Film studies make use of both, qualitative as well as quantitative methods (Korte, 2004). While there is a large variety of qualitative approaches to analyze movies (cf. e.g. Monaco, 2009; Sikov, 2010), most quantitative attempts seem to be focused on the analysis of the length and frequency of a film’s shots, which are understood to be the “single definable elements which can be nominated and described” (Salt, 2006: 14). After Barry Salt’s

For a concise review of Barry Salt’s work on quantitative movie analysis cf. Buckland (2008).

Studies that take into account quantifiable parameters other than shots are, however, rather rare. Among the few examples are Hoyt et al. (2014), who describe a tool that can be used to visualize relations between the characters of a film. Another example can be found in Ewerth et al. (2009), who present a toolkit that allows researchers to automatically detect shots and camera motion, superimposed text, faces and audio signals. While the latter example is rather focused on the automatic annotation of quantitative features, other projects, such as Lev Manovich’s (2013) „Visualizing Vertov“, focus on the presentation and visualization of quantitative parameters.

In this paper, we suggest to enhance the existing, shot-focused approaches to quantitative movie analysis, by considering additional parameters, such as language (cf. Forchini, 2012) and color use (cf. Flückinger, 2011). We present a prototype that can be used to automatically extract and analyze these parameters from movies and that makes the results accessible in an interactive visualization.

Language use as well as the use of colors in movies have long been known in the area of qualitative film studies. We argue, that these parameters are equally suited for a quantitative approach and present an experimental prototype that can be used to quantify the language and color used in different movies. Much like Clement et al. (2008), who discuss the drawbacks and opportunities of computer-based methods in the field of literary studies, we believe that digital, quantitative movie analysis tools can be helpful in “offering provocations, surfacing evidence, suggesting patterns and structures, or adumbrating trends”. As the prototype allows researchers to investigate potential correlations between color usage and corresponding language in a movie, it can be used to examine questions such as the following:

We believe the real strength of a quantitative approach that makes use of language and color information lies in a mix of “distant watching” (cf. Howanitz, 2015) and close watching, i.e. characteristic language-color patterns identified in specific movies can be used as a query to search other movies for similar patterns. Our next steps will therefore go into the direction of an information system that allows researchers to search and compare a collection of movies according to language and color characteristics.