From monolingual annotations towards cross-lingual resources: An interoperable approach to the analysis of discourse
Abstract
in this paper, we perform a cross-lingual analysis of discourse phenomena in German and Czech, using two corpus resources annotated monolingually within two different frameworks: functional Generative description (Sgall et al. 1986) for Czech, and textual cohesion (Halliday & Hasan 1976) for German. We take advantage of the existing resources reflecting systemic peculiarities and realisational options of the languages under analysis. in our previous work (Lapshinova et al. 2015), we have shown that the annotations of the involved resources are comparable if abstract categories are used and only the phenomena with a direct match in German and Czech are taken into consideration.
Our analysis is a first step towards unifying separate analyses of discourse relations in Germanic and Slavic languages. at the same time, it demonstrates that the application of 'theoretically' different resources is possible in one contrastive analysis. This is especially valuable for nlp, which uses annotated resources to tra