I've wanted to build a mixed-script Korean input method for more than ten years, and I finally started building it.

The project is called Bibim. Korean, Japanese, and Chinese share a lot of Sino-derived words written with Chinese characters, even when the readings differ. Bibim uses those characters as the bridge: if you're writing Japanese and can't remember the Japanese reading of 博物館 (“museum”), you can type the Korean reading (bangmulgwan) instead, and it will recover the characters for you.

I finished the design doc today. If you're into CJK writing systems, East Asian languages, or input method development, I'd love feedback.

https://codeberg.org/hongminhee/bibim-prototype

@hongminhee this looks VERY interesting! I run into this problem more than I’d expect. Glad it’s not just me!

@hongminhee that’s so cool!! I’ve long wished for something like this:

1. I know one or more Japanese readings of some Kanji.
2. I am now learning a lot of the Korean Hanja readings (currently reading 마법천자문) and often don’t know the Japanese readings for those.
3. The macOS Hanja input method is terribly under-documented and just alright (Japanese input methods are fantastic my comparison). And as far as I can tell it doesn’t really exist on iOS.

Super cool project ❤️

cc @andrewzah

@hongminhee I have been dreaming about something like this! You are amazing.

Bibim is the perfect name for it, too!

@hongminhee Yk I don't speak any of htese languages but one day i could make this a linux IME just for how cool it would be

Finally have a working prototype of Bibim to show off.

The demo in the video: typing “daxueshengeun hakubutsukane ganda” in Latin letters produces “大學生은 博物館에 간다”, Korean for “The university student goes to the museum.”

The input is a mix of three languages. daxuesheng is the Mandarin reading of 大學生 (university student); hakubutsukan is the Japanese reading of 博物館 (museum); -eun, -e, and ganda are Korean. The input method finds 大學生 through the Chinese phonetic path and 博物館 through the Japanese one, then stitches them together with Korean particles into mixed-script Korean output.

Bibim treats Chinese characters as a shared logographic layer across CJK languages, so you can reach any word from whichever pronunciation you happen to remember.

Try it yourself: https://hongminhee.codeberg.page/bibim-prototype/.