While edited texts published in the media, scientific texts and literature are expected to comply with certain language norms, it is evident that texts in social networks often do not meet the same standards. Will this have a long-term effect on written language in general and on the quality of texts produced by digital natives in particular?

Our project aims at exploring ways of empirically addressing this question. So far, scientific debate was based on a relatively limited amount of data concerning the effects of digital media on writing skills. We intend to provide an objective basis by developing a model that takes into consideration different aspects of quality, which are relevant for dialogical communication in digital and multimodal social media, and relates to the tradition of reflecting on language quality in rhetoric and text linguistics. We further plan to develop methods for measuring coherence as an aspect of quality and to evaluate them with the help of small case studies. In this way we hope to gain some insights into how constantly reading and writing short messages within dialogues affects the competence of planning and drafting longer monologues, thus contributing to filling a research gap. The project uses digital text collections (corpora) of learner texts and web genres that have been prepared and evaluated with the methods of corpus linguistics. New methods (e.g. crowd sourcing and data mining) will also be tested.