Quantifying Comics

How do readers integrate text and images to make sense of comics? Or: What formal characteristics distinguish today’s long-form graphic narratives and serial comics? These are two of the diverse questions that guide the early-career research group “Hybrid Narrativity: Digital and Cognitive Approaches to Graphic Literature,” based at the universities of Paderborn and Potsdam in Germany since Spring 2015. Established with major funding from the German Federal Ministry of Education and Research (BMBF), the group assembles an unusual mix of people for a humanities project: computer and cognitive scientists work closely with literary critics and cultural historians. Our goal, of course, is to find out more about comics – and, in particular, contemporary graphic narrative in English, including graphic novels, memoirs, and journalism. Yet we aim to do so with the help of empirical methods that produce research results which can be replicated and falsified. In the process, we are developing infrastructure that, we hope, will prove useful to other researchers: an XML-based markup language and editor software that allows for the annotation of all visual and verbal aspects of comics; an online database and digitized corpus of 250 book-length narratives that covers the years 1975-2015; as well as a second corpus which records readers of comics with the help of high-speed eyetracking cameras. Much of this research remains work in progress but you can download a beta version of our editor here. The database, which includes relevant information on authors and publication details, will become available in early 2017.

The computational analysis of comics still represents a novel and challenging undertaking. The digital humanities are undergoing rapid growth but don’t pay much attention to comics. Therefore, we’re adapting existing approaches for our specific needs. These include image plots, which were pioneered in art history, and social networks, which have a long history in fields as diverse as physics and sociology. Our adoption of eyetracking research attempts to bridge the divide between textual analysis and reception studies. Given that empirical comics studies remains a small, albeit growing, area of research, we also aim to bring together interested scholars. Over the last two years, we’ve organized workshops and masterclasses with Lev Manovich and Neil Cohn, and our first larger conference – titled, “The Empirical Study of Comics” – will take place from 7-9 February 2017.

Where does that leave the research questions with which we began? The need to build infrastructure means that we’ve been doing a lot of ground work so far. Learning how to construct reliable corpora and designing software takes time. Still, initial results have been encouraging. We never imagined that readers spend 80 to 90 per cent of their time on text. In fact, only characters’ faces or objects relevant to the narrative get a look in. However, those who read comics regularly allocate more time to scanning images for contextual clues. Our digitized pilot corpus has also yielded first, tentative, insights. Graphic memoirs, a subgenre that supplies many of the best-known examples of contemporary comics, seem to be distinguished by a larger amount of narrative text than any other subgenre and feature significantly less dialogue. In addition, their protagonists are more visually dominant and central to the plot. These distinctions between subgenres did not hold when we looked at female characters. Graphic novels such as V for Vendetta and City of Glass feature hardly any or no conversations between women. The situation proves even more severe when ethnic minorities are taken into consideration: the graphic novel remains a largely white format. As we go along, we expect more robust result the larger our corpus becomes – and, hopefully, some additional surprises.

dunst images