Meet the Grantee: Thomas Wallnig
Mathematics and statistics have the power of definition over everyday knowledge. Is it then heretical to ‘embrace’ these statistical methods for your work? After all, these are possibilities that are not available to a traditionally working historian.
SB: We can see where our thinking comes from, our categories, how they are inescapable because they are historically imprinted on us as habitus. We try to use digital algorithms to make structures visible that are not visible to the classical reader and that are hidden in the texts. In this way, we reveal a ‘historical unconscious’, the categories and structures which we think by, which can explain how we deal with historicity, how we deal with the relationship between constancy and change. Intercultural comparison is also about these categories: To what extent are they valid, to what extent are they theoretically inherent in the science of history, or to what extent are they only culturally based and thus relative? It's a highly interesting question that can only be grasped in this way.
TW: But that also means that the way we historians critically illuminate our material is something highly surprising for data scientists; or for all those who assume the stability of categories and the inescapable accuracy of models. It takes a lot of convincing to coders to make clear that the task of humanities scholars is not to enter seemingly stable categories into an algorithm. ‘State’ is not a stable category, ‘Germany’ is not a stable category, it cannot be modelled as ‘same-as’ over four centuries. These are assumptions about the world, but they are not in the world itself. It is a human construct very much bound to time and culture.
So we don’t ‘embrace’ data science, but we engage with it amicably but critically, discussing and sharing the basic assumptions behind the modelling.
And what does that mean for your role as a historian? How do these possibilities of data science affect your academic work?
TW: Large parts of the historian community have not yet engaged with data sciences. But the institutional pressure for data management is increasing, so compliance will be enforced in the medium term. The younger professors will have to deal with this. It is a matter of making one’s own contribution transparent, of making your research data available. I have to be familiar with the different methods and licensing procedures when I develop something from others and put it back into the large, ever-growing pool of available data. These are the skills of discoverability, usability and re-usability of data.
At the same time, I continue to believe in the medium of the book because it is necessary to work on a well-dimensioned research question. There is a difference between simply putting a few results on the web and taking on a question and trying to process it. Every person who has written a book knows that this is a maturing process that takes them one step further in their knowledge and also gives others the opportunity to follow their step. It is a travel guide through data. But: I don’t need to work through a collection from A to Z when almost everything is available digitally, I just can’t look at the totality of all data anyway, and it’s wiser to take openly accessible data from similar contexts for a targeted comparison. So the designs will change, but what shouldn’t change is the creation of an inner tension in the conscious generation of a narrative arc.
SB: We can now handle hundreds of thousands of pages that have been accumulated over a century. Even handwritten texts can now be transcribed and digital possibilities are opening up that far exceed what is humanly possible. For human beings do not function like algorithms, but the way they think is constantly changing. If I have read two texts, I have to re-read the first one because the second has changed my conception, and then I have to start all over again. The machine, on the other hand, has the advantage of its inflexibility, the advantage of applying the same algorithm to all texts and arriving at comparative results that can be contextualised again.
The technical possibilities of digitisation allow, for example, an analysis of all known German textbooks. We would look at this national autobiography of the last 400 years, and we could see how some discourses have disappeared, or sunk, such as the discourse on France, which was very much alive in the 17th century because France was the central power in Europe at that time. This is my subjective view, one could criticise, but I can also prove it now, with a longitudinal section by way of digital processes.
Part 3: On academic work across disciplines and the ambivalence of research mobility