Learning a Korean word from scratch, with a note on AI
« previous post | next post »
While attending an international conference on the application of AI to the study of the Silk Road and its history, at which most of the papers were delivered in Korean, I was struck by the frequent occurrence of one distinctive word: hajiman. For some speakers, it almost seemed like a kǒutóuchán 口頭禪 ("catchphrase"). I had no idea what it meant, but its frequency led me to believe that it must be some sort of function word. However, the fact that it is three syllables long militated against such a conclusion. Also its sentence / phrase final position (though not always) made me think that it wasn't just a simple function word.
I kept trying to extract hajiman's purpose / meaning from its position and intonation (usually not emphasized, almost like an afterthought).
When, during coffee / tea breaks I asked some Korean colleagues about it, their reply — "Oh, hajiman" (with an offhand smile) only added to the word's mystique.
Finally, after the conference was over and I had access to my laptop computer where I could do some research on this puzzling Korean word, I found out that it is merely a conjunction meaning "but; however". That kind of blew my mind, because that wasn't close to the meanings I had been hypothesizing for hajiman judging from the linguistic environments in which I was hearing it.
Here's what lexical resources were telling me for hajiman 하지만:
but conj.
while conj.
whilst conj.
after all adv.
whiles conj.
again adv.
but conj. — four varieties
however adv.
I asked a few Koreans who know Mandarin well how they would translate hajiman, and they said rán'ér 然而 ("however; but; yet"). That fits better with the environments in which I was often hearing it being spoken at the conference, a kind of lingering afterthought, "and yet…". Some individuals used it so much that it was like a catch / pet phrase, as I mentioned above.
As for holding conversations with people on the street, in shops, etc. who did not know English, that was not too much of a problem in Korea, because it seemed that almost everyone had some sort of electronic translation device (usually an application on their phone) by means of which they could communicate.
I should note that Koreans are very big on AI and they are very good at it. Some of them treat AI as their "assistant" who carries out a wide variety of tasks for them — research, writing, design, and so forth. They even contemplate whether they might fall in love with their AI companion and whether two AI "characters" they create might fall in love with each other!
Finally, unlike my colleagues who are wary of AI and to one degree or another forbid their students from using AI (it is a very touchy subject at Penn), Koreans in higher education encourage (or even require) the use of AI.
Selected readings
- "Is there no / any longer a reason / need to learn a foreign language?, part 2" (7/28/23)
- "Is there no / any longer a reason / need to learn a foreign language?" (7/14/23)
- "Alien encounters" (9/15/16) — esp. the fourth comment
- "No, no, no; yes, yes, yes" (2/2/25) — pet phrases
- "Catchphrases, mantras, and verbal tics" (11/16/22)
- "'Bāphre bāph!' — my favorite Nepali expression" (8/12/18)
- "A Chinese analog to English 'you know'" (11/22/19)
Thomas said,
May 16, 2025 @ 8:12 am
For me too, the most memorable feature of the Korean language are exactly these kinds of words. And they come in whole series like 그렇지만, 그러면, 그래서. Of course all the Korean I know is just from listening to K-pop and watching Korean dramas.
Victor Mair said,
May 16, 2025 @ 9:20 am
Interesting, Thomas. All three of the words you cite are decidedly polysyllabic — geuleohjiman, geuleomyeon, geulaeseo — yet they would be rendered in English as monosyllables: "but, then, then".
KWillets said,
May 16, 2025 @ 11:44 am
Agglutination in action — the sense is carried by the affix, and 하다 is practically a placeholder. -지만 can be affixed to any verb, and resorting to 하지만 may be a symptom of overly thick discourse.
Yves Rehbein said,
May 17, 2025 @ 3:56 pm
I only know Jp. hajime 始め "begin!" from Judo tournaments.
Interestingly, shǐ 始 in Sino-Japanese shi し, Sino-Korean si 시 just sounds like the translation for 하지만 in Japanese しかし shikashi: "Originally 然し […]" https://en.wiktionary.org/wiki/しかし which see for more.
I don't get anything out of it, but the pattern reminds me of the negative conjunction [be]-ba-[be] construction in "Syntax of Sentence-final Particles in Chinese" (The Cambridge Handbook of Chinese Linguistics); I have no access and can't quote correctly.
-지만 -jiman [t͡ɕima̠n] aligns well with 시 si (compare 지), as far as 하 ha- is linked to 하다 hada, "Of native Korean origin, […]" https://en.wiktionary.org/wiki/하다
Yves Rehbein said,
May 17, 2025 @ 4:07 pm
Correction: 然し shikashi is "Originally 然り (shikari, […]" (WT l.c.), obviously. The difference is largely lost on me.
KWillets said,
May 18, 2025 @ 5:19 pm
-지만 itself appears to decompose into two smaller common particles, so I confirmed in Martin that it is what he calls the suspective -ci, marking hypothetical or negated action, and the particle man, meaning "only, just". Like "but" or "furthermore" it downplays the preceding statement and amplifies the following one.
The negative imperative 하지마 (don't do that) sounds nearly the same and shares the first two morphemes, but the final one (ma instead of man) is different, fully prohibiting the verb marked with 지.
So you learned three words, or so.
Keith said,
May 19, 2025 @ 9:14 am
This is interesting to me, because at first glance I thought of the word 하지마 that I used to hear shouted by the Korean moms around the school that my children attended.