It has been two weeks since I started my research at CREATE and still the irony strikes me sometimes that I, of all people, am specialising in digital text analysis. As a bibliophile and – among friends – an infamous luddite, I kept a safe distance from anything that was associated with ‘humanities computing’ or digital humanities for years.
It is even more ironic that in my project I will test topic modeling technologies on the dissemination of pre-Enlightenment thought, Spinoza in particular. Spinoza operated in a socially and politically instable situation (the Dutch Republic between 1650-1677) and was fully aware of the political consequences of his revolutionary words. He lived up to his motto ‘Caute’ – be cautious – by expressing his radical mind in an utmost ambiguous and avoiding manner. Being a child of the humanistic tradition he questioned the reliability of words and recognised the power of rhetoric. Due to aggressive censorship, Spinozist sympathies had to be formulated with a likewise caution and linguistic virtuosity. Hence, the irony can be found in the transformation of such wisely chosen words into indiscriminate most frequent words lists (which is what topic models essentially are). Isn’t that a philological sin? Because of its philosophical and linguistic sophistication, one could easily argue that the Radical Enlightenment is the worst test case for modeling topics in early modern texts.
However, one could equally argue – as I will do – that Spinoza is the best test case for this project’s methodological evaluation. Let’s just start with the hardest part: if we can enable our technologies to see but a glimpse of a complex political-intellectual movement like the Radical Enlightenment, they will arguably have less difficulties understanding other early modern texts. And on an abstract level: the opportunities for intellectual, cultural and literary history are endless if computer intelligence can identify and understand the intellectual structures in digital corpora. It is highly ambitious, but the possibility that we might bring this large quest one small step further, fascinates me.
I lost my scepticism of the digital humanities when I discovered that the field itself has cultivated and internalised this very attitude. During my research master in Dutch literature and while attending several courses in Digital Humanities at University College London, I was taught a methodological self-consciousness that I had not seen before in the humanities. I was interested in character diversity in Dutch novels and employed topic modeling to measure it. However, I soon had to acknowledge that I needed different techniques after the information appeared to be too complex to model. That different technique turned out to be crowdsourcing, and I developed a crowdsourced and data publication platform for measuring character diversity in a large corpus of novels: Personagebank. After careful evaluation of my methods, I eventually found the right one.
At CREATE, I hope to continue this search for methods that fit our purposes in humanities research. It is exactly this combination of methodological scepticism and scholarly ambition that I have grown to love in digital humanists. I am excited to have found a project and a research programme in which these two values come together.
By Lucas van der Deijl, Pre PhD Fellow on TOPIC