What actually is lost in machine translation?

Ever since its release, “Lost in Translation” movie never ceases to encourage people to talk about the best of their lost in translation moments. Language is quicksilver and slippery—the meaning of its words cannot be fixed, and the grammar rules cannot be generalized. And even if one tries to do, the results are either drastic or hilarious. You may have come across the ambiguous statement “Dogs must be carried on an escalator” that linguaphiles could not stop ranting about on social media. The statement is not grammatically incorrect but is grammatically ambiguous leading to so many interpretations that the writer’s actual intended meaning seems lost.
This is just one example. Most of the similar glitches happen in the world of translation where the intended meaning of writers and speakers remains vulnerable. Working with several languages, it becomes a challenge for translators to maintain the integrity of the work, thoughts, and objectives of the writer. And nowadays, newbie authors have started using automatic/machine translators for preparing the first draft of the translated book. Where serving justice to the source book through proper translation is a Herculean task for human translators, how come the machine translators can get even an inch closer to precision?
Machine translation may fuel up your productivity, save some considerable bucks, but will not improve quality. For this, a professional and veteran translator is more a need than choice. To explain it a little more, let’s see what all is lost in machine translation:
• The tenor of idioms/charades
The brain-teasers in words are way too much for translation memory software and tools. They cannot precisely translate idiomatic phrases and pun intended. Even A.I. and M.L. powered software cannot figure out the intended meaning in the source language and convey its real essence in other target languages as this task requires contextual understanding. For instance, machine translation of English idiom “a stitch in times saves nine” often results in “time bite at nine saves,” in Croatian— a complete disaster!
• Brand salience and recall
In business communication, blunders that machine translation creates are beyond explanation. A multinational firm that has to communicate with stakeholders of a different nationality, translation of content in foreign languages is a must. Many a time, companies are founded with a brand name in a particular language and may need to translate the same into multiple languages so that people of other speech community can understand and recognize the brand. And, software are terrible at translating a brand name. For example, the French global alliance for internal packers and movers, Fédération Internationale des Déménageurs Internationaux.
(FIDI) is translated by machines in English as International Federation of International Removals, where it should be International Federation of International Movers. You see, incorrect replacement of just one word can change the face of a brand. To keep such crisis at bay, human translators use their wit and intelligence which they developed through learning and evolving. On the flip side, the imitation of human intelligence is retrofitted in the translation software that has its own repercussions just as we discussed above.
Technologists are doing their best to teach software and tools how to learn, behave, and work like us, but the nature of human language is and will always take over the binary language of O and 1.