Doctoral thesis

Humanitarianism is a recently professionalised domain in which competing conceptualisations of key humanitarian notions coexist. In recent years, multiple initiatives like the Humanitarian Encyclopedia (HE) have emerged to foster shared understandings of humanitarian concepts with a view to improving humanitarian coordination. The HE is an ongoing project that aims to define descriptively 129 humanitarian concepts that are particularly ill-defined, fuzzy or controversial. Due to intra- and inter-organisational differences among humanitarian actors, the HE seeks to offset biases associated with the academic and professional backgrounds of entry authors by informing entries with conceptual analyses based on lexical data. To this end, the HE combines expert knowledge with knowledge facilitated by linguists into encyclopaedic entries with a focus on conceptual variation, i.e., how the understanding of concepts varies among humanitarian actors.

The main goal is therefore to design a method for systematic conceptual analysis that (1) communicates the conceptualisation of notions in humanitarian discourse to entry authors, (2) that disaggregatesthe extensions and intensions of concepts, and (3) that detects and represents conceptual variation. This method will be partly informed by Faber’s Framebased Terminology whereby concepts will be modelled by means of conceptual propositions, i.e., entity-relation triples linked by a semantic relation, which will be substantiated by lexical data obtained from knowledge-rich contexts (KRCs) curated from a corpus of humanitarian documents.

Thanks to corpus metadata, the modelling of concepts will consolidate conceptual propositions, KRCs and document metadata into a single dataset, thereby operationalising conceptual variation as the variation of semantic relations and entities in comparable conceptual propositions enriched with metadata. In addition, the method will harness the quantitative dimension of conceptual variation by incorporating corpus-based reference metrics to differentiate prominent conceptual characteristics from marginal ones.

A model for result reporting will also be designed to incorporate data visualisations that will enable users to disaggregate conceptual components by organisation type, year of publication and region, among other metadata. Since a high degree of lexical heterogeneity is expected, conceptual descriptions will have to be classified inductively into sizeable and comparable semantic categories through qualitative data analysis techniques. In short, this project aims to create (1) a method for conceptual analysis based on lexical data, (2) a model for conceptual representation that accounts for conceptual variation, and (3) a model for results presentation that is supported by data visualisation.