Already having diverged from Korean by the 15th c., Jeju features many terms without obvious cognates in the mainland.

Of these, I focus on two anatomical terms—yakaki ‘neck’ and twokmolop ‘knee’—and explore potential etymologies based on historical sources and modern reflexes in various languages.

yakaki

The Jeju word yakaki ‘neck’ occurs in variation with yakayki and yakaci. Here, yakayki is a transparent derivation from yakaki via umlaut, and yakaci can be understood as an alternative suffix form *yak- + -aci c.f. yakaki < *yak- + -aki. Thus, we must investigate the origin of this root *yak-.

For this, I would like examine two Korean1 roots for ‘mouth’, ip and *aku. The former is not only found intact in Modern Korean but also thoroughly attested in Late Middle23 and even Early Middle4 Korean. On the other hand, the latter is more speculative; given the standard Modern Korean akali ‘mouth’, I propose a comparative examination of various dialectal (and historical) forms as follows5.

  1. akwu (NK6), akwu-thwui (NK), akwu-thingi (NK), akwu-thwong (NJ, SJ, NK), akwu-chang (SJ, NK), akwu-ttak (SJ), akwu-thayngi (NJ)
    1. akwuy (JJ), ip-akwoy7
    2. aku-ttak (SJ)
  2. akali (SJ, JJ, NP, NK), akwuli (SJ, JJ)
    1. akal-thwong (NK), akwul-ci (NJ)
  3. akwuci (SK, NJ, SJ), akwoci (SJ)
    1. akamci (SK)
  4. ak-thwungi (NP), ak-pal (SJ), ak-pali (NK), ak-cayki (NP)

The above list shows a grouping of the different word forms based on shared characteristics. Group 1 forms all appear to suggest a root akwu. Group 1.1’s akwuy, then, would best be analyzed as akwu + -i, with akwoy being its ablaut form. Group 1.2’s aku is similar to Group 1 except it lacks the rounding on the vowel; more on this later.

Groups 2 and 3 forms appear to be transparent derivations from a root *ak(V) with the suffixes -Vli and -Vci, respectively. Group 2.1 forms appear additionally to have undergone elision of the final -i and another step of compounding with suffixes -thwong and -ci. Group 3.1’s akamci would be best understood as a form of *akaci with nasal epenthesis.

Finally, Group 4 forms suggest a root *ak.

From this, we can see two things. First, akali and all of its alternate forms examined above can be explained through a common root *akV. Second, this common root *akV has a propensity to become *ak- in compounds, not only before vowels but also before consonants. This tendency makes it less likely that the original form would have had a labialized vowel as in akwu~*akwo. It stands to reason, then, that the original form was most likely *aku.

With that, I further propose that *yak from above is a compound of ip and *aku. In Middle Korean, we see a hypothetical cognate ip-akwoy, meaning ‘lip’8. We have reason to suspect that p-lenition was a productive change within Jeju, with words like iyaki ‘conversation’9, mipap ‘cooked rice’10, and -path ‘field’ having W > w exclusively in Jeju: iwak ‘conversation’, miwap ‘rice’, and was ‘-field’. Applying this to *ip-aku yields a hypothetical reflex *iWaku~iwaku.

Next, we consider *iWa~iwa > *ya. The elision (or assimilation) of w in between i and a does not seem particularly far-fetched, with examples such as Northern Pyeongan piat- ‘to spit’ (consider Middle Korean piWath-11 > *piwat(h)- > piat(h)-) and iyaki ‘conversation’ above. Such a change yields *yaku; we have already established above that the final u vowel in *aku has a propensity to be elided or absorbed in compounds, finally yielding the alternate *yak- as in yakaki.

Admittedly, ‘lip’ and ‘neck’ is not a common polysemy. However, considering broader senses of each root, ‘mouth’12 and ‘throat’13, we see that the colexification of this semantic pair is indeed attested.

Thus, I put forth the derivational chain Korean ip ‘mouth’ + *aku ‘mouth’ > Jeju *ipaku > *iWaku > *iwaku > *iaku > *yaku ‘mouth’ > ‘throat’ + -aki > * as the etymology of Jeju yakaki ‘throat, neck’.

twokmolop

Next, looking at this Jeju word for ‘knee’ and its variants, three things immediately stand out:

  1. -molop reflects Korean mwuluph ‘knee’ (< Middle Korean mwulwuph ‘knee’)
  2. There are many variant forms, including twokmeli, twok-mwulep, and twok
  3. There are compounds twokmolop-kkwang, tokmoli-kkwang, and toksayki-kkwang (and several other variations thereof), all meaning ‘kneecap’

From points 1 and 2, it is clear that twok-molop is a compound of twok and molop, both components independently meaning ‘knee’. Regarding point 3, the suffix -kkwang commonly appears in words for bones; specifically regarding toksayki-kkwang, the component sayki is also found in words such as kwol-paksayki ‘skull’, further supporting the idea that twok is an independent root meaning ‘knee’.

The alternate o vowel in tokmoli- and toksayki- are somewhat curious. However, given that these are the only two instances of it, it is probably best analyzed as the product of a local sound change (perhaps due to assimilation or neutralization) from the original twok form.

Thence, we can evaluate potential cognates. Most notably, Korean has twokani ‘kneecap, (animal’s) knee’, which seems like an obvious candidate; however, there is a striking lack of both historical and dialectal forms.

One may note some Northern Jeolla dialects, which feature twoka~twokani~twokaci ‘large clay pot’. twokani in particular would seem to be twok + **-ani, transparently linked to standard twok ‘large clay pot’ (< MK twòk14). While there is no direct semantic relationship between the two homophonous forms, the morphological parallel allows us again to extract a root twok 'knee'.

Now looking outside of Korean, a tempting candidate at first sight is Manchu takiya ‘animal’s knee’ and takitu ‘leather knee pad’. We can reasonably reconstruct a root *taki- meaning knee, noting also that the usage of takiya15 mirrors that of Korean twokani. This form is potentially related to Mongolian takim ‘back of the knee’, which in turn I posit to be a doublet of takiyi- ‘to bend’; considering ‘back of the knee’ and ‘to bend’ to come from the same root *taqi-, we would expect the original meaning of this root to be ‘to bend’, and for the ‘knee’ sense to be a later development that was borrowed into Manchu.

This is a close semantic parallel to Korean wokum ‘back of the knee (human)’, which originally meant bent parts of one’s body16 but is now almost exclusively used to mean the back of the knee. Further, while twokani’s poorly attested history hinders diachronic analyses, twokani and wokum are often seen as semantic counterparts in modern usage, the former being used for people and the latter for animals. However, we again find contrary phonological evidence; even overlooking the final -i, the expected Middle Korean reflex of a hypothetical Middle Mongolian *taq- would be *tak~thak~takh~thakh-, not twok-. This makes the connection less plausible.

Instead, we will investigate Mongolian toyiɣ ‘kneecap’. Unlike taqi-, toyiɣ is not only a semantic match but—seemingly—also a phonological one. While it is difficult to ascertain the precise phonological processes surrounding Middle Mongolian loanage into Korean, we can comparatively study other apparent loans to evaluate whether a Mongolian toyiɣ could have resulted in a Middle Korean reflex *twok.

Middle Mongolian t- is often associated with Korean th- as in them(w)ul < temür, thwikwon < toyiɣun, tharak < taraɣ, and thyellik < terlig. However, there are instances of Korean t- as well, such as takal < taqa and twokwom < toqom. The latter of the two, twokwom < toqom, is especially promising, since it is a Jeju dialectal form17. It also brings us to our second point, which is wo < o: This correspondence is attested in numerous other borrowings as well, including kwotwoli < godoli, kwongkwol < qongɣor, nwoyen < noyan, and many more. Skipping ahead a bit, our third point—the loanage of final -ɣ > -k—is also attested: tharak < taraɣ, aktay < aɣta, swulwuk < sürüɣ, and aymak < aimaɣ. This leaves us with just -yi- to consider.

-yi- poses somewhat of a challenge. For our proposed correspondence *twok < toyiɣ, we would need -yi- to map to ∅. I was only able to verify two examples of Mongolian borrowings containing -yi-: thwikwon < toyiɣun and *cyapwul18 < čayibur. In the former, -oyi- clearly maps to the diphthong -wi-, but in the latter, -ayi- seems to map to -a-19. This is by no means conclusive evidence that a yi > ∅ loanage occurred; there are too many variables in the phonological environment, namely the already existent y- glide in cy as well as the lack of a checked coda. Nevertheless, I posit that the deletion of -yi- either during or following the hypothetical borrowing of toyiɣ is not implausible.

With this, I put forth the derivational chain Middle Mongolian toyiɣ ‘kneecap’ > Middle Korean *twok ‘kneecap’ + *molop ‘knee’ > * as the etymology of Jeju twokmolop ‘knee’.

Afterword

It must be stressed that a lot of the theorization thus far lacks a solid evidential basis. This is somewhat of a natural consequence of working with such little data; regardless, it is my wish that the methodologies and ideas hitherto presented provide a basis for future research. At the least, I enjoyed the creative exercise.

Special thanks to the East Asian Linguistics Discord server for inspiration and support. Please feel free to drop in and say hi :-)

References

경남방언연구보존회 (2017). 경남방언 사전. 경남: 경상남도.

고경재 (2022). 12~15세기 국어 모음체계 연구ㅡ모음추이 이론의 재확립ㅡ
(박사학위논문). 고려대학교.

구본관 (1999). 축소 접미사에 대한 연구. 국어학 (국어학회), 34,
109-141.

김이협 (1981). 평북방언사전. 경기: 한국정신문화연구원.

문세영 (1940). 수정 증보 조선어사전. 서울: 조선어사전간행회.

서보월 외 (2019). 경북 북부지역 방언사전. 서울: 한국문화사.

소강춘 외 (2021). 전라북도 방언사전. 전북: 신아출판사.

이기갑 외 (1997). 전남방언사전. 전남: 전라남도.

제주문화예술재단 (2009). 개정 증보 제주어사전. 제주:
제주특별자치도.

최영선 (2015). 鷄林類事의 음운론적 연구 (박사학위논문).
전남대학교.

Norman, J. (2013). A comprehensive Manchu-English dictionary.
Massachusetts: Harvard University Asia Center.

Ryzmski, C. et al. (2020). The database of cross-linguistic
colexifications, reproducible analysis of cross-linguistic polysemies.
Scientific Data, 7, Article 13.

胡增益 (1994). 新满汉大词典. 新疆: 新疆人民出版社.

Санжеев, Г. Д., Орловская, М. Н., & Шевернина, З. В. (2016).
Этимологический словарь монгольских языков. Москва: Институт
востоковедения Российской Академии Наук.

  1. The “Korean” label is intentionally ambiguous. The values are to be treated as proto-forms for all attested reflexes (Late Middle Korean and later) rather than reconstructions of some specific historical Korean language.↩︎

  2. 용비어천가09:40b - 마ᅀᆞᆫ 사ᄉᆞ미 등과 도ᄌᆞᄀᆡ 입과 눈과 遮陽ㄱ 세 쥐 녜도 잇더신가↩︎

  3. 월인석보04:34b - 百千無量劫에 몸과 입과 ᄠᅳᆮ과ᄅᆞᆯ 조히 닷ᄀᆞ실ᄊᆡ 이런 모ᄆᆞᆯ 어드시니↩︎

  4. 계림유사 - 167. 口曰邑↩︎

  5. I personally suspect that akami ‘gill’ and akwungi ‘furnace’ also share the same root. Including these in the analysis would have expanded the number of data points; however, since I did not note any noteworthy characteristics in the dialectal forms of akami or akwungi that were not already captured by those of akali, they were deliberately omitted as to limit the scope of discussion.↩︎

  6. Northern and Southern Jeolla are abbreviated NJ and SJ, Northern and Southern Kyeongsang as NK and SK, Jeju as JJ, and Northern Pyeongan as NP.↩︎

  7. 훈몽자회37a - 吻 입아괴〮 믄〯↩︎

  8. 훈몽자회37a - 吻 입아괴〮 믄〯; 吻 is a character meaning ‘lip’, ‘corner of the mouth’.↩︎

  9. This word does not have a p in the central language, even its earliest attestation in the early 17th century having a glide: 서궁일기74b - 녜아기 사마 보려 ᄂᆡ인들이 잠간 긔록ᄒᆞ노라​. However, many dialectal forms have a p, suggesting that the iyaki form is the product of lenited p > W > w > y.↩︎

  10. While the standard language technically only recognizes ssalpap for this meaning, mipap is a transparent compound of Sino-Korean 米 mi ‘rice grain’ (equivalent to native ssal) and pap ‘steamed grain’.↩︎

  11. 월인석보17:71b - 妙覺明體ᄂᆞᆫ 샤ᇰ녜 제 ᄆᆞᆯ가 十方ᄋᆞᆯ 머구므며 비ᄫᅡ타 業을 조차 發現ᄒᆞᄂᆞ다 ᄒᆞ시니↩︎

  12. While the cognate ip-akwoy appears to mean ‘lip’ or ‘corner of the mouth’ in Middle Korean, since *yaku here is ultimately a compound of two roots meaning ‘mouth’, there is no reason why the original meaning could not have been ‘mouth’.↩︎

  13. Both Korean and Jeju colexify ‘neck’ and ‘throat’. In other words, yakaki (as well as its Korean counterpart mok(-aci)) can mean both ‘neck’ and' ‘throat’.↩︎

  14. 두시언해19:9b - 酒盃ㅣ ᄆᆞᄅᆞ니 도기〮 곧〮 뷔〯도다〮↩︎

  15. Both Manchu takiya and Korean twokani refer to the knees of animals as opposed to humans; these words may have filled a certain semantic gap.↩︎

  16. For instance, a sense meaning the inside of your elbow is still seen, albeit much less commonly.↩︎

  17. In fact, Jeju seems to possess a relatively high number of such Mongolian loanwords, making a Mongolian-based etymology for twokmolop synchronically appropriate.↩︎

  18. Here, *cyapwul is actually seen as cyampwul. However, given the phonemic environment (Vp) and the lack of a nasal in čayibur, the m is appropriately analyzed as nasal epenthesis, yielding *cyapwul as the original form. This is also consistent with the earlier Sinographic transcription 者佛.↩︎

  19. The glide in the cy- onset reflects Middle Mongolian č-.↩︎