Nowadays, professionals in the translation industry use a variety of resources and practical approaches when providing a particular language service. Tasks that were initially done on paper entered the era of digitization, subsequently being done on computer screens, and at the same time, specialized computer software was created, thus making it possible to import texts into editing tools. These editing tools offer us countless features that enable us to streamline localization processes and, for some time now, we have seen that machine translations can be generated based on translation corpora. Translation has always been a necessary activity and it is able to adapt and evolve to new stages of technological development in order to take advantage of new discoveries and optimize performance. In fact, today the translation profession is not only related to the mastery of languages and specific knowledge about specialized areas, but a strong working knowledge of computer software and computer-assisted translation tools is required in many jobs, especially in large language service agencies, which usually digitize most of their projects to be able to feed translation memories. In this expanse of resources, there is one that has gained notoriety and has become quite controversial in terms of its use: machine translation. Of course, this resource isn’t new to the scene and much has been written about it. Moreover, major progress has been made to improve the quality of its features. For our part, we recommend reading the article Types of machine translation, in which we explain this type of translation and what it doesn’t entail, and we briefly summarize some of its forms. Nevertheless, nowadays there doesn’t seem to be any consensus as to when its use is the most appropriate, since there are “raw” machine translation tasks and other “refined” machine translation tasks. That’s why we are going to focus on the practical applications of machine translation in today’s article.
In essence, the premise of machine translation is to offer an instant translation generated from results based on translation corpora. The results of these translations depend on many factors. Fundamentally, if it is a specialized text, but a general translation engine is used, the results produced will likely not meet the terminological expectations of a specific assignment. Likewise, if a translation engine for a specialty area different from the one to be applied is used, this same terminological suitability may be affected. Furthermore, machine translations still cannot guarantee quality terminological consistency. This is known as “raw” translation, that is, translations automatically generated and used as such, without modifications. On the other hand, the use of suitable translation engines makes it possible to produce translations that transmit certain types of information accurately and satisfactorily for specific communicative situations. So, which ones are we talking about?
“Raw” translation is instantaneous and therefore useful for when we want to immediately know interlinguistic content for clarification purposes, in other words, to have a general idea of what a text is about. In fact, there are many cases in our daily life in which we need a specific and general translation to understand textual material. For example, when we browse through different publications on social media, we usually have the option to automatically translate the captions, user comments, the title of an article, etc. This text doesn’t need to be accurately translated for later use, but it serves the purpose of quickly providing a translation to the interested person. The translation fulfills its function and is then “abandoned”. Another similar case is when we receive a message in another language and we want to know its content in order to determine whether it’s relevant or not. This may include emails and communications between companies that speak different languages and wish to advertise their services; nevertheless, this is an increasingly uncommon trend and correctly localizing textual content is usually chosen. One area in which these “raw” machine translations are also quite common is that of product reviews. Shopping portals usually incorporate this function so that their users know what others think about their products, in order to encourage them to buy. These translations, however, may not always be entirely useful, since products can be very technical and, as mentioned earlier, many translation engines are not prepared to provide functional translations. For this reason, machine translation, understood as a resource to create quality, functional communication between two cultures, must be accompanied by a post-editing service. This service involves the work of a translation professional who is responsible for reviewing the automatically generated translation and making the appropriate changes so that the final result is as close as possible to a translation that started from scratch. To do this, the translator will focus on aspects such as accuracy (all the material has been translated), correct terminology (use of the appropriate terms within the specialty area) and consistency (a word or expression in the source has the same translation in the translated version, whenever necessary). In doing so, translators can get the most out of this resource and streamline and optimize their performance by incorporating it into their standard procedures. Post-editing, despite not being effective in the most creative areas, is extremely useful in technical specialties or in more voluminous and repetitive projects. For more information, we recommend reading the following articles: Advantages and disadvantages of post-editing (client’s perspective) and Advantages and disadvantages of post-editing (translator’s perspective).
Image reference:
https://www.eni.com/en-IT/digital-transformation/machine-learning-investments.html
