There is never only one way to analyze the phonological structure of a language. The beauty of phonemic transcriptions is that one is allowed to be quite abstract in this analysis. For Korean, I’ve sketched out my idea for the phonemic transcription in the page IPA transcription of Korean, but it is possible to simplify this further by reducing the number of basic sounds in the inventory. Here I consider a more abstract, even minimalistic phonemic transcription of Korean.
Vowels and glides
In the phonemic transcription used elsewhere in this site, the following symbols are used for the vowels and glides of Korean: vowels: a ɛ ʌ e o ø u y ɯ i glides: j w ɰ The correspondence with hangul (Korean alphabet) orthography is as follows: ㅏ a / ㅐ ɛ / ㅑ ja / ㅒ jɛ / ㅓ ʌ / ㅔ e / ㅕ jʌ / ㅖ je / ㅗ o / ㅘ wa / ㅙ wɛ / ㅚ ø / ㅛ jo / ㅜ u / ㅝ wʌ / ㅞ we / ㅟ y / ㅠ ju / ㅡ ɯ / ㅢ ɰi / ㅣ i One way to reduce the number of basic symbols is to look at the design logic of the hangul symbols. For instance, ㅐ ae /ɛ/ was originally a combination of ㅏ a /a/ and ㅣ i /i/, and originally represented a falling diphthong. We can treat such historical diphthongs as sequences of vowels plus the glide /j/, which gives the following: vowels: a ʌ o u ɯ i glides: j w ㅏ a / ㅐ aj / ㅑ ja / ㅒ jaj / ㅓ ʌ / ㅔ ʌj / ㅕ jʌ / ㅖ jʌj / ㅗ o / ㅘ wa / ㅙ waj / ㅚ oj / ㅛ jo / ㅜ u / ㅝ wʌ / ㅞ wʌj / ㅟ uj / ㅠ ju / ㅡ ɯ / ㅢ ɯj / ㅣ i This reduces the number of vowel phonemes from ten to six, and the number of glide phonemes from three to two. We eliminate the need for the glide /ɰ/ by treating ㅢ ui as a vowel followed by a glide, /ɯj/, though today it is indeed pronounced as a glide followed by a vowel, /ɰi/. Similarly, ㅚ oe and ㅟ wi are written as /oj/ and /uj/ respectively, though today they are overwhelmingly pronounced as glide-vowel sequences /we/ and /wi/ [ɥi] (though normative grammar still favours the conservative monophthongal pronunciations /ø/ and /y/ as the basic pronunciations, treating /we/ and /wi/ as optional variants). This minimalistic approach is good for giving a more systematic view of the vowel system of the normative conservative pronunciation of Korean. Instead of requiring the glide /ɰ/ that only ever appears in the combination /ɰi/, we can explain this combination along with various monophthongs or glide-vowel combinations as the results of an underlying vowel and the glide /j/, which can be attached to any of the vowels except for /i/. However, because this phonological analysis is based on historical pronunciation and is very different from the pronunciation of today’s Korean, it would be very confusing for anyone who is not already familiar with the sounds of today’s Korean and has some awareness of the historical sound changes of Korean. Furthermore, today’s Korean is going through some changes such as the merger of ㅐ ae /ɛ/ and ㅔ e /e/ (or /aj/ and /ʌj/ in the minimalistic analysis we are discussing), the systematic nature of this analysis will continue to be eroded.
The phonemic transcription used elsewhere in the site recognizes nineteen consonant phonemes: consonants: m b p pʰ n d t tʰ l z̥ʰ s ʣʲ ʦʲ ʦʲʰ ŋ ɡ k kʰ h ㄱ ɡ / ㄲ k / ㄴ n / ㄷ d / ㄸ t / ㄹ l / ㅁ m / ㅂ b / ㅃ p / ㅅ z̥ʰ / ㅆ s / ㅇ ŋ / ㅈ ʣʲ / ㅉ ʦʲ / ㅊ ʦʲʰ / ㅋ kʰ / ㅌ tʰ / ㅍ pʰ / ㅎ h We can be much more economical, treating the fortis series of stops and affricates as geminates of the plain series and the aspirated series as clusters of the plain series and /h/. Then we can reduce the number of consonant phonemes from nineteen to just ten, slashing it nearly in half: consonants: m p n t l s ʦ ŋ k h ㄱ k / ㄲ kk / ㄴ n / ㄷ t / ㄸ tt / ㄹ l / ㅁ m / ㅂ p / ㅃ pp / ㅅ s / ㅆ ss / ㅇ ŋ / ㅈ ʦ / ㅉ ʦʦ / ㅊ ʦh / ㅋ kh / ㅌ th / ㅍ ph / ㅎ h I first though of using the voiced symbols /b, d, z, ʣ, ɡ/ for the stops and affricates, but decided not to make it too alien and settled on the voiceless symbols /p, t, s, ʦ, k/ instead. And since we’ve already stepped too far into the realm of phonemic abstraction, I’ve stripped away the various diacritics for phonetic details from the phonemic symbols, writing /s, ʦ/ instead of /z̥ʰ, ʣʲ/. To put it bluntly, these phonemic transcriptions are not meant to give information directly about the pronunciation at all, just the underlying structure of the sound system of the language that may give rise to the pronunciation.
Let’s see this alternative phonemic transcription in action, and for fun, let’s give both the morphophonemic representation that is closer to the spelling and a form that is closer to the surface pronunciation. We can write /ʔ/ for fortition that is not reflected in the spelling. Here we give the alternative phonemic transcription between bold slashes (/ /), followed by the more conventional phonemic (/ /) and phonetic ([ ]) transcriptions used elsewhere on the site.
- 꺾쇠[꺽쐬/꺽쒜] kkeoksoe /kkʌkksoj → kkʌksoj/ /kʌɡ.sø/ [kʌk.sø, -swe]
- 먹히다[머키-] meokida /mʌkhita/ /mʌ.kʰi.da/ [mʌ.kʰi.da]
- 사과 sagwa /sakwa/ /z̥ʰa.ɡwa/ [z̥ʰa.ɡwa]
- 아빠 appa /appa/ /a.pa/ [ap.pa]
- 채찍 chaejjik /ʦhajʦʦik/ /ʦʰɛ.ʦiɡ/ [ʦʰɛʦ.ʦik]
- 폈다[펻ː따] pyeotda /phjʌːss-ta → phjʌːt-ta/ /pʰjʌːd.ta/ [pʰjəːt.ta]
- 필승[-씅] pilseung /philʔsɯŋ → philssɯŋ/ /pʰil.sɯŋ/ [pʰil.sɯŋ]
One advantage is that it is somewhat easier to see some of the logic behind contractions and regular sound rules:
- 아니하고 anihago /anihako/ /a.ni.ha.ɡo/ [a.ni.ɦa.ɡo]
- 않고[안코] anko /anhko → ankho/ /an.kʰo/ [an.kʰo]
- 하였다[-엳따] hayeotda /hajʌss-ta → hajʌt-ta/ /ha.jʌd.ta/ [ha.jʌt.ta]
- 했다[핻ː따] haetda /haːjss-ta → haːjt-ta/ /hɛd.ta/ [hɛt.ta]
For example, it is easier to see why /anihako/ contracts to /anhko/, or why the latter turns into /ankho/ through simple metathesis. This can be slightly more obscure if we just say that /a.ni.ha.ɡo/ contracts into /an.kʰo/. Granted, this isn’t much of a difference, but it could be helpful in thinking about the phonology of Korean. In fact, if you are acquainted with the Yale romanization of Korean, then this alternative scheme of phonemic transcription should be familiar. The logic of the Yale romanization, which is used extensively in linguistics, is similar to this analysis. For example, the examples given above would be written in Yale romanization as follows: 꺾쇠 kkekksoy, 먹히다 mekhita, 사과 sakwa, 아빠 appa, 채찍 chayccik, 폈다 phyessta, 필승 philqsung, 아니하고 anihako, 않고 anhko, 하였다 hayessta, 했다 hayssta. The Yale romanization goes even further in the simplification of the vowel inventory, by treating the back rounded vowels ㅗ /o/ wo and ㅜ /u/ wu not as basic vowels but as combinations of the glide /w/ and the basic vowels ㆍ /ɒ/ o and ㅡ /ɯ/ u. The vowel phoneme ㆍ /ɒ/ is no longer used in Standard Korean and does not figure into the phonemic analysis above. This means that the implicit system of the Yale romanization is as follows, at least if we replace our /o, u/ above with /wɒ, wɯ/: vowels: a ʌ ɒ ɯ i glides: j w ㅏ a / ㅐ aj / ㅑ ja / ㅒ jaj / ㅓ ʌ / ㅔ ʌj / ㅕ jʌ / ㅖ jʌj / ㅗ wɒ / ㅘ wa / ㅙ waj / ㅚ wɒj / ㅛ jwɒ / ㅜ wɯ / ㅝ wʌ / ㅞ wʌj / ㅟ wɯj / ㅠ jwɯ / ㅡ ɯ / ㅢ ɯj / ㅣ i But the Yale romanization actually writes ㅛ as yo instead of ywo, ㅠ as yu instead of ywu, and ㅟ as wi instead of wuy (this last bit would imply that ㅟ should be analyzed as /wi/ instead of /uj/—this raises an interesting question of which is more historically correct, to which I don’t have the full answer). The priority seems to have been to use as few letters as possible without creating ambiguities rather than strict internal consistency (indeed, for modern Korean where ㆍ /ɒ/ isn’t used, the Yale romanization simply writes ㅗ as o instead of wo). The Yale romanization’s analysis with five basic vowels was probably forced by the restriction of having only five basic vowel letters in the Roman alphabet, with y being chosen to represent a glide, rather than by the nature of the Korean sound system itself. The alternative phonemic analysis discussed in this post is much too abstract and quite removed from the surface pronunciation of Korean, so it won’t be of much use when the goal is simply to describe how something is pronounced in Korean. The phonemic transcriptions on this site are aimed at straying not too far from the phonetic transcriptions. But it is possible that the exact level of phonemic analysis may be reviewed in the future, and we may incorporate some of the ideas here, though not all.