Seite - (000102) - in Knowledge and Networks

94 This clustering approach is of particular interest because it is able to sum up thousands of statements, speeches, and resolutions in order to produce an artificial map of political regions. The map does not necessarily signify that two member states in the same group always behave in the same way or share common views on the vast range of topics examined in the UNGA. However, it does provide an initial delineation that should be kept in mind for in-depth study. Other approaches appear valuable and can provide relevant results, such as investigating resolutions’ sponsors by topic or similarities among resolutions to see whether or not the same type of resolution (on human rights or economic develop- ment) tends to produce highly interconnected graphs. However, studying voting behavior provides only a partial picture of relational patterns in the UNGA because only one third of resolutions are included. Whereas studying voting behavior is a traditional way to view relations in the UNGA, the study of speeches remains less developed even though the General Assembly can be considered an arena of words rather than decision-making. Each year, representatives from nearly all countries (Taiwan excepted) issue statements on a broad range of issues, and these speeches can be considered the official posi- tion of those states’ governments. But states are not the only speakers, and the study of speeches supports multiscalar analysis. Speech Dynamics in the UNGA: A Network Approach Various individuals make statements in the UNGA: representatives of states (nearly 80 % of all statements during a session for the period under consideration), repre- sentatives of groups, of institutions from the UN system, and, less often, representa- tives from NGOs (e.g., the Red Cross) or institutions external to the UN system (e.g., the European Commission). In the last two decades, a change in speech dynamics has occurred. Groups issued more and more statements on a growing range of topics, and states’ representatives have supported their statements more and more often, two patterns that could be a relevant indicator confirming the politi- cal world regionalization hypothesis. Minutes of meetings from 1990 to 2010 were reviewed to collect all speeches that affirmed “state A supports group a,” and these data were studied as bipartite graphs. Occurrences of such speeches were trans- formed into a state-group matrix in which each case indicates the number of times a state A supported a statement made by a group a. The poor quality of documents from older sessions prevents automatic word-searching, and data-gathering was therefore manual and time consuming. L. Beauguitte