Speaking of civilians: Automated text analysis of the United Nations’ framing of complex humanitarian emergencies

Abstract

Since the end of the Cold War, the protection of civilians has increased its weight on the United Nations (UN) agenda. This article (1) maps the evolution of civilian protection within the UN framework as an indicator of its shifting priorities, (2) identifies breakpoints in the prevalence and character of the protection discourse, and (3) explores how internal processes of policy development and real-world triggers (namely conflicts and peacekeeping operations) shaped this transformation. The article uses Structural Topic Modeling (STM) to analyze an original corpus of Security Council and General Assembly resolutions on complex humanitarian emergencies since 1990. The analysis uncovers two distinct forms of protection, labeled as “Ground protection” and “Political–legal protection,” which are characterized by contrasting temporal and geographic trajectories. Moreover, critical junctures in the protection rhetoric (during the years 2000, 2005, and 2008) coincide with policy watersheds rather than conflict outbreaks or trends in peacekeeping deployment. This article offers a comprehensive analysis of the intricate evolution of civilian protection using text-as-data methods, which uses its theory-building design to encourage further explorations on the interplay between internal and external factors in shaping its progression within the UN framework.

Desde el final de la Guerra Fría, la protección hacia los civiles ha aumentado su peso dentro de la agenda de la Organización de las Naciones Unidas (ONU). Este artículo: (1) traza un mapa de la evolución de la protección civil en el marco de las Naciones Unidas como un indicador de sus prioridades cambiantes, (2) identifica los puntos de ruptura en la prevalencia y el carácter del discurso en materia de protección, y (3) estudia cómo dieron forma a esta transformación tanto los procesos internos de desarrollo de políticas como los desencadenantes del mundo real (es decir, los conflictos y las operaciones de mantenimiento de la paz). El artículo utiliza el Modelado Temático Estructural (STM, por sus siglas en inglés) con el fin de analizar un corpus original de resoluciones del Consejo de Seguridad y de la Asamblea General en materia de emergencias humanitarias complejas desde 1990. El análisis revela dos formas distintas de protección, denominadas «protección sobre el terreno» y «protección jurídica política», la cuales se caracterizan por trayectorias temporales y geográficas que contrastan entre sí. Además, las coyunturas críticas en la retórica de la protección (ocurridas durante los años 2000, 2005 y 2008) coinciden con los puntos de inflexión de las políticas y no con los estallidos de conflictos o con las tendencias en el despliegue de las operaciones de mantenimiento de la paz. Este artículo ofrece un análisis exhaustivo de la intrincada evolución de la protección civil. Para ello, el artículo utiliza métodos de texto como datos, los cuales utilizan su diseño, basado en la creación de contenido teórico, con el fin de fomentar nuevos estudios acerca de la interacción entre factores internos y externos en la configuración de su progresión dentro del marco de las Naciones Unidas.

Depuis la fin de la guerre froide, la protection des civils a pris de l’importance dans le programme des Nations Unies (ONU). Cet article (1) représente l’évolution de la protection civile dans le cadre de l’ONU en tant qu’indicateur d’un changement de priorités, (2) identifie les points de rupture dans la prédominance et la nature du discours de protection, et (3) examine le façonnement de cette transformation par les processus internes d’élaboration de politiques et les facteurs déclencheurs du monde réel (c’est-à-dire, les conflits et les opérations de maintien de la paix). L’article emploie un modèle thématique structurel (Structural Topic Modeling ou STM) pour analyser un corpus inédit de résolutions du Conseil de sécurité et de l’Assemblée générale concernant des urgences humanitaires complexes depuis 1990. L’analyse dévoile deux formes de protection distinctes, la « protection sur le terrain » et la « protection juridico-politique », dont les trajectoires temporelles et géographiques s’opposent. De plus, les tournants observés dans la rhétorique de protection (en 2000, 2005 et 2008) coïncident avec des raz-de-marée politiques, et non avec le début de conflits ou des trajectoires dans le déploiement de missions de maintien de la paix. Cet article propose une analyse exhaustive de l’évolution complexe de la protection des civils à l’aide de méthodes de texte comme donnée. Sa construction théorique vise à susciter de nouveaux travaux de recherche quant à l’enchevêtrement de facteurs internes et externes qui façonne sa progression au sein du cadre de l’ONU.

Keywords:

• Peacekeeping
• protection of civilians
• STM
• text-as-data
• United Nations

Acknowledgments

The author thanks Edoardo Baldaro, Jutta Bakonyi, Paride Carrara, Matteo Dian, Jessica Di Salvatore, Mirek Tobiáš Hošman, Francesco N. Moro, Luca Pinto, Lorenzo Zambernardi, all SICSS participants in Bologna (especially Marco Albertini and Aidar Zinnatullin) as well as the two anonymous Reviewers and Editors of International Interactions.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

* Replication data are available through the International Interactions Harvard Dataverse: https://dataverse.harvard.edu/dataverse/internationalinteractions. Please refer to the author for any inquiries.

1 The terms civilian and non-combatant are used interchangeably to reflect the United Nations standard.

2 The United Nations Operations in the Congo (1960–1964) engaged in protection activities, but the mandate was not formalized by the Council.

3 As customary in topic modeling, this article uses the terms concepts topic and (keyword) cluster interchangeably.

4 United Nations (https://peacekeeping.un.org/en/protecting-civilians).

5 United Nations (https://research.un.org/en/docs/resolutions).

6 This agenda-setting focus is also replicated in the ground behavior, as the UN deploys to the most challenging environments with human rights violations (e.g., Page Fortna 2008).

7 Contradicting this narrative, Melander, Öberg, and Hall (2009) find that targeting civilians has declined after the Cold War.

8 Peacekeeping is understood as an “invention” of Secretary-General Dag Hammarskjöld, as it was originally not mentioned in the Charter.

9 This concept builds on the topical shifts of attention by Bagozzi and Berliner (2018). Even Hanania (2021) uses breakpoints, but does not elaborate on them theoretically.

10 United Nations (https://www.un.org/en/library).

11 The appendix contains pre-processing validity measures based on the preText package (Denny and Spirling 2018).

12 Identified based on theoretical knowledge and through a review of word pairs with a 𝜆 z-score > 100 based on the quanteda textstat_collocations function (Benoit et al. 2022).

13 All figure and table indications preceded by “A” refer to the online supplementary material.

14 The different mandates of the UNSC and UNGA can question the decision to use a single corpus. A supplementary analysis is presented in the supplementary material to address this concern (Figures A18–A20). While some topics are driven more by one institution than the other, both speak in similar terms about civilian protection. Moreover, the binary distribution between “Ground protection” (Topic 3 and 5 for UNSC, Topic 2 for the UNGA) and “Political–Legal protection” (Topic 1 and 2 for UNSC, Topic 1 for the UNGA) persists also in the two separate STMs.

15 ß describes the “base 2 logs of the probability of that word being in that particular topic, divided by the average probability of that word appearing across the other topics” (Hanania 2021, 11).

16 United Nations (https://peacekeeping.un.org/en/mission/monusco).

17 Human Rights Watch (https://www.hrw.org/publications).

18 Breakpoints exist also in 1993 and 2016, but their effects based on the regression coefficients are significantly weaker.

19 As UNPI covers only the period up until 2015 and no similarly granulated information for later years is coded in other data sets, this year is used as a cutoff. This is deemed acceptable, as no significant juncture materializes thereafter.