Meet the guy trying to turn emoji into a legitimate, usable language | Hacker News Reader

Meet the guy trying to turn emoji into a legitimate, usable language

washingtonpost.com

24 pointschmullig9 comments

Egypt beat him to it; originally ran in the New Yorker: https://pbs.twimg.com/media/CDbYv9RWMAA-7rK.png

Seems to me that writing using Emojis requires significantly more effort than writing it using just words. Besides the comical factor, I just do not see the advantage here. Correct me if I am wrong, but there would be certain phrases that you could not say in Emoji, it would struggle with slang and other made up words. I probably would not back this, but I cannot deny this is not cool and unique, I will be keeping my eyes on this to see if it goes anywhere.

One note of likely relevance to the HN crowd, if the Kickstarter[1] is successful, we're definitely planning to release an API + the raw data.

[1] https://www.kickstarter.com/projects/fred/the-emoji-translat...

This would certainly make emojibot[0] a lot easier (and cheaper)! :)

[0] https://github.com/ChimeraCoder/emojibot

You're proposing a project involving translating between two languages... do you have a linguist on board?

>There are literally no other image-based languages.

Written Chinese started as an image-based language.

If you go back far enough, what language didn't? All writing systems are derived from or borrowed from a script that in its ancient past was pictographic. Braille is the only exception I can think of.

Give emoji a few thousand years, maybe less, and it could become a proper language. I think it's already starting to evolve. The eggplant symbol has been repurposed to mean "penis". (So I've heard.)

But I think the main roadblock to the machine translator described in the article is that emoji are currently used as an adjunct to peoples' native language. Does a Japanese speaker use emoji the same way as a Russian speaker? Or German or Farsi? I suspect the answer is they don't. So trying to train an AI using texts from all these different speakers will be trying to build a Tower of Babel.

To write fish in chinese you write a pictogram. That pictogram is an image of a fish, except the image has evolved over the millennia to be unrecognisable.

The English word "fish" on the other hand, is not a pictogram, and its evolution has not been primarily visual. Instead it has undergone evolution in pronunciation, which has indirectly affected how the written word "looks".

No, to write the word for fish, you write the character pronounced "Yú" which happens to be 魚. Chinese speakers don't see that and think "that looks like a fish". They think "that sounds like Yú". Just as when you see the letter F you think of the sound, not that it looks like a hook: Phoenician 𐤅 - wāw. (Or if you know that the Latin root is piscis: 𐤐 - pē, that means mouth.) And before you point out that a Chinese speaker would describe 魚 as "the character for fish" remember that an English speaker when asked to spell a word might say "F as in fish."

Like I said, all written scripts if you go back far enough started with pictograms. But that was many millennia ago. Their evolved descendants are divided into alphabetic or syllabic scripts, both using the symbols to represent sounds but with a different granularity.

And really, Firefox spell-check, you're not going to recognize "millennia"? What's wrong with you?

Reply on news.ycombinator.com