With the evolution of computer language and fields such as artificial intelligence, machine translation became popular and widely used, being seen by some as a quick and inexpensive solution for translation demands. However, it is really just a tool that should only be used to obtain a general idea of a topic, especially helping individuals who are inexperienced in the target language. Its use also requires caution, as it can induce the user to misinterpretations, translating just the words without considering the context, and leading to ambiguity and inconsistency throughout the text.
According to the European Association for Machine Translation, machine translation – or automatic translation – is a mode where translations are entirely performed and automated by a computer system. It should not be confused with other types of translation systems such as Machine-Aided Human Translation (MAHT) and Human-Aided Machine Translation (HAMT), which may involve the use of Computer-Aided Translation (CAT) Tools. These tools can help with the process of post-editing an automatic translation, by conducting searches in dictionaries and allowing the use of glossaries and translation memories. However, they still require the services of a professional translator.
There are two types of machine translation: direct – which translates a language pair using a unidirectional approach that is very literal and somewhat limited, and indirect – in which the analysis of the source and target languages is independent. In either case, interference, revision and handling by a translator are essential.
A study based on a survey conducted with students of the Federal University of Santa Catarina presents the pitfalls of using these tools without the application of subsequent revision by an expert translator. The students were presented a text in Portuguese addressing a familiar subject – in this case, the city of Florianópolis, Santa Catarina – and its versions translated by an automatic translation tool. Note some of the results obtained:
Original text: Floripa agrada vários tipos de turistas.
Automated translation/English: Floripa like various types of tourists
Automated translation/Spanish: Floripa, como diversos tipos de turistas.
Original text: Lagoas, morros verdes e pelo povo hospitaleiro, o famoso Manezinho da Ilha
Automated translation/English: Lakes, green hills and the friendly people, the famous Manny Island.
Automated translation/Spanish: lagos, colinas verdes y la gente amistosa, la famosa Isla de Manny.
In these examples, errors involving cultural localization and semantics were observed, elements which cannot be perceived by the machine in an automatic translation process, since its ability to interpret the text cannot be compared to that of an expert translator, who must know the source language and the subject being discussed, and who conducts a careful research to obtain enough information in order to ensure that the message is fully conveyed in the target language.
As a promoter of multilingual communication, ZAUM invests in the most modern translation tools, encouraging its translators to learn them and always seek the latest information on the subject. However, it is crucial to point out that translation quality is not based on a word-for-word translation. In order to ensure its quality, the translation must be conducted by a professional team of project managers, proofreaders and, most importantly, specialized translators.