ADAPT Researchers publish article in Machine Translation for Conquering Language Barriers

18 February 2022
Sheila Castilho and Natália Resen published article in Special issue Machine Translation for Conquering Language Barriers

ADAPT Research Fellows Dr Sheila Castilho and Dr Natália Resende recently published an article in the Special issue Machine Translation for Conquering Language Barriers of the open access journal Information. The article “Post-Editese in Literary Translations”, investigated the post-editese phenomenon, i.e., the unique features that set machine translated post-edited texts apart from human-translated texts.

Castilho and Resende analysed two books: Alice’s Adventures in Wonderland and The Girl on the Train to investigate whether the post-editese features can be found on the surface of the post-edited (PE) texts. While also examining how the features found in the PE texts differ from the features encountered in the human-translated and machine translation versions of the same source text. 

“The post-editese phenomenon was manifested on the surface of the post-edited texts as there were differences between those and Human Translated versions.”

They found that there was a clear and distinctive difference between literary genres when it came to this post-editise phenomenon. Machine translation, as they discuss at the beginning of the article, is perceived as “less useful for creative texts than for other text types”. Which appears to hold true from their analysis of their two texts as they uncovered that texts whose author’s style was full of figurative language were more difficult for machine translation, while “texts that emphasized action over language style are less challenging.”

The post-editese features found in the machine translated output might warp the author’s style of writing which could influence the final product of the post-edited texts.

Their work found that the main characteristic of post-edited literary texts was their similarity to machine translation output in terms of lexical density, use of pronouns and sentence length. While noting that this scenario could be blurred by the sub-genre of literary texts.

Altogether they argue that translators when post editing should be aware of the priming effect that raw machine translation can produce on lexical and syntactic choices. That the post-editise closeness to the machine translated output might warp the author’s style of writing which could influence the final product. Which is therefore a major challenge that translators must face when post-editing literary texts.

To read the full Journal Article, Click here >