Picturing Politics: Techniques, Challenges and Opportunities for Semi-Automated Online Imagery Collection in Discursive Research

This paper investigates new ways of researching images as visual symbolic representations. Research on visual representations in politics, political geography and International Relations lives by the mantra of context being all-important. However, in an overwhelmingly digital landscape, an image rarely has a fixed context, but surfaces in multiple places. Stock photos are used as illustrations by multiple outlets, images are paired with provocative headlines or catchphrases on social media, or are shared by private persons, politicians and organisations. The reproducibility of digital images makes it easy to copy and paste an image into an entirely new context, that may be diametrically different from the original. An article using a photo as an illustration is unlikely to be the only context in which the picture appears. Researchers interested in visual representations should pay attention to this shift and treat the image as an artefact in its own right rather than secondary to written text. Images are not objective accounts of the empirical world ‘out there’. They are limited and particular representations, constructing political meaning in a distinct, visual stream of discourse that interplays with, but is not secondary to, textual discourse. In an ocular-centric culture, images enjoy a privileged position as truth-claims to support political arguments. One image is chosen over another because it resonates with or symbolises a certain value, opinion or narrative, be it descriptive, informative or provocative. Technological developments have seen the role of images metamorphose from secondary to text to becoming mobile and powerful elements of discourse: political scientists need a method to respond to this shift and treat images as components of a selective visual representation of reality. This paper proposes a toolkit for gathering data that allows researchers to study the worlds created by visual representation. To introduce a screen between researchers and data collection, that allows researchers to decouple images from immediate contexts, I propose a basic, open-source Python script to automate advanced Google image search queries to gather images within a given domain, requiring only initial and occasional manual intervention. By giving a step-by-step account of this process from search query to database ready for analysis through a case-appropriate framework, the paper invites readers to evaluate, critique and expand on this technique. Alongside conceptual and methodological pointers, reflections and challenges, the paper provides examples from a project carried out using this method of image collection in conjunction with a political geographic analytical framework, where I examined representations of the Arctic as a political space. I sought to identify thematic (economy, environment, geopolitics) as well as spatial (local, global) representations through images and maps respectively, and how these differ between actors. By being transparent on process, the goal of the paper is not to provide a rigid procedure, but to trigger future experiments with the technique, foster academic debate on the role of images as discursive artefacts in their own right, and evaluate challenges and opportunities for automatic data collection in discursive research.