Operationalising and Representing Conceptual Variation for a Corpus-driven Encyclopaedia

Abstract

Detecting conceptual variation among humanitarian actors in textual sources is one of the challenging objectives of the Humanitarian Encyclopedia. This article proposes a method to operationalise and represent conceptual variation. Conceptual variation is a phenomenon whereby individuals and organisations show different understandings of the intensions and extensions of concepts. Despite the existence of a shared vocabulary, humanitarian concepts are presupposed to be affected by conceptual variation due to the recent professionalisation and diversity of the sector. In a pilot study, the four humanitarian principles (i.e., HUMANITY, IMPARTIALITY, NEUTRALITY, and INDEPENDENCE) were analysed with a hybrid methodology that combines Frame-based Terminology and Content Analysis. Definitions were extracted from a corpus of humanitarian documents, coded inductively to unveil definitional elements, and consolidated with corpus metadata to associate them with specific types of humanitarian organisations. Finally, a conceptual profile for each concept was represented by plotting its definitional elements and the number of occurrences on radar charts. Occurrences were subsequently disaggregated by organisation type to reveal differences between humanitarian actors. Several cases of conceptual variation were preliminarily detected. Minor cases of semantic overlap were also identified. Our preliminary results suggest that this method can detect and represent conceptual variation satisfactorily.