A Machine Will Never Replace A Human Being?

“In the translation industry, a machine will never fully replace a human being”, claims Timea Balajcza, the founder of Balajcza Linguistic Services, a translation firm based in Warsaw. Her claims are based on years of market experience: the company which was set up in 2010 currently provides translations for over 250 language combinations and cooperates with a global network of 1500 translators. However, this does not mean that Balajcza Linguistic Services does not tap into opportunities offered by modern technologies.

“We use two different CAT programs, but mainly SDL Trados”, says Timea Balajcza. CAT stands for “computer-assisted translation”, whereas SDL Trados is professional software, valued on the market. Benefits of using CAT programs are obvious to the company founder. She reels off a list: “increased speed of translation, optimisation of translation quality, ensuring consistency of terminology”. Basically, to work with such a tool is as follows: files sent in by the client are uploaded into the program which the translator is logged into. On the basis of a database, the so-called translation memory, the program generates prompts which the translator may use or ignore. Each translation prepared using the software or imported into it is added to the memory. It should be mentioned that the software producer, through a cloud solution, offers access to combined translation memory to all users. However, due to data security considerations, Balajcza Linguistic Services exclusively uses a database built on the basis of its own translations. In the case of a company such as Balajcza, that has been operating on the market for eight years and has been translating over 2.5 million words each year, the company’s own translation memory is a significant resource and considerably speeds up the completion of some orders. “Particularly in the case of repeated orders from the same client, where we deal with specific specialist terminology, having such a database is priceless”, emphasises Timea Balajcza.

The possibility of using automatically generated prompts is not the only function of the software that increases the company’s effectiveness. The technology used enables several user-translators to access the same product at the same time. “Our translators often work simultaneously on various parts of the same project. The results of their work are then put together, and the cohesion of the translation is maintained”, explains the company founder. This function is particularly useful in the case of a more extensive and complex project.

The Balajcza translation firm decided to use a server-based version of the software because it offers the highest data security and confidentiality among all solutions currently available on the market. Cloud solutions or any free tools are no match for it in this respect. The server on which documents of Balajcza’s clients are stored is under the company’s direct control. Translators employed by Balajcza, and also external ones with which the company cooperates, have access to the program. Thus, there is no need to transfer documents using traditional data transfer channels, such as e-mail, which do not guarantee security. “Our translations include legal, marketing and financial texts. I can’t imagine that a professional translation firm could use free CAT software if you consider current data confidentiality requirements and high contractual penalties.”, says Timea Balajcza. Not all clients realise, however, the level of risk involved. “I know cases where the use of such free software by a company ended in the texts that were sent for translation leaking out into the internet”, she recalls.

Although both terms are used interchangeably in the colloquial language, computer assisted translation is not the same as machine translation. The difference consists in the degree of human and machine – or computer software – intervention in the process of translation of a given text. With CAT software, the computer program only aids the translators in their work, increasing their efficiency. The machine translation, on the other hand, takes place without human participation, exclusively with the use of algorithms. One of the best known examples of machine translations at the moment is the free Google Translate services which, according to Google’s own information, “quickly translates words, expressions and websites from Polish into over 100 other languages and vice versa”. Tools of this type are favoured by users of mobile devices and work well in simple, everyday communication, especially in more popular language combinations. The Balajcza translation firm, following market expectations, recently introduced slightly cheaper machine translations into its offer, combined with the proofreading by a translator. The scope of use of this type of translations, however, is very limited: they take place only at the client’s explicit request and do not include confidential documents, texts intended for publication or for other important purposes. Balajcza provides machine translations on the basis of a paid program that is more advanced than Google Translate. Although not only private entities but also the public sector, for example the European Commission, invest in the development of machine translations, their quality is still inferior to that provided by humans. The European Commission’s page, in the section extolling benefits of the CEF eTranslation machine translation service, we can read: “Supports the work of translators, reducing the burden of routine translation and enabling them to focus on very specific or important sections of documents”. In this case, Google Translate provides a rather good translation into Polish: “Wspiera pracę tłumaczy, zmniejszając obciążenie związane z rutynowym tłumaczeniem i umożliwiając im skoncentrowanie się na bardzo konkretnych lub ważnych sekcjach dokumentów”. The founder of the Balajcza translation firm, Timea Balajcza, entertains no illusions as to the limitations of machine translations when live, far-from-routine business language is concerned: “I cannot imagine that texts such as restaurant menus or marketing releases, in which intricacies of the language and the tone of the message are of importance, could ever be entrusted to a machine”. Time will tell whether she is right.