Is machine translation for me? 5 questions to ask before deciding to use machine translation

machine_translation

Machine translation (MT) is, according to its definition in Cambridge Dictionary, the process of changing text from one language into another using a computer. It is also known as automated translation, given that there is no human translator involved –at least in the first draft– and the translation is produced automatically. But which questions should be asked before deciding to use machine translation?

1. What is the content about?

The content that needs to be translated is of utmost importance when deciding to use machine translation. Technical documentation that has a lot of repetitive content or has been previously translated in large volumes is a type of text that could be suitable for machine translation. On the contrary, using MT in marketing and other types of creative content would not be suggested as subtle nuances may be lost and errors could be introduced without the assistance of a professional translator. For instance, wordplays and puns which are specifically written to catch the readers’ attention are usually not grasped by machine translation engines.

2. What is the intended use of the translation?

Your needs and the intended use of the translation are definitive factors when it comes to using machine translation or not. If the text to be translated is of high visibility, e.g. if it is addressed to existing customers or potential clients, then the outcome would not be safe to be published before prior review and editing by human professional translators. For instance, if you own a website, integrating a machine translation app would not be advisable without further linguistic review by a human translator. A similar approach could lead to customer confusion or even profit loss. Nevertheless, if the translation is intended to be used internally or just to get the gist of the original text, then MT without a review could be an option. However, this approach should be used with caution, as there is no such thing as a perfect machine translation.

3. Does quality matter?

With the exception of instances where a text is machine translated just to understand its rough meaning (as in the case of Google Translate that many of us have used for languages we don’t speak), translation quality does matter. And using machine translation does not necessarily mean that the outcome will not be of good quality. Just like in any other translation project, the first draft, which is in this case produced by a machine, needs to be reviewed by a human translator, who has good knowledge of the subject and specialization field. This post-editing step, as it is called in the translation industry, can guarantee high quality and readability of the translation. Content that has been machine translated could be published for almost any use, as long as it is guaranteed that it has been post-edited to the extent needed.

4. Is the content confidential?

When it comes to confidential documents or texts that contain sensitive information, the use of machine translation should be thoroughly examined. The reason behind this potential issue is that machine translation engines used to produce automatic translations are fed with the content that is being translated. So, the content that is being machine translated is also used to train the engines for future projects. According to the procedure used, the recipients of the translated text may suggest corrections, so that any errors may be avoided in the future and the output quality is improved in the long run. Given this, the provider of the MT engine needs to explicitly state how your content is processed, so that any involuntary sharing of sensitive information is avoided.

5. Will machine translation cost less than standard translation?

The short answer is most probably yes, but it depends. Machine translation usually speeds up the translation process and by extension its cost. However, depending on various factors such as the language pair, the content type, the complexity of the text at hand and the extent to which an engine may have been trained already, the cost of MT may vary. Factors such as the above, determine if the post-editor is going to spend more, less or the same time as doing the translation from scratch, so the cost should be affected accordingly. For higher output quality, if a proprietary MT engine is intended to be used, then any cost-saving should be examined in the long-run, as this is a process entailing many steps before it can run smoothly. In this case, machine translation should be seen more as an investment than an immediate cost-reduction approach.

Finally, it is true that the technological advancements in this field are important and the quality of MT output has been improving over the last years. What is left to do is just ask the right questions before deciding to launch a machine translation project.

If you are exploring the use of machine translation click here, we’d be more than happy to exchange ideas on the topic and share our insights and know-how.