International Relations Online Working Paper Series

2012/1
Paper ( pdf - 737 kB)

Corpus-Based Content Analysis:
A Method for Investigating News Coverage on War and Intervention

Access to large digital text samples in various languages and the development of general-purpose tools in natural language processing holds great promises for comparative content analysis in the social sciences. However, boosted size, multiplicity of text genres and languages, and automation sharpen some classical problems inherent in the content analytical method. They reinforce the question of how certain (pre-defined) conceptual categories can be assigned to passages of text without violating the meaning that is implied by the entire text item through its linguistic structure. The paper proposes corpus-based content analysis as a method to achieve high internal validity and increase in effectiveness through semi-automation. We suggest adopting the concept of ‘semantic fields’ as a heuristic tool for inference-making. It offers criteria that help to relate semantic information retrieved with the help of corpus-linguistic methods and computational tools to specific conceptual categories of a social science research project. An illustration of how to apply the method is provided for the conceptual categories: ‘military intervention’; ‘EU-Europe’; and ‘actorness’.

2011/4
Paper ( pdf - 260 kB)

Rethinking Morgenthau in the German Context

After a lapse in interest and popularity, realist approaches to International Relations (IR) have begun to spark interest again, even in Europe. This paper elaborates the origin of Hans J. Morgenthau’s realist thinking. Given his various power-oriented, legal and moral arguments against Wilson’s doctrine of democracy promotion and democratic interventionism, the Vietnam War, and in favour of an Israeli state (3.), my thesis is that there is a primal and entrenched critical and normative dimension in Morgenthau’s thought from the beginning. His particular understanding of international law and morality as limitations of power extends beyond a rationalist and conservative ‘Cold War’ theory about power politics (as Realism is presented in prevalent readings). The core argument of the paper is that the complex relation between power, law and morality inherent in Morgenthau’s thinking, and the contradictions coming along with this (2.), are characteristic of his complete works (Gesamtwerk). In order to deal with his ‘paradoxical’ thinking, we should rethink young international lawyer Morgenthau in his German Context (Erfahrungshintergrund). Surprisingly, less work has been done on the evolution of Realism in the inter-war period (which E.H. Carr calls the “Twenty Years’ Crisis”) in general, and Morgenthau’s life and work in Germany in the 1920s and 1930s in particular which is also neglected in the literature about Realism and the history of IR so far. But, Morgenthau’s Weltbild can only be truly understood if and when the German years in which he was socialized intellectually and politically are considered as the roots (or the constitutive moment) of his realist theory (4.). Moreover, an understanding of Morgenthau’s brand of Realism is necessary to understand both the history of Realism as a 20th-century school of thought as well as the development of our discipline IR.

2011/3
Paper ( pdf - 209 kB)

Distinguishing Specialised Discourse:
The Example of Juridical Texts on Industrial Property Rights and Trademark Legislation

In this paper, we have given an overview of computational linguistic tools available to us, which can be used to produce raw material for the lexicographic description of a specialised language. The underlying idea of our method is the following: what is significantly more frequent in a domain-specific text than in a general language reference text may be a term (or collocation) of the domain. In the near future, our tools will be integrated in a web-based environment in order to make them available for textbased research, e.g. in the humanities, whenever needed. The researcher interested in term or phraseology candidate extraction of a certain domain would identify and upload texts to be searched, and the tools would be running on servers of e.g. computational linguistics centres. The researcher would select tools to be applied and receive the analysis results over the network.

2011/2
Paper ( pdf - 153 kB)

How to get rid of the Noise in the Corpus: Cleaning Large Samples of Digital Newspaper Texts

Large digital text samples are promising sources for text-analytical research in the social sciences. However, they may turn out to be very troublesome when not cleaned of the ‘noise’ of doublets and sampling errors that induce biases and distort the reliability of content-analytical results. This paper claims that these problems can be remedied by making innovative use of computational and corpus-linguistic procedures. Automatic pairwise document comparison based on a vector space model will bring doublets to light, while sampling errors can be discerned with the help of textmining procedures that measure the ‘keyness’ of a document, i.e. the degree to which it contains or does not contain keywords representing the research topic.

2011/1
Paper ( pdf - 481 kB)

Semi-Automatic Core Sentence Analysis:
Improving Content Analysis for Electoral Campaign Research

Most automated procedures used for the analysis of textual data do not apply natural language processing techniques.While these applications usually allow for an efficient data collection, most have difficulties to achieve sufficient accuracy because of the high complexity and interdependence of semantic concepts used in the social sciences. Manual content analysis approaches sometimes lack accuracy too, but, more virulently, human coding entails a heavy workload for the researcher. To address this high cost problem without running into the risk of oversimplification, we suggest a semi-automatic approach. Our application implements an innovative coding method based on computational linguistic techniques, i.e. mainly named entity recognition and concept identification. In order to show the potential of this new method, we apply it to an analysis of electoral campaigns. In the first stage of this contribution, we describe how relations between political parties and issues can be recognized by an automated system. In the second stage, we discuss facilities to manually attribute a positive or negative direction to these relations.