I have long been troubled by Korean 부처 puchhŏ and Japanese hotoke 'Buddha' which must be related but not in a straightforward way. Here's my attempt to explain how they came to be.

When Buddhism came into the Korean peninsula in the 4th century AD, the first word for 'Buddha' there must have been Chinese 佛 *but, short for 佛陀 *butda, a borrowing of Sanskrit Buddha-. This word was localized as *put with the addition of *-ka-i 'ruler' (cf. Koguryŏ 皆 *kɛ 'king', perhaps from *ka-i, the Puyŏ title suffix -加 *-ka, Paekche -瑕 *-ka 'king', Shilla -干 *-kan 'king' [Lee and Ramsey 2011: 48], Khitan <qa> and <qa.ɣa.an> [Kane 2009], and Old Turkic and Written Mongolian qan and qaɣan [see Vovin 2007 on the ultimate origins of this term]). *put-ka-i is similar in structure to Old Turkic and Written Mongolian burqan 'Buddha' (lit. 'Buddha khan'). (The -r of bur reflects Late Middle Chinese 佛 *bvur 'Buddha'.)

In Paekche, *putkai became *potkai, and was borrowed into pre-Old Japanese as *potəkai which then became 保止氣 potəkəy in Old Japanese. Unfortunately, I don't know of any other evidence for *u lowering to *o in Paekche; if such a change occurred, it must have predated the borrowing of the Chinese character readings that were later transmitted to Japan. (11.24.0:57: Did Paekche have vowel rotation: e.g., > *u > *o?)

In Shilla, *putkai became *putke. Later *tk became th and *e broke to yə, resulting in Middle Korean 부텨 puthyə. In Modern Korean, thy fused into chh [tɕʰ]: 부처 pucchŏ.

Some Sino-Korean readings with < -yə may also have undergone the same monophthongization and breaking:

西 *sai (cf. Go-on sai) > *se > 셔 syə >

This begs the question of where Middle Korean ㅐ ay came from. Did only some *ai monophthongize to *e, and if so, why? Is Middle Korean 개 kay 'dog' from *kaCi with a *-C- that blocked monophthongization before it was lost? THE -IFFERENCE BETWEEN NORTHEASTERN LATE MIDDLE CHINESE DIALECTS

Since at least 1999, I have thought that Korean borrowed from a northeastern Late Middle Chinese dialect. The Khitan Empire encompassed northeastern China and so Khitan must have also borrowed from a northeastern Late Middle Chinese dialect. Standard Mandarin today is based on the Beijing dialect, and what is now Beijing was once the Southern Capital of the Khitan. So one might think that loans from Chinese in Korean and Khitan come from the ancestor of modern standard Mandarin: i.e., a prestigious northeastern Chinese dialect.

But there are subtle clues indicating otherwise. In my previous post, I wrote about the different Mandarin and Korean initials corresponding to *dʐ- in the Middle Chinese rhyme dictionary tradition. The very limited Khitan data I had on hand suggested that Khitan borrowed from a language similar to the ancestor of standard Mandarin, not from the source of the Chinese loans in Korean.

Looking at the Mandarin, Khitan, and Korean nitials corresponding to *dʑ- in the Middle Chinese rhyme dictionary tradition also confirms that conclusion from last night:

Class Sample sinographs Middle Chinese tone Mandarin Khitan Korean Notes
1 Any sh- <sh> /s/ Before Late Middle Chinese *-ɨ
2 Level ch- <c> Mandarin 尚 chang is extinct
3 Regular
4 ? /c/ Cf. Korean 振 /cin/
5 尚 (again!), 署 Nonlevel sh- <ś> /s/ Regular

The Mandarin initials are predictable:

sh- before *-ɨ (class 1)

ch- before level tone rhymes (classes 2-4)

sh- elsewhere (class 5)

This pattern is close to that for *dʐ- except that the elsewhere initial is sh-, not zh-. I don't understand why an affricate weakened to a fricative before *-ɨ and nonlevel tone rhymes.

Khitan had two initials for 尚 (class 2 and 5; Takeuchi 2007: 29) presumably depending on tone. (Khitan was atonal, but of course the source of its Chinese loanwords was tonal.) Pulleyblank (1991: 50) listed chang as a level tone reading of 尚 in 尚羊 'ramble', but I cannot find any other source indicating that it is still in use: e.g., zdic.net lists the reading of 尚羊 as shangyang.

Korean has a fricative with the exception of 辰 /cin/ (class 4) which has an unaspirated initial unlike Mandarin chen; its reading may be by analogy with 振 /cin/ and may be of Korean origin since its prescriptive 15th century reading in 東國正韻 Tongguk chŏngun is <ss.i.n> like 臣 (class 3) which is its homophone in the Middle Chinese rhyme dictionary tradition. Is there any modern northeastern Chinese dialect with fricatives for *dʑ- and *dʐ- in the Middle Chinese rhyme dictionary tradition corresponding to Korean /s/? SA(L)SA-FLAVORED CHACHA

I wonder how many people learned the Mandarin word 查 cha 'investigate' from the name of 查查 ChaCha.com.

Standard Mandarin ch- and -a normally correspond to chh- and -a in Chinese borrowings in Korean, so the expected Korean reading of 查 is *chha. However, the actual Korean reading of 查 is sa with s-, not chh-. (Hence the asterisk on *chha indicating that it is not attested.) So 查查 would be read as sasa in Korean. Why does Korean have s- instead of chh-?

The initial of 查 in the rhyme dictionary tradition is *dʐ- which corresponds to three consonants in both Mandarin and Korean:

Class Sinographs
(all from Baxter & Sagart 2011 except 查)
Mandarin Korean Notes
1 none! ch- chh-
2 儕棧 ch- Cf. Korean 齊 che,chŏn
3 查槎巢祡豺牀愁崇鋤 s-
4 zh- chh- Cf. Korean 聚 chhwi
5 轏助 ch- Cf. Korean 孱 chan
6 乍狀 s-
7 士仕事 sh-

The Mandarin initials are predictable:

sh- before *-ɨ (class 7)

ch- before level tone rhymes  (classes 2-3)

zh- elsewhere (classes 4-6)

The Korean initials are only partly predictable:

s- before *-ɨ (class 7)

s- nearly everywhere (classes 3 and 6)

ch(h)- in classes 2, 4, and 5 by analogy with other readings of characters with the same phonetic components

I cannot explain why the high-frequency class 5 character 助 has irregular ch- unlike the low-frequency character 鋤 containing it as a phonetic. Its final -o is also irregular; 鋤 has the regular rhyme -ŏ.

Does Korean s- reflect a northeastern Late Middle Chinese *ʂ- corresponding to *dʐ- in the rhyme dictionary tradition?  It is a shame that I can only find Khitan transcriptions for two syllables with the rhyme dictionary initial *dʐ- (Takeuchi 2007: 27, 33, Kane 2009: 258):

Sinograph Rhyme dictionary Khitan Mandarin Korean
chong sung
shi sa

So far - which is to say not far at all - the Khitan initials match the Mandarin pattern. Were other syllables of this type transcribed with Khitan <sh> or <ś>-graphs?

11.22.1:07: I can't find 查 in the prescriptive Korean dictionary 東國正韻 Tongguk chŏngun, but its variants 楂 and 槎 have the reading <ss.a.Ø>. This indicates that a fricative initial was considered prestigious.

I expected to find 助 and 鋤 in the section for <ss.ə>-syllables (disregarding tones), but in fact both were listed in the section for <cc.o> syllables.

Korean <cc.ò>/<cc.ó> and cho for 助 seem to be descended from Old Chinese *dza(-s) whereas the rhyme dictionary reading dʐɨəʰ is from Old Chinese *rɯ-dza-s with a presyllable.

Cantonese [tsɔː] for 助 also seems to be from Old Chinese *dza-s; a descendant of *rɯ-dza-s would have been *[tsɵy]. It seems that northeastern and southern dialects of Old Chinese independently retained presyllableless *dza-s.

On the other hand, Vietnamese trợ for 助 was borrowed from a descendant of a southern Old Chinese dialect with *rɯ-dza-s. Old Chinese *dza-s would correspond to *tộ in Vietnamese. NON-NASAL-INITIAL NEGATIVES IN INDO-EUROPEAN Wow, that title is full of nasals.

I normally expect Indo-European languages to have negatives descended from *ne: e.g., Sanskrit na, Russian ne, and English not. (See Wiktionary for more examples.) Standard Irish 'not' fits the pattern.

Looking at this part of the Wikipedia article on Ulster Irish,

In Ulster the negative particle cha (before a vowel chan, in past tenses char - Scottish/Manx Gaelic chan, cha do) is sometimes used where other dialects use and níor.

I wondered where cha 'not' (which is also in standard Irish) came from. I thought it might be from Proto-Celtic *kʷā-nī which looks like a compound of *kʷā- (some sort of question word?) and *nī 'not', but it turns out to be from Old Irish nícon < 'not' + con 'that' (Thurneysen 1980: 538) with 'not' as the first syllable, not the second! I assume its initial fricative ch- [x] weakened before a lost ní-:

nícon [niːkon] > *[niːxon] > cha [xa]

In the now-extinct dialect of Rathlin Island, cha was reduced to [a] (Holmer 1942: 37)!

Many years ago I was surprised that North Germanic languages have e/i-negatives (e.g., Icelandic ekki and Swedish icke) instead of n-negatives like German nicht and Dutch niet. Wiktionary derives the e/i-negatives "[f]rom Old Norse ekki ("nothing"), from eitt (neuter of einn ("one")) + negative suffix -gi, -ki." So ekki etc. are like English none from 'not' plus 'one' but with the elements in the opposite order. But what is the origin of the negative suffix, and what other words contain it?

11.21.2:19: -gi/-ki "comes ultimately from the PIE [Proto-Indo-European] indefinite suffix *kʷid" (Fortson 2010: 375).

11.21.2:36: I assume that Swedish inte 'not' and intet 'nothing' also contain 'one' like icke, but why do they have -t- instead of -k-?

I think Swedish ej 'not' is from Old Norse eigi 'not' < ei 'ever' + -gi (Fortson 2010: 375). These 'not'-less words for 'not' are products of Jespersen's Cycle; they originated as intensifiers for 'not' and came to mean 'not' themselves: cf. French pas 'not', originally 'step'.

11.21.1:01: While I'm on the subject of Germanic, where does German kein 'no' come from? Wiktionary derived it from Old High German nihein, and Jäger (2008: 200) in turn derived that from Proto-Indo-European *ne-kʷe 'not-and' plus 'one'. If the ni- fell off, hein would be left, not kein, and I have never heard of fortition of h- to k- in German. So why does the German form have k-? And why does its Dutch cognate geen have voiced g- [ɣ]? (The voiceless northern Dutch pronunciation [x] of g- is a recent innovation and not a retention of Proto-Germanic *x.)

11.21.2:41: If I understand this Wikipedia reference desk entry correctly, kein has k- because when earlier nich ein 'not a' lost its ni-, initial ch- was not permissible and was replaced with k-. Did Dutch speakers do something and replace ch- with g- after something like nich een 'not a' lost its ni-? GOSNELL: A HOSIER IN DISGUISE?

I assumed Gosnell was an English name like Cornell and Parnell, so I was surprised to see that ancestry.com listed it as Irish:

from an Irish adaptation (Góiséir) of English Hosier, the name of a family in Munster in the late 16th century.

Is Gosnell an Anglicized version of an Hibernized English name? I would have expected an Anglicized version of Góiséir [goːʃeːɾʲ] to be something like *Gosher. Gosnell has an -n- lacking in Góiséir, and its -ll corresponds to -r.

Modern Irish has h- for foreign h-, but Góiséir has a Russian-like substitution of G- for H- (cf. Russian Gollandija 'Holland'). Are there other examples of Irish g- for foreign h-? (There are no native Irish words with initial h- in their base forms.)

11.20.00:24: I forgot to mention that Edward MacLysaght wrote in The Surnames of Ireland that the derivation of Gosnell from Góiséir is "very doubtful". Could it be an unrelated English name? Are all the Gosnells in Great Britain* of Irish ancestry?

*Maps showing where Gosnells lived in Great Britain in 1891 can be viewed by clicking on "Name Distribution of Gosnell Families" at ancestry.com.

11.21.00:31: I don't see any examples of Irish g- corresponding to English h- in this admittedly incomplete list of loanwords. PREINITIAL GLIDES IN OLD CHINESE?

After posting "An I-nigmatic Reading", I realized that an intermediate stage between Old Chinese *iba and *bia for 夫/扶 could have been *jba.

Japhug rGyalrong has jC-clusters, though not jb- (Jacques 2004: 43). Those clusters are from Proto-rGyalrongic (PGR) *lC-clusters as well as *jC-clusters (Jacques 2004: 332). No PGR consonant is preceded by both *l- and *j- in Jacques' reconstruction; *j- is only before dentals and the velar *ŋ- whereas *l- is elsewhere: e.g., *lp- and *lk-. I suspect all jC-clusters go back to *l(V)C-sequences: e.g.,

PGR *ləl- > *λəl- > *jəl- > jl-

(I also suspect j- goes back to PGR *l- rather than PGR *lj-. Jacques' table of PGR laterals on p. 266 has no *l-.)

I don't understand why Jacques (2004: 336) reconstructed PGR *jŋ- instead of *lŋ- as the source of Japhug jŋ-, particularly since he reconstructed PGR *lŋk- as the source of Japhug jŋ- corresponding to Somang jk- on p. 304. Also, he wrote on p. 271 (emphasis mine):

Nous reconstruisons les préinitiales j- du japhug comme *l- en PGR devant les labiales (et peut-être aussi les vélaires).

Jacques' (2004: 331) PGR reconstruction also has *wC-clusters. Maybe initial *uC- became *wC- in Old Chinese. If so, Old Chinese preinitial *w- may have had a wider distribution than PGR *w- which is only before liquids.

I have been writing the vowels of Old Chinese presyllables as and respectively representing a higher unstressed short vowel and a lower unstressed short vowel. If Old Chinese was like Pacoh as described by Watson (1964: 144), could have been *[i] or *[u], and could have been *[a]. Although the height of presyllabic vowels left traces in later Chinese, I do not know of any way to determine whether a higher presyllabic vowel was front *[i] or back *[u].

On the other hand, *-a has a variety of 'brightened' reflexes in Tangut ranging from low -a to high -i and -u (Matisoff 2004). The high reflexes may have been conditioned by high front presyllabic *[i] and *[u]: e.g.,

'year': 1vɨi < *vi < *Ci-βia < *Ci-pa (cf. Japhug tɯ-xpa 'id.')

'nose' 2nii < *Ci-naCH (cf. Japhug tɯ-ɕna 'id.')

'ear': 1niu < *nu < *Cu-na (cf. Japhug tɯ-rna 'id.')

11.19.2:28: Added 'nose'. Matisoff (2004: 20) wrote:

The high back vowel in the Xixia [= Tangut] form [for 'ear'] is unexplained. As a wild guess, we might claim that this abnormal development was due to a desire to avoid homophony with NOSE [which has a high front vowel in Tangut instead of a high back vowel].

In my pre-Tangut reconstruction, the development is normal: a high back vowel in the presyllable conditioned a high back vowel in the main syllable.

One problem with my explanation is that I would expect a lot of Tangut Cu from pre-Tangut *CuCa which is not an exotic sequence and therefore must have been common. However, Matisoff only listed five examples of Tangut -u from *-a, and two are dubious*. Were most presyllabic *u were lost by the time vocalic transfer' (i.e., the shift of vocalic features from the presyllable to the main syllable) took place?

*I doubt that

2ʐɨu 'fish'

is from *ŋya or that

2dziụ 'rain'

is from *g/r-wa. I would reconstruct their pre-Tangut sources as *Cɯ-džuH (I don't see any evidence for the low vowel of his Proto-Qiangic *dza) and *S-dzuH. The initial of 'fish' lenited to -ʐ- in intervocalic position. AN I-NIGMATIC READING

In the History of the Northern Dynasties, two names are listed for the capital of Paekche:

固麻 Early Middle Chinese *koʰ mæ (a transcription of the Paekche word for 'bear'; see my previous post)

居拔 Early Middle Chinese *kɨə bɛt

Gari Ledyard (1975: 247) wrote,

Giving the first character [of 居拔] a Korean "kun" reading of i yields the reading *Ibal or *Ipar.

A "kun"* reading in this context is a native Korean morpheme that is a rough translation of the Chinese morpheme represented by a Chinese character. Since 居 *kɨə meant 'dwell' in Chinese, I would expect i to mean something similar in Korean. But I don't know of any Korean 이 i 'dwell'. 이- /i/ 'to be' is an exact phonetic match but a loose semantic match; 있- /iss/ < 잇 /is/ 'to exist' is an even looser phonetic and semantic match. The only hun reading I know of for 居 is 살- sal- 'to live'. (The i reading of 居 does, however, coincidentally resemble the Japanese kun reading i- < wi- 'exist [animate]' for 居.)

Ledyard connected his *Ipar to Japanese 磐餘 Iware < Ipare with an anomalous second character 餘 that is never read -re elsewhere but

just happens to be also the second character of the name Puyŏ (夫餘 or 扶餘 [= Buyeo]) - one begins to see the plausibility of a connection with that name too. The character had, according to Karlgren, a voiced dental initial in Old Chinese - that is, an initial d-; this could well have served as a transcriptional equivalent for a trilled -r- of the Japanese and Korean type. I am inclined to regard theof Iware as a vowel-harmony variant of the suffix -ra meaning clan or tribe or nation, and that the name Puyŏ possibly orignating in a form like -re *Pora or *Para might boil down to something like "Rocklings" - a more than appropriate name for a people whose legends tell us of births from under rocks, of rock princesses and rock boats.

In modern Old Chinese reconstructions, 餘 has an initial *l-, so I viewed 夫餘/扶餘 as a transcription of *Bala which has reminded me of Japanese *para 'field'. Here is a wild speculation. My Old Chinese reconstruction requires a high-voweled presyllable *Cɯ- in 夫/扶 to account for its later vocalism. What if that presyllable were *ʔi-? Then 夫/扶 would be *ʔiba, a match for the *iba that some would reconstruct as the Proto-Japonic source of iwa 'rock'. Iware, then would be from*Ibala 'Puyŏ' plus a Japanese suffix *-i; *-ai later fused into -e. Here is how *ʔibala could have come to be read as Puyŏ in Korean:

Old Chinese *ʔibala > *ʔibialia > *bialia > *bɨalɨa > *buajɨa > *buojɨə > *pujɨə > borrowed into Old Korean as *pujə = Puyŏ

Ledyard (1975: 248) listed "earlier names for the peoples of the Liao and Eastern Manchurian area" in Peter A. Boodberg's unpublished Old Chinese reconstruction from c. 1960-61:

Sinograph Boodberg This site's Old Chinese
*part *Cɯ-pat
*pharn *phar
浡 (cf. later 渤海 Parhae) *bert *Cʌ-bət (*Nʌ-pət?)
GSR 320 *phard ?*phas

The final letter indicating the final sinograph in Karlgren's (1957) Grammata serica recensa is missing in Ledyard's footnote, so I can only guess that my equivalent of Boodberg's *phard would be *phas. (GSR 320 is a series for Old Chinese *Pas-syllables.)

All of Boodberg's reconstructions - including *barn for 磐 (= my *ban) - contain *Par matching Ledyard's *Para. I don't know the logic underlying Boodberg's *-r-. Only some of my reconstructions roughly match *Para (if one regards *-t as a transcription of a foreign flap *-r).

Going back to 居拔 *kɨə bɛt, could it be a transcription of a Paekche cognate of Middle Korean kʌβʌr < *kʌbʌr < *kʌpʌr 'village'? (11.17.22:37: That Paekche cognate could be the source of Japanese 郡 koori < kəpori 'district'.)

*訓 Kun is a Japanese term; its Korean equivalent is 訓 hun.

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