Algorithmic iconography: intersections between iconography and social media image research

Drainville, Raymond A (2018) Algorithmic iconography: intersections between iconography and social media image research. Doctoral thesis (PhD), Manchester Metropolitan University.

Preview

Available under License Creative Commons Attribution Non-commercial No Derivatives. Download (130MB) | Preview

Abstract

The objective of this thesis is to develop a methodology for examining a large dataset of visual materials coming from social media in a nuanced, historical fashion. Iconography is a qualitative methodology originating in the history of art that identifies, categorises, and interprets works of art, placing the result in historical perspective. This thesis develops a mixed methodology that unites iconography with approaches common in social media studies in order to give the analysis of imagery on social media historical depth and nuance. The method incorporates social media commentary, which provides insights into users’ interpretations of the pictures they share, in a level of detail that traditional iconography rarely achieves. An adapted iconographical method can provide evidence for why the pictures users share are interpreted as they are, and why some are shared more than others. I demonstrate iconography’s usefulness with a case study, centring upon an historically important dataset: the most-shared tweets containing pictures of Alan Kurdi and other refugees during a two-week period beginning when Alan was found on 2 September 2015. Reaction to pictures of Alan shared on Twitter spilled into the real world and led to global demonstrations on behalf of, and increased support for, refugees. This thesis explores the mechanisms by which these pictures crystallised popular sympathetic reactions for refugees at an important point in contemporary geopolitics. To aid in the iconographic analysis of this data, I have created a digital tool called a “datasheet” (https://doi.org/10.23634/MMUDR.00621172) to help analyse visual and textual patterns. Working in tandem with data analytics software, the datasheet provides other forms of otherwise difficult-to-discern “top-down” insights. By using these tools, it is possible to uncover pictorial preferences, waves of sharing over time, cross-cultural interpretative patterns, and longstanding interpretative themes such death as sleep, children as angels, oversized monumentality, standard representations of refugees, the fixing of ambiguous imagery through anchoring text, and hostile representations of refugees as threats. Many of these themes returned after the time of data collection as the political climate surrounding refugees soured in 2016. The iconographic approach and the organisation of this datasheet provide a foundation for future analyses of imagery shared on social media.