Machine or computer-assisted translation – your choice!

Machine or computer-assisted translation – your choice!

Although their names may sound quite similar, machine translation and computer-assisted translation are two completely different things. In fact, they are used for quite opposite reasons and their results may differ significantly.

Translation systems based on Artificial Intelligence are becoming ever more popular. Occasionally, they replace a human completely, and sometimes they just make the job easier. For this reason, translation specialists are happy to use these tools. The more content is translated and post-edited, the better the quality of their output.

Fast, cheap, less accurate

Machine translation is provided automatically by software utilising linguistic algorithms, grammar rules and a repository of previously translated texts.  The main advantage of this solution is its low cost and speed, which enables the processing of large amounts of content in a short time. The quality, however, may leave a lot to be desired. If the translation is to be further used or processed, it is strongly recommended that it is checked. This would be the role of a human specialist trained in working with machine-translated content. This is what we call post-editing.

Therefore, machine translation is perfect if you only want to roughly familiarise yourself with some content, without much attention to details and stylistics. You no doubt know some free online machine translation tools such as Google Translate or Bing Translator.  For a few years now, they use AI mechanisms based on neural networks. Thanks to this, the quality of machine translation has significantly improved. It is based on a single language model, regardless of the text type and topic. Therefore, it is not recommended for businesses as it usually needs specific terminology and industry knowledge. Moreover, we need to always bear in mind, whether we agree for both source and target texts to be processed, including their sharing, permanent storage, transmitting and further processing.

For confidentiality and quality reasons, it is much better for businesses to use subject-matter related, commercial machine translation programmes. These tools are created for a single subject area, and sometimes even a specific document type or client. They use revised, area-specific translation memories, glossaries of terms and advanced algorithms.

A computer assisted… translator

Despite the fact that the quality of machine translations has increased significantly over recent years, many linguists would simply prefer to translate a text from scratch, rather than proofread automatically generated content. Among many other factors, it is crucial to simply understand the meaning behind a specific text. Moreover, a translator has to know the context, the non-standard project requirements that the client might have, the intended use of the document, specialist terminology and specifics of the industry.

Although linguists may not use MT support, they might employ other software. Computer-assisted translations are carried out using so-called CAT tools (Computer-Assisted Translation). Once translated, a given text is stored in a translation memory. The idea is to divide the text into segments, with source text on one side and target text on the other. These data are then used for translating new documents for a particular client or specific area. Basically, the idea is similar to searching for a word or phrase in a browser: you start typing and you are given some hints as to what you might be looking for by the search engine. That is what makes the work easier and faster.

However, that is not the only advantage of CAT tools. They also include various plug-ins to facilitate the translation process. These range from grammar and spelling checkers to number format verification, dictionaries or a preview of both source and target text. There is also a project management module that makes it possible to divide the assignment between two or more translators and check their progress in real time. All of the above-mentioned functions not only increase the speed of the translation workflow but also improve its quality and consistency.