understanding came incrementally: first 10%, then 30%, then 60%. I may never reach 100%, but 70–80% is already functional for what I’m trying to achieve. Progressively, you realise you’re not supposed to know everything, but it is essential to understand in depth the tools you’re working with for your project—and then the learning curve becomes less steep.

Well-roundedness

If possible, invest time in external courses and workshops related to your project. Even if you don’t use every skill immediately, you’ll build the confidence to explore new tools later. In experimental projects, you often have to manage every part of the pipeline: data creation, structuring, training, interpreting. For me, that meant becoming familiar and autonomous with everything from HTR to XML-TEI to deep learning and visualisation.

Being autodidact, taking initiative, and having (not all time consuming) side projects can make the learning curve more manageable. And being open to seemingly unrelated knowledge will only enrich your thinking.

Identity crisis and liberty

Be prepared that you may never fully feel like a “Humanities researcher” again (and this is something to grieve but also to celebrate), nor fully like a computer scientist. Depending on the phase of the thesis you’re in, it can be various percentages of both. This hybrid identity can feel unsettling, but it doesn’t mean you’re less of a Humanities researcher—you just have extra features that make your identity and approach different from what was expected for years. And your Humanities background is never erased—it’s always present, shaping the questions you ask and how you approach answering them.

Finally, a DH thesis project can be (to an extent) what you make it. There are ways to scale up the technical aspects or focus more on results and interpretation. In that sense, there is a healthy margin of movement to orient your project where it feels comfortable doing high-quality scientific work.

A PhD is in itself a journey of scientific self-discovery, where everything you learn is an investment in your future (research and beyond) self.

2. Supervision, Collaboration, and Community

Supervision dynamics

In the Humanities, your supervisor is often THE authority for a big part of your PhD. In CS, supervisors are often more like advisors than supervisors. Many times, they won’t have the answer to a technical question—and that’s okay. It requires a shift of mindset: you must feel comfortable taking initiative, documenting and clearly explaining your rationale and experiments as well as failures (especially failures!), and it’s about building a strategy together rather than expecting pre-defined solutions.

Building community and asking for help

I’m very pro integrating Humanities students into CS labs full-time. It’s intimidating at first, but it’s the most effective way to learn how the field really works. No one works alone in CS, and not everything is learned only through courses and lectures. Collaboration is essential: sometimes a colleague will spot a bug you’ve been stuck on for days or show you a new approach. I used to believe I had to know and do everything by myself to feel legitimate, but learning to ask for help when you need it is essential. For me, my amazing computer vision colleague Yannis taught me (how to learn to develop) many essential skills: debugging code, refactoring and using modules, using Colab with GitHub, visualisation tricks in matplotlib, making a webpage — and these are only some I remeber out of many. I only hope going on and on about marginalia and networks of knowledge, allographs and medieval scribes has enriched his way of thinking as well.

Interdisciplinary partnerships

Partnering Humanities and CS PhD students for joint projects can be extremely effective. But there is a delicate balance: both parties need tasks where their expertise is equally crucial—not where one side’s knowledge is merely in service of the other. The best collaborations, for me illustrated through our learnable-handwriter project, are those where everyone feels their contribution is respected and necessary for a quality result, and where contributions are clear and help advance respective PhD projects. Such an environment encourages both parties to understand each other’s work in depth and broaden their perspectives on their respective objects of study (networks, letterforms, texts, and so on). Ultimately, it is also a means of learning to communicate a research question in terms that are meaningful to different disciplines.

Networking

It may sound cliché, but networking doesn’t have to be formal or awkward. At the intersection of Humanities and CS, I have come to realise that we are all figuring things out in various degrees. Maybe an "extrovert's" take but : talk to people at conferences, reach out to researchers whose work you admire, share your ideas. You might end up making friends and collaborators along the way.

3. Well-being and Working Conditions

Scarcity mentality vs. abundance mentality

In the Humanities, we often internalise a scarcity mindset: be grateful for any resources or attention, feel lucky to even survive in academia. But (as in all fields but focusing on under-ressourced Humanities PhDs) good research requires real investment: a stable salary, regular meetings with supervisors, a well-organised lab, access to computing resources, funding for conferences and workshops. Having this material support helps you shift from survival mode to abundance mode: believing that you deserve resources and have options (all without feeling guilty for not being productive all the time, which takes me to my next point).

Mental health

It might look like everyone is an overachiever and can handle anything because we’re young and doing what we love, but doing a PhD is very mentally taxing, especially when you’re in a precarious position. There is a shared notion that to succeed we need to struggle, but hard work doesn't equate struggle. If you're struggling, speak up. If it’s not to your supervisors because you don’t have that type of relationship, then to your peers and family. Most of us share feelings of being imposters, of homesickness, of feeling alone, of not feeling capable enough. If these feelings are overwhelming, talk to a mental healthcare specialist. These are the best years of our lives: we get to develop our critical thinking in difficult times, do research, and invest in ourselves to become better humans through science, and no one/no situation should deprive us of this. (An advice from my supervisors that took time for me to follow): listen to your body, take vacations, no one will point the finger at you for taking time off to recharge. You can do better research with a clear mind (and definitely debug code better!).

4. The Future

Going from a Humanities to a Digital Humanities profile will already be valued both in academic and non-academic environments, and this can give you a sense of safety. I went from thinking the unimaginable task of doing a PhD in Latin (a field I adored) after my Master’s—which would mean going through agrégation, fighting for a contrat doctoral, working while finishing my thesis, struggling for a few post-doc or ATER positions—to feeling like I have many more opportunities in academia. Most importantly, this doesn’t make me feel any less of a Latinist, but rather equipped with tools to serve my field and community better. Even though I had never imagined myself not doing research, knowing that this possibility exists (data architecture, consulting, research engineering) gives me some solace.

5. Final Thoughts

Ultimately, I believe that for a Humanities student to thrive and truly integrate into a digital environment—and to go beyond merely applying pre-existing frameworks or automating tasks—everyone involved must invest time, energy, and, above all, goodwill. The student needs to be curious, disciplined, humble, and genuinely passionate about their project, especially if it’s experimental. This passion and excitement are what carry you through the inevitable frustrations. Advisors, in turn, need to dedicate time to guiding, explaining, listening, and engaging in dialogue with the student. It’s imperative that they remain present and actively support the process.

I’m grateful to have had advisors and colleagues who trusted me, challenged me, and supported me while I was/am growing as a researcher. A big thank you to my advisors Dominique Stutzmann and Mathieu Aubry, my colleague-in-crime Yannis Siglidis, my ENC professors Thibault Clérice and Peter Stokes, and the IMAGINE lab team—Sonat Baltaci, Raphael Baëna, Syrine Kalleli, Ségolène Albouy, López Rauhut, Romain Loiseau, Tom Monier, Julien Gaubil (it’s an ever-growing list, really).

If you want to discuss or add anything to this list, don’t hesitate to reach out: matenia03[@]gmail[dot]com. And if you’re a new DH PhD student, best of luck!

Last updated: July 2025

Design and source code from Jon Barron's website