The analysis of the depiction of heads of state and "common people" is part of the DFG-funded projects "Automatic Identification and Classification of Persons as Key-Visual-Candidates" at Jacobs University Bremen (Prof. Dr. Peter Ludes) and the Center for Computing and Communication Technology (TZI) of the University of Bremen (Prof. Dr. Otthein Herzog). The aim of the project is the analysis of visual conventions that help to identify and distinguish heads of state and common people. These visual conventions are operationalized with help of techniques from automatic image recognition. Ideally, both types of actors can be identified automatically. Algorithms of automatic image recongition would go beyond the identification of simple geometric shapes. Instead, the social role of an actor would be identified automatically. With regard to communication science, the projects builds on and extends existing research by Prof. Ludes. It deals with questions like: How is "power" and social hierarchy represented visually? Is there a trend towards a more informal depiction of heads of state over the years? Are the visual conventions universal across cultures?
The analysis is based on TV annual reviews of the last decade from the USA and from Germany. TV annual reviews summarize the most important events and pictures of the year. For this reason, TV annual reviews contain typical depictions.
Potentially relevant characteristics of the footage has been coded and analyzed with the help of the video coding software Kivi. Kivi has been developed for this project by Jan Müller. It can be shown that heads of state are predominantly present in reports on domestic- and foreign policy. They are depicted as active. "Common people", in contrast, are shown in thematic contexts like economy, human interest, and celebrations. They are depicted as comparatively passive (see figures 1 and 2).
A multiple correspondence analysis permits a closer investigation of the depiction of heads of state and common people. It can be shown that US heads of state are relatively often depicted in military contexts. This is reflected in the style of clothing and in the situation shown. German heads of state, in contrast, are frequently depicted "e;at work"e;. This includes depictions of a head of state in assemblies, showing him or her giving a speech (without the actual content of speech), on the corridor or his or her way to assemblies. A formal clothing style (suit, tie) is characteristic of the depiction of German heads of state.
Automatic face recognition as been used in an attempt to automatically distinguish heads of state and common people. It has been hypothesized that heads of state are depicted as larger and more central. An analysis of the spatial and temporal distribution of the heads of the actors yields patterns that are specific for particular situations (Abbildung 4). These patterns are analyzed with cluster analysis. This analysis has been done by Martin Stommel of the Center for Computing and Communication Technology. Unfortunately, it turned out that heads of state and common people cannot be distinguished on the base of these criteria. Kivi permits the marking of areas in single frames from the videos. This feature is used to measure the size of the depictions of the actors in the video (see figure 5). The hypothesis that heads of state as the more prominent and more powerfull actors are depicted on average larger cannot be confirmed. Instead, it is the presence or absence of an interview situation that determins the size of the depiction.
Müller, J. & Stommel, M. (2011). Heads of State and Common People: Perspectives from the Computer and Social Sciences. In: Peter Ludes und Otthein Herzog (Hrsg.), Algorithms of Power (in preparation). Power. Berlin: Lit Verlag.
Stommel, M. & Müller, J. (2010). (2011). Das bewegte Bild: Filmanalyse – Methoden der automatischen, computergestützen Bilderkennung und Analysen. In: Thomas Petersen und Clemens Schwender (Hrsg.), Die Entschlüsselung der Bilder - Methoden zur Erforschung visueller Kommunikation. Köln: von Halem.
Müller, J. & Stommel, M. (2010). Using automatic face recognition to identify presentation patterns of power and hierarchy in U.S.-American and German TV annual reviews. Paper presented at the 3rd European Communication Conference, Hamburg.
Müller, J. (2009). Symbols of Statehood, Legitimation, and Power in the Depiction of Heads of State in German and US-American Annual Reviews. Paper presented at the 9th Conference of the European Sociological Association, Lissabon.
Descriptions, authors, and licences of the pictures of the collage from left ro right and from top to bottom: