Google Translate Doesn’t Work? Google and Language Learning
I often hear people complain about Google translate. “It isn’t accurate”. “It isn’t reliable” and so forth. I am a fan of Google Translate as well as Google Text to Speech(TTS) and here is why.
We live in the age, both of global connectivity, and of rapid advances in information technology. Since the age of 60, over the last 17 years, at a stage in life when, supposedly, our memory and cognitive abilities decline, I have learned or have been learning 10 languages.
The fact that I was able to make meaningful progress in languages as different as Russian, Korean, Portuguese, Romanian and Greek, as well as Arabic and Persian, is the result of modern information technology. In particular I owe a big debt of gratitude to Google, Google Translate and Google Text to Speech, both of which are a big part of my language learning. Let me explain.
Google Translate, Multi-Purpose Dictionary
Computer-based translation technology, such as Google Translate, relies on comparing massive quantities of language content between pairs of languages, in order to establish patterns, or the probability that certain words or phrases in one language will correspond to some other set of words and phrases in another language.
The accuracy of machine translation is dependent on the amount of language content compared. The greater the volume of content, the more accurate it becomes. Thus Google Translate is usually better in languages to and from English, where more such content is available, or for languages with similar word order and structure. Furthermore, the accuracy is constantly improving. In all of the languages where I use Google Translate, it has been improving by leaps and bounds. It is more accurate now than even a few years ago, as more and more corpora of language content are processed using artificial intelligence, or AI.
I sometimes hear criticisms about the accuracy of machine translation, such as Google Translate. But this criticism is either from people who expect perfection, or from people who feel machine translation threatens the job of language professionals. But I am just a humble language learner with no expectations of perfection, and am used to accepting uncertainty as I go about discovering a new language. I just love Google Translate.
Learning With Google Translate
I’m learning Arabic and Persian, at LingQ. It has been a long road, reading a new and still quite unfamiliar writing system. I can look up the meaning of individual words or phrases using Google Translate. In many cases the meaning is accurate, and where it isn’t I just move on. I keep reading in the knowledge that with enough exposure, listening and reading, the language will become clearer. The inaccuracies in Google Translate are not long term problems, but rather short term irritants.
Most often I am reading on my iPad using LingQ. LingQ offers a selection of dictionaries which provide more accurate translations than Google translate of individual words. I use these. Context Reverso is my favourite since it also offers help on verb conjugation as well as providing examples of the word in use. However, for a sense of the meaning of a phrase I need to rely on Google translate.
Google Translate is the fastest off the mark but not always the most accurate or thorough for individual words. Where Google Translate really shines, however, is for phrases and sentences. I can highlight a phrase or whole sentence and Google Translate provides an instant translation.
This is extremely useful, because very often the translation of individual words doesn’t give me a real sense of the meaning of the phrase or sentence.
Google Text to Speech.
When I highlight a word, a phrase or a sentence, I hear it pronounced for me in text-to-speech. This is valuable for all languages, but especially for Arabic and Persian, which use a writing system that is still quite unfamiliar to me. With the help of this, albeit artificial, pronunciation, I can read through material on my iPad or online that would be difficult to read in a conventional paper book. Google helps to bring this material alive for me as I struggle to improve in a new language. Where possible I study and read material where I also have access to natural audio. It is always more pleasant and more reliable to hear a natural voice. The TTS is helpful but still not perfect!
Using Google Translate to Generate Tailor-Made Learning Content
Most of my learning activity is based on listening and reading. The words and phrases that I am learning are those that I find in the texts that I read and listen to. Often, however, I find myself wondering how I would express certain concepts in the language I am learning. I often can’t remember even the most basic vocabulary, or connector phrases, like “in my opinion” or “in other words”. Sometimes I just want to see how longer texts would be rendered into the language I am learning.
In order to acquire the vocabulary I’m looking for, I just write out some text in English, using the vocabulary and expressions I want to learn. Then I copy this to Google Translate to get the translation. The resulting text in my target language is then imported into LingQ as a lesson. I can then save the relevant words and phrases to my personal vocabulary database, and use text-to-speech to better pronounce and learn these phrases.
Google Translate as a Communication Tool
Some people wonder if the availability of machine translation and text-to-speech technology will make language learning obsolete. I don’t think so. I want to communicate with people from different language groups personally, face-to-face, in a real environment. I don’t want to talk through a computer.
Nevertheless, I find that Google technology helps me to communicate in writing, in emails or on Internet forums. This is true even for languages that I speak quite well, but where I’m not 100 percent confident about the grammar or spelling. I will simply type or dictate something into Google Translate and then take the resulting translated version, correct any obvious errors or inconsistencies and use that. I don’t have to worry about spelling. I don’t have to worry about accents or other idiosyncrasies of the language. And the person who receives my text, translated by Google and corrected by me, thinks that I’m quite fluent in the language.
I can interact with people in the language I’m learning, send messages, read answers, and respond using my computer and Google Translate. This form of written communication in the target language helps me learn the language. Needless to say, this works best if you are at a level in the language where you can make the necessary corrections and feel confident that what you are sending makes sense.
Google to Find Language Content
Google provides me with the ability to search for content of interest in the language I am learning. This can be newspaper articles, magazine articles, podcasts or e-books, audiobooks or Youtube videos which all of which I can import into LingQ using our Import Browser Extension. I can simply type
“Iranian Movies” or “Jordanian TV” , or the name of a famous person or author into Google or Youtube and I will find learning content of interest.
Google the search engine enables me to access a library of language content on a wide range of subject matter, at varying levels of difficulty, that far exceeds anything to be found even in the best university library. And it is all digital and can be accessed using modern information technology for enhanced learning.
Google for Grammar Information
If I need an overview of the grammar of a new language, I go to Google. If I need verb conjugation tables, noun declension tables, information on the use of pronouns etc. I just go to Google and type in the relevant terms, “Arabic verbs”, or decline the verb “avoir” , for example. The resources are almost limitless, and instantly available. I can target specific grammar questions or look for PDF versions of summary descriptive grammar resources for different languages. I have found some truly amazing resources in this way.
So I see Google Translate, the Google search engine, text-to-speech, and other innovations in information technology as a boon to language learning, but not something that will replace the need or motivation for language learning. We just need to look for the best ways to use this technology. Then we can continue to learn languages well into our old age.
And the next time you are not happy with Google Translate’s results, or you hear someone complain that Google Translate doesn’t work, just think of the benefits that this technology brings us.
16 comments on “Google Translate Doesn’t Work? Google and Language Learning”
I want to add that with the Google translate app you can make use of the handwriting tool which for some people and some languages allows you to draw the letters and Google translate can do word recognition. As well the camera tool has recognition capability. For example click on the camera icon and have it translate a sign in the airport or the restaurant menu. Can’t understand someone, have them speak into the app. Use it to listen to podcasts where you can pause. These are just some tools available.
As far as I know, Google Translate (GT) uses English as a hub. Thus, to go from Dutch to German, GT goes D–>E–>G . This significantly reduces the accuracy of the translation.
While GT is good or very good for Spanish, German and Russian to and from English, it is practically useless for Korean to English.
I found your article to be right on target. For one thing, I think average users have higher expectations, than they should, of Google Translate simply due to lack of knowledge. When someone understands the complexity of creating something as linguistically intricate as Google Translate, he understands that it would be impossible for translation software to do this job the way a person would. Now, another reason users expect too much from GT, and this reason also derives from lack of knowledge, is that users normally do not think thoroughly about how they are communicating with GT. There are specific observations that need to be taken into account. For example, I have used GT to translate from English to Russian and vice versa. The translation software will have some level of difficulty translating accurately compound or hyphened words, such as “figure out”. Instead of asking GT to translate “We need to figure something out. The issue’s bound to have a solution”, I would rephrase it and ask it to translate “We need to find a solution. There must be one” in the case of the Russian language. I would also avoid words such as “picture” and use “photograph” instead. This is also what I do in real life when I am speaking with someone whose English I suspect might have some limitations. I find that it is a bit arrogant to expect the other person to do all the work in figuring out an interpretation. It is ok to meet them half way and try to make the sentences as simple as possible. And it’s smart to use words that we might find to be similar in several other languages, such as photograph, because words such as “picture” may have different meanings in French, Italian, Spanish, so the likelihood that it will also have a different meaning in Russian than in English is very high. However, if we try to translate the word photograph, we will find that the translation of such words into other languages will be a word from similar Latin or Greek roots. Obviously if you are bilingual, you know exactly what I’m referring to. It helps if you are bilingual or have traveled a bit and dealt with language issues before, to understand what a speaker from another language might need us to use in order to help him interpret our message or question. Now, having said all that, I have to say that I have tried Bing translator, and I was very impressed. I used compound words and fixed phrases that GT would normally miss. Guess what? Bing translator got it right 9 times out of 10. I thought that was pretty amazing.
In my opinion, if you want to get a translation right through a translation software, the safest thing to do is to think of any translation tool as a relatively inexperienced translator; someone who will make mistakes unless you do your part to make things easier for him.
Agreed, we have to make it easy for the translator. Through trial and error we can usually get good results. Cheers.
Concurred, we need to make it simple for the interpreter. Through experimentation we can generally get great results.
Actually it doesn’t work. Try to translate any word in Greek and the only thing it will do is to convert the Greek alphabet into Latin alphabet but still in Greek.
I would just like to mention one outstanding German online translator: https://www.deepl.com/translator.
It doesn’t yet have as many languages as the Google translator, but there are some serious comments that attest DEEPL to a better translation than Google, Microsoft and Facebook can do.
I love google translate too. It can be very useful. I supplement google translate with lingoes dictionary which is an offline dictionary. lingoes dictionary has the option of functioning as apopup dictionary by setting up whatever hotkey you want. I usually set it to double left click so if google translate fails me I resort to lingoes for instantaneous relief (I’ve been doing this for Spanish/French).
Japanese has great pop-up dictionaries. 2 most notable ones are rikai-sama and yomisama
Thank you for the great article – I totally agree with the points you have raised.
I was previously rather sceptical about GT (especially since it used to provide much more rudimentary translations). I used to think of it as a tool for those who are too lazy to put into practice a language they are learning; or a crude tool one has to make do with for translating from really exotic languages.
Recently, I have changed my mind. I am currently reading a book in Chinese and my understanding is about 70-80%. Most of the time the words I do not know are simply stylistic ornaments, which do not convey much meaning and would not appear in everyday speech anyway – I am planning to acquire them gradually, but it is not my top priority.
GT is perfect for dealing with these unfamiliar words. I first read a few paragraphs of the book and then translate that fragment using GT to verify my understanding. If there is an unfamiliar word, which is frequently used, with time I notice it when I first read the Chinese text and can check its meaning using GT.
An alternative approach to GT would be to translate the text word-by-word using a dictionary – which would be much less fun, would probably take too long to be practical and would not give me the advantage of going through a lot of text to develop the feel for ‘word statistics’. By using GT I can get a better idea on how important a word is (based on how often it appears in text) and pay more attention to it.
I have not yet used GT for writing much, but your suggestion is certainly valuable!
Google Translate used to be dreadful at Turkish to English, I think partly because Turkish is a non-Indo-European language with an agglutinative structure, but in recent years it has greatly improved.
naver is better than google for korean!
Due to language complexity, Google, Bing and other machine translators cannot replace human translators. In the different languages, there are words with dual meanings and also what works in one language does not necessarily work in another. In addition, Machine can’t relate words to context. Translation is something when quality matters more than anything.
I think “Lingue” is better tool than GT but it have few languages Germany English and Spanish.
Lately I have been following very closely the advances in the field of Artificial Intelligence, and I am amazed by the latest achievements of AI systems: BioMind has beaten a team of 15 doctors at diagnosing brain tumors and predicting the expansion of brain hematomas. AlphaGo beat the Go world champion, and then was beaten by a superior AI system: DeepMind. However, I thought language translation was safe from the “assaults” of AI (at least for the next two or three decades), being that natural language is far more complex that Go or reading CT imaging.
Fast forward to last week. I was assigned to translate a psycho-neurological report and decided to give Google Translate a try to see how it is doing nowadays: it left me speechless. Mind you, I am not easy to impress. I am a linguist with 20 years of experience as professional translator and proofreader (English-Spanish-English). I have worked in Mexico, Puerto Rico, Spain and the United States for one of the most important translation organizations in the world, alongside linguists from all over Latin America and Spain obsessed with grammatical and syntactical correctness. This is to say I know my stuff.
I fed Google Translate full paragraphs of this highly technical report and it showed a deep understanding of morphology, syntax, linguistic and situational context, and semantics. Many sentences were impeccable. Some had flaws (especially when dealing with polysemic words) but it was nothing a proofreader could not work with. Not only did Google Translate save me time, it showed me that my profession is not as safe as I thought. I give it between 5 and 10 years for this AI system to make a translation as good as a human one (of course such translation will need proofreading, but so does a human translation anyway). Time to move to interpretation, which adds a whole layer of complexity to the task and will be safe from automation for at least the next two or three decades (gulp!)
Google Translate is good for simple translations but if you need reliable and technical translations, TechDico is better and it translates in 28 languages.
Google translate is a good option for translation. But many a times, we cannot totally rely on their results. One problem that I face is that most of the times it gives literal translations. We just need to have clear concepts. Only then can we modify google translation results.