Translation is usually defined as the act of transmitting the language of the source text into the language of the target text, taking into consideration cultural and linguistic differences. It gained a significant role starting from the 1940’s, especially during the Second World War. Back then, translators were highly needed to translate spying documents mainly between the U.S.A. and their most feared enemy at that time, the Soviet Union. Even after this era of conflict, the importance of translation increased in that it was needed in the field of economy. The first proposals for machine translation (MT) using computers were put forward by Warren Weaver, a researcher at the Rockefeller Foundation, in his July, 1949 memorandum. These proposals were based on information theory, successes of code breaking during the Second World War and speculation about universal underlying principles of natural language.
Machine translation may appear as a good way to save money and time. Such translators first analyse the structure of each term or phrase within the source text. They then break this structure down into elements that can be easily translated, and recompose a term of the same structure in the target language. This method may seem correct, but the quality of the translation is much poorer than a human translation, because the structure of each language is different and this is something that most machine translators don’t take into account.
In some cases, machine translators may even provide a literal, word-for-word translation, which leads to an undesirable result. Indeed, the translation, even though it may be comprehensible, will not sound natural at all, may comprise a certain number of grammatical mistakes, will not be well-structured, and may not keep the original meaning. Moreover, some words can have various possible meanings in the target language depending on the context. This is why a human analysis is necessary. A perfect understanding of the source text is essential, and machine translators don’t have this because they don’t have a global vision.
Essentially, machine translation can be used for small, non-crucial projects, for which you don’t necessarily need a perfect translation, but only a general understanding of it — for internal purposes, for example. But for important projects, which will reach a global audience, human translation is essential and much more reliable to ensure a high-quality job, and to make sure the message you want to deliver can be perfectly understood by your audience. Technology, far from replacing humans, is instead a tool that helps them keep up with a surging demand for high-quality translation. “Translation memory” (TM) was the first significant useful tool, and “machine translation” was the next step. Computers learn from huge databases of already-translated text to make ever-better guesses about how to render whole chunks from one language into another. For example, machine translation tools like Google Translate, Bing Translator and Yandex Translate are on the rise – and they are getting smarter.
Even so, any attempt to totally replace human translation by machine translation would certainly face failure due to a simple reason: there is no machine translation that is capable of interpretation. For instance, it is only the human translator who is able to interpret certain cultural components which may exist in the source text and which cannot be translated in equivalent terms, which is what automatic translation does, into the target language. In addition, it is widely agreed upon that one of the most difficult tasks in the act of translation is keeping the same effect left by the source text in the target text. Automatic translation, in this regard, has proved its weakness most of the time, when compared with a human translation. The human translator is the only subject in a position to understand the different cultural, linguistic and semantic factors contributing to leaving the same effect in the target text.
The differences between MT and human translation are obvious. For example, the following were translated
– from Romanian into English:
Romanian original: “A se uita ca vițelul la poarta nouă.”
Human translation: “To be perplexed.”
Google translation: “A bull at the gate looked like new.”
– from Bulgarian into English:
Bulgarian original: “АКОЩЕЕГАРГА, РОШАВАДАЕ”
Human translation: “In for a penny, in for a pound.”
Google translation: “If you will be Gargano shaggy be”
ImTranslator translation: “If it’s a crowd, it’s a mess”
It is an undeniable fact that automatic translation is regarded as a tool for quickly producing a great number of translated texts. Nevertheless, the quality of such translation is still much debatable. An automatic translation, for instance, cannot usually provide a definite translation for words that bear different vowelized forms such as the Arabic term /kotob/ which in English means ‘books’. In many translation programs, when translating from Arabic into English, the term is confused with the other Arabic term /kataba/, which corresponds in English to the verb write. No human translator would make the same mistake due to their ability to read words with different diacritic marks or vowels. In some cases, automatic translation cannot even provide equivalent terms in the target language, leaving them as they are in the source text.