Post-editing machine translation as a student project, exemplified by literary fiction

Abstract

The article examines post-editing machine translation (PEMT) as a key relevant skill that needs to be nurtured in aspiring translators amid the rapid development of translation automation technologies. Following a brief historical background of machine translation (MT), from the very first mid-20^th century experiments to neural networks, the author turns to literary fiction in the context of PEMT tasks for students. Taking into consideration that literary texts in particular abound in the main MT “pitfalls” — words with emotional and evaluative connotations, idiomatic expressions, emulation of natural colloquial speech, and more — fiction is capable of providing rich illustrative material for identifying, systematizing, and correcting MT errors. The author cites a personal case study, wherein third-year BA students majoring in Translation and Translation Studies successfully performed PEMT on Nadezhda Teffi’s short story, The Princess’s Ruby (1905), which had been automatically translated from Russian into American English. While working with the Phrase translation platform, students encountered various MT errors, the classification of which is presented in this study. Apart from a few isolated issues with the transliteration of proper names, this article identifies three major groups of reoccurring errors in the literary text’s MT: punctuation errors (55 instances), lexical and semantic errors (67 instances), and grammatical errors (13 instances). The author accompanies these errors with corrections proposed by the students during PEMT and, in some cases, offers alternatives. The study repeatedly stresses that the recurrence of the aforementioned errors is tied to the unique features of literary fiction, which can justify including PEMT of literary texts into the roster of practical tasks for translator training.