Can machine translation replace humans?
Updated: Jan 18
Google translate is a handy and valuable translation technology. Using it became a daily routine for many. Most of us use it to translate a word or phrase. However also, many people use it for work purposes across various industries.
In many cases, it only serves to give you an idea of context, and often, that is enough. The translation can be literal and sometimes inaccurate.
The problem arises when this machine translation is used for a more serious purpose. In those situations, you may find yourself with significant mistakes.
The misinterpretation of the Google translator has become very famous. The one occurred recently, and has affected the new official tourism page of the Santander City Council.
Everything was prepared for its presentation in style at the International Tourism Fair (FITUR) currently being held. And it has not been able to premiere with more publicity, although, in this case, negative. "Let them talk about you, even if it's bad," Oscar Wilde said. Better not.
The organization decided to use the Google translator "temporarily" to save costs. Among other pearls, the Botín Center became The Loot Center or Pillage Center in Google Translate, while the Historic Helmet was translated as a helmet, yes, but for the head. The result published on their website was ridiculous. It has raised complaints and discontent from the translation industry.
“However, for example, Airbnb has been providing machine translated listings for years. They added a clause that the listing is translated by machine and users accepted the possible embarrassing errors.”
Relying only on machine translation without proofreading is still not possible in most cases. This does not mean that you need a professional translator. Many teams and companies are able to proofread the machine translation in-house. Talking to people in various industries we see this as a growing trend coming with increased efficiency within translation and localization process.
There various ways to proofread the instant translation:
Copy and paste to Notes/Docs/MS Word, make necessary changes yourself
MUST TRY! If you need translations into more than one foreign language, then we suggest trying TranslateWise, where teams and translators can work together, instantly proofread and post-edit machine translation.
Plus, update already translated content across all languages simultaneously.
Here’s how TranslateWise looks like:
Putting ourselves at the level of a machine that does not understand different cultures is confusing, but the play has gone wrong given what we have seen.
Therefore, if what you need is a "real" translation to carry out or present for a job or expand your reach, you'll need a proofreader." Only a person will know how to interpret the other language's expressions and use the most appropriate words for that context or sector.
How reliable is Google translator?
Indeed many have a need to know the translation of a word while carrying out exercises in a language. In these cases, the most common is to search our answer directly in the well-known Google Translate. This online tool is widespread, and it is so easy, simple, and fast. But is it reliable? Up to what point?
For those of us who have studied languages (whether in a career, in vocational training, to obtain a linguistic degree or only in ESO), it has been very convenient for us to find a translation in this way, so handy and so direct (Translation appears instantly!). However, there is always a point in our learning process where we realize that all that glitters is not gold and that this tool is not as accurate as we thought.
When does this happen? When you start to understand a language. When you translate a whole paragraph and realize that most of the expressions appear without any sense, as out of context, then you know that you have to "spin" them on your own to understand the text. Isn't this a waste of time? Some teachers realize that words do not seem to have many relationships or cohesion since we find examples of essential phrases, more advanced verbs in infinitive without conjugation, and even literal translations from Spanish.
What if. There are translation errors (galore!), And such that even a person who is in the lowest level of English, for example, can notice and distinguish. Here are some very clear examples:
If you try to translate the word "AROMA" into English, you will find yourself "TO ROME" [To Rome (you know, towards the city, no smell or fragrance or anything like it).
If you want to wish someone a happy bridge in a letter or email, you will see" HAPPY BRIDGE" and not "HAPPY LONG WEEKEND". This is one of many examples of literal translations, and Google Translate does not work very well when it comes to translating expressions.
Here comes one that brings us so many headaches. Teachers always explain that "ACTUALLY" is a false friend (words that seem like one thing and mean quite another). Most students confuse this word with their false friend (CURRENTLY), when it means "REALLY", "IN FACT ..." Can you guess what appears in Google Translate when we search for the translation of "ACTUALLY" into Spanish? Exactly, "CURRENTLY"! A big mistake where they are… Be careful with that!
We like to make others aware that the abusive use of this apparently simple and straightforward tool is not recommended. Yes, it is true that it can be used to resolve a specific question, that is true. However, it loses effectiveness when we try to translate more professional or only longer texts. We should not abuse this tool, and as it has been proven and it is widely known that there is NO professional linguistic support behind it; therefore, it is NOT entirely reliable.
How does a machine translation engine work?
If we claim that we have all used Google Translate, we are not generalizing. The platform lives in the process of continuous improvement, and this has been demonstrated in recent years. It is an indispensable tool that has gotten us out of trouble, but how does this technology work?
Currently, Google Translate works with 103 different languages and is used by more than 200 million people daily. And, even though it was released in 2007 and has been with us for more than ten years, we have been surprised by its operation. Neither linguists nor philologists, we reveal to you what is behind the scenes of Google Translate.
Dive into translated texts
Managing 100 languages must be a really exhausting task for Google programmers. Well, it is simpler than it seems a priori if the translation is achieved through automatisms. Google Translate works with machine learning through examples. It is explained by Peter Norvig, Director of Research at Google, in one of its blogs.
In the early days of the tool, the programmers made an effort to ensure that the technology understood every word, considering each language's complexity. What happened is that they found a job impossible to materialize—millions of words, billions of combinations, and different rules for each language.
The solution was to prepare the translator to immerse itself in the web. The programmers thought that the best way to find patterns and examples was to locate them in the millions of already translated texts available on the web. In this way, the translator could focus on complete sentences and consider the context, not treating each word individually.
The challenge for voice translators: they go one step further
And if the implementation of a functional translator has been quite a journey, future challenges are not short. Google now sets its sights on achieving instant and viable voice translation through artificial intelligence. Pulling AI seems the only solution to finish off a technology that cries out for perfection. Although it is already feasible to use this tool, it does not achieve as accurate precision as Google Translate.
Cultural traits end up completely interfering with how a sentence is translated. And not only that, intonation or intention also alters the meaning of words. The idiomatic qualities directly infer the translation result and make it difficult for the Google translator algorithm to date.
“Google translator continues on the podium of the most chosen.”
The google translator does not stop incorporating cutting-edge technology, which makes it the most chosen worldwide when translating languages on your website or mobile application using text, voice, or even images.
It has around 100 languages and is the first source of reference for understanding what words mean in another language, thanks to its efficiency and fidelity.
An interesting detail is that google translate feeds on the activity that users carry out on the platform. In other words, the faithful interpretation is due to the database generated by the millions of translations requested by users who use it every day.
In recent years, Google has incorporated the best in terms of Artificial Intelligence for both searches and translations, such as voice recognition.
Some time ago, it incorporated the possibility of making transcriptions in real-time, allowing users to speak in a specific language and request the application to reproduce it in another language chosen at that moment. It is undoubtedly a significant advance that will enhance its quality of occupying the podium among 2020 online translators.
Although this technological advance mentioned is still a prototype, it is estimated that it will be able to be executed from any mobile device and enjoyed free of charge to carry out all kinds of translations in the coming months.
Today, Google translate facilitates the translation of writing, speech and has a conversation mode. It also has the option to activate your camera so you can take a picture of a poster, restaurant menu, or whatever comes to mind while walking, to be translated quickly—an ideal function for travel lovers who want to know other languages.
There is no doubt that an online translator can facilitate many situations: not only is it recommended to carry out all kinds of transcriptions, but you can also learn to speak another language effectively and efficiently. Which one will you choose to start your learning in a different language?
Machines updated to be more accurate
You might have noticed that the quality of the translation leaves much to be desired on some occasions, mostly if the translation is being done from a colloquially written text. Actually, on the internet, we do not usually find many web pages that write correctly. When we pass them through the translator, it is seldom almost impossible to understand the meaning. Google is aware of this problem that affects the quality of translations and has been working to solve the problem, optimising the application and web service to work more correctly.
Google claims that it has just introduced a new technology that enables machine translations called Neural. This new automatic system takes context into account when translating, instead of literally translating individual words as it did before. Logically, this new way of carrying out translations positively affects translations' efficiency, reducing the number of disparities and inconsistencies when we use it.
According to the company itself, the Google translator is now 30% more accurate both in iOS and Android applications and through the web service.
Neural's system is currently only available in nine languages, including Spanish, English, French, German, Chinese, Portuguese, Korean, Japanese, and Turkish. The company eventually wants to offer this new service in 103 languages currently supported by the application.
We keep monitoring the progress of machine translation technologies and learn new tips to enable mistake-free translation experience.
Hopefully, this article will help you to navigate better in understanding machine translation.
Thanks for reading!