An Annotated Corpus of Argumentative Microtexts

We present a freely available corpus of argumentative "microtexts", featuring short and dense authentic arguments, annotated according to a scheme for representing text-level argumentation structure. The corpus consists of 112 German texts plus professional English translations that preserve linearization and argumentative structure. We provide statistics of the variety and the linguistic realization of argumentation structure in the corpus. We hope the data release serves the needs of datadriven approaches to argument mining and qualitative analysis alike.