New CTS article on Human and Machine Translation
Dr Ana Frankenberg-Garcia's paper entitled 'Can a corpus-driven lexical analysis of human and machine translation unveil discourse features that set them apart?' has been published in Target.
The paper looks at how human and machine translation differ, using a parallel corpus of Portuguese published fiction of 72 thousand words aligned with professional, published English translations and neural machine translation. Lexical discrepancies in the two translation corpora were extracted using keyword analysis, and then examined qualitatively through parallel concordances.
The study shows that keyword analysis is not only sensitive to known problems of machine translation such as lexical inconsistency and pronoun resolution, but can also provide valuable insights about other aspects of the translation of discourse deserving further research.
The article is open access and freely available following this link.