ИСТИНА

We describe a methodological and technical framework for conducting qualitative and quantitative studies of linguistic research questions over diverse and heterogeneous data sources such as corpora and elicitations. We demonstrate how LLOD formalisms can be employed to develop extraction pipelines for features and linguistic examples from corpora and collections of interlinear glossed text, and furthermore, how SPARQL UPDATE can be employed (1) to normalize diverse data against a reference data model (here, POWLA), (2) to harmonize annotation vocabularies by reference to terminology repositories (here, OLiA), (3) to extract examples from these normalized data structures regardless of their origin, and (4) to implement this extraction routine in a tool-independent manner for different languages with different annotation schemes. We demonstrate our approach for language contact studies for genetically unrelated, but neighboring languages from the Caucasus area, Eastern Armenian and Georgian.