Four Common Misconceptions about Machine Translation

Four Common Misconceptions about Machine Translation


Since machine translation (MT) is still a relatively new technology, it remains unexplored by many. Given how much hype is being created about it by the media, it is important to know among all that excitement that not all of the information that we come across is actually true.

That is why we decided to sort the facts from fiction and address the four most common misconceptions related to MT.

1. Machine translation will eventually replace human translators

This is not only one of the oldest but also one of the most popular misconceptions. Many will say that it is also the most worrying one as well. There is lots of media coverage marvelling at how good machine translation results are becoming. Some coverage even speculating that machine translation quality is so good that human involvement will no longer be needed in the future.

But let’s not forget that machine translation is not always the answer and it does not work in all contexts.

When we think of more creative tasks such as transcreation, where wordplay, puns and double-meaning play a crucial part, it is very unlikely that machines will be able to reach that level of sensitivity and awareness. There’s also the matter of taking cultural nuances into account when transcreating content  – something that we currently have no means of training machine translation engines on.

Whilst it is true that the research supporting the increase in machine translation quality is moving at an unprecedented pace, there is currently no bulletproof evidence that machines are capable of producing target language output that’s equal to human output. This Common Sense Advisory article further details the discrepancy in the quality of the human and machine translation.

It is also important to stress that language service providers themselves have expressed no intention of removing linguists from the translation process. Instead, they focus their efforts on adapting the existing tools to better suit the new ways in which linguists work. This includes integrating machine translation suggestions into translation software, and turning  linguists into language consultants, who are in full control of the final translation outcome.

2. Human translation is always a better option than machine translation

Undoubtedly, human translation comes with more assurance in terms of quality. In scenarios where project budgets are generous and timescales ample, it is certainly a preferable solution to choose human translation, or even transcreation, over machine translation or post-edited machine translation.

However, there will be times when the budget or the timeline of a project, or both of those at the same time, will be so limited that it would be impossible to execute the project using just human resources. If there are 15,000 words to translate into a language but the client can only afford to allocate two days for it, one translator will not be able to cope with such a volume.

In those cases, a reasonable solution would be to machine translate the source text and have a trained linguist post-edit the MT output.

In this scenario, the machine does the “heavy lifting” and takes care of 60-70% of the process, whilst the linguist does the rest – editing, polishing the translation and making it their own final product.

Often, offering machine translation opens new doors to language service providers and helps to start relationships with clients, that with time, might develop further and eventually include other services such as human translation, transcreation and copywriting.

3. Machine Translation tools, especially Google Translate, return nothing but hilariously wrong translations

We are all guilty of this one to a certain degree. Although the translation industry is very clearly going through a period of change in terms of the attitude towards machine translation, there is plenty of people within – and outside of the industry – who think that machine translation is always wrong.

The Internet is full of examples of horribly wrong translations produced by machines. Translation Party is a fun website powered by Google Translate that explores how many automated translation rounds it takes to get the same sentence in English and Japanese.

The reality however is that in certain circumstances machine translation is indeed our best option and without it, we would be completely unable to communicate and interact with the environment.

For instance, when we are on holiday in a country where we don’t speak the language and that language does not have any similarities with the languages that we do speak. Without some sort of automated translation available at hand, we would be at complete loss; and Google Translate has built an impressive array of features that make it easier to manage in such situations. For instance, it is possible to scan a sign, a menu, a handwritten note with your phone and Google Translate will immediately translate it into your own language.

What’s more, there are new online machine translation platforms emerging that return better and more reliable translations than Google Translate. One of them is DeepL, a portal which we talked about previously in one of our previous newsletter articles. The results it returns for the Polish language are accurate and fully understandable more often than not.

4. Machine translation is all about the man vs machine competition

It is easy to think that it is either machines winning over humanity or humanity winning over machines.

However, reality has proven that the best outcomes are possible when forces are joined and the benefits of both worlds are combined.

In the context of the language industry, this means linguists using machine translation as an aid – as a productivity tool that enables them to be more efficient in their work and process more words than they otherwise would be able to. The linguist stays in charge of the final output and while the machine produces the raw target output, it is the humans who decide how useful it is.

This often means that linguists choose to use post-edited machine translation when handling more repetitive and less creative content. That way they are able to free up more time that they then can spend on working on more engaging and creative tasks such as transcreation or copywriting.

Ultimately, it is all about promoting communication and if machines facilitate its flow in certain contexts, then it would not be fair or justified to discard machine translation as an option.

 

Related posts

Subscribe to our newsletter