NLP-enhanced shift analysis of named entities in an English<>Spanish intermodal corpus of European petitions

This chapter aims at presenting an NLP-enhanced corpus-based analysis of the translation and interpreting shifts observed in the named entities (NEs) of PETIMOD, an English<>Spanish intermodal corpus of written and oral mediated texts from the Committee on Petitions of the European Parliament. Our main assumption is that shifts in institutional genres mostly occur in the transfer of NEs, and that NLP techniques such as automatic Named Entity Recognition (NER) can be applied to systematically extract and compare examples of these shifts, leading to the (possible) verification of translational and/or interpretational constraints. Results show that traits like normalisation, transformation and simplification depend not only on the language direction or the mediation mode, but also on the semantic category (person, organisation, etc.) of the NE involved. Further studies are needed in order to correlate observed shifts with different NE taxonomies.