Machine Translation’s Role
in Literary Translations
By Angela Gabrielle Fabunan
Imagine you’re a poet in Southern India back in the 5th century and you just finished a poem in Tamil. You want others to appreciate your masterpiece, and so you translate it into Sanskrit or maybe Bengali.
Can the translation be true to the original? Can it convey the same emotions and message that the Tamil original has? Probably, yes. A scholar living in 5th-century India might be able to translate a poem effectively but what about a modern AI? Can it effectively translate a literary work filled with figures of speech into another language? Those questions have been bothering writers, poets, and translators lately. This article will look into the role of machine translation when it comes to translating literary works and how I view it as a translator.
Machine translation today
Before we go further about literary translation and my experiences as a translator, we need to look into machine translation.
Machine translation is the process of using software to translate text from one language to another. There are different types of machine translation in use. Rule-based machine translation, for example, is built on linguistic rules of the two languages involved. If the text is in Spanish, for example, and it is being translated into English, rule-based translation will rely on the linguistic rules of the two languages to create the new text.
Then there is the statistical machine translation. This technology relies on the statistical analysis of bodies of work in both languages used. There are other types of machine translation, and all are constantly evolving and undergoing fast development.
The version of machine translation that most people are familiar with is Google Translate. This technology is being updated quickly, and they just recently added 24 new languages to the system. The additions include indigenous American languages. That news illustrates the fast pace at which machine translation is being advanced. It’s not surprising that soon it can do so much more.
But at this stage of the machine translation development, can it be used reliably for machine translation?
What makes literary translations different from other forms of translation
There are several things that set literary translation apart from other types of translation. The first one than will come to mind is the length. Translating a novel, for example, takes a lot more work than translating a birth certificate. Not all literary translations are like that, however. A poem can be just a single page, after all.
But then there’s the question of staying true to the original and then having the ability to convey the same message in the target language. A good example of this is Erich Maria Remarque’s bestselling novel, known in the English-speaking world as All Quiet on the Western Front. Its title in the German original was Im Westen nichts Neues which a directly translated would be Nothing New in the West.
While the direct translation actually works on some level as it conveys the sense of irony that was intended by the author, All Quiet on the Western Front does that even better. It is more cynical in how it conveys since it is on a supposedly “quiet” day that the author and so many others died in the trenches of France and Belgium during World War One.
All Quiet on the Western Front also became an expression. Can you picture Nothing New in the West becoming an expression too? Not really. It doesn’t have that ring.
Now consider translating a document. The aim here is to just be as faithful to the original as possible. Another challenge of translating literature has to do with names. Translators are faced with the choice of whether they should translate names that might seem off or retain them which in some cases will make no sense but they can decide depending on the situation. AIs doing machine translation might not be able to discern when it is best translate names and when not to.
My translator experience
I have been handling translating literary and creative pieces and overseeing content channels for Tomedes, a translation agency founded in 2007. Our translation agency handles different types of translation and language projects, which means I can compare document translation and literary translation.
Literary translation is both easy and difficult. At least that’s how I find it. I find it easy because it gives me some level of freedom that’s not available in other forms of translation. But at the same time, it can be difficult, because while I do have freedom, it’s limited. I have to subordinate my voice to that of the original author. As a poet or a writer, that’s not easy. I have to walk a fine line between giving the piece my own twist and letting the original author speak through me.
Then there’s the question of style. There are cases when doing a direct translation will ruin the literary merit of the piece. A good example is when I translated Jane Commane’s poem, Midlands Kids into Filipino for Harana Poetry Magazine. The first line of the poem read:
“We were raised in cars,
grew up on the back seats.”
Using machine translation, the translation would read:
“Pinalaki kami sa mga kotse,
lumaki sa mga upuan sa likod.”
While it is accurate as a direct translation, it loses its potency as a poem because the translation for the words “raised” and “grew,” “pinalaki” and “lumaki” have the same root word “laki”. So reading it in Filipino would sound repetitive. While repetition is a style used in poetry, it is not enhancing the translation in this case.
So I translated the line this way:
“Pinalaki kami sa mga kotse,
sa upuan sa likuran”
I omitted the counterpart of the word “grew” since the word “pinalaki” carried over to the second part of the line.
The role of machines in literary translation
Seeing this, do I think that machine translation can perform literary translation on their own? Maybe. We cannot underestimate what can be achieved by technology. What I’m seeing though is that machines or AI are still a long way from doing literary translation on their own. Maybe with the help of a translation agency that offers machine translation post editing, machines can be useful for working on literary pieces but full independence is still a long way off.
Angela Gabrielle Fabunan is the content manager for Tomedes, a leading international translation agency handling B2B-client language projects. She oversees content channels of Tomedes, including multilingual content about MT post-editing.
Introduction and background After attending GALA 2022 in April, I began to wonder if there was a gap in the way different stakeholder groups within…→ Continue Reading
Subscribe to stay updated between magazine issues.