This morning I found the spelling


<hrāḥlak> [ʃá lɛʔ]

for 'Charlotte' in the Burmese Wikipedia. I don't know anything about how foreign names are spelled in Burmese, so I wonder

1. Is ရှ <hr> [ʃ] the conventional spelling of foreign [ʃ]? Is ယှ <hy> [ʃ] only for native words?

2. Why does the first syllable have a high tone indicated with း <ḥ>? What determines tones of loanwords from atonal foreign languages?

3. Why did *a front to [ɛ] before back *-k? In Lhasa Tibetan, a fronted before coronals but not velars:

*-ak > Burmese [ɛʔ] but Lhasa Tibetan [aʔ]

*-at > Burmese [aʔ] but Lhasa Tibetan [ɛʔ]

4. Is foreign final t always approximated as Burmese [ʔ]? What is the treatment of other foreign final stops? *ZITÖR

Today i realized that Whitman's (1985: 235) reconstruction of a Proto-Koreo-Japonic word for 'thread' sounds like a drug name - if drug names had metal umlauts. What would Zitör be for?

I want to look at that word more closely than I did in my previous entry. Below I show how its segments line up with those of its descendants:

Proto-Koreo-Japonic z i t ö r
Proto-Korean s ì C (r)
Middle Korean s ǐ  r   (r)
Modern Korean sh i  l   (l)
Old Japanese   i t ə  
Modern Japanese   i t o  

1. *z

As Vovin (2010: 186) noted, the correspondence of Middle Korean s- to Old Japanese zero is sui generis and hence not a sound foundation for reconstructing a proto-phoneme *z-.

Whitman's choice of *z- reminds me of Martin's (1987: 36) reconstruction of Proto-Japonic *z- for a small class of words which have a zero initial in isolation but have s- in compounds: e.g.,

ame ~ -same < *zama-i 'rain'

ine ~ -shine < *zina-i 'rice plant'

However, there is no zero ~ s- alternation for 'thread': i.e., there are no compounds with -shito < -sitə.

According to orthodox Japonic tone theory, 'thread' is reconstructed with a *low-high melody which matches that of Proto-Korean *sìCV́. At first glance, the initial low tone in Japonic could imply an earlier lost voiced consonant (not necessarily *z-:  e.g., *j-, *ɣ-, *ɦ-) or a zero initial: i.e., *i- instead of *ʔi-. However, Shuri forms with a long vowel (ʔiichu ~ ʔiichuu as well as ʔichuu) may point to a Proto-Japonic *ʔii whose long vowel conditioned a low tone.

Modern Korean sh [ɕ] is from Middle Korean s before i and y.

2. *i

This vowel is all that the Korean and Japonic forms undisputably have in common. The rising tone of the Middle Korean vowel arose from a low-high tone sequence in Proto-Korean. Proto-Koreo-Japonic short *i does not match the long ii of Shuri (section 1). Would Whitman regard length in Shuri as secondary?

3. *t

There is no Korean-internal evidence for *-t- in 'thread'. It is likely there was a consonant between the *low and *high-toned vowels of Proto-Korean, but if that consonant became Middle Korean -r, it could have been *-l- or *-r- as well as *-t-. I don't know how Whitman would have explained the fate of Proto-Koreo-Japonic *-t- which seems to correspond to later Korean zero since I assume his Proto-Koreo-Japonic *-r became Middle Korean -r. See section 5.


ö is an old symbol for the Old Japanese vowel I reconstruct as ə, so Whitman might say that Old Japanese preserved the Proto-Koreo-Japonic vowel that was lost in Korean. There is no Korean evidence for the identity of this vowel; the Middle Korean rising tone only tells us that there was once a high-toned vowel. Vovin reconstructed the Proto-Korean final vowel as *ɯ́ which regularly corresponds to Old Japanese ə (see Whitman 1985: 126 for examples).

5. *r

I assume Whitman reconstructed Proto-Koreo-Japonic *-r on the basis of Middle Korean -r. There is no Japonic evidence for a final consonant. However, like Vovin, I think Middle Korean -r is from an earlier intervocalic *-t- or liquid rather than an original final *-r. Hence I have placed the Korean final liquids in parentheses in the *r column since I think they really belong in the *t column. THREADED PAN-ŬL

Homefront Six mentioned four Korean words yesterday:

냄비 naembi 'cooking pot' (on my to-blog list since last December!)

shil 'thread'

바늘 panŭl 'needle'

선풍기 sŏnphunggi 'electric fan'

All of them involve technology (obviously not always of the Silicon Valley kind) and all of them have proposed or actual Japanese connections.

Normally I assume that all technological terms went from Korea to Japan until modern times when the direction was reversed.

선풍기 sŏnphunggi 'electric fan' is an example of a modern loan from Japanese 扇風機 senpūki 'fan wind machine'. Like many such loans, it is a made-in-Japan Chinese root combination that is read in Sino-Korean.

Conversely, I would expect the lower-tech word 냄비 naembi 'cooking pot' to be the source of Japanese nabe, but in fact the direction seems to have been the other way around.

Nabe is from Old Japanese 名倍 nambəy* which has an Japanese-internal etymology:

na 'side dish' (now in sakana 'fish' < orig. 'sake** side dish') +

mbəy < (genitive suffix) + pəy 'pot'

i.e., 'side dish's pot'

Ōno proposed Sino-Korean 瓶 pyŏng 'bottle' as a source, but the semantic match is loose, and the phonetic match is even looser.

Could pəy 'pot' have been borrowed from some earlier Koreanic word related to Middle Korean pʌ́y 'boat' (cf. the use of vessel for food vessels as well as boats in English)?

I cannot find naembi or the earlier form 남비 nambi in Yu's (1964) dictionary of premodern Korean or even in Gale's (1897) dictionary. However, Gale (1897: 484) does have an entry for 람비 (濫沸) rambi 'frying pan, small dish for rice'. Is that an attempt to give a Sino-Korean etymology for a Japanese loanword? (Sino-Korean word-initial r- is normally read [n] before a.) The characters mean 'overflow' and 'boil'; they seem to have been chosen primarily for sound and only secondarily for their meanings. Martin et al. (1967: 308) listed nambi but not naembi. However, Naver regards nambi as an error. The first vowel of naembi fronted to assimilate to the following i; cf. 애기 aegi from 아기 agi 'child'. (Naver regards fronted aegi as wrong even though it says fronted naembi is right! In any case, the nonfronted forms are older.)

Nambi obviously must have entered Korean at some point before 1897. My guess is that the final -i indicates a period before Korean developed e from earlier əy. Perhaps nambi was borrowed from a Late Middle Japanese nãbe during the Japanese invasions of the late 16th century. At the time, Korean had no [e], so the -i of nambi was an attempt to imitate the -e of Japanese nãbe. Lee and Ramsey (2011: 264) date Korean [e] from "the end of the eighteenth century".

Unfortunately Naver only lists one nonstandard apparent cognate 갈레비 kallebi for na(e)mbi without identifying a specific dialect as a source. That looks like a compound of kal '?' and nebi from naembi with n assimilating to the preceding l.

I vaguely recall someone proposing that 실 shil < Middle Korean sǐr 'thread' was borrowed from Old Chinese. Did I discuss this in my 1997 article evaluating proposals for early Korean loans from Chinese? I can't find my copy of that article. Oh well, no great loss since I know a lot more about Korean and Chinese than when I wrote it 19 years ago. (It wasn't published until two years later.) Long ago, such a proposal didn't seem outrageous since Karlgren (1957: 256) reconstructed Old Chinese 絲 'silk' as *si̭əg. One could imagine *-g leniting to *-ɣ which is close to *-ʁ and could be borrowed as *-r. However, we now know that

- 絲 ended in *-ə or *-ɯ; there was no *-r-like consonant

- Middle Korean sǐr is from an earlier Korean disyllabic *sìCV́; the rising tone is the contraction of a low and high tone on the two original vowels, and the medial consonant could have been *r, *l, or *t.

Pan Wuyun and Zhengzhang Shangfang reconstructed 絲 with a cluster initial *sl-. I don't know the reasoning for that, but if they are correct, then 絲 is not much like sǐr at all unless one resorts to metathesis (and even then the vowels don't match):

*slɯ > sǐr!?
Then again, maybe the i is an epenthetic vowel:

*slɯ > *sìrɯ́ > sǐr

Whitman (1985: 235) reconstructed a Proto-Koreo-Japonic *zitör as the source of both Middle Korean sǐr and Old Japanese itə 'thread', but Vovin (2010: 186) argued against that. I agree with Vovin that the initial is a problem; there are no other cases of MK s- corresponding to Old Japanese zero.

Vovin (2010: 96-97) also rejected Whitman's (1985: 209) proposed common origin for Middle Korean pànʌ́r (> modern Korean panŭl) 'needle' and Old Japanese pari 'id.' Even if those words are somehow related, I bet that 'needle' is a Koreanic loan into Japonic; I doubt the words were inherited from a shared ancestor. Ditto for 'thread'.

*This spelling is from Man'yōshū 3824. Shoki has the odd spelling 儺迷 nambe with a different final vowel in the personal name Wonambe as well as 那倍 nambəy in the personal name Minambəy.

**The root of sake is saka; sake is from *saka-i. The root appears in compounds like saka-na 'fish'. AN AW-TERNATIVE SPELLING IN LAO

Modern Lao spelling is quite straightforward unlike Thai with a few exceptions (Lew 2013: 10):

Four special graphemes based on etymological spelling are still in use [in Lao]. These regard the sequences /-am/ and /-aw/, and the sequence /-aj/ with two alternants.

/-am/ has two spellings:

ຳ <āṃ> (for native words)

Here າ <ā> is not for long /aː/ but is for short /a/; this may reflect Khmer ាំ <āṃ> for /am/

ັມ <am> (for Indic loans: e.g., ກັມ <kam> /kam/ < Pali kamma 'karma')

And /-aj/ has two spellings:

ໄ <ai> < *-aj

ໃ <aɨ> < *-aɰ

I presume ໃ <aɨ> was retained since it appears in high-frequency words like ໃນ <naɨ> /naj/ 'in'; cf. the retention of high-frequncy irregular spellings in Japanese (i.e., は <ha>, へ <he>, を <wo> for [wa e o]).

But I only know of one spelling for /-aw/:


a combination of ເ <e>, ົ <ŏ>, and າ <ā> corresponding to

Thai เ <e> + า <ā> for /aw/

and Khmer ៅ <e> + <ā'> (not ា <ā>) for *aw

cf. Khmer ោ <e> + <ā> for *o (not /aw/ as in Thai which has a separate symbol โ <o> for /oː/ with no modern Khmer counterpart - was there an ancient Khmer source for that character?)

Neither the Thai nor the Khmer script has a character corresponding to Lao ົ <ŏ> which normally represents short /o/ in closed syllables. Why does Lao have that extra character? Did Lao ever have a sequence *ເ <e> + າ <ā> corresponding to Thai เ <e> + า <ā> for /aw/? (Lao does have the sequence ເ <e> + າະ <āḥ> for /ɔʔ/ corresponding to Thai เ <e> + าะ  <āḥ> for /ɔʔ/.) How did <e> + <ā>-type sequences come to represent o/au in Indic scripts: e.g., Devanagari (below)?

<ka> /ka/
cf. Khmer <ka> *kɔː
के <ke> /keː/
cf. Khmer កេ <ke> *keː
कै <kee> /kai/
cf. Khmer កៃ <kai> *kaj
का <kā> /aː/
cf. Khmer កា <kā> *kaː
को <kāe> /koː/
cf. Khmer កោ <keā> *koː
कौ <kāee> /kau/
cf. Khmer កៅ <keā'> *kaw

Is there another spelling for /-aw/ in Lao?

2.27.0:17: Neither Hoshino and Marcus (1981: 174) nor Diller (1996: 463) mention a second spelling for Lao /-aw/. Does the spelling ັວ <aw> for /aw/ exist? (Its Thai counterpart ัว <aw> represents /ua/, not /aw/! The Lao spelling of /ua/ is ົວ <ŏw> with ົ <ŏ> instead of ັວ <a>. Was Lao /ua/ something like [uo] when that spelling was devised?) RA-ITING (LA-ITING?) IN THE LAND OF APRICOTS

I forgot to look for other Tungusic words for 'apricot' in my previous entry. I was hoping to find more forms that could bridge the gap between Jurchen <gui.fa.ra> (?) and Manchu guilehe [gujləxə], but Cincius (1975 I: 168) only lists Nanai gujləxə moni 'apricot tree', regarded as a loan from Manchu. (What is moni? I assume it isn't simply 'tree', as 'tree' is moo in both Nanai and Manchu.)

(2.25.23:40: Too bad Onenko's 1989 Nanai dictionary has no entries for gujləxə or moni.)

I also forgot to mention last night that


also represented the stem of the Jurchen verb <ra>/<la> 'to make, write' cognate with Manchu ara- 'id.' Unfortunately Cincius (1975 I: 48) has no cognates elsewhere in Tungusic. Since I know of no prefix a- (and since prefixation is un-Manchu), my guess is that Manchu ara- is more conservative, though it is attested later, whereas pre-Jurchen *ara- underwent apheresis and lost initial a-. I wonder whether the result of apheresis was ra with an un-Altaic initial r- or la.

The character <ra>/<la> also has a dotless variant in the Da Jin deshengtuo songbei inscription of 1185 (see Jin 1984: 147):

I wonder if the dotless and dotted versions originally represented <la> and <ra> (or vice versa). The dotted version could have come to dominate due to hypercorrection. THE LAND OF APRICOTS

In August 2012, I asked,

Does the use of 杏 for gui 'nation' [in the Khitan large script] indicate that the Khitan word for 'apricot' was (nearly?) homophonous with gui 'nation'?

Today Andrew West answered my question. The Khitan word for 'apricot' was probably gui or something like it because he pointed out that similar words for 'apricot' are attested in later languages:

Mongolian ᠭᠦᠢᠯᠡᠰᠦᠨ güilesün

Jurchen <gui.fa.ra> or <gui.fa.la>

Manchu ᡤᡠᡳᠯᡝᡥᡝ guilehe

Andrew speculated that the Jurchen form may actually be guilafa:

Kiyose (1977: 103) and Kane (1989: 204) also proposed the inversion of the last two syllables in the Bureau of Translators vocabulary which is the only extant source of the Jurchen spelling of the word; the Bureau of Interpreters vocabulary only has a Chinese transcription 貴 of a monosyllabic word gui. Could the Ming Dynasty Jurchen dialect in the Bureau of Interpreters vocabulary have preserved a Khitan loanword in its original form sans suffixes (or final syllables retained in Mongolian but lost in Khitan)?

The syllables after gui do not match in the three non-Khitan languages:

- Mongolian -lesün and Manchu -lehe have 'feminine' vowels but Jurchen -lafa (or -fala, etc.) has 'masculine' vowels; this implies that Jurchen gui had 'masculine' vowels (and hence was gūi in a Manchu-style romanization).

- Manchu -he is from *-ke, whereas Jurchen *fa is from *pa

- Jurchen la might be ra, as the phonogram

is ambiguous according to Ligeti (see Kiyose 1977: 71). Jin (1984: 155) generally read it as <ra>, but read the word for 'apricot' as both <gui.ra.fa> (p. 155) and <gui.la.fa> (p. 130, 264, index p. 30). The Chinese transcription 歸剌法 *guilafa is ambiguous, as there was no sinograph pronounced *ra.

Rozycki (1984: 131) derived Manchu guilehe and Jurchen guwifala (his spelling) from "Early Mo[ngolian]"; he did not comment on the second and third syllables. Would he consider Khitan to be Early Mongolian? (Khitan is thought to be para-Mongolic but could be considered 'Mongolian' in an extremely broad sense.) He mentioned "Kitan" as valuable for "the establishment of both absolute and relative chronologies for sound change in Mongol loans into Manchu", but did not include Khitan in his list of abbreviations of languages (p. 15), suggesting that he may not have used the term Khitan in his lexicon of Manchu words of Mongol[ic] origin. (I have not checked all his entries to verify that.)

Was the Khitan large script character 杏 already identified as gui by 1984? Kane (2009: 169-171) did not mention the character in his brief survey of Khitan large script studies up to 2000, though omission is not evidence of absence. DUCK LEGS AND DUKE'S GRANDSONS

Japanese ichō 'ginkgo' has three unusual spellings:

銀杏 ('silver apricot'; can also be read ginkyō, the source of ginkgo)

公孫樹 ('Gongsun tree'; can also be read Kōsonju; kō is 'duke' and son is 'grandson')

鴨脚樹 ('duck leg tree'; looks as if it should be read *ōkyakuju)

The last spelling reflects the belief that ichō is from Mandarin 鴨脚 yajiao [jatɕjaw] 'duck leg'. But I think that derivation is improbable since I would expect that word to be Japanized as *yachau or *yachao. There was no sound change that would reduce *ya to i in Japanese. Moreover, the shift of *au to ō was already complete in Japanese by the time *jakjaw became [jatɕjaw] in Mandarin, so an early modern Japanese *yachau would not become modern Japanese ichō.

There is another etymology deriving ichō from 一葉 'one leaf'. I presume ichō is supposed to be a contraction of Middle Japanese it 'one' and 'leaf'. This is phonetically plausible but makes no semantic sense.

The derivation of ichō from i 'sleep' and chō 'butterfly' has the same problem.

I initially thought that the i of ichō could be from in, a Japanization of a northern Chinese reading of 銀 *jin 'silver' from the last millennium, but at no point was northern Chinese 杏 *xiŋ > [ɕiŋ] 'apricot' ever pronounced anything like chō or an earlier Japanese source for chō like *tyau or *tefu.

I conclude that the etymology of ichō is unknown. The long ō points to Chinese as the most probable source. Perhaps the word was borrowed from some colloquial Chinese term (*jin ... 'silver ...?') whose second half is unknown.

Tangut fonts by Mojikyo.org
Tangut radical and Khitan fonts by Andrew West
Jurchen font by Jason Glavy
All other content copyright © 2002-2013 Amritavision