Korean "gapjil"

« previous post | next post »

Watching the embedded video in this article, "Korean Air Chairman Fires Two Daughters Over Rage Incidents" (Bloomberg News [April 22, 2018, 8:45 PM EDT]),

er pronunciation is correct, as it is spelled 갑질(甲질) gapjil (Revised Romanization).

Your ears are working just fine. The lenis consonant "ㄱ" that  RR transcribes with a <g> is phonetically unvoiced, and even slightly aspirated, in initial position, even though it's fully voiced in medial position. But because voicing of obstruents is not a phonemic distinction in Korean, Koreans quite literally don't hear the difference that you and I, with our English-speaker ears, hear so clearly.

Krista Ryu:

The Korean g, "ㄱ" is actually /k/ (aspirated) so most foreigners (English speakers) tend to hear it as a k. However, there is another consonant  "ㅋ" which is even more aspirated and thus is written as k to distinguish is from a softer ㄱ which then became written as g. In fact, there is no consonant that actually sounds like a g, a voiced velar stop (manner of articulation is occlusive) in Korean. I think it is natural that you hear it as a k. A similar phenomenon is observed with the Korean d ㄷ (which sounds like a t to English speakers), and the Korean j ㅈ.

Bill Hannas:

It's a romanization problem.  In Korean, initial k/g is the same unvoiced, unaspirated velar consonant.  Neither English 'k' (in initial position, where the sound is aspirated), nor 'g' (voiced but unaspirated) are exact "matches".

Shelley Shim:

I think she still uses the word "gapjil", with a "g" instead of a "k", but her placing emphasis on the word makes it seem closer to "k". There was also an article today that talks about Hanjin Group's Lee Myung-hee (the wife of Hanjin Group's CEO Cho Yang-ho, also the father of the two Cho sisters) supposedly performing "gapjil" (or "삿대질" (sat-de-jil) as it says in the article (it was literally translated as "finger-pointing" when I looked it up in the dictionary) to a female employee from Ilwoo Foundation, which is the foundation that Lee Myung-hee supervises.

Different ears hear sounds differently.


  1. KWillets said,

    April 25, 2018 @ 10:47 pm

    It's the same romanization which your anonymous colleague so accurately slagged in 2013. One should devoice initial consonants, but only a Korean speaker would know that. It's a Gatch-22 situation.

  2. David Marjanović said,

    April 26, 2018 @ 4:40 am

    Listening to the video, I'm as surprised as you: this isn't a mere voiceless lenis (like Mandarin g, Hindi/Thai k, Spanish c/qu or southern German g) – it's [kh]…

    Why is everyone talking about voice? Voicing of plosives is quite unreliable in English, especially in initial position.

  3. ~flow said,

    April 26, 2018 @ 12:22 pm

    Ever since coming into contact with Korean and Hangeul, I've strongly felt that writing k- kh- for initial ㄱ ㅋ is an unfortunate choice, as is using -k for final ㄱ. Likewise for the other plosive consonants. To me, 갑질 is best rendered as gabjil. Both the phonotactics and the modern orthography of Korean are quite involved with many different things—retrograde and prograde assimilations, resyllabifications, alterations between liquids and nasals—going on all the same time, and I thinks its best to just stick closely to Hangeul orthography to avoid complicating matters further.

    Catering to an imagined, specific audience that is thought to be too unsophisticated for a transliteration that uses features not present in their first language is, IMHO, generally not a good idea (on a related note, German newspapers stick to writing Kioto, Tokio, Schanghai, and, brace yourselves, Pjöngjang; more or less *all* other place names of Japan, China and the Koreas follow English / international customs, i.e. y for the palatal glide and j for the affricate. Striving to make it easier for the reader, both newsanchors and readers then start to struggle real hard and fail when some report combines names like 'Pjöngjang' and, say, 'Pyeongjang').

  4. KWillets said,

    April 26, 2018 @ 12:44 pm

    The Wiki page on positional allophones is a good reference for this. Listening to the video again, I don't hear any aspiration, but I'm no expert.

    Another thing that's a bit ironic is that the reporter spells and pronounces her own name as "Jihye" with the voicing intact; this is the same initial character (ㅈ) as the subject of the story, written/pronounced as Cho further in. It's code-switching.

  5. David Marjanović said,

    April 26, 2018 @ 4:58 pm

    I haven't seen Schanghai in quite a while, but Tokio, complete with a pronunciation in three syllables, is here to stay. Same, incidentally, for Kenia [ˈk(ʰ)eːnia].

    Concerning Pjöngjang, ö [œ] isn't such a bad approximation to [ɤ] (and was indeed used to transcribe Mandarin in the early 20th century; Pinyin e as in de, ge, se, she). Too bad the Korean eo hasn't actually been [ɤ] in a long time.

  6. ~flow said,

    April 27, 2018 @ 4:45 am

    @David Marjanović—to be clear, I find that as such using letter ö to write the first vowel in Pyeongyang is not a bad choice per se; it's certainly better than the digraph eo, which is clumsy, easily confused, and hard not think of as a diphthong (which it is not). Letter ö is also certainly internationally more common and therefore more accessible than ŏ. I also don't find it unacceptable to approximate the place name as [pjœŋjaŋ] in a German context. The problem with letter ö, though, is that Korean—or at least some dialects—does have a vowel sound [ø] (as in the surname 최 Choi), which is what you'd most expect of ö to stand for. It's this collision that makes ö for 어 an unfortunate choice.

    Taken together it means that when a Mrs Choi Jihye from Seoul goes to visit Pyeongyang then a German paper will write those names as Choi Jihye (or Chö Jihye, as the case may be), Seoul, Pjöngjang. Professional news anchors will typically afflict damage to each of these names one letter at a time.

  7. Jichang Lulu said,

    April 28, 2018 @ 11:44 am

    Korean stop phonemes are realised as voiceless in initial position. If you hear a voiced stop [b d] at the beginning of a word, chances are it's actually a realisation of a nasal initial /m n/. (Since /ŋ/ doesn't occur initially, there's no initial [g].)

    As for aspiration, Korean has three degrees. Let's focus on the velar stops.

    Aspiration can be quantified by voice onset time (VOT), measured from the moment the stop is released. The greater the VOT, the more aspirated the stop. Negative VOT means ‘voiced’. English /k/ has mean VOT around 60~80 ms, depending on the study. Mandarin k (in pinyin) is more strongly aspirated, averaging 90~100 ms. Mandarin g has ~30 ms (thus is still voiceless, and close to the value of e.g. Russian k). English /g/ can go from ~20 ms down to a negative value (i.e. it can be voiced).

    Now Korean, notated in Revised Romanisation as in the OP:

    – ㅋ k (‘aspirated’) as in 커피 keopi ‘coffee’. This is usually strongly aspirated, as in Mandarin or more (often above 100 ms for older speakers). An English or Mandarin speaker should hear k.

    – ㄱ g (‘lenis’) as in 갑질 gapjil ~‘arrogance’ (as in the OP), 김치 gimchi ‘kimchi’. This can have VOT ~ 60 ms. The range overlaps with that of English k. Based on VOT alone, this is closer to English k than g. That's indeed how Victor and many others perceive it: also as k.

    – ㄲ gg (‘fortis’) as in 까치 ggachi ‘magpie’. Around 20 ms, thus still voiceless, but clearly in the g range of English and Mandarin. (A speaker of Russian, Dutch or French could perceive k if VOT was the determining factor.)

    Aspiration is not the only distinction between the three stop (and affricate) series, but it does explain why Korean g rather sounds like an English k.

    An interesting thing about (at least Seoul) Korean is that the ‘aspirated’ (e.g. k) and ‘lenis’ (e.g. g) are merging in VOT, while still being distinguished by tone (lenes like g have a lower tone). Here's a study on tonogenesis in Seoul Korean by Yoonjung Kang.

    Another interesting issue regarding Korean aspiration is its irregular occurrence in the Sino-Korean vocabulary (the aspiration of the Korean reading of a Chinese character can't be predicted in general by the aspiration of the (Middle) Chinese reading). Sun-Mi Kim's thesis discusses Sino-Korean aspiration in detail.

    The aspiration of the Japanese fortes (k can be around 60 ms, i.e. in the Korean g range) was discussed in the comments to an earlier LL post (“A Stew with a Consonant Shift”).

  8. Giodisseo said,

    May 3, 2018 @ 10:39 am

    So, to pick the k-series across CKJ, English(/German?), and French/Italian/Dutch [lowercase c/g], in decreasing order of voice onset time we should have:




    That's, of course, only if we agree to 1. lump them all in the same articulatory category; 2. forcibly assign an average unique value of VOT to what are mostly overlapping ranges (each velar listed will allophonically overlap with at least one other velar above and/or below it); and 3. disregard intralinguistic trends and variants.

    Big ifs, but perhaps a handy schematic for language learners to keep in mind.

RSS feed for comments on this post