Review: Google Translate’s Word Lens (Japanese-to-English)
Most Japanese learners probably heard the news when Google Translate added Japanese-to-English translation to its Word Lens feature in early 2017. If you’re not familiar with the feature, it’s designed to read text in one language and translate it to another language in real time. I was really excited when Japanese-to-English translation was announced, but my early experiences with Word Lens were disappointing.
Google Translate has had another photo-based feature for a while. You take a picture of something, the app quickly scans to look for text, then you use your finger to highlight the exact text you would like to have translated. In order to put Word Lens to the test, I took some pictures and compared the translations between Word Lens and the traditional photo translation system.
From left to right, we have: a normal photo with no translation, a screenshot of Word Lens in action, and a screenshot of the traditional photo translation feature being used on the same sign.
Word Lens obviously did great translating “Nagoya Castle” and did an okay job with Osu Kannon Temple. It struggled a bit with the names of the museums, but the old feature translated this part perfectly.
For those of you who don’t know, peanut butter doesn’t really exist in Japan. Instead, there are many other peanut-based spreads, like this can of peanut whip. Ignoring the concerning part at the top where Word Lens says “this dies,” we can see that “peanut butter blend the sweet taste” is elegantly translated as “Savory taste blended with deep roasted peanut butter” in the traditional photo translation.
For two more examples and my final thoughts, keep reading:
