Exploring Cities in Crime: Significant Concordance and Co-occurrence in Quantitative Literary Analysis Janneke Rauscher1 Leonard Swiezinski2 Martin Riedl2 Chris Biemann2 (1) Johann Wolfgang Goethe University Gruneburgplatz¨ 1, 60323 Frankfurt am Main, Germany (2) FG Language Technology, Dept. of Computer Science Technische Universitat¨ Darmstadt, 64289 Darmstadt, Germany
[email protected],
[email protected], {riedl,biem}@cs.tu-darmstadt.de Abstract Although the number of research projects in Dig- ital Humanities is increasing at fast pace, we still We present CoocViewer, a graphical analy- observe a gap between the traditional humanities sis tool for the purpose of quantitative lit- scholars on the one side, and computer scientists on erary analysis, and demonstrate its use on the other. While computer science excels in crunch- a corpus of crime novels. The tool dis- ing numbers and providing automated processing plays words, their significant co-occurrences, and contains a new visualization for signif- for large amounts of data, it is hard for the com- icant concordances. Contexts of words and puter scientist to imagine what research questions co-occurrences can be displayed. After re- form the discourse in the humanities. In contrast to viewing previous research and current chal- this, humanities scholars have a hard time imagining lenges in the newly emerging field of quan- the possibilities and limitations of computer technol- titative literary research, we demonstrate how ogy, how automatically generated results ought to CoocViewer allows comparative research on be interpreted, and how to operationalize automatic literary corpora in a project-specific study, and how we can confirm or enhance our hypothe- processing in a way that its unavoidable imperfec- ses through quantitative literary analysis.