Towards a corpus-based distinction between language-specific and universal features of mediated discourse

After investigating the various ways in which apparently “original” communication is in fact mediated intralingually, interlingually and intersemiotically, the present paper aims to verify whether the various forms of mediated discourse have features in common and, if so, whether these shared features derive from specific language-pair interference or whether they are universal across languages, as is increasingly being demonstrated in relation to translational phenomena and in the area of English as a Lingua Franca (ELF). The hypothesis of recurring patterns of linguistic behaviour that are generated by the process of mediation taking place in various modes of communication is tested on the EUROCOM corpus, a corpus of draft texts written in English by non-native speakers within the European Commission (EC) and subsequently revised by native speakers of the EC Editing Unit. The initial results of our investigation into these mediated English texts support the view that, like translators, non-native authors and editors tend to adopt domesticating strategies such as simplification, normalisation, explicitation and levelling-out in an attempt to make texts more easily accessible and intelligible to target audiences. The universal features of mediated discourse observed are then compared to the findings of the research on computer learner corpora, which act as a sort of litmus paper thanks to their predominantly language-specific nature.