From ‘text as text’ to ‘text as data’: the research potential in International Relations

ABSTRACT

Introduction:

The purpose of this article is to present the potential of ‘text as data’ analysis in research in International Relations. From the discrepancy in the proportion of publications that adopt this textual analysis approach, we presented the leading methodologies in the dimension ranging from ‘text as text’ to ‘text as data’ to highlight the existing opportunity. Aiming to stimulate production in the area, we developed a practical and replicable case, where we answered the following question: In the face of the Israel and Palestine conflict, how do the representatives of Brazil, United States, Israel and Palestine emotionally express themselves through their speeches at the Security Council meetings United Nations (UNSC)?

Materials and Methods:

Through a structured literature review, we presented the current state of the field of textual analysis in the scope of International Relations. A practical case was developed in a fully replicable way in the R language applying the sentiment analysis to the 77,857 given speeches at the UNSC from 1995 to 2019.

Results:

We demonstrated the potential of ‘text as data’ to carry out systematic analyses of large document collections and their complementarity to qualitative studies that try to obtain scientific conclusions through the ‘text as text’ approach. With the practical example, we show how the ‘text as data’ approach provides a well-informed and consistent analysis of actual events and inflections expressed in the content of speeches without the need for intense prior manual dedication.

Discussion:

The current state of the research in International Relations shows an evident discrepancy in the production of research with the ‘text as data’ approach compared to the ‘text as text’ approach. This scenario may result from the absence of training in contemporary methodologies and the lack of intensive computational approaches in the carried-out research. The article faces this challenge through a replicable theoretical and methodological exercise.

Keywords
text as data; text as text; content analysis; UNSC; sentiment analysis