10.3.13.23:27: A *FUR-EIGN LAND

I thank Andrew West for finding another typeset version of the quote from Youyang zazu about olive trees with the same unusual character 厂+虛 as Laufer (1919: 415):

齊墩樹出波斯國，亦出拂菻國，拂菻呼爲齊(厂+虛)(音湯兮反)。

'The *dzjej-ton tree comes from *Pa-sɨ. It also comes from *Fur-lim. In *Fur-lim, it is called *dzjej-thjej (with the sound *th- + -jej).

波斯 *Pa-sɨ is Persia. I wonder how old that transcription is. Could it date back to a period when it was read *Pa-si or *Pa-sie?

As for 拂菻 *Fur-lim, according to Laufer (1919: 436),

[...] for we know, and again thanks to Hirth's researches, that the Chinese distinguished two Fu-lin (modern Mandarin equivalent of Late Middle Chinese *Fur-lim), - the Lesser Fu-lin, which is identical with Syria, and the Greater Fu-lin, the Byzantine Empire with Constantinope as capital.

Why would Syria and the Byzantine Empire be called *Fur-lim? Laufer (1919: 436-437) wrote,

As to the origin of the name Fu-lin, I had occasion to refer to Pelliot's new theory [1914], according to which it would be based on Rōm, Rūm [...] by falling back on the ancient phonology of Chinese, we may hope to restore correctly the prototypes of the Chinese transcriptions. Pelliot starts from the Old-Armenian form Hrom or Hrōm, in which h represents the spiritus asper of the initial Greek r. In some Iranian dialects the spiritus asper is marked by an iniital vowel: thus in Pahlavi Arūm, in Kurd Urum. The ancient Armenian words with initial hr, as explained by A. Meillet, were borrowed from Parthian dialects which transformed initial Iranian f into h: for instance, Old Iranian framana (now ferman, "order") resulted in Armenian hraman, hence from Parthian *hraman. Thus *Frōm, probably conveyed by the Sogdians, was the prototype from which Chinese Fu-lin, *Fu-lim [in Laufer's reconstruction], was fashioned. In my opinion, the Chinese form is not based on *Frōm, but on *Frim or *Frīm. Rīm must have been an ancient variant of Rūm; Rim is still the Russian designation of Rome. What is of still greater importance is that, as has been shown by J.J. Modi, there is a Pahlavi name Sairima, which occurs in the Farvardin Yašt, and is identified with Rum in the Būndahišn; again, in the Šāhnāmeh the corresponding name is Rum. This country is said to have derived its name from Prince Selam, to whom it was given; but this traditional opinion is not convincing. A form Rima or Rim has accordingly existed in Middle Persian; and, on the basis of the Chinese transcription *Fu-lim or *Fu-rim, it is justifiable to presuppose the Iranian (perhaps Parthian) prototype *Frim, from which the Chinese transcription was made.

The form Rīm also occurs in Arabic as a result of an Arabic-internal process of ʔishmaam 'delabialization' "which explains why ū rhymes with ī in Koranic Arabic" (Kaye 1987: 671):

uu > yy > ii

Does this mean that Rim-type names for Rome such as Russian Рим are ultimately from Arabic?

I don't quite follow Laufer's argument, perhaps because I know almost nothing about Iranian languages between Avestan and modern Persian. (Here's a vocabulary table showing the diversity of Iranian languages past and present.) If Parthian shifted *f- to *h-, why would it shift *Hrim 'Rome'to *Frim?

I wonder if *Frim could be a hypercorrection of *Hrim: i.e., an erroneous 'restoration' of a nonetymological *f- on the basis of other words in which *hr- really was from *fr-.

A simpler explanation is that some language borrowed Greek ῥ- hr- as *fr-: cf. how Welsh Llwyd was anglicized as Floyd as well as Lloyd.

Laufer's reconstruction of 拂 as *fu lacks a final consonant still preserved in Cantonese fat, Taiwanese hut (Taiwanese has no f-), etc. I reconstruct 拂 *fur with *-r instead of *-t because the prestigious northwestern prestige dialect of the time (a source of early Chinese loanwords into Tangut) had shifted *-t to an *-r reflected in Tibetan transcriptions. The initial of 拂 may have been *(ph)f-. Coblin (1994: 375) noted that Tibetan phru was transcribed as 拂廬 which he reconstructed as *p(f)hur-lø.

Why transcribe foreign fr- and phr- as fur-l- with -l-? Late Middle Chinese had no *r-initial syllables or zero-initial syllables, so it was not possible to transcribe *Frim and phru as

*(ph)fu-rim or *(ph)fur-im

*(ph)fu-rɨw or *(ph)fur-ɨw (there was no syllable *u in LMC)

Transcriptions like 拂音 *(ph)fur-ʔim and 拂憂 *(ph)fur-ʔɨw would have a *-r-ʔ- sequence less like foreign -r- than the *-r-l- sequences of 拂菻 *Fur-lim and 拂廬 *p(f)hur-lø.

ADDENDUM: In modern Mandarin, 菻 'artemesia' is read with the third tone (< Middle Chinese rising tone), but Jiyun lists a departing tone reading *lìm for 菻 in the name 拂菻, a 胡西域 'barbaric western region'. Is this special tone for 拂菻 significant? Is it an attempt to reflect a long vowel and/or a breathy vowel in the foreign original? (But I've never heard of any Iranian language with breathy vowels.) The normal rising tone of 菻 originated from Old Chinese *-ʔ which may have also conditioned a short, creaky vowel unlike the long *ii of *Friim. The departing tone from OC *-s via *-h may have conditioned breathy vowels that may also have been longer than vowels in rising tone syllables.

Andrew pointed out that 厂+虛 isn't a typographical error in Laufer (1919: 415) since that unusual character doesn't exist elsewhere and would have been specially made for the text.

Last night, I proposed that 厂+虛 in a Youyang zazu transcription of a foreign word for 'olive' was

a distortion of 虒 LMC [Late Middle Chinese] *sɨ [厂 'cliff' atop 虎 'tiger'] which in turn could have been an error for 嗁 LMC *djej or 遞 LMC *djèj whose readings are much closer to LMC *thjej.

To confirm this, I would ideally need to see a YZ text with 虒 and an even earlier text with 嗁 or 遞.

This evening, I realized that two stages of distortion may not be necessary.

齊(厂+虛 with the Late Middle Chinese fanqie 湯 *thaŋ + 兮 *ɣjej = *thjej)

could have originally been *齊虒 *dzjej sɨ. Why write *thjej as 虒 LMC *sɨ? Perhaps the foreign word was something like *zeyθ(e), and 虒 with the fanqie 湯 + 兮 could have been an attempt to record the un-Chinese consonant (syllable?) *θ(e). There were no *-e or *-ej in LMC, so *-jej was the best approximation of a foreign *-e.

The choice of the LMC *-ɨ rhyme graph 虒 could imply a final consonant in the foreign original. 虒 LMC *sɨ is close in sound to the 思 si [sz̩] used to transcribe final -s in

成吉思 Old Mandarin *tʃhiŋkisz̩ > Chengjisi for Cinggis 'Genghis'

马克思 Makesi [makhəsz̩] 'Marx'

But the only 'olive' word I know of with *θ is earlier Hebrew זית zayiθ 'olive' (now zayit). Hirth thought the transcription represented Aramaic zaitaa. Could that have been pronounced [zajθaa]? According to Wikipedia, t and θ are "near allophones" in Aramaic (in general?) and Syriac, the most likely Aramaic source of *齊虒, had a letter ܬ that was pronounced as both [t] and [θ]. Gordian III was said to have been killed at a place called Zaitha meaning 'olive tree'. Wikipedia equates Zaitha with "Qalat es Salihiyah", which in turn was equated by ARTstor with Dura-Europos in Syria. This implies that *zaiθa was a Syriac word for 'olive'.

Summing up, perhaps Syriac *zaiθa (maybe more like *[zejθ(e)]?) was transcribed into LMC as *齊虒 *dzjej sɨ with a fanqie 湯 *thaŋ + 兮 *ɣjej = *thjej, and the second character *虒 was later distorted into 厂+虛.

Perhaps 厂+虛 is a typographical error in Laufer's book that is not in the Youyang zazu itself. (No. See my next post.)

The Wikisource version of YZ has 齊虛 without 厂 'cliff'.

I would still like to see the actual text of YZ.

Thanks to Andrew West for identifying the mystery Chinese character from my last post:

齊(厂+虛 with the Late Middle Chinese fanqie 湯 *thaŋ + 兮 *ɣjej = *thjej)

I have not been able to find 厂+虛 among the 106,230 characters in the Taiwanese government's dictionary of variants.

I have already mentioned that the fanqie is odd because it seems to represents a syllable that already existed in Late Middle Chinese (梯 *thjej). Why not transcribe the word as 齊梯? Maybe the fanqie is an attempt to represent a non-LMC syllable like *the(j). Or the fanqie was created by a speaker of an LMC dialect in which 梯 did not sound like 湯 + 兮.

In any case, 厂+虛 is an odd character for *th(j)e(j) because it consists of

厂 LMC *xàn 'cliff'

虛 LMC *xy 'empty'

厂 might be an arbitrary element added to indicate that 虛 was not read normally: cf. similar arbitrary additions in other phonetic characters: e.g., 亻 'person' in 伽 LMC *ga representing the Sanskrit syllable g(h)a(a). (加 without 亻 'person' was LMC *kæ.)

虛 LMC *xy, however, sounds nothing like LMC *thjej, whereas there was a vague phonetic resemblance between 伽 LMC *ga and 加 LMC *kæ. Moreover, I don't know of any other phonetic characters with 厂 as an arbitrary addition.

I suspect that 厂+虛 might be a distortion of 虒 LMC *sɨ, which in turn could have been an error for 嗁 LMC *djej or 遞 LMC *djèj whose readings are much closer to LMC *thjej.

Although 虒 LMC *sɨ does not sound like 嗁 LMC *djej or 遞 LMC *djèj, it is their phonetic. The three had similar readings in Old Chinese:

虒 OC *sle > Late OC *sie > Late Early MC *si > LMC *sɨ

嗁 OC *le > Late OC *de > Late Early MC *dej > LMC *djej

遞 OC *leʔ(-s) > Late OC *deʔ/h > Late Early MC *dejʔ/h > LMC *djèj

10.3.9.23:27: NOT YET OUT OF Z-Y-T

I thought I was done with 'Z-town', but I misunderstood Andrew West's post on its etymology. He thought Quanzhou became known as Zaytuun because the first two-thirds of its Chinese name

刺桐城

'prickly tong city'

pronounced something like Taiwanese tshi toŋ siã in Quanzhou?

sounded like the Arabic word zaytuun 'olive'. Similarly, I wonder if Guangdong was called Canton because its name sounded like the English word canton, an unrelated borrowing from French. If so, then Zaytuun and Canton would both be substitutions of preexisting similar-sounding words rather than transcriptions of Chinese.

Tonight I discovered that some version of zaytuun 'olive' was transcribed in 段成式 Duan Chengshi's 酉陽雜俎 Youyang zazu (Miscellaneous Morsels from Youyang, c. 853 AD) as

齊墩

Late Middle Chinese *dzjej ton

*dzjej was the best available transcription of zay (or more likely zey) since LMC had no *zaj or *zej. LMC also had no *tun, so *ton was the best available transcription of tuun.

Berthold Laufer (1919: 415) noted that Hirth in 1910 derived 齊墩 from Persian zeitun, presumably from Arabic zaytuun. Laufer listed the following cognates of zaytuun

Aramaic zaitaa

Hebrew זית zayit

and derives them from a common Semitic *zeitu (presumably with a trilateral root *z-y-t).

Youyang zazu also lists another word for 'olive' whose transcription Laufer read in modern standard Mandarin as ts'i-t'i (= pinyin qiti). But I can't make out the second graph which looks like 廬 Md lu without a dot on top:

齊(厂+盧?)

This page has 卢, the simplified form of 盧 Md lu < LMC *lo and Wikisource has 虚 Md xu < LMC *xy with the fanqie 湯 *thaŋ + 兮 *ɣjej = *thjej, equivalent to Md tang + xi = ti. (I don't understand why this fanqie is needed since there were graphs for LMC *thjej > Md ti: e.g., 梯.)

The only Md ti-graphs with similar shapes known to me are 嗁 (which has 口 on the left!) and 屜 (with totally different components), both LMC *djej.

This second transcription was linked to

Aramaic zaitaa (by Hirth; the others below by Laufer)

Grusinian (= Georgian) and Ossetic (an Iranian language not related to Georgian) zet'i

Armenian jeet, dzeet 'olive oil', zeit 'olive'

Arabic zayt

If the original was zaitaa, I would expect it to be transcribed as LMC *dzaj ta.

If it was like zet'i, I would expect LMC *dzjej t(h)jej (since there was no LMC *ti). LMC *dzjej djej does come close.

If it was like zeet, zeit, or zayt, I would expect LMC *dzjet (since there was no LMC *zjet).

Just to confuse matters further the 太平廣記 Taiping guangji (Extensive Records of the Taiping Era; 978 AD) lists the word as 齊匫 LMC *dzjej xot with the fanqie 陽 *jɨaŋ + 兮 *ɣjej = *jej. Why write *jej with 匫 *xot? I suspect that 匫 LMC *xot is a corruption of the mystery graph in Youyang zazu. 陽 LMC *jɨaŋ in the fanqie is almost certainly a corruption of 湯 LMC *thaŋ.

10.3.8.23:59: SATIN FROM Z-TOWN

On Saturday, I mentioned the unusual Xiamen colloquial reading bɔŋ for 墓 'grave'. That graph also has a Xiamen literary reading mɔ. Xiamen is close to Taiwanese, and Quanzhou is the source of northern Taiwanese (e.g., the Taipei dialect), so I assume that Quanzhou is also close to Xiamen.

Tonight, Andrew West pointed out that Hphags-pa script Christian tombstones in Quanzhou have mu for 墓 'grave' instead of something like boŋ (or mo; I know of no letter for ɔ distinct from o in Hphags-pa). I am a bit disappointed because it would have been neat to see a non-Mandarin Chinese variety written in Hphags-pa. But at least this tells us that Hphags-pa Chinese spread all the way to south China. I wonder if the inscriber simply converted Chinese characters into Hphags-pa Chinese spellings without actually being conversant in spoken early Mandarin, just as a Cantonese monolingual from Hong Kong can use a dictionary to convert Chinese characters into pinyin romanization.

In Marco Polo's time, Quanzhou was known as Zayton. Andrew explained the origin of Zayton:

Quanzhou was known to European travellers to China as Zayton (in various spellings), from the Arabic name for the city, Zaitún, meaning an olive tree. However the Arabic Zaitún is in fact a corruption of the Chinese name for the coral tree (Erythrina variegata), citong 刺桐 "prickly tong". [See these posts on tong trees. -A] Because Quanzhou was famed for its coral trees, since the Tang dynasty (618-907) it had been known informally as the "city of coral trees" (citong cheng 刺桐城), and this literary epithet was later adopted as Zaitún "olive tree" by Arab traders.

For centuries after Marco Polo's account of the splendours of Zayton were first made known to Europe, no-one knew exactly where Zayton was, and some even thought that it was a spurious, made-up name. It was only at the end of the 19th century that Zayton was conclusively identified as Quanzhou.

The name Zayton has passed into the English language as the word "satin", which derives from the Arabic zaitúníah (via medieval Italian zettani, Spanish aceytuni or French zatony), being the name for the rich satins from Zayton, for which Quanzhou was renowned [...]

The strange thing about Arabic زيتون zaytuun 'olive' is that it has an initial voiced z- corresponding to a voiceless affricate tsh- Although I don't know the Quanzhou reading of 刺桐, the Taiwanese government's online dictionary lists tshì-tông as the Taiwanese reading of the variant spelling 莿桐. Hanyu fangyin zihui (1962: 42) lists the Xiamen readings tshu, tshi, and tshĩ, all with tsh-. The Hphags-pa Chinese reading of 刺桐 is *tshiduŋ in Coblin's (2007) reconstruction. Why was something like *tshi Arabized as zay- instead of si?

- The initial z- might reflect an otherwise unattested Chinese reading for 刺 with *dz- from an earlier *N-tsh- with a nasal prefix. However, if the source variety of Chinese had voiced *dz- for 刺, it should also have had voiced *d- for 桐 and I would expect Arabic zayduun with -d-.

- The -ay may have been pronounced [ej] and was an attempt to approximate a Chinese *-e. 刺 was Late Old Chinese *tshieh and could have had an *-e reading in the source variety of Chinese.

- In theory, زيتون <z-y-t-w-n> could represent Arabic

zaytuun

ziituun

zaytawn

ziitawn

Did the earliest attestations of zaytuun have a fatḥa and a damma (زَيتوُن) to clarify that <y> and <w> represented -ay- and -uu- instead of -ii- and -aw-? If not, could Arabic -ay be a misreading of -ii? And could Arabic -uu- be a misreading of an -aw- (phonetically [ow]?) approximating a Chinese *o or *ɔ?

The correspondence of Arabic z- to the most probable Chinese initial *tsh- reminds me of the use of z in German to represent [ts]. Why is [ts] written as z in German? Is that due to Italian influence?

3.9.21:48: I misunderstood Andrew's view of zaytuun. See my next post.

10.3.7.23:59: OUT OF VU

Last night, I listed the Hphags-pa Chinese (HC) reading vu for 牧 'to herd' and 目 'eye'. vu is a typographical simplification of Coblin's 2007: 130 reconstruction *ʋu. No Mandarin dialect - or Chinese language of any branch listed in the 1962 edition of Hanyu fangyin zihui - has v- or w- for these morphemes. Are there any Chinese languages today with a nonnasal initial for 牧 and 目? Did vu merge with mu in the descendant(s), if any, of the dialect(s) underlying HC?

In HC, 無 舞 務 物 and other morphemes read as wu in standard Mandarin and as u or vu in other Mandarin dialects also had the reading vu. Did some vu become mu while others became u ~ vu ~ wu? That is unlikely since only Middle Chinese *muk morphemes (e.g., 牧 and 目) would have become mu - even though HC speakers would not have known the Middle Chinese readings that merged into HC vu:

Is it possible that HC vu < MC *muk left no descendants? Perhaps vu-readings for 牧, 目, etc. stood out as unusual and were replaced by mu-like readings borrowed from other dialects.

Perhaps a more probable instance of *v- > m- in Chinese occurred in Cantonese. Sino-Vietnamese was borrowed from some early Cantonese-like language, and SV v- corresponds to Cantonese m- before certain rhyme classes: e.g.,

文 'literary': SV văn : Ct man

武 'military': SV võ : Ct mou

Maybe the SV and Cantonese initials are both later simplifications of an earlier cluster *mv- which also had a nonnasal counterpart *pf-:

Coblin (1994) reconstructed *pf-, *pfh-, *bv-, and *mv- for northwestern Middle Chinese, a source of early loans into Tangut*.

German is the only attested language I can think of with pf-. It doesn't have mv-. Does any language today have both pf- and mv-? UPSID lists only four other languages besides German with labiodental affricates:

pf-: German, Beembe, Teke

pfh-: Beembe

bv-: Aghem, Mambila, Teke

All four are Bantoid languages.

*3.8.0:24: Gong found one example of a possible *bv- word borrowed into Tangut with b-:

匍 *bvu >

TT4731 bɨu R2 1.2 'crawl'