Skip to content

Do Google Translate`s New Space Age Features Render Language Learning Useless?

I`m an EFL instructor at the university level in Japan, which involves marking a lot of essays. A surprising percentage of the essays submitted by first year students are near-total gibberish. I write “Google Translate?” on the essays and hand them back with no score. The students are always stunned. How did he know I used Google Translate?

Google Translate is amazing

Google Translate has been an impressive tool for quite a while, but this week Google announced two enhanced features that have made our jaws drop.

The first feature is the ability to scan signs or text in another language and translate it in real time. That means you just aim the camera at the sign or text, and it will be translated without having to take a photo of it. Perviously you had to take a photo of the text, then use your finger to highlight the words you wanted to translate. The new enhancement will make scanning faster, and just make us say “Wow” a bit more loudly.

The second feature enhancement allows you to interpret between two languages, meaning that the app can detect which of the two languages is being spoken and translate back and forth. Previously, you could only interpret in one direction at a time, so this new enhancement will increase the level of interactivity and the speed of it. Again, it makes me say “Wow”, quite loudly.

A danger to language learners?

Will the continuous progress of Google Translate devalidate all of the hard work that language learners put into their studies and practice? Will it devalue and make our skills less appreciated? Will it reduce the amount of joy we get from language learning?

To all of these questions I would emphatically answer “no”. Google Translate is amazing because (a) you can translate between so many different languages, (b) it`s automated and instant. But humans are far more amazing than Google Translate, and without a real human negotiating between the two languages, automated translation tools will always be very, very limited.

Human beings are more amazing

When human beings speak a language and communicate, we are constantly making judgements that help us create meaning.

  • the human brain is optimized for the subtleties of language learningWe check words against the context of the communication, because words mean different things in different situations. Automated tools handle this terribly.I tried using Google Translate with Japanese, and I spoke a simple sentence “doumo Yuki desu, yoroshiku onegaishimasu” which I would understand as “Hi, my name`s Yuki. Nice to meet you”. What did Google Translate spit out? “Thank you very much, it is snow.”  Yuki is a very common Japanese name, but also a homonym of the word for snow. Japanese is full of homonyms which confuse the hell out of Google Translate. But do Japanese people have any trouble with homonyms? Not really, because each sound is connected to a number of different Kanji pictographs, and when they hear a sound they check it against the context, then retrieve one of the Kanji pictographs in their minds – the one that best matches the context. That`s how the know the meaning of the words that sound the same. What does Google Translate do? It just spits one out. Why? Because it`s not human, therefore it`s much less amazing.
  • We understand nuance by examining word choice and judging how it reflects unspoken ideas/feelings of the speaker. Automated translation is always less than precise with vocabulary, so even if the core literal meaning gets communicated, the nuance will likely be unclear. Nuance can mean the difference between a compliment and a subtly disparaging remark, or the difference between a benefit and a drawback.
  • We judge how much we need to communicate based on how much shared knowledge there is between ourselves and the other speaker. Japanese culture for example is a very high context culture, meaning that the people share a great deal of common ideas and assumptions. So when they communicate, they often leave things out because they assume that you know. When you study the language and culture you get used to understanding things that are understood or implied but not spoken. An automated interpreting tool has no chance of filling in these gaps for you.
  • Humans constantly judge what is most important to pay attention to in order to understand meaning. People are bombarded by countless pieces of information every second, and we are natural experts at deleting any information that is not useful and focusing on what is most important. We do this with language, too. We pay attention to people`s vocal tone and volume, to the context, judge which words and ideas are the most important and construct the meaning in our minds. Human interpreters don`t translate word for word, they listen and construct a picture, then they recreate the same meaning in the other language. They ignore certain elements from the input language. Google translate tries to translate every element of the input language without bias, which results in some funny and nonsensical outputs.
  • People can understand a wide variety of dialects, accents, and unusual voices. We have a pretty amazing ability to listen to people producing fairly different language than we ourselves would use, but still understand it. This is related to the above point, because part of what we do is listen for what is similar to our own speech patterns and ignore the differences. Then with more exposure we learn the patterns behind the differences. And if we don`t understand, we`re able to do something translation apps can`t do: clarify what the person means. Google Translate has a lot of trouble with regional accents, let alone dialects with differences in vocabulary and grammar. That`s because the app doesn`t have the ability to judge approximations, to make decisions  on what information to delete, and to understand the patterns underlying differences.

To be able to do what humans do so naturally, Google Translate will have to go deep down the road of artificial intelligence over a long period of time. Even then, I don`t know if it`s possible. They are competing with Mother Nature`s greatest creation: the supercomputer known as the human brain.

Google Translate Japanese to English example

Some languages are beyond the scope of what Google Translate can do

I`ve used Japanese as an example mostly because Japanese is a very different language than English (and most other languages), so it is easy to find the limitations of Google Translate using Japanese.  Directly translating between languages that have very different syntax always results in a lot of mucky sentences.

This is probably why so far, the only languages which Google has introduced the new features for are English <–> French, German, Italian, Portuguese, Spanish, and Russian. Aside from Russian, the other languages are ones that are fairly straightforward to translate to and from English.

 People appreciate the human touch

Will my language skills become less valuable on my resume?

I don`t think so. People value human interaction, and in today`s high tech world I think it`s more valuable not less. People will get annoyed and fed up with automated translation if companies try to replace humans with machines.

Google Translate is one of those things that can be a value-added, if it is used in situations where it`s a pleasant surprise. For example, if you get your first ever customer from Azerbaijan and he can`t speak English. If you use the app to communicate with him it would probably make him happy that you`re making the effort to reach out to him. But imagine that 50% of your customers were from Azerbaijan and you tried to rely on Google Translate to interact with all of them. That would probably work against you, because you would be failing to meet expectations and neglecting your core customer base. You would be much better off learning some Azerbaijani or hiring someone who can speak it. Because the human touch, the ability to communicate directly with someone and share an experience with them in their language has tremendous value.  So even if automated translation becomes more commonplace in businesses, it will likely only be for functions for which there was previously no translation. It won`t likely replace employees.

Employers like smart and flexible people. Recent studies have proven that certain parts of the brain actually grow larger when you learn a second language. You also develop a flexibility of thought, an ability to understand and adapt to other ways of thinking, when you speak more than one language. In today`s economy, ideas and creation and hugely important, which requires people to think about problems from various angles. Multilingual employees can provide the valuable perspective of someone who can shift between different ways of seeing the world.

Language study is an incredibly rewarding hobby

language study books for german, spanish, italianEven if translation apps became perfect and we no longer needed to study languages, who cares – people do a lot of things that they don`t need to do. Why? Because they want to do them. Is basketball necessary? No, but some people love it and dedicate lots of time to do it.

Some of us love studying languages. We love to learn the patterns and crack the code, we love to communicate with amazing people from all over the world in an authentic human way, we love to master ourselves and put plans in place to achieve goals, and we love to gain global insight by delving into other cultures and their ways of thinking. Learning languages is fun and fascinating, and it`s a deeply rich hobby that is not going away anytime soon. As in, ever.

That`s the message I try to communicate to my first year students. If they hand me a gibberish essay courtesy of Google Translate, I know that they haven`t yet caught the language bug. They haven`t yet discovered the fun and fascination, the rush of overcoming challenges. The first step for them is learning that quick fixes will never be a substitute for the real thing.


  • LangFocus

    I`ve also tried Google Translate with Tagalog/Filipino since I started studying it a few days ago, and Google Translate between English and Tagalog/Filipino is just as messed up as English Japanese translation. Lots of concepts simply can`t be translated and need to be understood then reconstructed in the other language.

  • Paul

    Always a good exercise to make a sentence in your own language, auto-translate it to some other language, and then translate it back.

    Or go for 3 or more stops. Enjoy the gibberish!

  • MartinRz

    Very nice text. I’ve noticed that people rely too much on the G-translator, especially when posting on the internet. This exercise with translating the sentence in both directions is very good, but I’d like to ask if you tried another one: translation between two languages which have little or no similarity to English. It gives interesting effects. It happened to me that I helped a few times my friends on one of social networks to correct mistakes in translations between romance and slavic languages. I’m not 100% sure, but I came to an idea that the algorythm translates everything to English and then to destination language, what makes twice as much trouble as simple to- or from- English translation 😉
    Of course one source of mistakes in this case is that we are loosing the context during the first translation and it’s impossible to guess it during the second one. The other one is the language logic – for example double negation which works strictly mathematical way in English or German, in Slavic and Romance group means rather strong negation – when I say (in exact words) ‘I don’t do nothing’ in Slavic or Romance language (at least these few I know from both groups) it means ‘I don’t do anything’ in English. The translator sometimes does the work properly (sentences corrected by users) and sometimes not (and one of negations dissapeares). In some cases the final sentence is the negation of original one. One of my Polish friends wrote that he had no time to write, and the result was that he had a time.
    The other causes are the words which are unisex in English and have different male and female forms in other languages (e.g. ‘friend’) – our female friends are converted to males when using Google Translator 😉
    The most embarassing one I remember: One of my friends from Slovakia (male) was doing some renovation at home. And he wanted to write to our Spanish friend (female), that he had no time to respond to her post earlier because the whole night he was busy in the kitchen… I have no idea what phrase he used bur the result was the verb ‘masturbar’ (‘whole night in the kitchen’ has been translated properly)… The girl’s answer was ‘:-O’

    Sorry for mistakes in my English (no translating tools used). You’re doing good work (here and on Youtube) 🙂

  • LangFocus

    There are some funny videos on Youtube doing that. I remember one guy rapped the “Fresh Prince of Bel Air” theme after running the words through Google Translate several times, and it got weirder and weirder each time. lol

  • TalkClip

    Hi Paul, Thanks for sharing your wide experiences in learning languages and your videos has inspired me to learn more. And these days, learning languages is easier because the Internet provides us with so much more material. With messaging and social media, we use our smart phones more and more. And Google Translate is amazing and I use the app very often.
    I like to share with you my experience with Google Translate and combining it with speech recognition technology. Speech recognition is an amazing technology. But when you are using many different languages with different people on social media, the current apps are clumsy in switching between different languages. Out of that frustration, I wrote an Android app “TalkClip” that aims to make speech recognition easier to switch languages on the smart phone. I wonder if you and other readers, who often switch between languages, encounter the same difficulty? And if so, how do you cope? And if you do have suggestions for me, I can improve my programming too. Thanks.