DeepL supports Japanese, causes number of Japanese language students to plummet

Now with DeepL Japanese and Chinese support (more info here), I thought it might be a good time to check it out and see how it compares with some of the other translation services out there. 

Deep what?

I had heard of DeepL before, but I hadn’t paid much attention to it because it didn’t support Japanese (the language I would want the most) and other East Asian languages such as Chinese and Korean. However, I knew that the small company was focused on delivering really good translations compared to what we are used to normally.

Now that Japanese and Chinese support has been added to DeepL, I thought it might be fun to see how it stacks up against Google Translate. 

Let’s throw a few things at both DeepL and Google Translate and see which version comes off sounding like an English speaker wrote it.  

Test 1 (Entry from Nintendo’s Japanese Wikipedia page) 

For the first test, I took the first part of the history of Nintendo from its Wikipedia page. 

Below is the text I pasted into both translators. 

任天堂は1889年(明治22年)に創業した老舗企業で、娯楽に関する様々な事業を展開している。創業以来多くの種類の玩具を製作しており、特に花札やトランプは創業初期から現在に至るまで製造・販売を続けている。 

First up, Google translation Japanese

Google Translation of Nintendo article

The first issue to notice is the Meiji plus year number was translated directly. It’s not wrong and was translated as is. The next thing you will notice is the use of “we” later in the translation. This starts to read as if Nintendo themselves wrote this entry. I would say this is an okay translation considering it’s machine-based. 

DeepL Japanese translation

I pasted in the same blob of text and here are the results. 

DeepL Japanese translation of Nintendo article

From the beginning, the text begins with when Nintendo was founded. One interesting thing to note here is the removal of the Meiji year since it’s basically the same thing as 1889 anyway and is generally not useful to English speakers. 

Unlike the Google translation, the statement does not use “we” and reads as if a third party wrote about Nintendo, which is the typical style seen on Wikipedia. The writing seems a bit more natural in this translation, especially when you look at the final sentence. “early days of the company” seems easier to digest than “from the beginning the present” that Google produced. 

Test 2 (news from Yahoo JP News) 

Getting away from gaming, I decided to take some text from one of the recent news articles on Yahoo Japan. It’s in relation to a popular mini-manga that was on Twitter and Instagram.  

Here is the original excerpt for your pleasure. 

漫画家でイラストレーターの、きくちゆうき氏が22日午前10時2分からツイッターを相次いで更新し、19年12月19日から毎日、ツイッターとインスタグラムで連載を続け、20日午後7時20分に最終回を更新したコマ漫画「100日後に死ぬワニ」の読者に感謝のメッセージを送った。今後も創作活動は続けていく考えを示した上で、ラストシーンは連載開始当初から決まっていたことも明らかにした。 

Google Translate Japanese translation

Google Translation of Yahoo News article

Google starts off strong here until it suddenly shifts to using “I” as if the artist himself were talking, which was not the case. The translation did not spit out a plural version of reader and the last sentence goes back to using reporting style of “he”. While not exactly unreadable, it could use a lot of work to make it easier to absorb. 

DeppL Japanese translation

DeepL Japanese translation of a Yahoo JP article

DeepL decided to go with Cartoonist instead of manga artist, which makes the passage already more readable for those that might not understand what manga means. The plural version of readers is correctly used. Overall, I would say this entry is much easier to read than Google’s. 

Final Thoughts on DeepL Japanese support

I’m really amazed by how natural the translations that DeepL Japanese is able to produce compared to the results from Google. While the number of languages supported may be much smaller than other services, it does the few languages it supports really well if the Japanese translations are anything to go by. 

If you deal with any of the languages supported by DeepL, I suggest giving it a try and comparing it to your normal translation tool of choice and seeing how it differs.

It makes me wonder if people 10-20 years from now will not need to study Japanese since everything could be translated for them. Scary thought!

Link:  DeepL Translator

Leave a Reply

%d bloggers like this: