Talk:Korean phonology

From Wikipedia, the free encyclopedia
Jump to navigation Jump to search

Is the consonant table official?[edit]

I was wondering the source for the table of consonant pronunciations. I highly disagree with:
ㄱ - /k/
ㅂ - /p/
ㅈ - /ʨ/
ㄷ - /t/

In my opinion, it should be:
ㄱ - /g/
ㅂ - /b/
ㅈ - j sound like in english /dʒ/ (correct IPA symbol?)
ㄷ - /d/

I have experience with Korean and English since I was a kid and always disagree with the k,p,/ʨ,t pronunciations whenever I came across them in books and this article in particular. I recently conducted an experiment for fun by picking words that start with ㄱ,ㅂ,ㄷ, and ㅈ and asking native Korean speakers to choose what pronunciation sounds more accurate. For example, I would say "kalbi" and "galbi" for ㄱ and ask which was more accurate. Every person I questioned answered 100% to my consonant sounds system I personal think it is (g,b,j,d). Again, this wasn't done professionally, but it seemed obvious it wasn't the sounds from the current proposed IPA pronunciation table.

In my opinion, Korean Romanization systems have always been terrible and inaccurate, and those were suppose to have been devised by experts. Now I'm wondering if the IPA for Korean is just as bad.

Thoughts? KingKwon (talk) 07:57, 26 November 2007 (UTC)

'p' is used because IPA 'b' is regarded, by convention, as fully voiced in the sense of zero or even negative voice-onset timing. it's not that English 'b' is 'voiced' and 'p' is not, but that (prevocalically) 'p' may be thought of as having a very positive VOT, and 'b' has a much lesser, positive, VOT; it's about the timing of the voicing, not whether it's there or not. The fact is that English 'b' is indeed rather closer to the Korean 'ㅂ' , and some transcription/romanizations take note of that. You can often see it in professional linguistic papers on Korean.
Romanizations aren't intended to be close transcriptions, but rather to give people an approximation of what something would sound like. (Again, in linguistic papers on Korean, by (native) Korean linguists, publishing from Korean universities, you can see an entire discussion without a trace of Hangǔl. Everything is in romanization.) Even a close IPA phonetic transcription is not fine-grained enough for any language; that's what 24-bit recordings in a quiet studio setting are for. Most importantly, this is an article about Korean phonology, not phonetics. Once we settle on /p/ as a phonemic glyph, the matter is settled. As far as phonology is concerned, one is talking mental representation, not verbal expression. It could involve sign-language for that matter. This article tends to mix a lot of allophonic variation in, however, making it hard to distinguish in places from a phonetic discussion. JohndanR (talk) 17:02, 25 May 2016 (UTC)
My first thought is that since English /b/ /d/ /g/ /dʒ/ are themselves devoiced (except intervocalically) that your perception that Korean ㄱ is identical in voicing to English /g/ is correct. According to Hyunsoon Kim in "The place of articulation of the Korean plain affricate in intervocalic position: an articulatory and acoustic study" (Journal of the IPA 2001): "...Korean plain consonants, except for the plain fricative /s/, get voiced in an intervocalic position..." (p 252). This mimics the laryngeal properties of English consonants represented by <b> <d> <g> etc and not that of Mandarin <b> <d> <g>, which are still voiceless intervocalically. Does this help? Ƶ§œš¹ [aɪm ˈfɻɛ̃ⁿdˡi] 08:24, 26 November 2007 (UTC)
Thanks for your reply, but I'm a bit confused, I thought /b/ /d/ /g/ /dʒ/ were the voiced versions of /p/ /t/ /k/ "ch"? So how does a /g/ get made unvoiced?KingKwon 00:32, 4 December 2007 (UTC)
[b] [d] [g] [dʒ] (square brackets) are the voiced versions of [p] [t] [k] [tʃ]. The issue here is that English /b/ is not really [b]. The IPA characters don't perfectly match for the respective English sounds. Ƶ§œš¹ [aɪm ˈfɻɛ̃ⁿdˡi] 01:19, 4 December 2007 (UTC)
I'm confused again. If the sounds are indeed the same for English and Korean, why does the IPA have [b] for English but not [b] for Korean in the initial position (and so on with g, d, dʒ). So I would expect "버스" and "bus" to both begin with a /b/ sound, but according to the chart, it would be a /p/ for Korean. If the b in English isn't a perfect match using IPA characters but still used, I would expect it to be used for Korean ㅂ as well. Thanks for your replies.
PS - Any recommendations on where I can look for more in depth information on Korean phonology? KingKwon 07:58, 4 December 2007 (UTC)
But when you start mixing loan words (""버스" and "bus" ") into the discussion it really gets hairy. The fact is that Korean native speakers, like speakers for any L1, can also import phonetic features that are not part of Korean phonemic inventory. Most Seoul speakers make little distinction between 'ㅔ' and 'ㅐ', but will often make the 'ㅐ' much like English 'ae' for some English loan words, to the point that even the most sophisticated Korean 101 textbooks will call the vowel 'ae' just so as to make a distinction, as well as provide a bit of uncomplificatification for the beginner. By comparison, English doesn't have a phonemic or even phonetic 'χ' (that is not an 'X', incidentally) sound, but educated speakers will quite easily pronounce (Johann Sebastian) 'Baχ' quite faithfully, as well as pronounce the Scottish 'loch' with the correct final ending.
To your concluding question: before looking further into Korean phonology, you need to read a basic discussion of phonology itself, and then come back. Phonology, at the segmental level, is about the mental representations of the most atomic level of linguistic meaning. It's not about the 'sound' of 'p' and 'b' (In fact, in Optimality Theory, either of those two representations could have just about any pronouncible sound under the sun), but that they distinguish a meaning. In theory, we could treat written English phonologically, and say there is a 'phonemic' (actually graphemic) difference between 'cup' and 'cub', even if there is no phonetic difference for someone deaf from birth.JohndanR (talk) 20:11, 25 May 2016 (UTC)
I'd say the IPA is inconsistent. English is treated differently than most other languages. Ƶ§œš¹ [aɪm ˈfɻɛ̃ⁿdˡi] 09:26, 4 December 2007 (UTC)
He KingKwon, its true that English is treated differently. But they treat it so much different, that the phonetician themselves believe that English actually HAS voiced stops. Im doing some research into this and hope to be published soon. If you're an avid linguist, as we all should be, give this a read . its an excellent paper on laryngeal features and is based on Kim (1970) research into Korean laryngealization. — Preceding unsigned comment added by User:Emelius7
Well, it does have voiced stops. /lap/ and /lab/ are phonemically contrasted, and the latter has full voicing of the final consonant at labial closure and can continue even after by buccal expansion until the resulting drop in glottal Δp extinguishes phonation.
I also generally agree. I have a friend whose family name is 조 and of course has always spelled it Cho, and eventually decided this doesn’t make sense and wants to spell it Jo. (P.S. I changed Aeusoes1’s angle brackets above to their html entities to prevent strange formatting down the page.) MJ (tc) 15:44, 26 November 2007 (UTC)
It doesn't matter whether you like it or not; Korean ㅂㄷㄱ are devoiced in initial and final position and voiced between voiced sounds; this means that it's simpler to assume the phonemes as being voiceless sounds. If you're unconvinced, keep listening; you will be able to notice slight aspiration in 가 sometimes. --Kjoonlee 17:10, 26 November 2007 (UTC)
It should matter if someone has a different opinion about something, otherwise things would never get revised. Even with a slight aspiration in 가, It still sounds more like ga than ka to me (although still not a perfect match for either letters). How exactly has the phonology of Korean been determined anyways? And is it possible the IPA's characters of /g/ and /k/ (etc) don't perfectly match for the respective Korean sounds? KingKwon 00:32, 4 December 2007 (UTC)
But opinions don't change observations. If you don't like the theory of gravity, does that stop apples falling from trees? No. If 가 sounds more like ga, then that means you're accustomed to the distinctive features of Korean. The phonology of Korean has always been described based on observations. /ɡ/ and /k/ are distinguished by voicing. 가 and 카 and 까 are all unvoiced in initial position, so they're all [k]. --Kjoonlee 00:33, 20 December 2007 (UTC)
Well, then my observation was that with initial consonants aspiration makes a greater difference than voicing, in the ears of Korean speakers, not just foreigners like me, because (as you yourself said below) it is a distinctive feature in Korean. Your assertion that unvoiced = [k] (IPA) is not the same as saying that unvoiced = ‘k’ (everyday English writing). Some are taking you to mean the latter, which is the cause of this confusing (and IMO not very productive) dispute. As I said two weeks ago (below), the table is partly misinformative as it is. MJ (tc) 14:41, 20 December 2007 (UTC)
That's not something that belongs in the consonants section of "Korean phonology", because that's Korean phonetics. If you account for the differences between phonology and phonetics, there's no error whatsoever. --Kjoonlee 16:53, 20 December 2007 (UTC)
I don't understand the rest of your comments, though, but the parts I understand all back up my own claims. --Kjoonlee 17:06, 20 December 2007 (UTC)
To be fair, other phonology pages don't have phonological processes completely sectioned off (look at Spanish phonology and Russian phonology for instance). I think a few phonetic notes below the table might help readers who are similarly confused. I'll see if I can't move a few of those phonetic notes up and you guys tell me if It's weird. Ƶ§œš¹ [aɪm ˈfɻɛ̃ⁿdˡi] 17:24, 20 December 2007 (UTC)
I think this may all come down to the fact at the OP is thinking that /b/ is the sound in boy, when in fact, boy is actually a /p/. Because of this, and the fact that Korean phonology doesn't have voiced initial stops, the current table seems well enough, though a little messy. User:Emelius7
On what basis do you make this claim? To my ear (and by holding my finger on my larynx) "boy" starts with a voiced plosive. English is my mother tongue.
He's essentially correct. Pronouncing words solus is not good linguistic practice. Phoneticians will look at full utterances in sentences, from many speakers, and use something like PRAAT to visualize the very things we are talking about. The [b] in 'boy' does not have full phonation at onset, but has a slight delay, just as the nucleus of 'poi' (Hawaiian food) has phonation after a long delay. That is the only phonetic difference that makes a phonemic distinction for English speakers/listeners using the words 'boy' and 'poi'. In a noisy environment, a listener might ask a repetition of the sentence "That poi was sitting on the table...". "Hm? Why was that boy sitting on the table...?" JohndanR (talk) 20:11, 25 May 2016 (UTC)

On less sure ground, I would say that the problem is that voicing is not a distinguishing feature in Korean, so that trying to match English and Korean consonants runs aground. Again, to my ear, voicing in Korean does not seem to be consistent from one speaker to another, so that I hear "kay" from one Korean and then from another, saying yes, that's right, it's "gay." Maybe the second person isn't actually voicing but doing something else that my mind/ears interprets as voicing, being the best it can do. (talk) 11:46, 15 February 2013 (UTC)

Regarding Romanization, McC-R's focus was on phonetic representation (for foreigners), while RR's focus is more on phonemic representation (for Koreans). People who've lived in Korea for a long time, who have now returned to Sweden, are amused everytime people write down Busan instead of Pusan. ("In Korean P and B are the same," they say.) --Kjoonlee 17:15, 26 November 2007 (UTC)
Regarding kalbi and galbi, this is because aspiration is a distinctive feature in Korean and voicing is not. In English, voicing is a distinctive feature whereas aspiration is not. Just try to say 달 탈 딸 (be careful to say 달 first) and ask English speakers if they heard [d]. --Kjoonlee 17:19, 26 November 2007 (UTC)

Also, it would be nice to have some links for Korean phonology. Anyone have any useful links? KingKwon 00:32, 4 December 2007 (UTC)

The problem with the consonant table is that it doesn’t distinguish initial, medial and final pronunciations, which obviously require different IPA transcriptions. That would require more space (perhaps even separate tables) but it’s necessary to really explain the phonology. As it is, the table offers as much misinformation as information. I haven’t time this week to make the change, but if no one else does by then I’ll have a go. MJ (tc) 20:57, 4 December 2007 (UTC)

I have added a long note before the first table, attempting to explain what information is there and what is not. I am surprised that I, as a foreigner, should be the one to add this disclaimer; it seems to dare editors to add the missing info. In my opinion, presening comprehensive Korean consonant phonology in the layout of the IPA table is cumbersome; showing sound changes by Korean alphabetical order would make the info much easier to grasp. MJ (tc) 15:30, 20 December 2007 (UTC)

Sorry, but that info did not belong there. "Pin" and "spin" are both use the same /p/ phoneme, but there's no need to mention [pʰ] in the consonant table at English phonology, since [pʰ] is just an allophone and aspiration is not a distinctive feature. Likewise, allophonic rules do not belong in a phoneme table for phonology. --Kjoonlee 17:17, 20 December 2007 (UTC)

Korean stress pattern[edit]

I always have a hard time finding a language's stress pattern. What, if it exists, is Korean's stress pattern exactly? For example, English stress pattern is Trochaic, with footing in all heavy syllables (2 moras), then foot all the rest except for final feet. — Preceding unsigned comment added by Emelius7 (talkcontribs) 00:49, 2 December 2011 (UTC)

"으" (eu) as a high back vowel? / "ㅚ" (oi/ø) as a front rounded vowel?[edit]

As far as I can remember, Korean linguistic books always represented eu by ɨ (high central unrounded), instead of ɯ (high back unrounded). The Korean version of this article (ko:한국어 음운론) uses ɨ. Can somebody verify this?

When I have time, I'll take a look at Korean books I have. Yongjik 06:50, 20 September 2007 (UTC)

The Handbook of the IPA uses a high back vowel, IIRC. Maybe this is a case of free variation? --Kjoonlee 13:06, 20 September 2007 (UTC)
There's a similar thing with Vietnamese. It probably has more to do with the preference of the author/linguist. Ƶ§œš¹ [aɪm ˈfɻɛ̃ⁿdˡi] 13:56, 20 September 2007 (UTC)

There has been a very productive discussion in the Korean talk page ko:토론:한국어 음운론, and the conclusion is that different scholars use either ɨ or ɯ, probably because the range of the phoneme lies somewhere in between. When I have time (read: in six months if we're lucky -.-) I'll copy here those references that were brought up in the discussion there. Yongjik 08:48, 29 September 2007 (UTC)

There's a strong tendency cross-linguistically for rounded vowels to be further back than unrounded vowels. For instance, in French and German, /y/ and /ø/ aren't really the rounded versions of /i/ and /e/, they're further back than that (though not quite central either). This is not universal, but is common enough that it's influenced transcriptions of, say, Vietnamese. If the Korean vowel were central, you might expect it to be slightly front of central, since it's unrounded. Instead, it's well back of central, and rather close to /u/, at least in the vowel diagram we're using in this article. kwami (talk) 00:58, 17 April 2008 (UTC)

Is there still consensus around ㅚ as a rounded monophthong? It's commonly presented in Korean textbooks as a glide and mid-front unrounded vowel: [we]. Is this possibly a dialectal difference, or simply one of those "linguists sitting in arm-chairs theorizing" instead of using phonetic production data? I haven't read Lee (1999), sadly, and I'm not expert by any means on Korean phonology, but I am curious about that vowel. Amieni (talk) 17:59, 14 May 2012 (UTC)

In my years of living here in Korea, I've always heard ㅚ as /ɥe/ or as /we/ from younger people. I HAVE heard it as /ø/ a few times, but nowhere near as commonly as the first two. Mexicocamboya (talk) 00:29, 3 July 2015 (UTC)

Initial ㅅ and ㅆ transcription[edit]

In my exposure to Korean speech, initial ㅅ is aspirated [sʰ] while initial ㅆ is like a plain [s] or [sː], not tensed like the other double-consonants. I have always heard 살 pronounced [sʰal] and 쌀 as [sːal]. Whenever I pronounced such a ㅅ-word with an English s- my friends would say it sounded like ㅆ. Is this possibly peculiar to the Seoul dialect (which is the only one I’ve heard)? MJ (tc) 06:15, 15 October 2007 (UTC)

It's very possible that the acoustic effect of tensing a fricative is different that tensing a plosive, even if the articulation is the same. I've not seen anything that suggests the articulation itself is different, though of course it could be. kwami (talk) 00:51, 17 April 2008 (UTC)
For lack of a better term, there's much more sibilance in ㅆ than in ㅅ. --Kjoonlee 05:14, 27 May 2008 (UTC)

There is much debate in the literature about the correct categorization of ㅅ, which definitely has an aspirated segment (it has a remarkably similar spectrogram to Burmese [sʰ], in fact). There is a detailed discussion here. There are dialects which merge ㅅ and ㅆ, pronouncing both rather like ㅅ, but the pronunciation of ㅅ with the aspiration is standard. ㅆ is also [s(ː)] in my idiolect in front of /a/; in fact, many Korean 'tense' consonants in modern standard pronunciation are not really distinguishable from simple or geminated unvoiced unaspirated consonants in most other well-known languages. Exceptions are ㅆ in front of front high vowels, and to an extent, ㅉ in all positions—in other words, [(t)ɕ͈]. Even the Russian щ [ɕɕ] is not as tense as the consonant in Korean 씨. --Iceager (talk) 21:53, 16 August 2010 (UTC)

According to Lee Ki-Moon and S. Robert Ramsey in their book, A History of the Korean Language (2011), they noted the behaviour of the plain plosives/affricates in contrast to the plain sibilant. Plain stops and affricates have a voiced allophone in medial position, but this is not true of the sibilant. The /z/ phoneme was lost around Middle Korean, apparently. Moreover, the sibilant is shown to have a greater aspiration than the plain plosives/affricates. ---Hmanck (Hmanck) 16:55 23 June 2010 (UTC-5) —Preceding undated comment added 20:56, 23 June 2012 (UTC)

tone/pitch accent[edit]

Is vowel length the only thing that's left of Korean tone/pitch accent? Is it still found dialectically? Should be covered in the article. kwami (talk) 22:14, 17 November 2007 (UTC)

Tone/pitch is distinctive in Gyeongsang dialect IIRC. Gaji can mean eggplant, branch, or a conjugation of "gada" depending on high/low pitch. Details may vary depending on the variety of the Gyeongsang dialect. --Kjoonlee 17:16, 27 November 2007 (UTC)

Korean pronunciation key?[edit]

It's been suggested that we agree on a system for indicating the pronunciation of Korean words that have Wikipedia articles, so that there isn't inconsistency for example in writing "ㅡ" as [ɯ] or [ɨ]. We could put up a key in Help space, and link to it from the transcriptions, the way we do now for Irish, Russian, French, Italian, Hebrew, etc. How does that sound?

Since we're writing primarily for an English-speaking audience, I personally think we should have a phonetic transcription that makes distinctions English speakers tend to hear, such as intervocalic voicing and [l] vs. [ɾ] (that is, things which are indicated in Wade-Guiles). However, I wouldn't want too narrow a transcription, which could just confuse people.

We would also need to decide on what to do with the round vowels which do not occur in all dialects. Perhaps Seoul pronunciation should be our guide in such cases.

It would also be nice to include vowel length, as we do in this article.

Any comments/suggestions/objections? kwami (talk) 01:10, 17 April 2008 (UTC)

What a good idea. Personally I'd say a "Seoul-dialect speaker trying his/her best to speak the standard dialect" sort of key with voice recordings would be fine. /me takes a look around him. --Kjoonlee 11:49, 18 April 2008 (UTC)
But I wouldn't bother with length; too much effort for too little gain. Maybe some examples for words like snow, eye, horse, speech would be nice, though. --Kjoonlee 13:19, 18 April 2008 (UTC)
Okay, take a look at Help:IPA chart for Korean. We can take up how to improve the key over there. I don't have any voice recordings to use, but it should be enough to give the reader a basic idea of what the IPA represents, and why there isn't a one-to-one correspondence with either hangul or RR. kwami (talk) 20:04, 25 April 2008 (UTC)


"/ɰi/ ㅢ is the only true diphthong in the Korean language"

Can someone explain the reasoning behind this? Is it maybe /ɯi̯/ and this is the only falling diphthong? kwami (talk) 13:16, 4 June 2008 (UTC)

Okay, deleted. kwami (talk) 16:52, 22 August 2008 (UTC)

Short Vowel Vs. Long Vowel Transcription[edit]

I am confused on the usage of stress markers as in /ɕiˈdʑaŋ/ vs. /ˈɕiːdʑaŋ/ since Korean stress is still highly debatable and according the Journal of the Acoustical Society of America -- November 1995 -- Volume 98, Issue 5, p. 2893, stress primarily relays on the phrase depending on the number of syllables in the phrase, syllable weight, and the position of the phrase in the sentence.

I would like to how how you incorporated these stress markers, because I am working on a linguistic project that has to differentiate between short and long vowels, and I am wondering if it has something to do with the syllabic stress in duosyllabic and polysyllabic words.

Thank you Finitoultero (talk) 03:24, 23 August 2008 (UTC)

It looks like it's any long vowel or else ultima. Presumably that only holds in citation form. I'm removing the stress marks until someone can come up with a source. kwami (talk) 08:11, 23 August 2008 (UTC)
I'll let you know if I find anything in citation form as well, as I am currently looking. Thanks for your input Finitoultero (talk) 00:26, 26 August 2008 (UTC)

I don't understand why the chart put ɯ for both 어/으. I believe this is wrong, the pronunciation is not the same unless the person speaks the Gyeongsang_dialect dialect. (talk) 13:29, 2 January 2009 (UTC)

Are you sure? I looked again and it doesn't put [ɯ] for ᅥ. It puts [ʌ] for short and [ɘː] for long. - Gilgamesh (talk) 14:24, 2 January 2009 (UTC)

/wi/ ㅟ as a diphthong or monopthong?[edit]

This article has /wi/ ㅟ listed as a diphthong; some books I have (mostly from the mid-90s) I have list that vowel as the monopthong /y/. Does anyone know if there's an accepted conclusion about this? Or is it something that hasn't been agreed on yet? —Politizer talk/contribs 02:14, 15 November 2008 (UTC)

If it's not covered here, check out hangul. AFAIK ㅟ is /y/ for some of the older generation in Seoul, and perhaps dialectically. Its standard value is /wi/. kwami (talk) 03:15, 15 November 2008 (UTC)

ㅟ as /y/ and ㅚ as /ø/ were the original standard pronunciations of these vowels. However, there seems to be ongoing sound change - a diphthongisation, yielding /wi/ and /we/ instead. However, this feature could be analysed as unrounding of the original vowel with a labial glide as a vestigial feature of the old pronunciation. If you look at vowels, you can see in the chart that /i/ is the unrounded counterpart of /y/, just as /e/ is the unrounded counterpart of /ø/. This unrounding feature has also been observed in Germanic dialects, including Yiddish. Compare Yiddish (shen) and German schön, both meaning "beautiful". Hmanck (talk) 18:11, 23 June 2012 (UTC-5)

Morphonemes and phonemes[edit]

The table of consonant "phonemes" uses pipes, presumably because these are actually morphophonemic. My question is: what, then, are the consonant phonemes of Korean? How do they differ from the morphophonemes? — Ƶ§œš¹ [aɪm ˈfɻɛ̃ⁿdˡi] 08:24, 27 November 2008 (UTC)

My bad! Those *are* the phonemes. However, they'd been made equivalent to hangul letters, which do not represent phonemes. ㅅ stands for several phonemes, such as /s/, /t/, and /n/. In the notes too it's clear that the things in pipes are simply phonemes, so my edits to the table are inconsistent. However, if we use the proper slashes in the table, the hangul would be incompatible. Would it be acceptable to delete the hangul? kwami (talk) 08:51, 27 November 2008 (UTC)
Yes, at least in that table. We'll have to go through the article to strike an appropriate balance between orthography and phonology, obviously we should have at least some of the former to be clear about the relationship but not at the expense of properly covering the latter. — Ƶ§œš¹ [aɪm ˈfɻɛ̃ⁿdˡi] 18:54, 27 November 2008 (UTC)

standard IPA for jamo clusters[edit]

One thing this article begins to analytically describe (but in many ways merely touch upon) is the IPA pronunciations of jamo clusters—consonant clusters between two vowels. Like, with the information provided, it can be gleaned that, in regular situations, ᄃᄉ (t-s) is [s͈], and ᇂᄌ (h-j) is [ʨʰ], etc. But it would be useful and helpful to delineate the regular pronunciations of every combination, because some combinations are not so straightforward to guess, especially when a consonant digraph at the end of a syllable comes before a consonant in the next syllable (three consonants collapsing). Unfortunately, I can't seem to find comprehensive tables for this online. But it really should be in this article, because it's an important part of how to guess the pronunciation of Korean words by their hangeul. I know that there are exceptions to these rules, but I'm talking about what guesses are normally regular. Anyway, I'm talking about a format like this, where the left column are consonants at the end of the syllable before, and the top row are consonants at the beginning of the syllable after. I'll fill in what I do know—feel free to correct me. Something like this should be reviewed by someone else (preferably with good sources to back it up), because I actually want to learn these details, but learners of a subject don't always make the best editors on a subject.

Pronunciation of final consonants based on following sound



















start   k t n n, ɾ, - m p s - ʨ ʨ͈ ʨʰ h
vowel   ɡ d ɾ b ʥ ɦ, -
ᆨg k̚k͈ k̚t͈ ŋn ŋm k̚p͈ k̚s͈ ɡ k̚ʨ͈ k̚ʨʰ k̚kʰ
k̚tʰ k̚pʰ k̚kʰ,
ᆪks k̚s͈ ?
ᆰlg lk͈ lkʰ
ᆫn n nk͈ nd nt͈ nn ll nm nb np͈ ns ns͈ n nʨ͈ nʨʰ nkʰ ntʰ npʰ nɦ, n
ᆬnj nk͈ nt͈ np͈ ns͈ nʨ͈ nʨʰ
ᆭnh nkʰ ntʰ npʰ nɦ, n nʨʰ nɦ, n
ᆮd t̚k͈,
t̚t͈ nn ll nm t̚p͈,
s̚s͈ d t̚ʨ͈ t̚tʰ,
ᆺs s
ᆽj ʥ ʨʰ
ᆾch ʨʰ
ᆯr l lk͈ ld lt͈ ll lm lb lp͈ ls͈ ɾ lʨ͈ lʨʰ lkʰ ltʰ lpʰ lɦ, l
ᆳls ls ltʰ?
ᆴlt ltʰ ltʰ
ᆶlh lkʰ ltʰ lpʰ lɦ, l lʨʰ lɦ, l
ᆷm m mk͈ md mt͈ mn mm mb mp͈ ms ms͈ m mʨ͈ mʨʰ mkʰ mtʰ mpʰ mɦ, m
ᆱlm lm
ᆸb p̚k͈ p̚t͈ mn mm p̚p͈ p̚s͈ b p̚ʨ͈ p̚ʨʰ p̚kʰ p̚tʰ p̚pʰ p̚pʰ,
ᆹps p̚s͈ p̚tʰ
ᆲlb lb lpʰ
ᆵlp lpʰ
ᆼng ŋ ŋɡ ŋk͈ ŋd ŋt͈ ŋn ŋn ŋm ŋb ŋp͈ ŋs ŋs͈ ŋ ŋʥ ŋʨ͈ ŋʨʰ ŋkʰ ŋtʰ ŋpʰ ŋɦ, ŋ
ᇂh k̚kʰ,
? ? ? p̚pʰ,
ɦ, - t̚ʨʰ,
ɦ, -

- Gilgamesh (talk) 06:37, 28 November 2008 (UTC)

I've made a start (k & n), but we should ask Kjonnlee to verify & fill in the gaps. kwami (talk) 08:28, 28 November 2008 (UTC)
Yes...thank you. There really needs to be an adequate comprehensive online source for this data, and so far the best I've found is a vague McCune-Reischauer transcription table with some final-jamo rows missing, and I don't know how accurate or reliable it is. - Gilgamesh (talk) 09:27, 28 November 2008 (UTC)
I'm going by the companion book for a set of NHK cassette tapes (Umeda Hiroyuki, 1985, Hanguru Nyūmon. NHK Pub). Most are derivable by analogy with their examples, but not all. The els may be retroflex in many cases, as you had them, but I seriously doubt they are before alveolar consonants, and I don't know where else. kwami (talk) 18:19, 28 November 2008 (UTC)
Okay, done for now. There's a hundred pages of dialog to go through, all transcribed in IPA, which might fill in some of the blank cells, but it may take me a while. kwami (talk) 19:53, 28 November 2008 (UTC)
This is helpful, but it really needs to be finished and polished somehow, and exceptions clarified. - Gilgamesh (talk) 21:26, 28 November 2008 (UTC)
One other thing: NHK has that |hk| is pronounced /kkh/, |sb| is /tpp/ or /bpp/. They don't have the reduced /kh/ and /pp/ forms. kwami (talk) 21:29, 28 November 2008 (UTC)
Confusing. This really needs clarification and verification. - Gilgamesh (talk) 21:47, 28 November 2008 (UTC)
What does the "coda" row you just added mean? kwami (talk) 22:05, 28 November 2008 (UTC)
Well, whereas the "start" row is the consonant at the beginning of a word, the "coda" row is an initial consonant after a vowel (no consonant cluster). - Gilgamesh (talk) 22:43, 28 November 2008 (UTC)
Okay, that just repeats the iung column then, so we don't need it. kwami (talk) 22:59, 28 November 2008 (UTC)
Actually, no. The ieung column is for final jamo pronounced in the vowel coda position. The start row is for initial jamo in the vowel coda position, and some of the initial jamo do not have equivilent corresponding final jamo versions. It certainly cannot hurt to be complete. - Gilgamesh (talk) 00:15, 29 November 2008 (UTC)
Okay, you mean as an initial after a vowel. kwami (talk) 00:26, 29 November 2008 (UTC)

A lot of the time a C after an |r| is fortis rather than voiced. Generally |rs| is /rss/, but also sometimes |rd| is /rtt/, |rg| is /rkk/, etc. kwami (talk) 00:48, 29 November 2008 (UTC)

But 일본 (Ilbon = Japan) is [ilbon], right? That ᆯᄇ is not fortis? How does one know whether fortis or not? Which is more regular? - Gilgamesh (talk) 01:36, 29 November 2008 (UTC)
I don't know. Perhaps it's lexical. kwami (talk) 01:47, 29 November 2008 (UTC)

Charts and rules[edit]

First impressions:

  1. A lot of the combinations seem unattested.
    • Some syllable-final clusters are only found in certain verbs or nouns derived from them.
      • "Syllable-final" ㄲ is only found in 볶다, 묶다, 볶음, 묶음, and so on. Verb conjugation would not use ㄹ. Proper names with ㄹ would not use ㄲ.
      • Same for "syllable-final" ㄵ, which is only found in 얹다, 앉다, 얹음, 앉음, and so on.
  2. Some combinations seem very unfamiliar.
    • k+kk seems like a long kk (with a long stop before the release) to a Korean's intuition. (부엌 가 is practically /부어 까/, BTW.)
  3. Some combinations are not clear-cut.
    • 김밥 is either /김빱/ or /김밥/ depending on speaker.

--Kjoonlee 03:56, 29 November 2008 (UTC)

IMHO a list of phonological rules would be clearer. --Kjoonlee 03:59, 29 November 2008 (UTC)

Well, some rules that aren't so clear to me. I'll enumerate the ones I can think of.
  • I read that |h.C|; metastasizes (at least in some cases), and becomes /Ch/ or /C.Ch/, so that if I understand correctly, |hk| is /k.kʰ/, |h.n| is /n.h/, etc. And |h.s| is /s͈/, correct? But how does a preceding |h| affect tense or aspirated stops? Do both also become /C.Cʰ/? And to be clear, word-final |h| is /t/, right?
  • Does |Cʰ.C| metastasize in the same way, becoming /C.Cʰ/?
  • Do combinations like |lC.C| become /l.C͈/ or /C.C͈/? Do |C.C| regularly collapse to /(C).C͈/?

Many of my confusions surround |h| as well as consonant clusters (especially three consonants in a row). - Gilgamesh (talk) 04:19, 30 November 2008 (UTC)

  • 놓다 becomes 노타, 낳다 becomes 나타, yes. But I can't think of cases of doubling and I don't think it can be generalized to other consonants, because there is no word-final |h|.
  • 쫓다 becomes 쫃따 and 끝 께 becomes 끋 께, so the preceding stop is unreleased and the following stop is tense, I think.
  • 읽다, 떫다, 읊다 become 익따, 떱따, 읍따. This can't be generalized, because verbs are a closed set. --Kjoonlee 04:53, 30 November 2008 (UTC)
    • Although you can of course say that stops are favoured over liquids in allophones of verb-internal consonant clusters. --Kjoonlee 04:54, 30 November 2008 (UTC)
      • 삶도 becomes 삼도. The exception is 핥다, which becomes 할따. (Probably because the stops have the same place of articulation.) --Kjoonlee 05:00, 30 November 2008 (UTC)
What about ᆳᄀ? Is that /lk͈/ or /lɡ/? And "Kjoonlee" is 큔리, right? - Gilgamesh (talk) 05:28, 30 November 2008 (UTC)
It's an obsolete combination only found in 돐 (baby's first birthday party), which is written 돌 nowadays. In the old days, 돐 금반지 would have been 돌 금반지. Kjoonlee is meant to be 케이준리, BTW. ;) --Kjoonlee 06:13, 30 November 2008 (UTC)
Keijulli... - Gilgamesh (talk) 15:42, 30 November 2008 (UTC)

Thanks for your input, Kjoon.

After ㄹ, I sometimes see plosives voiced, and sometimes fortis: |ㄹㅈ| as /ㄹㅉ/, etc. Is the latter a closed set of irregular verbs, or a productive rule? kwami (talk) 06:40, 30 November 2008 (UTC)

I think it's lexical. 날짐승 is 날찜승 but 길잡이 is 길자비. --Kjoonlee 06:52, 30 November 2008 (UTC)
But which is more regular? Or are they both about as regular as /ow/ is in English? (show, cow, mow, now, throw, plow, low, sow, stow, brow...) - Gilgamesh (talk) 15:42, 30 November 2008 (UTC)

Okay, what about ᆬC? Does ᆬᄌ become ᆫᄍ? What about ᆬ before other consonants? ᆬᄀ = ᆫᄁ or ᆫᄀ? I have reasons for all these specific questions. :P - Gilgamesh (talk) 19:57, 30 November 2008 (UTC)

Here's what I have for consonant assimilation (vowels are another story). (I include /j, ch/ in with 'plosives'):

  • Obstruents are tenuis unreleased before another obstruent. (/s, ss, j, ch, jj/ fall together with /d, t, tt/)
  • Tenuis obstruents are fortis after any obstruent, even if it's the silent member of a cluster like /nj/ or /ls/.
  • Tenuis plosives plus /h/ (either order, even if the /h/ is in an /lh, nh/ cluster) reduce to an aspirated plosive. (/sh/ becomes [th], but /hs/ is [ss]: mos-ha-da --> motada, joh-so --> josso.)
  • Plosives become nasals before nasals and /l/ (and /l/ becomes [n])
  • /l/ becomes [n] after any consonant but /n/ (/nl/ and /ln/ are [ll])
  • Final /d, t/ become /j, ch/ before (h)i or (h)y: mad-i "maji", mit-i-da "michida", kod-hi-da "kochida", dad-hyeo-seo "dachyeoseo". (Only across morphemes)

kwami (talk) 01:11, 6 December 2008 (UTC)

Personally, I think that both a list describing the rules, and a chart outlining the rules, would be good to compliment one another. Afterall, it's not like the language is a piece of software with wholesale fundamental changes every couple weeks or months—languages take decades and centuries to evolve. It should ultimately be feasible to use a chart to capture how Korean is now. - Gilgamesh (talk) 12:30, 7 December 2008 (UTC)


So, based on all this, we can probably postulate:

12 #



















# init   k t (n) m p s - tɕ͈ tɕʰ h
vowel ɡ d n ɾ b (ɦ)
ᇂ-h () k̚kʰ   t̚tʰ   nn   p̚pʰ   ss͈   (ɦ) t̚tɕʰ  
ᆨ-g k̚k͈ k̚t͈ ŋn ŋm k̚p͈ k̚s͈ ɡ k̚tɕ͈ k̚tɕʰ k̚kʰ k̚tʰ k̚pʰ
ᆪ-ks k̚s͈ k̚t(ɕ)ʰ
ᆰ-lg lk͈ lkʰ lkʰ
ᆼ-ng ŋ ŋɡ ŋk͈ ŋd ŋt͈ ŋb ŋp͈ ŋs ŋs͈ ŋ ŋdʑ ŋtɕ͈ ŋtɕʰ ŋkʰ ŋtʰ ŋpʰ ŋ(ɦ)
ᆮ-d t̚k͈ t̚t͈ nn (ll?) nm t̚p͈ ss͈ d(ʑ) t̚tɕ͈ t̚tɕʰ t̚kʰ t̚tʰ t̚pʰ t(ɕ)ʰ
ᇀ-t t(ɕ)ʰ
ᆺ-s s
ᆽ-j tɕʰ
ᆾ-ch tɕʰ
ᆫ-n n nk͈ nd nt͈ nn ll nb np͈ ns ns͈ n ndʑ ntɕ͈ ntɕʰ nkʰ ntʰ npʰ n(ɦ)
ᆬ-nj nk͈ nt͈ np͈ ns͈ ndʑ ntɕ͈ ntɕʰ
ᆭ-nh nkʰ   ntʰ   npʰ     n(ɦ) ntɕʰ   n(ɦ)
ᆯ-r l lk͈ ld lt͈ ll lm lb lp͈ ls ls͈ ɾ ldʑ ltɕ͈ ltɕʰ lkʰ ltʰ lpʰ l(ɦ)
ᆳ-ls lk͈ lt͈ lp͈ ls͈ ls ltɕ͈ lt(ɕ)ʰ
ᆴ-lt lt(ɕ)ʰ
ᆶ-lh lkʰ   ltʰ   lpʰ   l(ɦ) ltɕʰ   l(ɦ)
ᆷ-m m mk͈ md mt͈ mn mm mb mp͈ ms ms͈ m mdʑ mtɕ͈ mtɕʰ mkʰ mtʰ mpʰ m(ɦ)
ᆱ-lm lm
ᆸ-b p̚k͈ p̚t͈ p̚p͈ p̚s͈ b p̚tɕ͈ p̚tɕʰ p̚kʰ p̚tʰ p̚pʰ
ᆹ-ps p̚s͈ p̚t(ɕ)ʰ
ᆲ-lb lp͈ lb lpʰ lpʰ
ᆵ-lp lpʰ

Correct? - Gilgamesh (talk) 21:19, 9 December 2008 (UTC)

I wouldn't be comfortable with assuming some of those values. Also the second row is mislabeled: It's a vowel coda, not just any coda. kwami (talk) 21:25, 9 December 2008 (UTC)
No problem. The unassumable values can be greyed out. Which ones? I think I know some already—I'll grey those out. - Gilgamesh (talk) 23:36, 9 December 2008 (UTC)
We've also got Kjoon's concern that not all of these are actually found. We might be able to predict what they should be, but I'm not sure we should be telling our readers what they are. I can find attestations for many of them, but certainly not all, and I could never be certain that I'd found were the productive rather than lexicalized values. kwami (talk) 01:41, 10 December 2008 (UTC)
I suppose that's a sensible concern. Some of the more pedantic readers and editors like me, however, tend to crave orthogonal presentations that are as complete as possible. X3 The main reason I was asking all these questions was because I was programming a Hangeul generator in JavaScript/DOM, which also disassembles the hangeul result to parallel Revised Romanization of Korean and International Phonetic Alphabet. As complete a table as possible was needed for a project like this. - Gilgamesh (talk) 02:46, 10 December 2008 (UTC)

I plan to post a chart for the sequences I can attest to in existing IPA or hangul-equivalent transcription. A couple that were left out: |hs| = /ss͈/, and |nhs| = /ns͈/.

But isn't this OR?[edit]

Have these charts been published elsewhere? WP:NOR --Kjoonlee 01:34, 21 December 2008 (UTC)

Putting them in chart form is not OR, but filling in cells we can't attest is of course OR. If you could verify which are actually found, that would be great. kwami (talk) 01:59, 21 December 2008 (UTC)
But if we do put them in chart form, who do we source it to? --Kjoonlee 02:42, 25 December 2008 (UTC)
The same refs as if we don't put them in chart form. kwami (talk) 07:47, 25 December 2008 (UTC)
And which are those? --Kjoonlee 11:01, 27 December 2008 (UTC)

This edit was made because the previous info was unsourced and incorrect. I hope you can understand my concern... --Kjoonlee 03:20, 10 January 2009 (UTC)

I'm going through an NHK publication, Hanguru Nyūmon, a set of introductory Korean instructional tapes with a companion book that has running IPA transcription, to see which sequences I can verify. 'Course, that doesn't tell us which sequences are productive and which are lexical (//ls//, for example, which is sometimes /lss/ and sometimes /ls/), but it's a start. BTW, some of these only occur across word boundaries, so I'll miss those that don't involve assimilation. kwami (talk) 06:15, 28 April 2009 (UTC)
Okay, I took out the consonant sequences, which are uncommon, and also consolidated the 1st consonants by articulation, so that I don't have to attest to every cell. There are some gaps in what I've been able to find for fortis and aspirate C2, but since in all attested cases there is no assimilation with these consonants, I doubt there's anything to note. (In fact, we could probably remove these columns without any loss.) The one exception here is //t-ss// etc, which I'm assuming is [ss͈] just as //t-s// is, but I haven't attested it.
More CCC sequences would be welcome, especially /lp/ or /lpʰ/ followed by a labial consonant, or any CC sequence followed by a nasal consonant. kwami (talk) 07:25, 29 April 2009 (UTC)

The table is helpful as a guide, but it needs massive caveats since Korean pronunciation is not entirely predictable from the orthography. For example, there are exceptions to ㄹㄴ or ㄴㄹ becoming /ll/: 의견란(uigyeonran) [의ː견난] (uigyeonnan), 임진란 (imjinran) [임ː진난] (imjinnan), 생산량 (saengsanryang) [생산냥] (saengsannyang), 결단력 (gyeoldanryeok) [결딴녁] (gyeoldannyeok), 공권력 (gonggwonryeok) [공꿘녁] (gonggwonnyeok), 동원령 (dongwonryeong) [동ː원녕] (dongwonnyeong), 상견례 (sanggyeonrye) [상견녜] (sanggyeonnye), 횡단로 (hwengdanro) [횡단노] (hwengdanno), 이원론 (iwonron) [이ː원논] (iwonnon), 입원료 (ibwonryo) [이붠뇨] (ibwonnyo), 구근류 (gugeunryu) [구근뉴] (gugeunnyu) (examples taken from the official pronunciation rules). There is no clear-cut pattern; you just have to check the dictionary. As for whether following obstruents are tensed or not, there are exceptions to nearly every rule. The phenomenon of 'reinforcement' which not only causes the tensing of following obstruents but also may insert /n/ to give 색연필 (saegyeonpil) [생년필] (saengnyeonpil), 솔잎 (sorip) [솔립] (sollip) is wondrously complicated and only indicated in the orthography as 'ㅅ' (사이시옷: sai siot) in certain cases. --Iceager (talk) 21:24, 16 August 2010 (UTC)

more phonology, less orthography[edit]

A recent editor commented in edit summary that we're focusing too much on orthography and not enough on phonology. Though it is important to understand the patterns of how orthography results in phonological assimilations, maybe it's better to instead list the possible consonant combinations and which spellings typically yield them, since as we've already discussed, many combinations are never found in written Korean and their results are undefined. If I'm remembering correctly, the initials are /k k͈ n t t͈ ɾ m p p͈ s s͈ tɕ tɕ͈ tɕʰ kʰ tʰ pʰ h/, the finals are /k̚ n t̚ l m p̚ ŋ/, and the codas are /ɡ k͈ n d t͈ ɾ m b p͈ s s͈ dʑ tɕ͈ tɕʰ kʰ tʰ pʰ ɦ k̚t͈ k̚p͈ k̚s͈ k̚tɕ͈ k̚tɕʰ k̚tʰ k̚pʰ nɡ nk͈ nn nd nt͈ nm nb np͈ ns ns͈ ndʑ ntɕ͈ ntɕʰ nkʰ ntʰ npʰ nɦ lɡ lk͈ ld lt͈ ll lm lb lp͈ ldʑ ltɕ͈ ltɕʰ lkʰ ltʰ lpʰ lɦ mɡ mk͈ mn md mt͈ mm mb mp͈ ms ms͈ mdʑ mtɕ͈ mtɕʰ mkʰ mtʰ mpʰ mɦ p̚k͈ p̚t͈ p̚s͈ p̚tɕ͈ p̚tɕʰ p̚kʰ p̚tʰ ŋ ŋɡ ŋk͈ ŋn ŋd ŋt͈ ŋm ŋb ŋp͈ ŋs ŋs͈ ŋdʑ ŋtɕ͈ ŋtɕʰ ŋkʰ ŋtʰ ŋpʰ ŋɦ/. There's also DPRK-only /mɾ ŋɾ/, but those are artificial affectations, correct? So, we can make a table enumerating these combinations, and the known regular combinations that result in them, couldn't we? - Gilgamesh (talk) 06:58, 18 December 2008 (UTC)

"Coda" and "final" mean the same thing.
To the extent that the orthography is morphophonemic, this is phonology.
I think both a rule-based summary and a chart are useful. People understand things in different ways. kwami (talk) 02:01, 21 December 2008 (UTC)
If anything the problem here is that there's not enough IPA. I think it's worth making mention and describing the Korean romanization system somewhere (maybe not even here), but for a linguistic reference work, this should all stick to IPA as much as possible. This is especially necessary in the part on Korean vowel harmony. Pyry (talk) 18:05, 8 October 2011 (UTC)

I've been correcting the IPA in the Korean food articles, and realized that we've never gone anywhere with this. Shouldn't we? If I don't get comment, I'd like to add the chart to the article, and any unsupported cells can be deleted. kwami (talk) 10:05, 19 April 2009 (UTC)

What’s “normal”?[edit]

In the latest edit the use of the word normal grates on me. Why do people regard the initial and not medial pronunciations as “normal”? I think textbooks present them first just because they occur at the beginnings of words. This is the crux of why I don’t like how the consonant table shows only initial pronunciations, with the allophones relegated to footnotes, as if they were rare exceptions. If anything medial consonants are more frequent. But deeper than that, the voiced sounds seem much more “normal” to me in Korean phonology. It’s the initial sounds that are de-voiced.

Does anyone agree with me? I don’t have my hopes up, and I suppose even if people do agree, available references probably all present this information this same way, so switching around the presentation might even be considered OR. But I have to at least ask. MJ (tc) 18:46, 26 April 2009 (UTC)

It's the Clark Kent/Superman phenomenon with phonemes. You've got to pick a symbol to represent an abstract phoneme and you either pick the most common allophone (such as /j/ in English despite a voiceless variant being possible) or the most typographically expedient (such as Spanish /b d g/, which are more often approximants). What this does, though, is force an interpretation of a phoneme as having a prototypical phonetic realization that is then shifted because of acoustic and articulatory pressures rather than a group of sounds that are phonetically similar and perhaps replace each other upon affixation. In the case of Korean, the idea here is apparantly that the consonants are prototypically voiceless but that they assimilate voicing from the adjacent vowels and consonants. I don't know if this is an assumption and what sorts of evidence can be brought forth to justify it, but we can simply word it "/p, t, tɕ, k/ are voiced [b, d, dʑ, ɡ] between voiced sounds but are otherwise voiceless." — Ƶ§œš¹ [aɪm ˈfɻɛ̃ⁿdˡi] 19:20, 26 April 2009 (UTC)
That doesn't make sense because ONLY /b/, /d/, /dʑ/, /g/ get voiced in those positions, the emphatic and aspirated consonants do not. I see why that's the case for aspirated consonants since voiced aspirated consonants are rare, but emphatic consonants could easily be voiced, yet they are unvoiced between vowels. Then what, you propose that /n/, /m/ are NASALIZED between vowels? I'd rather use the medial symbols for the phonemes, and then say that Korean has a system that lowers the sonority of initials, like /n/, /m/ -> [d], [b] and [b, d, dʑ, ɡ] -> /p, t, tɕ, k/ rather than the other way around. -iopq (talk) 03:36, 1 November 2018 (UTC)
It depends on how you formulate the phonological rules. If you say stops or plosives, then it would be too general. But if you say that the rule applies only to plain consonants, it would only be /p, t, tɕ, k/) and not the aspirated or tense stops. Languages have these sorts of rules all the time. — Ƶ§œš¹ [lɛts b̥iː pʰəˈlaɪˀt] 15:27, 2 November 2018 (UTC)
I agree with MJ. An analysis of the current practices in the phonologies of some common Latin-script-using languages suggests that using /p t tɕ k/ for ㅍㅌㅊㅋ and /b d dʑ ɡ/ for ㅂㄷㅈㄱ is acceptable. Further, there is already a precedent in the Danish phonology.
The English phonology page tells, "The voiceless stops /p t k/ are aspirated [pʰ tʰ kʰ] at the beginnings of words...." This suggests that it is not compulsory for a phoneme to have the same symbol as the allophone that occurs word-initially. Moreover, "Depending on dialect, /r/ may be an alveolar approximant [ɹ], postalveolar approximant, retroflex approximant [ɻ], or labiodental approximant [ʋ]." Thus a phoneme symbol may even differ from all allophones'.
Several more phonologies contain examples of phoneme symbols differing from the IPA glyphs of the corresponding default allophones. As Ƶ§œš¹ previously mentioned, "The phonemes /b/, /d/, and /ɡ/ are realized as approximants (namely [β̞], [ð̞], [ɣ˕] ...) in all places except after...." in the Spanish phonology. In the German phonology, "The voiceless stops /p/, /t/, /k/ are aspirated except when preceded by a sibilant", but the aspiration is not represented. Finally, the Turkish phonology has "Voiceless stops are aspirated in initial and medial position" similarly to the Korean counterparts. These Turkish phonemes also lack aspiration markings.
In all above phonologies, digraphs are used only for the affricates, and except German, only for the post-alveolar. Perhaps it is acceptable and even encouraged to sacrifice the phonetic precision for conciseness, contrast (arguably, "p" looks more different from "b" than from "pʰ"), and, as Ƶ§œš¹ suggests, typographic expediency.
But then again, the change might not necessarily imply a loss of precision. The Aspiration (phonetics) page tells, "In Danish and most southern varieties of German, the "lenis" consonants transcribed for historical reasons as ‹b d ɡ› are distinguished from their "fortis" counterparts ‹p t k› mainly in their lack of aspiration." In fact, the table of allophones in the Danish phonology page shows that these letters also name their corresponding phonemes. As there exist precedent and current use in other languages, the suggested change is perhaps not that controversial. It is also arguably more neutral and robust: it represents the aspiration contrast at word-initial occurrences (interpreted the Danish way), and also the voicing contrast at intervocal occurrences (interpreted the IPA way). The current scheme is inadequate for the second contrast.
To summarise, the loose observance of precision in the naming of phonemes in other common languages suggests that MJ's proposed change should not be precluded, and the common tendencies for conciseness, intelligibility, and ease of input encourage the change. In addition, precedents exists in other languages, whose usage, in addition to IPA's, imply a better fit for the Korean phonology. In conclusion, I second the change. Yes, I know this is all OR, all from Wikipedia stuff, and so likely won't leave a dent. Just wanted to concretise the issues that came to mind. Like MJ, I'd appreciate pointers to empirical evidences. — Preceding unsigned comment added by (talk) 12:57, 2 July 2011 (UTC)

The pronunciation of the vowel ㅓ[edit]

This article, and other reference books that I have read, give the IPA pronunciation of the vowel ㅓas /ʌ/. I was surprised when I first saw this, because when I lived in Seoul and Gyeonggi-do, the Korean speakers around me pronounced this vowel as /ɔ/ (equivalent to the French "short o" vowel used in the French word "bol"), not as /ʌ/. Since I have moved to Daegu, I notice that everyone here pronounces ㅓ as /[ə]/. I haven't met any Korean speaker who pronounces ㅓas /ʌ/, as in the English word "cup". Has anyone actually heard it pronounced /ʌ/?

However, I have found that Koreans pronounce the English "short u", as in "bus", not as /ʌ/, but as /ɔ/. In fact, the Korean word for "bus" is "버스", which means that Koreans use the vowel ㅓto represent the English sound /ʌ/. Maybe this is the source of the transcription of ㅓ as /ʌ/, i.e., perhaps when Koreans first heard the "short u", with sound /ʌ/, which I believe they do not have in their phonology, they approximated its sound with the Korean vowel ㅓ, which they actually pronounce as /ɔ/ in Seoul and Gyeonggi-do, but /[ə]/ in Gyeonsang-do. This led them to write "버스" for "bus". Then, when they made reference works to describe their vowels in IPA, they used /ʌ/ for ㅓbecause they were erroneously pronouncing the IPA /ʌ/ like their vowel ㅓ, as /ɔ/ or /[ə]/.

I am considering adding a note about this difference in pronunciation between the north and south of South Korea, and replacing the present IPA equivalent of ㅓin this article from /ʌ/ to /ɔ/ or /[ə]/. Before I do this, however, I would like to hear others' opninions on such a change. —Preceding unsigned comment added by Wandering-teacher (talkcontribs)

(edit conflict) Iksop Lee an Robert Ramsey discuss this in "The Korean Language" (2000) on page 65. In discussing /e/ (ㅓ), it mentions that the prescribed "Standard Pronunciation" is [ə], but it actually varies on the environment between [ʌ] and [ɨ]. It goes on to say the this allophonic difference is only maintained by older speakers, and "younger speakers, in contrast, tend to pronounce the vowel in all cases with the tongue slightly retracted, and with a phonetic value near [ɔ]." Regards, Bendono (talk) 08:57, 4 May 2009 (UTC)
Wandering-teacher: I was surprised in the same way as you. My experience, mostly in the 1990s with speakers (mostly young, but some older) from Seoul, Incheon, etc., is consistent with your observation of /ɔ/. See my inquiry at the ɔ talk page. Maybe this is more of a generational difference than a regional one. But I agree that back-&-forth transliteration causes confusion. “Bus” is バス (basu) in Japanese, because [a] is the closest of their 5 vowels to /ʌ/. Many English words have found their way to Korean via Japanese, and so have been transliterated twice. And I observed a pattern of spelling with 어 words that are spelled with the ア (a) vowel in Japanese. ¶ My concept of 어 is actually more specific than the various interpretations of /ɔ/ I find, and the English word awe of course has much more regional variation than that. It’s hopelessly vague to define IPA symbols based on English words and then try to apply them to other languages. To me 어 is a simple, clear vowel exactly halfway between 아 and 오, and easier to locate as a point in a continuum than English ah, awe, owe or even IPA /a, ɔ, o/. MJ (tc) 20:34, 4 May 2009 (UTC)

The 1987 book that cames with my NHK intro Korean tapes transcribes it [ɔ] throughout. They say [ə] is primarily used by those over 50 (now 70!). They don't speak of dialects, but they're based on the pronunciation of Seoul.

The examples they give both have long vowels. Our short-vowel transcription of [ʌ] is already pretty much the same thing as [ɔ], since by it we don't mean the English vowel of cup conventionally transcribed by <ʌ>. kwami (talk) 18:03, 5 May 2009 (UTC)

We don’t? Certainly to me /ʌ/ has always indicated the vowel in but or cup or sun. This is the problem. Why has it changed to mean something else, and exactly what? I thought the purpose IPA was to prevent this sort of vagueness. MJ (tc) 09:45, 6 May 2009 (UTC)
Like anything would ever be that straightforward! There are national conventions for the use of the IPA that make it not entirely international. Check the position of Korean <ʌ> on the short-vowel chart of this article. It's a very back vowel, and rather open. It's precisely where <ʌ> is on the IPA chart. The only difference from [ɔ] is that it's not rounded. Now check out the position of <ʌ> at Received Pronunciation. It's an entirely different vowel: central, not back. It's where [ɐ] is defined on the IPA chart. That's why you'll often see /ɐ/ for the vowel of "run". It has the same location in American English, at least the LA variety that Ladefoged worked with. I understand that "u" may be a true [ʌ] in some regions of both countries, but it's conventional to transcribe this vowel as <ʌ> in English regardless of its actual position.
The IPA started out as a pronunciation key for teachers of English, French, and German in the late 19th century. As more languages were added, and the need was felt for more letters, and some of them shifted around. (This happened again more recently, when <a> was redefined from a central to a front vowel, not that there's much difference in that part of the vowel space.) I'm guessing that the <ʌ> convention for English became established when <ʌ> was a central vowel, and wasn't updated when <ʌ> became a back vowel in 1932. But I'm just guessing; perhaps an influential founding English phonetician pronounced his /ʌ/ as a back vowel. Anyway, it's just accepted practice to use <ʌ> for "run", just as we use <r> for the r of that word, even though it's not a trill: [ˈrʌn]. You can more precisely transcribe "run" as [ˈɹʷɐn], but that seems to be considered somewhat pedantic. But in Korean, there is no such convention, and here <ʌ> means a back vowel—unless of course it's long! (Check the long-vowel chart of this article.)
There are other national idiosyncrasies. The French "mute e", for example, is transcribed [ə] even though it's a rounded vowel. It would more accurately be transcribed [ø], but <ə> has long been the convention for any indistinct reduced vowel, and so is used for both French de and English the despite the fact that they aren't pronounced the same. kwami (talk) 12:15, 6 May 2009 (UTC)
Isn't this why there are three modes of describing Korean phonology? |ɤ| for hangeul structure (frozen from Middle Korean), /ʌ/ for modern pronunciation phoneme, and [ɔ] for actual nearest-to-exact pronunciation. Using pipes, slashes and brackets, respectively. This also applies to the associated "diphthong" ᅦ: |ɤ͡i|, /e/ and [e̞] respectively. - Gilgamesh (talk) 12:45, 6 May 2009 (UTC)
Middle Korean would be asterisked, , because it's a reconstruction. The pipes are morphophonemic, which hangul generally (tho not always) represents. BTW, I changed pipes to double slashes per a discussion elsewhere, where people thought double slashes were more intuitive/accessible than pipes: the mnemonic for //a// would be that it is "more phonemic" than phonemic /a/. But I don't like the way it displays on my browser: |a| seems more elegant, and //a// distracts the eye enough to be almost disruptive. Should we change back, do you think? kwami (talk) 13:22, 6 May 2009 (UTC)
Ahh, my mistake. Anyway, pipes look better IMO. Anyone disagree? - Gilgamesh (talk) 14:44, 6 May 2009 (UTC)
But your idea was right. If we consider long and short ㅓ to be variants of the same vowel (which is reasonable), then they'd get the same symbol phonemically, say /ʌ/ and /ʌː/. That could be used for any dialect of Korean, or generation of speaker, that distinguishes them. But phonetically we could be more precise, say [ɔ] vs. [əː] (or whatever the values are). kwami (talk) 14:56, 6 May 2009 (UTC)
I agree that pipes look better than double-slashes. I’d like to see some documentation of their usage at IPA#Usage. As for /ʌ/ vs. [ɔ] I vote for more precision wherever applicable; I would think it definitely applies in a phonology article. And thanks, kwami, for the other explanations. I will take some time to understand it all. MJ (tc) 15:19, 6 May 2009 (UTC)
But in a language with regional accents (sometimes with very significant pronunciation difference), is it even possible or practical to be that precise? It seems more reasonable to use representative phonemes for a conservative, enunciated pronunciation. So we ask ourselves empirical questions: Which is more conservative, /ʌ/ or /ɔ/? If /ʌ/, then does anyone even use it anymore? As for long and short vowels, it is my understanding that it's not really a problem for them to have different vowel articulation positions. Afterall, there are other languages with short/long vowel distinctions where the different long and short vowels are often indicated in IPA with different vowel articulations. Many of the Germanic languages and Insular Celtic languages come to mind (with English being a very extreme example because of the Great Vowel Shift), but there's also languages like Cantonese where this is also the case, and to a more limited extent Vietnamese. - Gilgamesh (talk) 16:11, 6 May 2009 (UTC)

Not really on-topic, but /ɔ/ for 어 reminds me of North Korean dialects (some? all? I don't really know), where 어 is more rounded than in Standard South Korean. --Kjoonlee 18:51, 6 May 2009 (UTC)

Yes, I'd heard that. I had the impression that northern speech has it more rounded, while southern speech has it less rounded. - Gilgamesh (talk) 19:24, 6 May 2009 (UTC)
On another off-topic note, it seems remarkable to me how much work I've injected into this project, considering I'm not Korean, never been to Korea, don't know any conversational Korean, and my sole connection with anything remotely Korean is my pedantic intensive study of obscure linguistics topics (entirely out of boredom) and an admiration of King Bowser Koopa (named after 국밥 gukbap). - Gilgamesh (talk) 19:28, 6 May 2009 (UTC)

Thank you everyone for your contributions. It is the first time I post anything and what you have written is all very interesting. Wandering-teacher (talk) 08:55, 2 June 2009 (UTC)

Official IPA list[edit]

Just curious where is the source for Hangeul sounds and their IPA equivalent? Is there an official source or anything? Also does anyone know the exact difference between a tɕ sound and an English J (ʒ) sound? Just curious since ㅈ is a tɕ and not ʒ. Thanks --Bluesoju (talk) 11:08, 7 June 2009 (UTC)

I don't think you're the first person to ask about the source on the IPA transcriptions, so we should probably provide inline citations for that. The source is the Handbook of the International Phonetic Association. I've answered your second question at Talk:Voiceless alveolo-palatal affricate. — Ƶ§œš¹ [aɪm ˈfɻɛ̃ⁿdˡi] 17:12, 7 June 2009 (UTC)

Unanswered questions[edit]

In Talk:Jeju dialect and Talk:Ulsan. I don't expect them to really get answered for months if not years, unless mention them in a place with more watchers. - Gilgamesh (talk) 08:33, 30 November 2009 (UTC)

Vowel length citation?[edit]

I'm curious where the vowel length distinction comes from? I have not conducted a thorough search of references, but in the 5 papers I've read on Korean vowels, none mention a length distinction. From Swan & Smith 2001, Lee's article on Korean speakers actually specifically states that there is no distinction between long and short vowels - "Instead of a long/short vowel distinction, Korean uses rising and falling intonation and the pause." Flexlingie (talk) 20:14, 30 December 2010 (UTC)

It used to be marked in hangul, and is still marked in Korean dictionaries, though it may be more robust regionally than in Seoul. Google Books has several good refs. — kwami (talk) 02:00, 31 December 2010 (UTC)
This has a nice account of the loss of distinction in younger speakers. However, it doesn't appear to consider dialectical variation. — kwami (talk) 02:34, 31 December 2010 (UTC)
Currently doing research on this righ tnow, check out this book: (talk) —Preceding undated comment added 04:19, 6 December 2011 (UTC).


Does Korean have any kind of stress? The article doesn't mention anything about it at present. Pokajanje|Talk 22:17, 18 November 2012 (UTC)

It may have pitch accent, depending on dialect and generation. — kwami (talk) 06:17, 29 March 2013 (UTC)

Tenuis becoming aspirated[edit]

I notice at least with younger speakers that the tenuis stops are becoming aspirated, at least in initial position, sometimes more so than the "aspirated" consonants, and that speakers have no idea whether they're aspirating or not. Is the distinction collapsing? Is it preserved some other way, such as vowel length or a new tonal distinction? — kwami (talk) 06:13, 29 March 2013 (UTC)

It seems to me that the distinction is there in the tone contour of the succeeding vowel. I've no source to back this up, though--someone else is gonna have to go digging. --Lfdder (talk) 22:20, 30 March 2013 (UTC)
Never mind, I did. The following passage is from [1]:
"Low vocalic f0 provided the most salient information for lax stops; tense and aspirated stop identification depended on a combination of VOT, f0, and H1 – H2 characteristics. The perceptual dominance of f0 over VOT for lax stops is consistent with the size of the f0 differences in word- (and phrase-) initial position, as well as the prominent role of the resulting tonal patterns in Korean intonational phonology."
[1]: Kim, Mi-ryoung; Beddor, Patrice; Horrocks, Julie (2002). "The contribution of consonantal and vocalic information to the perception of Korean initial stops". Journal of Phonetics. 30: 77–100. --Lfdder (talk) 22:32, 30 March 2013 (UTC)
Thanks! So we're back to a tone difference. — kwami (talk) 23:53, 30 March 2013 (UTC)

Monophthong transcriptions[edit]

I've noticed that kwami changed the transcriptions in the table to more accurately represent the way the vowels are pronounced in Seoul (Standard) today, but AlexanderKaras has reverted most of these saying that they're too narrow and that "these are just minimal pairs". I'm guessing he means that we don't need to worry about narrowly transcribing anything other than the vowel in question. Anyway, I think that it would be better to have two tables: one in line with Lee (1999), and another one after vowel/length mergers. — Lfdder (talk) 22:35, 4 April 2013 (UTC)

IPA tense diacritic[edit]

I've adjusted the description of ⟨◌͈⟩, the diacritic for tense consonants. It may resemble an (ASCII} "double straight quotation mark", as the text had it, but its Unicode name is "COMBINING DOUBLE VERTICAL LINE BELOW". I've made it explicit that this is merely resemblance, and added the Unicode code point and name into the explanatory note that was already there.

To discuss this with me, please {{Ping}} me. Thnidu (talk) 06:17, 30 December 2014 (UTC)

  • I just found the Unicode 0348 (combining double vertical line below) that describes the tense consonants. Does anyone have related references or references for the note 1 of this article?

Ana JE (talk) 17:31, 16 March 2015 (UTC)

The Unicode name is irrelevant for both Korean phonology and the IPA. — kwami (talk) 01:55, 3 July 2015 (UTC)

"The sibilant /sʰ/ has behavior of both the plain and aspirated stops"[edit]

So is it followed by the high tone, like aspirates, or the low tone, like plains? (talk) 09:00, 28 January 2016 (UTC)

Possible Vowel Explanation?[edit]

So, I see that there be a lot of confusion here as to certain vowel's qualities, and I came across an interesting possibility as to why that occurs. It basically says that /o/, /ʌ/, and /ɯ/ are the same as /u/, /a/, and /i/, aside from the back of the tongue being lowered, which is there to explain the spread lips of the /ɯ/ and the extra rounding of the /o/. Of course, I don't know how valid it is, especially since it is just a draft, but it seems like a good explanation, even if it fails to address the evolution of the Korean vowel system, such as the previous distinction between /e/ and /ɛ/. The author admits in a different article that he does not have the ability to properly test such hypotheses, which does make them a bit more sketchy, on top of being merely drafts. They can also be found on, for what that's worth. I don't really want to edit it in without consulting others, but do y'all think that it be worth mentioning in the article, especially to address the issue of other vowel qualities being assigned to the same phoneme?

--Blanket P.I. (talk) 23:17, 28 May 2016 (UTC)

Also, I've noticed some things on Wiktionary that might be worth integrating here. Firstly, /n/ becomes [ɲ] before /j/ (but not /i/, curiously). Secondly, rather than /l/, they list /ɭ/, which I somewhat confirmed with my South Korean friend (I asked him whether his tongue point forward or upward at the end of 한글, to which he answered upward), and it does sound different from the English "clear" /l/. Thirdly, they show /ɭ/ as becoming [ʎ] before /i/ and /j/, which I can again somewhat confirm, this time through listening to words like 어울리다, compared to words like 할머니. Should we note these in the article? I highly doubt that they'd be too complicated for the article, given that it talks about the palatalization of /t/ before a suffix beginning /i/ or /j/ (though I imagine that that be somewhat more common). Also, I ain't sure on this one, but can't /ŋ/ move to the beginning of the next syllable, if it begin with a vowel, such as in 병원장? (I only chose that one because it has the voice clip.) If so, shouldn't that be clarified in the article, rather than saying that /ŋ/ can never be initial?

--Blanket P.I. (talk) 14:33, 4 June 2016 (UTC)

A few things more things to mention, perhaps? Charles B. Chang did a study comparing Korean perception of native fortis consonants with foreign unaspirated voiceless consonants. It should be noted that they did not make the same study with Korean lenis consonants and the same foreign unaspirated voiceless consonants. Still, this may be in line with this blog's mention of the lenis consonants being the deviants, not the fortis ones. Lastly, there seems to be a change in how tense consonants are being pronounced. Should that be noted as well? I would imagine so, but I thought I should ask, just in case. (I know such can't be inserted, but I wonder if they are assimilating to the outside world, now that they are part of a very global society?)

--Blanket P.I. (talk) 21:37, 24 June 2016 (UTC)

Superscript on m and n!?[edit]

In the IPA table and the IPA notations for different examples n + m are regularly given as m͊ + n͊. These superscripts are not explained here and I did not find an explanation for them on the IPA article either. - They should be explained. What is the difference between n + m and m͊ + n͊? Is it for nasalisation? — Preceding unsigned comment added by (talk) 15:04, 27 August 2016 (UTC)

It’s very hard to make out on my screen, but it’s a “combining not-tilde above”, U+034A, indicating denasalization – thus the (m~b) and (n~d) notations. This is explained in the 3rd-to-last paragraph under Consonants. MJ (tc) 23:19, 27 August 2016 (UTC)

Thank you for your fast reply. - Yet, alas. You do not explain anything. I had seen the not-tilde by copying m͊ + n͊ and increasing the font. And I had also seen the 3rd-to-last paragraph under Consonants. - Here it is again. -

"/m, n/ tend to be denasalized word-initially.[14] Often, they are not actual stops either, but sometimes l, a stop release burst is audible: 그런데메밀 /kɯlʌnte memil/ → [kɯɾʌnde bemil].[15][note 2] /ŋ/ appears only between vowels and in the syllable coda."

Absolutely no explanation for the superscript. The superscript does not even appear in the exammples given. This is extremely bad. -

And apart from that. The sentence "Often, they are not actual stops either, but sometimes l, a stop release burst is audible:" is not really to be understood anyhow. Maybe a period after l could help. — Preceding unsigned comment added by (talk) 14:42, 28 August 2016 (UTC)

alveolar fricative s () as /s/, not /sʰ/[edit]

In Korean phonology, the alveolar fricative s () is considered a "plain" consonant (or yesasori 예사소리), along with b (), d (), g () that show moderate aspiration. (These stops are transcribed as /p/, /t/, /k/ and never as /pʰ/, /tʰ/, /kʰ/ between slashes, especially since another voiceless set called geosensori (거센소리), namely p (), t (), k (), are transcribed as /pʰ/, /tʰ/, /kʰ/.) Unlike /pʰ/, /tʰ/, /kʰ/, the alveolar fricative shows no aspiration word-medially. Also, the alveolar fricative s is often realized as a voiced fricative [z]. (e.g. gamsahamnida (감사합니다) / realized as [kɐm.zɐm.dɐ]) Cho, Jun, & Ladefoged (2002) states that "The plain /s/ generally undergoes intervocalic voicing word-medially, in contrast to what has commonly been assumed. About 47% of the tokens have fully voiced /s/ and about another 40% of the tokens are voiced over more than half the frication period." It is well known that Korean /p/, /t/, /k/ often undergo intervocalic voicing whereas /pʰ/, /tʰ/, /kʰ/ do not. I suggest the alveolar fricative s () should (at least between slashes) be transcribed /s/, not /sʰ/. --Vindication (talk) 05:12, 24 January 2017 (UTC)

Affricates and North Korean dialects[edit]

The affricates j (), jj (), ch () are transcribed as /tɕ~dʑ/, /t͈ɕ/, /tɕʰ/. The sounds are indeed palatalized in modern Seoul Korean and other South Korean dialects, but is it also the case in North Korean dialects? They may be /ts~dz/, /t͈s/, /tsʰ/ unless preceded by a palatal (semi-)vowel. --Vindication (talk) 07:07, 24 January 2017 (UTC)

I think this is distinctive of Pyongan dialect phonology (King 2006), and not part of standard 문화어. All the DPRK news broadcasts I've seen have fairly thoroughly palatalised j (), jj (), ch (). Michael Ly (talk) 20:29, 28 April 2017 (UTC)

The vowel charts need to be updated.[edit]

Seoul Korean vowels have undergone a variety of changes. Vowel length is reported neutralized, /y/ and /ø/ diphthongized, /e/ and /ɛ/ merged, /o/ raised, /ɯ/ fronted, and so on. Vowels of North Korean dialects have also undergone some changes, towards different directions. The article needs updated vowel charts, preferably illustrating both North and South Korean vowels. I'd also like to know if I need Moxfyre's permission to replace the current charts she/he/they made with new (not-yet-existing) ones. --Eumseonghakdo (talk) 18:44, 27 February 2017 (UTC)

No permission needed, @Eumseonghakdo:. The vowel charts are under the same license as other wikipedia content. —Moxfyre (ǝɹʎℲxoɯ | contrib) 00:11, 28 February 2017 (UTC)
@Moxfyre: Thank you for the reply. --Eumseonghakdo (talk) 00:00, 16 March 2017 (UTC)
@Eumseonghakdo: The problem is that, as far as I can see, the current vowel charts are based on the ones in the Handbook of the IPA. We shouldn't update them but make entirely new ones from scratch and only if there's a reliable source that features such charts. Mr KEBAB (talk) 18:19, 8 March 2017 (UTC)
@Mr KEBAB: Yes, that is the problem. Kang, Schertz, & Han (2015) (It is a program abstract of the 169th meeting of the Acoustical Society of America) offers a relevant chart (uploaded by the first author; It is perhaps the full document that the abstract summarizes), but the dots cannot be located (at least I can't locate them) in the new (still-not-yet-existing) chart without the full data (in figures), which are not presented in the above document. I've been waiting to see if some new relevant articles come up. --Eumseonghakdo (talk) 00:00, 16 March 2017 (UTC)
@Eumseonghakdo: That source offers formant vowel charts, rather than cardinal vowel charts such as those that we use in the article. I think we can have both in the article, following Norwegian phonology. Just be aware that they're two different things, and it would be WP:OR to create a cardinal vowel chart based merely on the formants. Mr KEBAB (talk) 00:33, 16 March 2017 (UTC)
@Mr KEBAB:, completely agree! It would be a great idea to have both formant (acoustic) and cardinal (articulatory) vowel charts in the article, if @Eumseonghakdo: can add them. It'd also be worthwhile to point out the age of the primary source for the current chart. —Moxfyre (ǝɹʎℲxoɯ | contrib) 02:36, 16 March 2017 (UTC)
@Moxfyre and Mr KEBAB: Sure. I'm aware of the difference, and thank you. When adequate numeric data or resources are available, I'll put a chart (or hopefully two, showing the vowels of both North and South Korean dialects) together with the existing one. --Eumseonghakdo (talk) 03:29, 16 March 2017 (UTC)

Voiceless (de-voiced) vowels[edit]

Does anyone have information on this? The sources I found seem to claim that Korean vowels are always voiced. However, I also found some videos that seem to show otherwise. For example in; pay attention to the following syllables - the text has captions, so it's easy to follow: time index 0:25 아픔에 (first 2 syllables); 0:29 얼마나 (first syllable); 1:35 그사연은 (first syllable); 1:36 가슴에 (first 2 syllables); 1:40 오늘도 (first syllable); 2:04 외로운 (first syllable). Is this a dialect variation? The entire vowels/syllables seem to be voiceless, not just aspirated. Selerian (talk) 07:49, 6 March 2017 (UTC)

I'd think that was a stylistic feature of the singer and/or the performance more than the language. If you listen to this version you'll hear a different approach to the same song. Michael Ly (talk) 21:02, 7 March 2017 (UTC)

Korean n-insertion[edit]

This article doesn't mention the n-insertion at all, which is a common phonological phenomenon in Korean. It generally occurs between two morphemes if the second one begins with /i/ or /j/. If the first morpheme ends in a vowel, it's indicated in writing with the addition of 'ㅅ' (사이시옷: sai siot). For example:

  • 색연필 (saek + yeonpil) [생년필] (saengnyeonpil)
  • 막일 (mak + il) [망닐] (mangnil)
  • 나뭇잎 (namu + "s" + ip) [나문닙] (namunnip)

Hzb pangus (talk) 22:45, 26 April 2017 (UTC)

@Hzb pangus: I would like to see this morphophonology (including 사이시옷) added too - what would be the clearest way to do it? A new section under consonant changes? Michael Ly (talk) 10:32, 20 June 2018 (UTC)
It's not insertion. It's actually dropping /n/ before /i/ and /j/ word-initially. Etymologically those words have /n/ -iopq (talk) 03:35, 1 November 2018 (UTC)


@Sotaque: The fact that having many allophones is 'not specific to Korean' is very obvious. That sentence doesn't imply that Korean is unique in that aspect, but that readers should be aware of the difference between slashes and brackets. That's literally the only purpose of that sentence. The 'context' tag is also inappropriate, as the sentence is self-explanatory. Mr KEBAB (talk) 10:04, 10 October 2017 (UTC)

" it (ㅅ) generally undergoes intervocalic voicing word-medially." Really?[edit]

I just found that my latest edit to clarify the matter was reverted, and I must dissent. As a native speaker I've never heard anyone pronouncing ㅅ as a voiced sound.

I checked the source, and it does say that ㅅ can be voiced for *some speakers*, but even that sounds dubious to me --- ideally, I would like better sources (say, a treatise in general phonology of Korean) rather than a single article that examines previously unknown phonetic variations. In any case, saying that some speakers may pronounce ㅅ as voiced is very different from saying that "ㅅ generally undergoes intervocalic voicing". And listing the letter ㅅ as a variation of /s/~/z/ is, IMHO, just plain wrong.

After all, phonemes go through all sort of crazy variations in all languages, but we normally don't list voiceless /n̥/ as an English phoneme even though that readily happens in certain situations. (talk) 01:30, 7 March 2018 (UTC)

Well, please read the source again. It does conclude /s/ *generally* undergoes intervocalic voicing. (See p. 219, section 5.1.2. Fricative /s, s*/.) With 46% of the /ㅅ/ tokens fully voiced, the voicing pattern of intervocalic /ㅅ/ is not different from that of /ㅂ/, /ㄷ/, /ㄱ/, and /ㅈ/. Native speakers of Korean including myself also tend to not actively hear voicing of any Korean obstruent, because voicing is non-distinctive in Korean. --Tisanophile (talk) 03:24, 7 March 2018 (UTC)
I've never heard this, I've heard it aspirated, though. Here:
I'm not a native speaker of Korean, so I can hear [z] just fine, and I cannot hear it here -iopq (talk) 04:22, 1 November 2018 (UTC)
To add, I just looked at the spectrogram of the first file and I am not going deaf. Here it is: it looks unvoiced just like I hear it. -iopq (talk) 05:04, 1 November 2018 (UTC)
Well, in that case, don't listen to me, but listen to the experts. Google books gives ample citations, including:
The sibiliant /s/ is normally unvoiced in all environments in contemporary Korean, but there is evidence for a voiced allophone [z] in northern dialects, especially in earlier times. - "Is Japanese Related to Korean, Tungusic, Mongolic and Turkic?" / Martine Irma Robbeets, p61
Some Korean sounds change in certain phonetic contexts. The top row of Table 11-1 shows p or b, t or d, ch or j, and k or g: the lax stops change from voiceless to voiced in certain phonetic contexts, such as -k- changing to -g- between two vowels, ... - "Writing and Literacy in Chinese, Korean and Japanese" / Insup Taylor, Martin M. Taylor, Maurice Martin Taylor, p189 (Notice the absence of ㅅ)
Third, all stop and fricative consonants are voiceless, except the lax stops that become lightly voiced between voiced sounds. - "The Korean Language" / Ho-Min Sohn, p153
Really, I have a hard time believing anyone needs to argue for this. (talk) 19:02, 19 May 2018 (UTC)
Oh, please. Two books on Altaic hypothesis and a 1975 book? None of the books is written by an expert (a phonetician) and what it seems like is that the books just failed to (or more likely, didn't care to; and possibly wasn't able to in 1975) mention the information on voiced fricative allophones like [ɦ] and [z], which are found in Korean language—that is, observed in corpus studies carried out by phoneticians. Because the phonetic details are not of great importance in their fields of study and in those books you cited. The details are however central to the analysis by Cho, Jun, and Ladefoged (2002), which can easily be verified with any contemporary Korean speech corpus. I'm saying you can actually see the voice bars on spectrograms yourself. As Tisanophile said, most naive native speakers do not generally perceive allophones because they are allophones, that do not contrast with each other in their language. It is natural that you don't notice the intervocallic voicing of any Korean obstrudent (by ear), but you can instead look into actual recordings (by eye). Many linguists who aren't phoneticians/phonologists do not tend to spend much time looking into spectrograms. --Comedora (talk) 02:00, 20 May 2018 (UTC)
I'm a native Russian/Ukranian/English speaker and I cannot hear [z] in the recorded examples I posted in the other part of the thread. All of my native languages distinguish /s/ from /z/ intervocally. -iopq (talk) 04:26, 1 November 2018 (UTC)
But I can hear in some recordings things like 이사 [iza] when spoken as a part of a sentence. So there's a difference between quick pronunciation and careful pronunciation. -iopq (talk) 08:38, 2 November 2018 (UTC)
I'll just note that Koreans usually use ㅈ to represent foreign /z/ and I believe /ʒ/ in loanwords, not ㅅ. Also as a native English speaker, when I listen to recordings and native speakers I never hear [z], only [s]. That being said, it's possible that there's becoming a transition in South Korea itself. I haven't ever gone to South Korea, and I hardly have the knowledge to listen to speakers of Korean closely for such sounds, anyway. Blanket P.I. (talk) 22:20, 28 May 2018 (UTC)

Wiktionary/IPA Allophones[edit]

I've noticed that there are a lot of allophones that are mentioned either on the Wiktionary pages, like 어울리다, or Wikipedia pages about consonants, like Alveolo-palatal consonant or Palatal nasal (which often direct one to Korean phonology, even though said page says nothing on the matter). Additionally, I've noticed that the semivowel /j/ (and probably /w/ and /ɰ/ too, if I'm hearing correctly) tends to combine with the preceding consonant phonetically (so /hjʌŋ/ turns into [çʌ̹ŋ] not [çjʌ̹ŋ], much like /hjud͡ʒ/ turns to [çud͡ʒ] in most dialects of English).

Now, the latter part I have no basis other than my own hearing (which is insufficient for our purposes), but I'm wondering if at least the allophones mentioned on Wikipedia and Wiktionary have a reference we could use to add that information to this page. I'll also note that there was (but is no longer) something in the article for Palatal lateral approximant which made /ɭ.ɭ/ (which Wiktionary uses over /l.l/, and seems to be all I've heard, including when asking a friend of mine where his tongue ended up at the end of /maɭ/) turn into [l̠ʲ.l̠ʲ] before /i/, so that might be worth researching into, too. Anyway, here's a potential table, with the emboldened allophones being ones I'm sure I've seen on one of these two websites:

Standard Before /i,j/ Before /ɯ,ɰ/ Before /o,u,w/
/k/ [kʲ] or [gʲ] (possibly [c] or [ɟ]?) [k] or [g]
/k͈/ [k͈ʲ] [k͈]
/kʰ/ [c͡ç] [k͡x] [kʰ]
/t/ (morpheme internal) [t̠ʲ] or [d̠ʲ] [t]
/tʰ/ (morpheme internal) [t̠ʲʰ] [tʰ]
/t/ (morpheme final) [d͡ʑ] [d]
/tʰ/ (morpheme final) [t͡ɕʰ] [tʰ]
/s/ [ɕ] [s]
/s͈/ [ɕ͈] [s͈]
/h/ [ç] or [ʝ] [x] or [ɣ] [ɸʷ] or [β] (why not [βʷ]?)
/ɭ.ɭ/ [l̠ʲ.l̠ʲ] (might also be [ɭ.l̠ʲ]) [ɭ.ɭ]
/n/ [n̠ʲ] [n]

There may of course be others. I believe two consonants in a row will both be palatalized if they're the same consonant (so 안녕 is [an̠ʲ.n̠ʲʌ̹ŋ]), but again, I don't have any source to back up my claim.

I think it might also be worth mentioning that when the syllable 의 is preceded by a syllable ending in a consonant then the /ɰ/ is sometimes still pronounced, such as in 논의, which is oft pronounced phonemically /no̞.nɰ̜i/. Additionally, /kj/, /k͈j/, and /kʰj/ often simply drop the /j/.

Feel free to add to that table!Blanket P.I. (talk) 22:14, 28 May 2018 (UTC)

Aspiration of /sʰ/[edit]

The aspiration of Korean phoneme /sʰ/ is very distinct and should never be ignored. The aspiration often makes /sʰ/ sound like [tsʰ]. I don't know why someone keeps deleting this. Mteechan (talk) 18:19, 16 June 2018 (UTC)

While /s/ is aspirated in word-initial position, it typically functions like a lax consonant, not an aspirated one. This can be seen in the fact that it is not aspirated word-medially. Furthermore, the aspiration of /s/ is more slight than that of the aspirated series. Additionally, it appears that it is beginning to become voiced word-medially, though I don't know if this is common outside of Seoul (as I have never heard it). The page does mention the aspiration of /s/, and notes that there is some controversy over whether it should be considered aspirated or lax. I also would not consider [sʰ] similar to [t͡sʰ], seeing as the affrication is more prominent than the aspiration (at least to my ears). Blanket P.I. (talk) 23:18, 10 July 2018 (UTC)

Another q on vowel length[edit]

The "Vowel" section says that short and long vowels have mostly merged. Which is the merged form of ㅓ more like, [ʌ] or [ɘː]? The article should make this clear. — Preceding unsigned comment added by (talk) 20:15, 2 September 2018 (UTC)

Difficult to say which one of [ʌ] (more open, more back) or [ɘː] (more close, more central) is more "correct", as 어 is a very wide-ranging vowel. Phonetically, 어 is often rounded to /ɔ/, and is being raised quite a lot in Seoul Korean [F1 = 470 Hz, F2 = 1370 Hz], approaching /o/ (causing a chain shift eventually, which is kind of happening now). In Pyongyang too, 어 is being raised, especially in female speech, as well as shifting back in the mouth. Michael Ly (talk) 16:16, 11 September 2018 (UTC)

Have ㅞ and ㅙ merged? Also, ㅖ and ㅒ[edit]

Have ㅞ and ㅙ merged along with ㅔ and ㅐ? Also, ㅖ and ㅒ. The diphthong table indicates not. The explanatory text should make it explicit one way or the other. — Preceding unsigned comment added by (talk) 21:02, 2 September 2018 (UTC)

Yes the diphthongs have also merged. Additionally, ㅞ and ㅙ have also merged with ㅚ (into /we/), in casual speech in Seoul especially. Michael Ly (talk) 15:15, 11 September 2018 (UTC)

Vowel table formatting[edit]

There is no evident explanation as to why some entries in the table are shaded, and why others have black borders. Please add an explanation, or remove the inconsistent formatting. —DIV ( (talk) 14:24, 25 April 2019 (UTC))

It could be useful -[edit]

to put Phonotactics befor the section of Consonant assimilation - because the assimilation is superseding on the phonotactical adjustments. — Preceding unsigned comment added by 2001:16B8:46AA:A200:F183:E34E:4D11:977C (talk) 17:19, 27 April 2019 (UTC)