Lecture: Exploring Automation in The Analysis of Creative Shifts Found in News Translations: Machine vs Human Translation

The project presents the results of a research including the translation of a collection of news from English to Chinese in two modalities: MT and HT. An empirical approach is presented to quantify the creativity in each translation with the analysis and evaluation of creative shifts (all translations that deviate from the source text) and reproduction (all translations that reproduce the source text) in news translation. Guidelines are provided to automatically determine the creativity shifts in parallel corpora and the evaluation of these shifts in comparison to human annotated corpora.