220.127.116.11:54: THREE GRASSESThe Old Japanese word 三枝 sakikusa 'grass with three-branched stems' has an unusual spelling:
三 'three' is never read saki except in this one word
枝 'branch' is read like 草 kusa 'grass'
Kusa 'grass' is a native Japanese word. However, saki- 'three' is of peninsular (i.e., Koreanic*) origin. There are two strange things about it.
First, Vovin (2010: 181) reconstructed Old Korean *səki 'three'. Old Japanese had both ə as well as a. Why wasn't OK *səki borrowed into Old Japanese as səki?
Second, why is a foreign prefix attested solely in a compound with a native word? It is as if Russian troj- 'three' appeared only in front of some Anglo-Saxon word rather than before the Russian suffix -ka in the loanword
The vowel problem
troika < R тройка trojka
OJ was in contact with Paekche, a 'sister' to Old Korean. So it's possible that Paekche and OK had related but somewhat different words for 'three'. The Paekche word for 'three' may have been *saki or *sʌki with a first vowel lower than the schwa of OK *səki. Are there other Koreanic loanwords with OJ a instead of ə corresponding to Korean ə? Such words may reflect Paekche vocalism.
This solution brings up another problem. OJ speakers borrowed many Chinese character readings from Paekche. Many of those readings had ə in OJ, but some had -ai where an -əi is expected: e.g., 每 mai corresponding to Middle Chinese *məjʔ. Perhaps 每 mai and saki 'three' were borrowed from Paekche after it had shifted *ə(C)i to *a(C)i. (It's also possible that MC *-əj was actually *-ʌj with a lower mid vowel.)
The compound problem
One might suggest that sakikusa was borrowed as a whole from Paekche, just as troika was borrowed as a whole from Russian.
However, kusa is reconstructible in Ryukyuan as well as in mainland Japanese, so *kusa can be reconstructed in Proto-Japonic. Could PJ *kusa be a loan from some Koreanic languge? There is no evidence for a *kusa-like word for 'grass' in Koreanic. The closest match is Middle Korean 곶 koc 'flower'. It would be a semantic and phonetic stretch to link PJ *kusa 'grass' to Middle Korean koc 'flower'.
- Are there precedents for 'flower' becoming 'grass', or vice versa?
- If an early Koreanic *koc were borrowed into Proto-Japonic, I would expect a PJ *kotV, not PJ *kusa.
Perhaps sakikusa originated as a clarifying compound:
Paekche *saki-X 'three-?' > *saki 'the three' (a kind of grass) > OJ saki-kusa 'saki grass'
(Cf. Eng a one, a five, a ten, a twenty, etc. referring to types of dollar bills.)
In ancient Japan, sakikusa was associated with good fortune. Was this belief also borrowed from Paekche, or was it the result of coincidental homophony between Paekche saki 'three' and the native Japanese word saki 'happiness'? There is another spelling 福草 'lucky grass' for sakikusa. Would it have made sense to the Paekche?
The Late Old Chinese reading of 三枝 'three branches' is *samkie. Its resemblance to saki is presumably coincidental.
ADDENDUM: 三枝 can also be a surname Saigusa or Saegusa with irregular medial -k- loss. Cf. kai < kaki in 垣間見 kaimami 'peeking', lit. 'fence-space-see'. Medial voicing (e.g., -k- > -g-) is common in Japanese compounds. But I don't understand why Saegusa has -e- instead of -i-.
There are other readings of 三枝 which have nothing to do with Koreanic: Mie, Mitsue, Mieda, Mitsueda. These names are all compounds of native Japanese roots: mi(tsu) 'three' and e(da) 'branch'.
*I use 'Koreanic' as a cover term for the non-Japonic language(s) spoken on the Korean peninsula. Korean is the descendant of one of these language varieties (Shilla). It is not clear whether the other varieties (e.g., Paekche) were mutually intelligible dialects of Korean or distinct languages.
18.104.22.168:20: 口 MOUTH + 寸 HAND = ?
I am intrigued by Chinese characters that represent more than one root. 吋 has two parts (口 'mouth' and 寸 'hand, Chinese inch'*) and two different meanings. What are its two meanings?
Hint: Why am I posting about 吋 after three posts about 叱 < 口 'mouth' plus 七 'seven'**?
Select the blank area below for the answers.
1. 吋 Mandarin dou is a synonym of 叱 Md chi 'to scold'. 口 'mouth' is semantic. 寸 is phonetic. It is probably taken from 討 Md tao 'to punish', whose phonetic 寸 is in turn an abbreviation of 肘 Md zhou < *tuʔ 'elbow', a combination of 月 'flesh' (not 'moon' here) + 寸 'hand'.
2. 吋 Mandarin (Ying)cun 'English inch' is a special spelling of 寸 'Chinese inch'. There's nothing English (英 Md Ying) about 口 'mouth' which is an arbitrary addition also found in
呎 Md (Ying)chi 'English foot'; cf. 尺 Md chi 'Chinese foot'
哩 Md (Ying)li 'English mile'; cf. 里 Md li 'Chinese mile'
These are the only characters that I know of with optional disyllabic Mandarin readings. Are there others?
These words can also be written as 英寸, 英尺, 英里.
*The normal Chinese character for 'hand' is 手. 寸 is a drawing of a 'hand' used for Old Chinese *tshuns 'thumb' (and by extension, 'Chinese inch' < 'span of part of a thumb'?). The dot 丶 represents the thumb.
**Although 叱 generally represents earlier Korean *s, Nam Phung-hyŏn thought it also represented *c(ɯr) in 鄕藥救急方 Hyangyak kugŭppang (mid-13th c. AD), "a compilation of herbal prescriptions for emergency treatment" that is "the oldest Korean medical treatise that has been preserved" (Lee and Ramsey 2011: 81).
Earlier Korean 叱 *-s is a genitive marker that has been compared to the Old Japanese genitive marker -tu even though the correspondence K *s : J t is less regular than K *c : J t. Perhaps the original Korean genitive was *-c [ts] which underwent irregular reduction to *-s, though this change was not reflected in its spelling (i.e., 叱 was not replaced by a Chinese *s-graph).
Alexander Vovin (2010: 49-53) independently came to a similar conclusion, reconstructing Old Korean *ci and "a possible dialectal (Paekche?) variant" *-cɨ. The vowels of his two forms may be based on the vowels of the two Sino-Korean readings of 叱: chil and chŭl. He regards Old Japanese -tu as a loan from Korean.
However, I would expect *ci or *cɨ to have been borrowed into OJ as -si, not -tu: cf. OJ sasi < K *cas 'fortress'. The correspondence of OJ t to K c may be characteristic of a layer of borrowings into Japonic long predating OJ, possibly going back to a period when Japonic was still on the peninsula. But even if the consonants can be reconciled, I don't know of any examples of K *i or *ɨ (*ɯ in my notation) corresponding to OJ u. So K *-c(i/ɯ) and OJ -tu may not be related.
22.214.171.124:00: THE 七 SEVEN-口 MOUTHED WOLF
The 叱 'scold' character used to write earlier Korean s was used to write the first syllable of the earliest attestation of 'wolf' in a Mongolic language:
Early Middle Chinese (EMC) *tɕhitno
for Xianbei *cino?
Janhunen (2003: 9) reconstructed Proto-Mongolic (PM) *cïno 'wolf' with a nonpalatal first vowel *ï. The EMC transcription *tɕhitno differs from this reconstruction in two major ways:
- the EMC transcription has *-i- corresponding to PM *ï
but note Sino-Korean chŭl** for 叱 implying EMC *-ɨt
- the EMC transcription has *-t- corresponding to zero in PM
Should we conclude that PM 'wolf' was actually *citno with a cluster *-tn- that was simplified in all other Mongolic languages? No. The Chinese practice of derography (the use of derogatory characters like 叱 EMC *tɕhit 'scold' and 奴 EMC *no 'slave') may obscure the phonology of the underlying word. Derographic transcriptions are not necessarily phonetically accurate. 叱 *tɕhit was the closest match with a derogatory meaning for the first syllable of 'wolf'. A nonderographic transcription of *cïno might have been something like
EMC *tɕhɨʔno 'tooth-child***'
Derographs for non-Chinese words were later used by non-Chinese to write their own languages: e.g., 叱 'scold' for earlier Korean s and 奴 for earlier Korean no and Japanese nu. (Hiragana ぬ nu and katakana ヌ nu are based on 奴. 奴 and 又 were also kugyŏl symbols for no and ro.)
The question of whether the Xianbei word for 'wolf' had palatal *i (as in later Mongolic) or nonpalatal *ï (preserved intact from proto-Mongolic) can only be resolved by looking for EMC *ɨ : Proto-Mongolic *ï correspondences in other transcriptions of Xianbei words.
In my last entry, I amused myself (and probably no one else) by proposing a (nonexistent) relationship between Korean 꾸짖- kkujich- < <skucic-> 'to scold' (used to gloss 叱) and its vague soundalike English scold. Proto-Mongolic (PM) *cïno has its own vague soundalike: Japanese inu 'dog'. It would be easy to derive both from a 'Proto-Altaic' (PA) *jino or the like:
'PA' *j- > *y- > Jpn Ø-
cf. Whitman's Proto-Koreo-Japonic *jip > Kor cip, Old Jpn ipe 'house'
'PA' *i > PM *ï (harmonizing with the following nonpalatal vowel *o)
'PA' *o > Jpn u
But it wouldn't be correct:
PM has a *j distinct from *c. I would expect 'PA' *j to become PM *j-, not the *c- of PM *cïno.
Serafim (1999) reconstructed Proto-Japonic *ʔeno 'dog' with a *ʔe- that cannot be reconciled with PM *cï- or 'PA' *ji-.
Although some Jpn u are from Proto-Japonic *o, the final vowel of Jpn inu may not be one of those -u. Vovin (2010: 33) reconstructed PJ *inu 'dog'.
*Chinese aspirates normally correspond to Mongolic voiceless obstruents: e.g., the word for 'wolf' was transcribed centuries later in The Secret History of the Mongols as Old Mandarin 赤那 *tʂhinɔ. (赤 'red' and 叱 'to scold' were homophones in Old Mandarin.)
**叱 has two Sino-Korean readings:
질 chil (with ch- instead of the expected aspirated chh-)
즐 chŭl (same unusual initial; with -ŭ- instead of -i- for Middle Chinese *-i-)
The unaspirated initials may indicate that these readings were borrowed before Korean developed phonemic aspiration.
***Do these (near-)homophones form a family?
女 OC *rna 'woman'
奴 OC *na 'slave' (with a drawing of a 又 hand on the right)
孥 OC *na 'wife and children' (with 子 'child' on the bottom)
駑 OC *na 'bad horse' (with 馬 'horse' on the bottom)
Not all words written with 女 or 奴 as phonetics are related: e.g.,
如 OC *na 'to resemble' (with 口 'mouth' on the right)
笯 OC *na 'cage' (with 竹 'bamboo' on top)
the second half of 蘮蒘 Late Old Chinese *kɨas-na 'a kind of plant' (with 艹 'grass' on top, 口 'mouth' on the right, and 手 'hand' on the bottom; not attested in earlier Old Chinese)
努 Late Old Chinese *naʔ 'to exert strength' (with 力 'strength' on the bottom; not attested in earlier Old Chinese)
怒 OC *nas 'anger' (with 心 'heart' on the bottom)
弩 OC *naʔ 'crossbow' (with 弓 'bow' on the bottom; probably a loanword from some Mon-Khmer language: cf. Khmer ស្នា snaa 'crossbow')
are nearly homophonous but not related.
126.96.36.199:02: 口 MOUTHS 七 SEVEN, RELATIONSHIP ZERO
Bart Mathias, a historical linguist specializing in Japanese, once joked that he could demonstrate that English refrigerator and Japanese 冷蔵庫 reizouko were cognates, even though the latter word actually has nothing to do with English. 冷蔵庫 reizouko is a compound of three roots borrowed from Chinese:
冷 rei 'cold'
蔵 zou 'storage'
庫 ko 'vault'
In the spirit of that joke, I could demonstrate that English scold and Korean 꾸짖- kkujich- 'to scold' (the native Korean translation of Chinese 叱 from this post) are cognates:
- Korean kk- comes from sk-, which matches the sc- of scold
- Korean u is a nonlow back rounded vowel like the -o- of scold- Korean -ji could be from *-lti from an even earlier *-ldi like the -ld of scold
(Asterisks indicate reconstructed forms.)
- Korean -ch doesn't correspond to anything in English, but it may be a suffix, since there is no -ch in Korean 꾸지람 kkujiram 'scolding'
Pre-Korean *skuldi- is a nice match for English scold. But this similarity is illusory.
Although it is true that Korean kk- is from sk- and -ji could be from *-lti, the earliest attested forms of 꾸짖- kkujich- are 구짖- <kucic-> and 구짇- <kucit->.
Initial <s-> was added to 구짖- <kucic-> and 구짇- <kucit-> later, fusing with the following <k-> to become the modern initial kk-. It is unlikely that *sk- became <k-> and then 'regrew' an <s->.
The spellings 구짖- <kucic-> and 구짇- <kucit-> predate the shift of 디 <ti> to ji. So if the word really had <ti>, its earliest spelling should have been 구딪 <kutic-> ~ 구딛- <kutit->, not 구짖- <kucic-> ~ 구짇- <kucit->. Early <ci> is from *Cci, not *lti or *ldi. (*C is an unknown consonant.)
Conclusion: Pre-Korean *kuCciC- is no longer much of a match for English scold. And even if it were a match, a single pair of lookalikes is not sufficient to demonstrate a 'genetic' relationship between languages. In fact, there are many good matches between Korean and English, but even they are insufficient to demonstrate a 'genetic' relationship between them because they are recent loanwords: e.g.,
스카우트 sŭkhauthŭ 'scout'
스커드 Sŭkhŏdŭ 'Scud' (missile)
스커트 sŭkhŏthŭ 'skirt'
The fact that Korean and English share a name for a modern kind of missile tells us nothing about their (nonexistent) prehistoric relationship.
ADDENDUM: The noun 꾸지람 kkujiram is first attested as <kuciram>. It looks like a noun derived from an unattested verb *kucira-.
The stem *kucira- is unusual because it violates vowel harmony: u and a belong to different vowel classes
|higher vowel||i (neutral)||ɯ||ə||u|
so I would have expected *kocira- or *kucirə-. (I suspect that the i associated with lower vowels was once *e: *kocira- < *kocera-?).
In early hangul texts with pitch marking, <kuciram> has a low-low-low pitch pattern whereas <kucic-> and <kucit-> have low-high patterns. This mismatch may indicate that the words are unrelated in spite of a shared low-pitched first syllable <ku>. Or did something alter the pitch of the second syllable in <kuciram> but not <kucic->/<kucit-> or vice versa?
Even if one assumes these words shared a root <kuci-> with a single pitch pattern followed by suffixes, I cannot explain the functions of <-ra> and <-c> ~ <-t>. The latter alternation occurs within Wŏrin sŏkpo vol. 17.
is Thai for 'McDonald's':
Even if you don't know Thai, you can use logic to figure out the Thai alphabet.
Compare Maekdoonan to
Mikaa'in*for the Russian surname Mikhail.
These two words should enable you to identify the Thai letters for m, i, k, l, and d.Hint 1: Thai letters are generally written from left to right, but there are exceptions.
Hint 2: Look for recurring patterns. The Thai letters for m, i, k, l, d all occur more than once.Hint 3: Don't confuse the Thai letters for k and d.
These words should help you identify the other letters:
โอ, นา, แนม, มัน, มิลล์
'oo 'pomelo', naa 'rice paddy', naem 'to add', man 'oil', Min 'Mill'
What is unusual about the Thai vowel symbols oo, ae, and i?
Why is the sound n spelled three different ways at the end of a word?
Does the symbol atop the letter ล in -ลด์ and -ลล์ stand for s?
-ลด์ in Maekdoonan 'McDonald's'
-น in man 'fat'
-ลล์ in Min 'Mill'
Last question: Which English name does this Thai spelling represent?
ADDENDUM: The Thai McDonald's site is promoting
with -ฟ-ว- -pw- for English -v-.
My understanding is that this pattern is only supposed to be used for ambisyllabic -v-: e.g., in
lipwiŋsatoon 'Livingstone' (example from thai-language.com)
the -v- of the English original is the end of liv- and the beginning of -ving-. li- has a stressed short vowel, and English doesn't have syllables ending in such vowels, so -v must be at the end of that syllable. liv- (pronounced like live) in English, but li- without -v isn't.
However, the -v- of Saver isn't ambisyllabic. Sa has a long a and needs no final consonant. Saver is not sav-ver. I would have expected
with -ว -w- for English -v-. Perhaps -ฟ-ว- -pw- for English -v- is by mistaken analogy with some word like lipwiŋsatoon for Livingstone which really does have an ambisyllabic -v-.
*I'd normally romanize this as Mikhaain, but I left out the -h- and inserted an apostrophe to make this exercise simpler.
188.8.131.52:54: HA-NDS OVER MOUTH
I found this blog when looking for examples of the 口訣 kugyŏl character sequence
丷 hʌ- 'to do' (an abbreviated drawing of a hand; see "Eight Ha-nds") plus
口 ko (an abbreviation of 古 ko 'old' used as a phonetic symbol)
for earlier Korean hʌ-ko (modern 하고 ha-go) 'do, and ...' The blog has photos of a 1990 edition of the 春秋 Spring and Autumn Annals in Classical Chinese with tiny kugyŏl notes added next to words to help Korean readers:
CLASSICAL kugyŏl CHINESE kugyŏl
Here's an example of kugyŏl from Lee and Ramsey (2000: 52) including 丷 hʌ- with a different verb ending -ヒ -ni:
In the multitude of the myriad things midst heaven and earth at that place man he最貴丷ヒ所貴乎人者ㄗ以其有五倫也亽
is the most noble and so: what is noble in man it is his possession of the Five Human Relationships it is.
This passage is from the Classical Chinese primer 童蒙先習 Tongmong sŏnsŭp (First Instruction for the Young and Ignorant). The kugyŏl serve as Korean training wheels. Students in ancient Korea were eventually expected to read Chinese without the assistance of kugyŏl.If Russian were taught to English using something like kugyŏl, beginners would see sentences like
Я am a лингвист.
Ja am a lingvist.
'I (am a) linguist.' (Russian has no 'a' and does not need 'am', but English does.)
𠮦 hʌ-ko 'do, and ...' is third from the right in this list of kugyŏl characters. Notice how a vertical stack of kugyŏl characters can look like a single Chinese character*:
丷 hʌ- atop 口 ko = 𠮦 hʌ-ko
This is reminiscent of the stacking in the Khitan small script, though the latter does not allow pure vertical stacking. Moreover, although Chinese characters always represented syllables in Korean, some kugyŏl characters could represent single consonants like some Khitan small script characters: e.g.,
kugyŏl 叱 and Khitan small script
both represent the consonant [s]**. The top half of even happens to resemble the hangul letter ㅅ s dating three centuries after the fall of the Khitan Empire in 1125***. It's not clear when kugyŏl was first created, but perhaps it influenced the Khitan small script - or vice versa. In any case, there must have been Koreans who were literate in the scripts of the neighboring Khitan Empire and the stacking and alphabetic principles were taken to the next level in the hangul alphabet invented in the 15th century.
When I first learned about kugyŏl back in the 80s, I was struck by its similarity to Japanese katakana and assumed that katakana must be based on a peninsular prototype. However, closer examination reveals differences as well as similarities: e.g. (examples from Lee and Ramsey 2000: 52),
Same shape, same readings, same sources of abbreviations
イ kugyŏl and katakana i (both abbrev. of 伊)
タ kugyŏl and katakana ta (both abbrev. of 多)
Similar shapes, different readings, different sources of abbreviations
丷 kugyŏl hʌ- (abbrev. of 爲) vs. ソ katakana so (abbrev. of 曾)
ロ kugyŏl ko (abbrev. of 古) vs. ロ katakana ro (abbrev. of 呂)
ヒ kugyŏl ni (abbrev. of 尼) vs. ヒ katakana hi (abbrev. of 比)
Although I have no doubt that writing in Japan has peninsular roots, I am less certain that katakana is directly derived from kugyŏl. The two systems may be siblings rather than mother and child. Katakana has no Koreanisms - character choices that only make sense in Korean: e.g., 丷 < 爲 for Korean hʌ- rather than for any reading of 爲 in Japanese: i, na, se, su, tame.
*In fact, 𠮦 hʌ-ko 'do, and ...' looks exactly like a rare variant of the Chinese character 召 'to summon', though I doubt the resemblance was intentional.
**I'm not sure why 叱 Middle Chinese *tɕʰit 'to scold' represents [s] in early Korean. Starostin reconstructed *sʰit as a possible Old Chinese reading of 叱, but his *sʰ- normally becomes MC *tsʰ-, not *tɕʰ-. Conversely, MC *tɕʰ- normally comes from *th-, but 七 'seven' is not a normal *th-phonetic. (The left-hand element 口 'mouth' is semantic.) The native Korean word for 'to scold', 꾸짖- kkujic-, has an earlier spelling <skucic-<, but its earliest hangul spelling is an s-less 구짖- <kucic> which still long postdates the use of 叱 for [s]. Did Korean once have a native word for 'to scold' with initial s-? Could such a word be the source of Japanese shikar- 'to scold'?
***The Khitan scripts survived the end of the Khitan Empire and were used by the Jurchen up through the end of the 12th century.
184.108.40.206:07: MANUAL SEE-MANTICS
There are three Chinese characters
䙸 䙷 㝶
consisting of drawings of
All three characters (䙸䙷㝶) meant the same thing. Can you guess their shared meaning? Select the blank space below for the answer.
an 目/罒 eye on 儿 two legs* = 見 'to see' +
a 寸 hand (now used for 'Chinese inch')
They are variants of 㝵 'to get' (from this post) which looks like
旦 'dawn' (sounds almost like its Mandarin reading dan!), a drawing of a 日 sun over a 一 horizon atop
a 寸 hand
'To get' is usually written 得 with 彳 'left step'** on the left. Why? And what does 見 seeing or the 旦 dawn have to do with getting? Nothing, apart from the fact that they are distortions of a 貝 cowry representing money in the original graph for 'get', a drawing of a 寸 hand grasping for 貝 money. 彳 was added later. Morohashi et al. (1992: 323) explained 得 as representing someone 彳 walking and picking up 貝 valuables with his 寸 hand.
ADDENDUM 1: The characters 䙸䙷㝶 appear in the Longkan shoujian, a dictionary of over 26,000 Chinese characters including rare variants compiled by a Chinese monk in the Khitan Empire.
ADDENDUM 2: 目 'eye' has a number of odd variants. Some redundantly incorporate 目 'eye'.
*罒 atop 儿 may also be a variant of 四 'four'.
something resembling 㓁 'net' (but with 冂 instead of 冖) atop 山 'mountain'
𠫕 = 乙 'second heavenly stem' + 厶 'private'
𡆲 = 囗 with 亡 'to die' inside
囗 with 人 'person' + 日'sun' inside (the name Mata Hari 'eye (of the) day' comes to mind!)
𡇡 = resembling 四 'four' with 目 'eye' inside and below the 儿
𥃦 = 人 'person' atop 目 'eye'
里 'village' (?) next to 目'eye'
**There is an expression 彳亍 Md chi chu 'walk slowly' written with 'left step' and 'right step'. Together they form 行 Md xing 'to go'.
Some speak of the 214 radicals of Chinese dictionaries as if they were the keys to Chinese writing, but some 'radical' assignments are arbitrary: e.g., 亍 is filed under 二 'two' even though it shares nothing with 'two' apart from two horizontal lines.The earliest attestation of 亍 by itself (as opposed to being a part of 行) that I can find is in Shuowen.
彳亍 was a reduplicative expression in Late Old Chinese: *ʈhiek-ʈhuok/h, perhaps from an unattested Early Old Chinese *hrek-hrok(-s) or *threk-throk(s). The latter could be cognate to 躇 *thrak 'to jump over, to pass over' with an e-o vowel pattern common in reduplicative expressions. Sagart (1999: 137) compared these e-o expressions to English riffraff, seesaw, and zigzag.
220.127.116.11:27: EIGHT HA-NDS
In my quest for the top half of 益 in Unicode or a shape siimlar to the Khitan small script character:
I found 丷 in in Andrew West's BabelMap freeware.
Google's top recommended search for 丷 was 丷怎么打 'How do you type 丷?' I'd like to know why people want to type such an obscure character.
zdic.net identifies 丷 as an old variant of 八 'eight', now pronounced ba in Mandarin. I'm surprised that it's available as a choice for ba in Windows 7's simplified Chinese input method. (It's listed last.)
丷 is also a kugyŏl character for the earlier native Korean verb stem hʌ- (now 하 ha-) 'do'*. The kugyŏl character 丷 has nothing to do with any Chinese word for eight. It originated as an abbreviation** of the Chinese character
a drawing of "a hand [爪] at the head of an elephant (possibly referring to the handicraft in ivory, so prominent in YIn time?)" (Karlgren 1957: 27). What do elephants have to do with doing?
Vietnamese voi 'elephant' resembles Old Chinese *waj 'to do'. Perhaps a word similar to voi was borrowed into Chinese from some Southeast Asian language and a drawing of an elephant 爲 was used to write an unrelated Old Chinese (near-)homophone 'to do' (Schuessler 2007: 510), just as a drawing of a scorpion 萬 was used to write an unrelated Old Chinese (near-)homophone 'ten thousand'. A scorpion is easier to draw than ten thousand dots, and an elephant is easier to draw than the abstraction 'to do'.
On the other hand, Morohashi et al. (1992: 555) identified the animal under 爪 as a monkey. Monkey see, monkey do?
爫, an abbreviation of 爲, is the Vietnamese nôm character for the native Vietnamese word làm 'to do'. 爫 retains two more strokes than its Korean counterpart 丷. There is also a more complex spelling 濫 recycling a nearly homophonous character 濫 for Sino-Vietnamese lạm 'to overflow'. There is even a hybrid spelling 氵+爫 (PNG image) combining the water radical 氵 of 濫 with the 爫 from 爲. One could consider 氵+爫 to be a phonetic-semantic compound:
氵 (the sound of 濫 lạm + 爫 (the meaning of 爲 'to do') = 氵+爫 làm 'to do'
*Middle Korean hʌ- 'to do' and Old Japanese su 'do' have been derived from a Proto-Koreo-Japonic *sho- (Whitman 1985: 235), but Vovin (2010: 186-187) rejected this etymology. He regarded the OJ stem of su as se- which could be from Proto-Japonic *sia- or *sai-. The latter resembles the longer MK stem hʌy-, so I am hesitant to completely reject this etymology even though I am still a skeptic of a genetic Koreo-Japonic relationship. Is there any language which borrowed 'to do' from another language?
**Here's a table with many more kugyŏl abbreviations.