Robots, translation and metaphor

We are living in times of constant change and mechanisation of every possible area of life. Books are replaced by Kindles, organs by 3D-prints, vacuum cleaners by robots and cashiers by automatic cash desks. Do translators need to worry?

Recently I had a conversation with a professor dealing with Artificial Intelligence that I cannot get off of my mind. It was certainly a queer encounter – a young Jane-Austen admirer adoring letters and old books that make you sneeze of dust when you approach them and an elderly professor adoring Artificial Intelligence and everything it provides humanity with. What began as a superficial pro- and con-discussion, quickly evolved into a consideration of the aspects of language that make it truly human and that might be irreplacable by any machine.

Machines have a seemingly unlimited capacity to learn new words, phrases and even whole contexts. In the near future, there will be robots teaching classes, nursing seniors and conducting consultations. But will there be robots completely replacing translators? I bet not. I bet not, because I want to believe in a human language that is resistant to technological progress. Facebook, Twitter and chatrooms actually lead to an approximation of our everyday language to a language that might be used by machines without larger barriers. Nevertheless, even though we are breaking down language to a minimum through these ways of communicating, we are still using aspects of language that, I believe, a machine can never be able to recognize.

There is at least one prominent stylistic device that might be truly resistant to mechanisation: metaphor. A metaphor is the transfer of characteristics of one area of reality to another. When we use metaphors like „He rolled the dice“ (in the sense of „take a chance“), we are giving away a lot of additional information about our mental structure, about our culture and our attitude. This metaphor would not be used in a culture where gambling is forbidden, it could be meant ironically and in some contexts it might as well be a literal statement. Well, one could say that one day machines will have such a huge pool of contexts, cultural information and nuances of meaning at hand that they will pick the right meanings for common metaphorical expressions and idioms. One thing that will be impossible for machines to grasp are creative metaphors. Imagine a conversation between a robot and a human:

„My dog really has a hard time deciding whether to destroy the pillow or the ball.“

„H-e w-i-l-l m-a-k-e u-p h-i-s m-i-n-d.“

„Well, on the one paw he really hates that pillow and the flowers on it. On the other paw, the ball should be stopped from rolling away from him.“

„---„

A robot cannot recognize the transfer of the phrase „on the one hand…on the other hand“ to the animal world and will not be able to make any sense of what he heard. A human being would probably answer with another humorous comment, or at least understand what was said. And there is another aspect that would be lost through robot communication – humour. How should a machine understand irony and comments that are not literal and how should a machine even produce something like that? And further, what would a life without humour and unpredictable conversations be?

Humour, irony, metaphor – all these non-literal uses of language are tools for expressing human emotions and attitudes. A robot can be programmed to sort and recognize these, but he can never truly understand and reenact them. In order to translate texts and utterances, every nuance and facet of the background of the message must be understood and felt. Even human translators are good translators only if they have a sense of style and fine-grained differences and if they are acquainted with the culture of the speaker/writer.

Culture-specific concepts are, even amongst human interlocutors, a very sensitive issue. Some are universal – happiness is in all cultures seen as an upwards movement. When it comes to the center of emotion in our body, things look different. In the Iranian culture, the liver is the seat of emotion – we don’t „fall in love“ and feel this in our heart, rather we „fall liver onto someone“. A machine would, thus, need to be able to 1. recognize the culture the interlocutor comes from and 2. know all cultural concepts manifested in the language.

Besides that, there are plays and literary pieces to be translated, including names and sceneries that need to be adapted to the target culture. A machine cannot have the sensitivity for finding the adequate equivalents to these expressions in a target language.

All in all, it seems impossible to replace human translators. If they do not implant a real brain in a robot and a heart (or a liver) for the reception of emotional waves, it will never be possible for machines to fully replace a human language specialist. It seems as if some aspects of humanity will just stay human.