Are post-editese features really universal?

This study aims at assessing whether or not differences between human and post-edited machine translations, so-called post-editese features, are common to all post-edited texts. We investigate the presence of four candidate post-editese features in corpora that include two language pairs and three different machine translation systems to find out whether or not different post-edited texts share the same characteristics, regardless of other factors. Our results show that the language pair, as well as the MT system, can have a great influence on these four features, thereby going against the assumption that they are common to all post-edited texts.