A scribe in Gyeongju (경주) jammed a tiny mark between Chinese strokes as if whispering into the margin. The king was waiting, but the note had to sound Korean first, Classical Chinese second.
Silla (신라) writers used systems called Idu (이두) and Hyangchal (향찰) to make Chinese characters do double duty. They picked some characters for meaning and some for sound, so you could squeeze Korean endings and particles into a sentence built from Han characters. That mix let poets write whole songs in Korean centuries before Hangul, and about 25 of those hyangga poems still survive as scrambled clues to how people actually spoke.
Court readers leaned on a different trick, Gugyeol (구결). You'd see tiny marks and extra strokes added to a Chinese line of a sutra. Those marks told a Korean reader how to reorder words and add endings so the line made sense in Korean. It worked like live subtitles, and it let officials read Chinese archives without sounding like they were speaking foreign grammar out loud.
Balhae (발해) did the same kind of trick, but with a local twist. Their diplomats and tomb epitaphs used Classical Chinese as the written face, while the choice of characters often hints at Mohe or Korean speech under the hood. Dae Joyeong (대조영) founded Balhae in 698, and the documents his state left are a puzzle box. Modern linguists read the odd character choices to pull bits of Balhae sounds and words back out of the stone and paper.
These systems left messy, human marks everywhere: hurried ink, scratchy marginal notes, a poetic line where the meaning and the sound argue. They're the place where written language gets weird and personal.
So if you think Hangul suddenly made Korean readable, remember scribes had been coding and subtitling Korean for centuries. Their little hacks turned Chinese pages into living speech, and we still text about it today.